QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

First published online May 19, 2008
Journal of Experimental Biology 211, 1757-1763 (2008)
Published by The Company of Biologists 2008
doi: 10.1242/jeb.015149

This Article Summary Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Related articles in JEB Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Kelman, E. J. Articles by Baddeley, R. J. PubMed PubMed Citation Articles by Kelman, E. J. Articles by Baddeley, R. J.

A review of cuttlefish camouflage and object recognition and evidence for depth perception

Emma J. Kelman¹, Daniel Osorio^1,* and Roland J. Baddeley²

¹ School of Life Sciences, University of Sussex, Brighton BN1 9QG, UK
² Department of Experimental Psychology, Social Sciences Complex, 8 Woodland Road, Clifton, Bristol BS8 1TN, UK

View larger version (125K): [in this window] [in a new window]   Fig. 1. Cuttlefish coloration patterns and their components. (A) Examples of four camouflage patterns from juvenile cuttlefish (mantle length 50 mm). From left to right the animals illustrate: a pale uniform pattern; a stipple with some mottle; a mottle with weak disruptive elements; and a high contrast pattern (but not typically disruptive). The animals also illustrate a range of skin textures. (B) A juvenile cuttlefish (Sepia officinalis) settled amongst pebbles, which is displaying components that are characteristic of both the disruptive and mottle body patterns. Increasing pebble size would cause the animal to emphasise components associated with the disruptive pattern, and decreasing pebble size would favour the mottle. The level of visual contrast within the image (e.g. between pebbles) appears to affect the overall contrast in the body pattern rather than the relative strengths of disruptive and mottle patterns (see Fig. 3A for further body patterns) (Mäthger et al., 2006; Kelman et al., 2007). (C) Some of these components numbered according to the scheme proposed by Hanlon and Messenger (Hanlon and Messenger, 1988) (see Fig. 2). In our principal components analysis (PCA) of cuttlefish, body pattern principal components (PCs) often approximate body patterns identified by Hanlon and his co-workers (Hanlon and Messenger, 1988; Mäthger et al., 2006). Here, features labelled on the left are often positively weighted in PCs that resemble the disruptive body pattern, and those on the right in PCs that resemble the mottle pattern (Figs 2, 3, 4). Features 5 and 40 are not clearly associated with either type of pattern.

View larger version (87K): [in this window] [in a new window]   Fig. 2. PCs derived after Varimax rotation (SPSS version 11.5) (Kelman et al., 2007) from scores of the expression of 32 chromatic and textural components (see Fig. 1) (Hanlon and Messenger, 1988) of 20 juvenile cuttlefish (mantle length 50–70 mm) placed in a 300 mm diameter arena (area 0.07 m2) on different backgrounds. Kelman and co-workers (Kelman et al., 2007) give further details of methods used for photographing the animals, scoring and PCA. Bold fonts indicate behavioural components that Hanlon and Messenger (Hanlon and Messenger, 1988) classified as body patterns. (A) PCs for pebble backgrounds of real pebbles under 5 mm Perspex, photographs of these pebbles at three contrast levels (see Fig. 3A,B) or photographs with 10 real pebbles (see Fig. 3C). For the 140 images that were graded (one per animal on each of seven backgrounds), three components (PC1–3) accounted for 47% of the total variance in the expression of the 32 behavioural components scored. A scree plot indicated that fitting further PCs was not meaningful. (B) PCs for checkerboard backgrounds (see Fig. 4). For the 60 body patterns that were graded, four PCs account for 59% of the total variance in the expression of 32 behavioural components. Fitting greater than four PCs was not meaningful. Positive values of three of the PCs correspond to body patterns identified by Hanlon and Messenger (Hanlon and Messenger, 1988): PC1 to the disruptive pattern; PC2 to the uniform stipple; and PC3 to the mottle pattern. PC4 involves two white components: white major lateral papillae (12) and white head bar (13).

View larger version (27K): [in this window] [in a new window]   Fig. 3. A comparison of body patterns displayed by 20 juvenile cuttlefish (A) in response to pebbles under 5 mm Perspex, with responses to photographic images of these pebbles, suggests that they are sensitive to real visual depth in pebble backgrounds (B). Placing 10 real pebbles on the photograph (C) suggests that a small number of light, but not dark, 3-D objects give a similar response to a natural substrate. (A) A photograph of the pebble background, and illustrations of body patterns that typify the three PCs identified in this study (see Fig. 2A). Positively weighted behavioural components for each PC are numbered as in Figs 1 and 2. (B) Mean weights (+s.e.m.) of PCs 1–3 for real pebbles under Perspex or photographs of the pebbles at three contrast levels. The photographic backgrounds were presented at three contrast levels: (i) to match the natural pebbles; (ii) enhanced by 25% in Adobe Photoshop; or (iii) enhanced by 50% in Adobe Photoshop. Weights of all three PCs increase with image contrast in the photographs, whereas real pebbles give the strongest weights of PC1 (disruptive pattern, Fig. 2) and negative weights of PC2 (mottle/stipple). This indicates that no level of image contrast in a photograph could produce a response matching that to real pebbles. A repeated measures ANOVA showed a significant effect of the background and amplitude of PCs 1–3 between the natural pebbles and all three photographic backgrounds (PC1, F3,57=19.61, P<0.05; PC2, F3,57=8.556, P<0.05; and PC3,F3,57=6.614, P<0.05). The ANOVA also revealed an interaction of background type (3-D vs 2-D) and the expression of PC1 and PC2 (F1,19=27.95, P<0.05). (C) Evidence that placing 10 pebbles at arbitrary locations on the 2-D background (area 0.07 m2) affects the body pattern. The presence of 10 light stones on the 2-D background enhances the expression of PC1 (disruptive; Student's t-test: t19=–3.547, P<0.05) and suppresses PC2 (mottle/stipple). With 10 dark stones there is no significant effect on PC1 (Student's t-test: t19=–1.811, P>0.05) but a suggestion that the expression of PC2 is suppressed. A mixture of five light and five dark stones gives similar responses to 10 light stones.

View larger version (8K): [in this window] [in a new window]   Fig. 4. Evidence that cuttlefish discriminate depth in checkerboard patterns. This study used the same 20 juvenile cuttlefish as in the pebble study (Fig. 3), and again we analysed the response of each animal to each of the three experimental backgrounds, giving a total of 60 images (Fig. 2B). (A) The cuttlefish were settled on a Perspex floor above a checkerboard in which the dark and light squares were either in the same depth plane (flat) or in different depth planes separated by 10 mm. (B) Light squares above dark were most effective in eliciting the disruptive body pattern (PC1; Figs 1 and 2). The expression of PC1 (disruptive) was stronger when the light check pattern was in the upper plane (Student's t-test, t19=–2.635, P<0.05). Similarly an ANOVA showed a significant interaction of the backgrounds (flat and depth with light checks on the upper surface) with the expression of PC1 and 2 (F1,19=0.5711, P<0.05). There is no apparent effect of the experimental treatments on PC3 and PC4 (not illustrated).

View larger version (12K): [in this window] [in a new window]   Fig. 5. A summary of how visual information controls cuttlefish camouflage. The animal detects local visual features, which include edge and depth information, and from these relatively low levels then classifies the background. For example, on the basis of whether it is a continuous surface, or made of discrete objects such as pebbles, and on the spatial scale of the pattern/objects. This classification determines the primary weightings (W1–3) of the components of the coloration pattern. Image contrast (and perhaps other low-level measures) then modulate the strength of the pattern. It is unlikely that the classification of `background type' is categorical, in the sense that an image has to be of one type or another, and this is why the animal is able to vary the relative levels of expression of the 40 or so chromatic components independently.