Behavioral Ecology doi:10.1093/beheco/arl015 Advance Access publication 19 June 2006 The effects of pattern symmetry on detection of disruptive and background-matching coloration Innes C. Cuthill, Martin Stevens, Amy M.M. Windsor, and Hannah J. Walker School of Biological Sciences, University of Bristol, Woodland Road, Bristol BS8 1UG, UK Two, logically distinct but sometimes compatible, mechanisms of camouflage are background-matching and disruptive coloration. In the former, an animal’s coloration comprises a random sample of the background, and so target–background discrimination is impeded. In the latter, object or feature recognition is compromised by placing bold, high-contrast colors so that they break up the prey’s body into apparently unconnected objects. Recent experimental evidence for the utility of disruptive colors, above and beyond that conferred by background matching, has been based on artificial prey with patterns lacking a plane of symmetry. However, it is plausible that the bilateral symmetry present in natural prey may compromise the efficiency of disruptive coloration, on account of the potency of symmetry as a cue in visual search. In this study, we tested this prediction in the field, by tracking the ‘‘survival’’ under bird predation of artificial mothlike targets placed on oak trees. These had background-matching color patches placed either disruptively or nondisruptively and with or without bilateral symmetry. We found that symmetry reduced the effectiveness of both nondisruptive and disruptive background-matching coloration to a similar degree so that the negative effects of symmetry on concealment are no greater for disruptive than nondisruptive patterns. Key words: bird vision, camouflage, crypsis, disruptive coloration, symmetry, visual search. [Behav Ecol 17:828–832 (2006)] amouflage is one of the commonest forms of defensive coloration (Cott 1940; Edmunds 1974; Endler 1991; Ruxton et al. 2004). As visual segmentation of a potential prey item from irrelevant background stimuli is aided by a mismatch in color or texture, it has been argued that background matching (termed crypsis by Endler 1981) is perhaps the most widespread form of camouflage. However, as recognized by some of the pioneers of the evolutionary study of animal coloration (Cott 1940; Thayer 1909), even perfect background matching can be compromised by a discontinuity in pattern at the body’s edge, potentially exacerbated by shadow. Thayer (1909) instead advanced a theory of ‘‘ruptive’’ coloration, now termed disruptive coloration after Cott’s (1940) more comprehensive treatment of the theory (see Stevens, Cuthill, Párraga, and Troscianko, forthcoming). Cott (1940; Merilaita 1998) proposed 3 key principles of disruptive camouflage: ‘‘differential blending,’’ where different color patches match different components of the background, ‘‘maximum disruptive contrast,’’ where adjacent pattern elements are highly contrasting in tone, and third, placement of some patches near the body’s edge (or across distinctive features such as eyes). These 3 principles reduce the probability that the prey’s outline is detected and its features are grouped (perceptually) as a coherent object of interest (Stevens and Cuthill forthcoming). Cott’s principles of disruptive coloration have only recently been tested, receiving both indirect (Merilaita 1998) and direct support (Cuthill et al. 2005; Merilaita and Lind 2005; Stevens and Cuthill forthcoming). In previous field experiments involving avian predators (Cuthill et al. 2005), we showed that artificial mothlike targets with differentially blending disruptive patterns survived significantly better than targets with nondisruptive background- C matching patterns. Furthermore, targets with highly contrasting disruptive patterns survived significantly better than equivalent low-contrast patterns, supporting the disruptive contrast hypothesis. However, this work left at least one important question unanswered (Sherratt et al. 2005). Is disruptive coloration effective when patterns are bilaterally symmetrical, as they often are in nature but not in Cuthill et al. (2005)? Many animals are bilaterally symmetrical, and because symmetry is a potent cue in visual search, symmetrical patterning is likely to reduce the effectiveness of camouflage (Osorio 1994; Cuthill et al. 2006; Merilaita and Lind 2006). In fact, Thayer (1909) recognized that ‘‘dual symmetry in pattern’’ is one potential major flaw in camouflaged species. However, in an experiment in which great tits (Parus major) foraged for patterned artificial prey, Merilaita and Lind (2006) found that not all symmetrical patterns were similarly disadvantaged. While symmetrical pattern elements might be particularly conspicuous, certain arrangements of patterns might reduce the costs of symmetry. As this conclusion was based on the relative success of just one pattern type, selected subjectively, this hypothesis deserves systematic investigation. We therefore compared the survival rates of artificial targets with disruptive or nondisruptive background-matching patterns, which were either symmetrical or asymmetrical about the vertical midline. We predicted that disruptive coloration should reduce the costs of bilateral pattern symmetry. The experiment also serves to test the counterproposal that disruptive coloration may be rendered ineffective as a camouflage strategy when bilateral symmetry is present (i.e., unlike in Cuthill et al. 2005), shedding doubt on the explanatory power of the theory of disruptive coloration in nature. METHODS Address correspondence to I.C. Cuthill. E-mail Address: i.cuthill@ bristol.ac.uk. Received 23 February 2006; revised 17 May 2006; accepted 17 May 2006. The Author 2006. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology. All rights reserved. For permissions, please e-mail: [email protected] The experiment aimed to determine the effects of symmetry on both disruptive and nondisruptive background-matching patterns and followed the same general procedure as in Cuthill et al. (2005); a more detailed treatment of the protocol and Cuthill et al. • Symmetry and camouflage the rationale for various details of experimental design and analysis can be found in Cuthill et al. (2006). We created cardboard triangular artificial ‘‘moth’’ targets, approximately 50 mm wide by 25 mm high. These were not designed to mimic any real species but were covered with patterns consisting of black and brown markings designed to match the background with respect to avian vision. These markings were chosen to mimic the ridges on mature oak bark. Patterns were samples of digital photos of oak tree trunks at 1:1 reproduction, taken with Nikon Coolpix 5700 camera, calibrated to linearize the relationship between radiance and the grayscale in each color channel (Párraga 2003; Stevens, Párraga, et al. forthcoming), and saved as uncompressed TIFF files. Images were converted using Image J (Abràmoff et al. 2004; Rasband 2005) to grayscale and thresholded at 50% to binary (black/white) images to provide, when printed onto brown card, barklike brown/black spatial variation. One or more pattern elements were selected, using the ‘‘copy’’ tool, from the thresholded photographs according to the criterion that they should be small enough to fit within the final triangular targets (Figure 1). The selection was then pasted into a triangular frame, placed either to touch the triangle’s edge (disruptive treatments) or not (nondisruptive treatments; see below). Whether a given selection was placed disruptively or not (and the treatments within those categories) was selected using random numbers. Different samples, from different trees, were used for each replicate target. Color matches of treatments to natural bark were verified using spectrophotometry of stimuli and bark as described in Cuthill et al. (2006). This was followed by modeling of predicted photon catches (Maddocks et al. 2001) of a typical passerine bird, the blue tit’s (Parus caeruleus) single-cone photoreceptors (Hart et al. 2000), using irradiance spectra from woodland shade under overcast skies, collected at our field site using an Ocean Optics USB2000 spectrometer fitted with a cosine corrector. Our acceptance criterion was simply that cone captures for the experimental stimuli fell within 1 SD. of the means of those for oak bark (for the reflectance spectra on which the calculations were based, both bark and paper, see Cuthill et al. 2006). There were 6 treatments in all (Figure 1): 1) disruptive asymmetrical, with the black markings touching the edge of the body, 2) disruptive symmetrical, where the markings on one side of the body were a mirror reflection about the body midline, 3) nondisruptive asymmetrical, where the black markings were randomly placed but did not touch the edge of the targets’ body, 4) nondisruptive symmetrical about the midline, 5) disruptive midline symmetrical, where the pattern elements were disruptive but also had to touch the midline of the body, and 6) nondisruptive midline symmetrical, where the markings were randomly placed but had to touch the midline of the body and not touch the edge of the target. Treatments 5 and 6 control for the fact that, on average, disruptive markings will be found further away from the midline of the body because they are placed near the target edges, and evidence from human studies indicates that object recognition of symmetrical patterns is facilitated when the patterns are closer to the midline (Swaddle 1999b). Therefore, to determine if this is the case with our study, we can compare the survival of the symmetrical midline to the symmetrical nonmidline treatments. Each target was different in its patterning, and so no 2 targets were the same, but as indicated above, because assignment of patterns to treatments was random, there should have been no systematic differences in features such as the area of black/brown. Analysis of a subset of the stimuli, using the measurement tools in Image J, indicated no obvious treatment differences (mean %black 6 SD in treatments 1–6, respectively: 32.60 6 12.24, 29.55 6 12.97, 829 Figure 1 Sample targets: 1) disruptive asymmetrical, 2) disruptive symmetrical, 3) nondisruptive asymmetrical, 4) nondisruptive symmetrical, 5) disruptive midline symmetrical, and 6) nondisruptive midline symmetrical. In the experiment, each replicate target had a unique pattern. 25.55 6 8.02, 29.50 6 11.67, 29.40 6 8.80, and 31.70 6 10.26; N ¼ 20 in each treatment; F5,114 ¼ 1.02, P ¼ 0.410; no Tukey pairwise comparison went below P ¼ 0.315, and the directions of differences between symmetric and asymmetric treatments were not consistent). Targets were pinned onto oak trees in the mixed deciduous Leigh Woods National Nature Reserve, North Somerset, UK (238.6#W, 5127.8#N) and their survival checked at approximately 2, 4, 24, and 48 h. Ten targets of each treatment were pinned to oak tree trunks in each of 10 trials, in a randomized block design, giving a total sample size of 600. Targets were placed along nonlinear transects of approximately 1.5 km by 20 m (targets were placed on ,5% of available trees), and were randomly allocated to trees, subject to the constraints that no lichen covered the trunk and no young trees of trunk circumference less than 0.9 m were used. Each block was undertaken on different days and in different locations of the field site. The experiment was conduced between November 2004 and May 2005. An edible component for birds was attached to each target, consisting of a dead (frozen overnight at 80 C, then thawed) mealworm (Tenebrio molitor larvae) pinned onto each target. Avian predation was revealed by all or complete disappearance of the mealworm. Species observed taking the prey in this and similar experiments in other field seasons included blue tits (P. caeruleus), great tits (P. major), European robins (Erithacus rubecula), chaffinches (Fringilla coelebs), and house sparrows (Passer domesticus). Other insectivorous species such as woodpeckers were observed in the area but were not actually seen taking prey. However, no systematic program of observation was carried out, this being incompatible with the quantity of data required. Other forms of predation could also be identified; spiders sucked fluids out leaving a hollow exoskeleton, and slugs left slime trails. Nonavian predation, the disappearance of the whole target, or survival to 48 h were treated as ‘‘censored’’ values in survival analysis with Cox proportional hazards regression (Cox 1972; Klein and Moeschberger 2003; Cuthill et al. 2006). Significance was tested with the Wald statistic (W), and after the establishment of an overall treatment effect, differences were investigated using linear contrasts (Rosenthal et al. 2000). The latter 830 Behavioral Ecology consisted of main effects of pattern placement (disruptive vs. nondisruptive treatments; 1, 2, 5 vs. 3, 4, 6), symmetry (1, 3 vs. 2, 4, 5, 6), proximity to the midline of symmetrical pattern elements (near vs. far; 2, 4 vs. 5, 6), and the interaction between pattern placement and symmetry. The latter term tests directly for a disproportionate cost of symmetry in disruptively patterned targets. Simple pairwise comparison of all treatments yields the same conclusions; we present the contrast analysis as it more fairly reflects our hypotheses (Rosenthal et al. 2000) and represents a more parsimonious account of the observed effects (a minimal adequate model). Effect sizes are quoted as odds ratios (OR), which should be interpreted as the probability of ‘‘death’’ in one treatment divided by that in the other, such that an OR of 1 represents no difference and an OR of 2 represents double the mortality per unit time in one treatment than the other. RESULTS There was a significant difference in survival between the treatments (W ¼ 37.918, P , 0.001, df ¼ 5; Figure 2). Linear contrasts indicated that those treatments with the disruptive patterns survived significantly better than the nondisruptive treatments (pattern W ¼ 15.028, P , 0.001, df ¼ 1; OR ¼ 1.307). Treatments with asymmetrical patterns survived better than those with symmetrical patterns (symmetry W ¼ 24.513, P , 0.001, df ¼ 1; OR ¼ 1.407). Contrary to expectation, symmetrical patterns with markings touching the midline did not suffer higher predation than those treatments with markings placed away from the midline (proximity W ¼ 0.368, P ¼ 0.544, df ¼ 1; OR ¼ 1.041). There was no interaction between pattern placement and symmetry (W ¼ 1.102, P ¼ 0.294, df ¼ 1; OR ¼ 0.930), and removal of this interaction term from the model made little difference to the parameter estimates and P values of the main effects. Although not an a priori hypothesis tested in our contrast analysis, we note that those targets with symmetrical patterns placed disruptively survived similarly to those targets with asymmetrical patterns placed nondisruptively (Figure 2). There was a significant effect of block (W ¼ 25.739, P , 0.001, df ¼ 9), meaning that there were differences in average predation rates in different parts of the woods on different dates. As this does not affect our conclusions about the treatment effects and we have no way of testing whether this was due to different weather conditions, bird densities, light environments, or levels of human disturbance, these average block differences are not discussed further. DISCUSSION The results support the findings of Cuthill et al. (2005) that disruptive markings provide a significant survival advantage against avian predators, compared with targets with nondisruptive background-matching patterns. Also, as in Cuthill et al. (2006), symmetry is a significant disadvantage in terms of predation risk for both disruptive and nondisruptive backgroundmatching patterns. These effects are additive, so there is no evidence that disruptive markings bear a higher cost of conspicuousness when arranged symmetrically. Indeed, symmetrical disruptive patterns survive at least as well as asymmetrical nondisruptive patterns (Figure 2), indicating that disruptive patterning reduces the costs of symmetry for camouflaged organisms. The result that the symmetrical patterns placed near the axis of symmetry survived as well as the nonmidline symmetrical patterns indicates that the detection of the targets was not enhanced by symmetry lying closer to the midline or the Figure 2 Survival curves of the patterned targets in the experiment: 1) disruptive asymmetrical, 2) disruptive symmetrical, 3) nondisruptive asymmetrical, 4) nondisruptive symmetrical, 5) disruptive midline symmetrical, and 6) nondisruptive midline symmetrical. Curves are the probability of surviving bird predation as a function of time, based on Kaplan–Meier estimates to account for censoring due to nonavian predation and survival to the end of the study period (SPSS Inc. 2003). The 2 long gaps without mortality correspond to overnight periods when targets were not checked. The disruptive symmetrical and nondisruptive asymmetrical treatment lines are indistinguishable between days 1 and 2 (24 and 48 h). mealworm. This is important because it indicates that the superior survival of the disruptive symmetrical treatments over the nondisruptive symmetrical treatments was not because of midline symmetry effects, where the disruptive patterns are on average further from the midline, but rather due to disruptive coloration itself. The lack of midline symmetry effects is not entirely surprising given that in our experiment target detection by birds probably occurred at a distance of several meters (IC Cuthill, M Stevens, personal observation) and the stimuli were relatively small, so minor differences in the placement of symmetrical patterns will have only a limited impact on detection rates. We note that midline-proximity effects in human search experiments are usually based on protocols where subjects view computer screens filling a large portion of the visual field (e.g., Barlow and Reeves 1979; e.g., Herbert et al. 1994). Also, the presence of the mealworm down the midline of the target was itself an indicator of the axis of symmetry and may have rendered any additional effects of the positions of the patterns negligible. It is clear that our results do not diminish the likelihood that symmetry close to the midline may be a revealing cue for larger prey, or when predators are closer to the substrate, as was the case in the experiments of Merilaita and Lind (2006). Furthermore, as they point out, mirror symmetry about the midline may create novel pattern elements that are themselves rare in the background and so distinctive. The design of our experimental stimuli, with the mealworm on top of the patterns, means that mealworm–wing contrast must be considered as well as the relationship between wing and background colors that was central to our study. Mealworms are yellow-brown and are lighter than either the brown or black color patches on which they were pinned. Therefore, in treatments where the black color patches were more likely to be placed centrally, the mealworm would have been more conspicuous than in those treatments where it was more likely to lie on brown patches. This is potentially a seriously confound because the edge treatments may owe their reduced conspicuousness to this effect rather than disruption of the Cuthill et al. • Symmetry and camouflage body outline. Fortunately, the manipulations of pattern positioning relative to the midline address this question as well as that of symmetry detection. As treatments with dark patterns far from the midline did not survive better than those positioned near the midline, mealworm-wing contrast seems to have had a minor effect compared with wing-background contrast. This is not surprising because the area of the wings was far greater than that of an average mealworm (wing area 625 mm2, mean mealworm area 65 mm2. See Stevens, Cuthill, Windsor, and Walker forthcoming). Disruptive coloration may help to reduce some of the negative survival implications of symmetrical patterns and enable animals to exploit backgrounds and environments toward which they have only a partial resemblance. However, symmetrical patterns are often still easier to detect. Also, observations that some insects adopt behavioral strategies apparently to introduce asymmetry, such as folding one wing over the other, coupled with our results suggest that there can be clear disadvantages to symmetry in camouflage patterns. This poses the obvious question of why so few animals with camouflage patterns are asymmetrical and, in those species such as cuttlefish (Sepia spp.) that can generate asymmetrical patterns, symmetrical camouflage is nevertheless adopted (Langridge 2005). No doubt, in many species, such as butterflies, the posture and wing-folding pattern means that both sides of the body will not be observed simultaneously, but this is not true of moths, many of which have a classic reliance on camouflage to reduce predation risk. A plausible argument for symmetry in camouflaged Lepidoptera is that there are genetic or developmental restrictions on pattern formation, if the latter is closely tied to morphology, as seems to be the case in butterflies (Monteiro et al. 1997). If wing coloration cannot readily be decoupled from the underlying morphology, then selection for symmetry of wing shape for flight performance is likely to exceed the camouflage benefit obtained from asymmetrical wing coloration (see discussion in Cuthill et al. 2006). However, there are documented cases of complete asymmetry in insect wing coloration (Barabás and Hancock 1999), indicating morphology and color can be developmentally uncoupled, so ascribing insect bilateral symmetry to an absolute genetic/developmental constraint seems unsatisfactory. A modification of the constraint hypothesis is therefore that only large levels of asymmetry bring a benefit in terms of camouflage, and for the usually accepted reasons (e.g., Dawkins 1976, 1996), such large mutations and developmental changes are statistically unlikely. In our experiment, the stimuli were either perfectly symmetrical or entirely asymmetrical. Real animals that appear at a glance to be highly symmetrical will have low levels of fluctuating asymmetry (Møller and Swaddle 1997), which selection could potentially enhance through relaxation of developmental canalization. To the best of our knowledge, only one study has quantified the level of asymmetry found in potentially concealing patterns in Lepidoptera (Forsman and Merilaita 2003), and this found only limited differences in asymmetry compared with signaling patterns. Crucially, it is unknown whether low levels of asymmetry in camouflage reduce detection chances, or are even detectable (see particularly Swaddle 1999a), so maybe, from a starting point of high symmetry, the strength of selection for asymmetrical cryptic coloration is negligible. Clearly, current theories of camouflage would be greatly strengthened by a firmer understanding of the mechanisms of symmetry perception and genetic investigations into the origin and plasticity of the genetic basis of pattern symmetry. The research was supported by a Biotechnology and Biological Sciences Research Council (BBSRC) grant to I.C.C., T. S. Troscianko, and 831 J. C. Partridge and by a BBSRC studentship to M.S. Our experiment was conceived and completed prior to the publication of Merilaita and Lind (2006), and we are grateful to Sami and Johan for freely discussing their research with us. Great minds think alike? Thanks to Sami Merilaita and an anonymous referee for their perceptive comments on the manuscript. I.C.C. and M.S. conceived the experiment, designed the stimuli, and wrote the manuscript; all authors took an equal role in the field experiment. REFERENCES Abràmoff MD, Magalhäes PJ, Ram SJ. 2004. Image processing with Image J. Biophoton Int 7:36–43. Barabás SP, Hancock EG. 1999. Asymmetrical colour and wing-folding in Tithrone roseipennis (Saussure 1870) a neotropical praying mantis (Mantodea, Hymenopodidae). Trop Zool 12:325–34. Barlow HB, Reeves BC. 1979. Versatility and absolute efficiency of detecting mirror symmetry in random dot displays. Vision Res 19:783–93. Cott HB. 1940. Adaptive coloration in animals. London: Methuen & Co. Ltd. Cox DR. 1972. Regression models and life-tables. J R Stat Soc B 34:187–220. Cuthill IC, Hiby E, Lloyd E. 2006. The predation costs of symmetrical cryptic coloration. Proc R Soc Lond B Biol Sci 273:1267–71. Cuthill IC, Stevens M, Sheppard J, Maddocks T, Párraga CA, Troscianko TS. 2005. Disruptive coloration and background pattern matching. Nature 434:72–4. Dawkins R. 1976. The selfish gene. Oxford: Oxford University Press. Dawkins R. 1996. Climbing mount improbable. London: Viking. Edmunds M. 1974. Defence in animals. Harlow, UK: Longman Group Ltd. Endler JA. 1981. An overview of the relationships between mimicry and crypsis. Biol J Linn Soc 16:25–31. Endler JA. 1991. Interactions between predators and prey. In: Krebs JR, Davis NB, editors. Behavioural ecology: an evolutionary approach, 3rd ed. Oxford: Blackwell. p 169–96. Forsman A, Merilaita S. 2003. Fearful symmetry? Intra-individual comparisons of asymmetry in cryptic vs. signalling colour patterns in butterflies. Evol Ecol 17:491–507. Hart NS, Partridge JC, Cuthill IC, Bennett ATD. 2000. Visual pigments, oil droplets, ocular media and cone photoreceptor distribution in two species of passerine: the blue tit (Parus caeruleus L.) and the blackbird (Turdus merula L.). J Comp Physiol A 186:375–87. Herbert AM, Humphrey GK, Jolicoeur P. 1994. The detection of bilateral symmetry: effects of surrounding frames. Can J Exp Psychol 48:140–8. Klein JP, Moeschberger ML. 2003. Survival analysis: techniques for censored and truncated data. New York: Springer Verlag. Langridge KV. 2005. Symmetrical crypsis and asymmetrical signalling in the cuttlefish Sepia officinalis. Proc R Soc Lond B Biol Sci 273: 959–67. Maddocks SA, Church SC, Cuthill IC. 2001. The effects of the light environment on prey choice by zebra finches. J Exp Biol 204: 2509–15. Merilaita S. 1998. Crypsis through disruptive coloration in an isopod. Proc R Soc Lond B Biol Sci 265:1059–64. Merilaita S, Lind J. 2005. Background-matching and disruptive coloration, and the evolution of cryptic coloration. Proc R Soc Lond B 272:665–70. Merilaita S, Lind J. 2006. Great tits (Parus major) searching for artificial prey: implications for cryptic coloration and symmetry. Behav Ecol 17:84–7. Monteiro A, Brakefield PM, French V. 1997. The relationship between eyespot shape and wing shape in the butterfly Bicyclus anynana: a genetic and morphometrical approach. J Evol Biol 10:787–802. Møller AP, Swaddle JP. 1997. Asymmetry, developmental stability and evolution. Oxford: Oxford University Press. Osorio D. 1994. Symmetry versus crypsis. Trends Ecol Evol 9:346. Párraga CA. 2003. Is the human visual system optimised for encoding the statistical information of natural scenes? [unpublished PhD dissertation]. Bristol, UK: University of Bristol. Rasband WS. 2005. Image J. Bethesda, MD: National Institutes of Health. Available from: http:/rsb.info.nih.gov/ij/. 832 Rosenthal R, Rosnow RL, Rubin DB. 2000. Contrasts and effect sizes in behavioral research. Cambridge, UK: Cambridge University Press. Ruxton GD, Sherratt TN, Speed MP. 2004. Avoiding attack. Oxford: Oxford University Press. Sherratt TN, Rashed A, Beatty CD. 2005. Hiding in plain sight. Trends Ecol Evol 20:414–6. SPSS Inc. 2003. SPSS for Windows Release 9.0. Chicago, IL: SPSS Inc. Stevens M, Cuthill IC, Párraga CA, Troscianko T. The effectiveness of disruptive coloration as a concealment strategy. In: Alonso J-M, Macknik S, Martinez L, Tse P, Martinez-Conde S, editors. Visual Perception (Part 2): Progress in brain research 155. Amsterdam: Elsevier. Forthcoming. Stevens M, Cuthill IC. Disruptive coloration, crypsis and edge detection in early visual processing. Proc R Soc Lond B Biol Sci. doi:10.1098/rspb.2006.3556. Behavioral Ecology Stevens M, Cuthill IC, Windsor AMM, Walker HJ. Disruptive contrast in animal camouflage. Proc R Soc Lond B Biol Sci. doi:10.1098/ rspb.2006.3614. Stevens M, Párraga CA, Cuthill IC, Partridge JC, Troscianko TS. Using digital photography to study animal coloration. Biol J Linn Soc. Forthcoming. Swaddle JP. 1999a. Limits to length asymmetry detection in starlings: implications for biological signalling. Proc R Soc Lond B Biol Sci 266:1299–303. Swaddle JP. 1999b. Visual signalling by asymmetry: a review of perceptual processes. Philos Trans R Soc Lond B Biol Sci 354:1383–93. Thayer GH. 1909. Concealing-coloration in the animal kingdom: an exposition of the laws of disguise through color and pattern: being a summary of Abbott H. Thayer’s discoveries. New York: Macmillan.
© Copyright 2026 Paperzz