AESTRACT The common cuttlefish, Sepia officinalis, is able to exhibit a wide yariety of body patterns. Fast studies have analyzed the animal's ability to camouflage itself using subjective methods. This study encorporated two objective methods, overall tone analysis and spatial frequency of pattern analysis, to examine how well the animal was matching its background. Un its natural background, sand, the cuttlefish matched both tone and pattern. On an unnatural background, such as a uniform gray, one animal matched tone, ame matched pattern and one attempted to match both tane and pattern. On analysis combining subjective method and the quantitative methods described in this paper offer a more definitive method for judging the degree to which the the animal can match the surrounding environment in behavioral tests. INTEODUCTICN The cuttlefish, Sepia officinalis, has fascinated biologists because of its ability to change its skin patterns within fractians of a zecond. Using the brown, yellow and red pigmented ordans in their skin (chromatophores), which are under neural control, the cuttlefish are able to exhibit a wide variety of hody pattemns. For example, cuttlefish use patterns to scare prey, to attract mates, and to camoflauge themselves. Early studies, by Bott (1938) and by Tinbergen (1939) concentrated on the cuttlefish's courtship patterns and Holmes (1940) described various other patterns. More recently, Hanlon et. al (1986) systematically characterised the elements and units that go to make up the body patterns. Hanlon divided the body patterns into a total of 34 chromatic components, 16 light and i8 dark and, in addition, described sis textural components, 8 postural components, and o locomator components. Textural components reter to the texture of the animal and the dermal papillae that can expand or contract to cause a rough or emooth surface. Postural components refer to the way the animal is positioned. and locomotor components to the motion of the animal. Althouch the animal can exhibit these components in almost any combination, Hanlon characterized 13 whole body patterns. Cott (1940) outlined the ways in which animals can conceal themselves from predators: colour resemblance, deceptive resemblance, obliterative shading, disruptive colouration, coincident disruptive colpuration and concealment. Hanlon (1986) describes how cuttlefish employ all the techniques of colour matching set out by Catt. Hanlon's study describes how the cuttlefish match in a descriptive, qualitative way; and he noted that his attempts to quanitty matching were unsatisfactory. The purpose of this study was to tind a more quantative, gjective way of examining the mechaniem by which Sepia officinalis match their background utilising recent advances in the field of digital imag processing. Computer analysis can be used to obtain numerical data in an accurate, rapid and reproducable manner (lnoue, 1786). MATERIALS AND METHODS Sepia were hatched in Galveston, Texas and brought to Hopkins Marine Station, Facific Grove, Ca. at around A menthe ot age. The animals used were between four and five monthe old at the time of testing. Typical mantle length of the animals was a cm. The three animals used for the study were kept in shaded, individual aquaria with running sea water 15" C. Fictures of individual animals were taken on three different backgrounds: (1) amall grained sand substrate, (E) a uniform light gray background and (S) a mock mubstrate cantaining rocks approkimately the mame zite as the animal, but in a variety of tomes. In an experimental tank, the test animal was placed in a 17 cm bowl containing of gne of the three substrates and given five minutes to aclimate. Fictures of the cuttlefish were taken with a m Wikon camera and twa electric flashes held on stands om sither side ot the tank. Ektachrome 100 alide film was used. The camera was tastened th a trippd approkimately mne meter above the tank. Since evem slight movement often seemed to bother the animals, the pictures were taken in a darkemed room and I stayed sitting quietly beside the tank fom the five minutes before the pictures were taken. In order to have some way of subjectively judging how well the cuttlefish matched their background, the pictures of the cuttlefish in the experimental tank were shown to 12 human judges' who were asked to rate the cuttlefish's camoutlage performance on a scale of O (a bad match) to 3 (a good match). Eody patterns of the animals were classified according to the system develcped by Hanlon (1984). The five patterne shown by the animals were: (1) Uniform light (see figure 2): to an overall light tone with very few of no dark chromatophores enpanded. (2) Stipple (see figure 4): an overall uniform tone that is slightly darker than l due to expansion of dark chromatophor. (3) Weak dismuptive (see figure 6): a wide variety of patterns, with the common characteristic that the body is braken into portions by transverse or lengitudinal elements. (4) Weak Zebra (see figure 5): a low contrast striped patterm also calledsandy-tripe' by Holmes (1740). () Uniform Dark (see figure 3): expansion of all dark chromatophores. To objectively measure the animals ability to match its hackground, the various slides were then digitized using a MegaVision 1OE4-XM computer and a Panasonic black and white vided camera, By selecting the function 'tyscannem' the colour slides T y array of pixels with werestored in the computer in a mach pixel assigned a value on a gray scale of 25 ( O being the blatkesta- eing the whitest). The gray scaleof the Hegalision was fpund to be linear at all but the darkest and whitest gray values (see figure 7). The Megavision function 'hpolygon' was used to output gray values within the area covered by the cuttlefish, and an equal size area on the adjacent substrate. The function 'hlist' was used to obtain the mean, standard deviation, miniumum and maximum values, skew and kurtosis of the histograms of the pixel gray scales from cuttlefish and background. Each histogram quantatively measures how well the animal was matches the oyerall tone of the zubstrate. A spectral analysis methed was used to determine how well the cuttlefish were matching pattern. Spectral analysis is oftem used to interpret time in terms of frequency (Chatfield 1784). Im this study, it was used to comvert distance in space (on the hack of the cuttlefish or on the substrate) into spatial trequency. The result of the analysis is presented as a speatrum. The abtissa represents the spatial frequency of periodic pattemns in inverme mm. and the ordinate is the power spentrum, a measure af the variance in gray scale associated with mach spatial frequency (Chatfield, 1984). The function 'profile' was used to draw a transect 9 pixels wide acros the back of the animal and an equal-length transect through the adjacent substrate. Each pixel corresponds to a distance in space of .Emm. The gray values obtained were then transferred tw an IEM XI personal computer. Here, the gray meale values were filtemed with a high pass digital filtem to remave the mean and the first five harmonics of the fundamental spatial frequengy. The filtering was done ta memoye the low trequency patterns associated with the averall length of the transect on the animal or substrate. Next, the power spectrum of a Tho pixel series was calculated and smoothed using a nine-point running average, and the objective measure of pattern matching was obtained. RESULTS Histogram pairs of gray scale (a measure of overall tome). ame shown in figures 8-13 for each test cuttlefish and its background. A student t-test was performed on the means of each pair. Due to the extremely large number of pixels in each histmgram (n240,000), even a difference in means of one was statistically significant. However, the human eye can only distinguish approximately Jo shades of gray (Takahashi, 1986). Theretere, the human eye can discern a difference in light intensity only if two regions of the image are more than 7 (2w6/30) apart on the gray scale. The means of the histograms, them, were comsidered 'different' if the means were farther than apart. Un the sand substrate, the difference in means for all three animals was less than sevem. On the gray substrate, all animals were greater than 7 gray scale units away from the background mean. The histogram of the cuttlefish on the rock background was compared to three other histegrams: that of one rock in the bowl. the whole area of the bowl and a cuttlefish sised area within the bowl. The cuttlefish judged the most difficult to see matched (difference in means 47) a rock in the bowl, but not overall tome or a cuttlefish-sied area in the bowl. Conversely, the mean of the histogram of the animal yoted most visible did not match any of the means. The results of the spectral analysis are shown in tigures 14-17. The cuttlefish and its background are plotted on thesame zet of axes. All the spectra exhibit a large peak corresponding to a spacing of 23.T mm. This peak was theught to be an artifact uf the filtering process and was therefore not included in the analysis. The cuttlefish which closely match their substrate according to the zubjective judging, show that the pattern spacings are within .omm, at most. Those cuttlefish judged nat to be matching have harmonics that differ by more than twoee animal +). IhE power spectra also compare the variance of the transect on the cuttlefish with that of the background. Cuttletisn photographed on sand show less variance than the substrate. The animals phatographed on the gray background show greater variance than the suhstrate, while those compared to the rack hackground mhow mither more or less variance than the mock aubstrate. he variance exhibited on the mocks does nat ampear to be asmociated with an individual animal or the trial, as each individual showed greater or leszer variance at different times. Ihe resulte of the individual animals body pattern choite is aummari zed in the bar graphs om hoth rack and gray substrates. All the animals except for animal n the gray background use more than one pattern when confronted with the same substrate. Un both the rock and gray substrates, five different body patterns were used. Similar bar graphs for the sand trials were not made, since all the animals matched the sand substrate well. and all showed the same overall body pattern. DISCUSSION Cuttletish on the sand substrate are able to match both tone and pattern and are subjectively well camouflaged. Since sand is the natural habitat of the animal it seems logical that all the animals would be able to match the sand backorpund best. When contronted with an unnatural substrate, rocks or gray, the cuttletish use various strategies to conceal themselyes. On the gray background, animal matched the uniform pattern well, but was highly visible because of lack of tone matching, (see figure 3) Animal matched the overall tone of the gray backargund to within one shade of gray, but displayed a pattern that was highly visible. (see figure 4) Animal matched tone to within 5 shades as well as matching pattern. he results on the rock substrate also indicate that different strategies can be used by the same individual. Individuals employ at least five different body patterns to mamoutlaque themselves on the unnatural substratee, Thus, when placed on an unfamiliar or unnatural background, the cuttlefish have the ability to respond in a variety of ways. The animal is able to examine his environment and then able to chosse the pattern and tome it will show. Evidently, the neural circuitry involved in chromatophore response is not "hard-wired". Similarly, the same cuttlefish does not always respond the same way to the same background (see figure 20). With the exception of S on the gray substrate, each of the three animals exhibited at least two different body patterns when placed on the same substrate repeatedly. Moreover, the patterns did not depend on how many times the cuttlefish had seen the particular zubstrate before. Ihis study demonstrates a method for quantifying matching. Uverall tone of the animal can be quantified using a digital image processor and pattern matching can be cbjectively determined by the power spectra. This objective, quantative method is especially powerful for judging matching on uniform mubstrates such am sand or the light gray background. Subjective analysis is still necessany, especially on a complex substrate (like a rock background), where it is not always evident how the amimal is trying to camouflage itself. Subjective analysis and the two cbjective methods outlined above, together provide a good basis for judging how well, and how, cuttlefish match their sumstrate, and the different strategies they use. ACKNOWLEDGMENTS I would like to thank Mark Denny and Stuart Thompson for all their advice, help and most of all, patience, throughout this quarter. Thanks also to Chris Fatton for all his help in photography as well as Steve Gaines and Mark Shibata for their computer help. Hoger Hanlon deserves many thanks for supplying the cuttlefish, many papers (including his own unpublished work) and for teaching all of us in the Spring Class in an entertaining and informative way. Finally, thanks to everyone in Bio 17SH Hopkins Spring Class for making this quarter nifty! ished. HEFERENCES Berns, M. and Walter, K. 1986 Digital image processing and (ed. Shinya ounep analysis. In Video Microscopy. Hew York: Flenum Fress, 584 pages. In Cephalopod life Boletzky, S.V. 1983 Sepia officinalis. Eyeles. Volume 1: species accounts. (ed. F.R. Boyle), pp. 1- London: Academic Fress, 475 pages. Chatfield, C. 1940 The analysis of a time semies: an Introduction (End ed) London: Chapman and Hall, 28 pages. Cott, H.B. 1940 Adaptive coloration in animals. London: Metheun amp Co. Ltd. Hanlon, R.T. and Hessenger, J.B. (1786) Adaptive coloration in young cuttlefish (Sepia officinalis L.): The morphology and develmpment of body patterns and their relation to behavior. Unp Holmes, W. 1940 the colour changes and colour patterns of Sepia mfficinalis L. Proc. Zool. Soc. Lond. A110, 17-25. Takahashi, 1785 Colom science (ed. B. Styles) End ed. FIGURE LEGEND Figure 1: Animalon sand background Figure 2: Animal on gray background Figure : Onimal on gray background Finure 4: Animalon gray backgrounc Figure 5: Animalon rock background Figure a: Onimal on rock background Figure 7: Gray scale versus relative intensity plot trom Megavision computer illustrating the linear gray scale. Fioure 8: Top: Histogram plot of Animal on sand measuring gray scale value and number of pixels. Bottom: Histogram of sand substrate. see tet. Figure 9: Histograms of Animal i on gray and gray substrate Figure 10: Histogram of Animal on gray and gray substrate Figure i1: Histogram of Onimal on gray and gray substrate Figure 12: Top Right: Histogram of animal l on racks. Top lett: histogram of large, square rock in the background. Bottom right: histogram of an area in the bowl the same size as the animal. bottom left: histogram output of whole bowl containing rocks. Fidure 13: Histograms of Onimal on rocks and various parts of the substrate. Finure 14: Fower spectral density function of animal i on sand. FEARE correspond to the distance between repeating gray values im pace. Figure I: Fower spectral density function of Animal i on gray zubstrate. Figure 16: Fower spectral density function of Animal on gray zubstrate. Figure 17: Fower spectral density function of Animal on gray zubstrate. Figure 18: Fower spectral density function of Animal l on rock bachargund. The graphs correspond to the animal's back and to one reck in the substrate (the large square one). he graphe corresponding to the area nest to the animal or the whole bowl nat shown. Figure 19: see explanation for figure 18. Fower spectral density function of Onimal . Figure 20: Top: comparison of the various body patterns shown on aray substrate by animals 1,2 and 3. Bottom: comparison of body patterns exhibited on the rock substrate. Both graphs illustrate the yaristy of patterns that are used by the animal to camoutlage itself. Fioure Ei: the percentages of patterns shown on the rock substrate and on the gray substrate. FIGURE FIGURE FIGURE 2 FIGURE 4 FIGURE 6 C — — — 1 ( FIGURE 7 atata- ktaa- + FIGURE 8 ANIMAL 3 ON SAND 1.0- 75 50 25 .00- 191 63 12. GRAY UQLUE 1.O-SAND .75 .50- 25 O0 191 12. 63 GRAY UQLUE MEAN 1SS STANDARD DEUIATION 22 255 MEAN 183 SIANDARD DEUIATION 23 255 FIGURE 9 OANIMALON GRAY 75 50- 25 MEAN 148 STANDARD DEUIATION 25 L .00 191 255 63 GRAY UQLUE 1.0- GRAY 75- 50- 25- MEAN 172 STANDARD DEUIATION S 00- 255 191 63 12 GRAY UQLUE FIGURE 10 1.0 AMMAL 42 ON GRAV 75- 50 2 O0 3 12 191 GRAT UQLUE 1.0— GRAY 75 50- 2 00 63 12 191 GRAY UALUE 255 MEAN ES STANDARD DEUIATION 43 MEAN 13 SIANDARD DEUIATION 2 55 FIGURE 11 1 O-ANIMAL 43 ON GRAV 50 25 MEAN 14 STANDAFO DEVIATION 22 00 191 255 127 63 GRAY UALUE GRAY 1.0- 75- 50 25 MEAN STANDARD DEUIATION O0 191 255 63 127 GRAT UQLUE FIGURE 12 ANIMAL ON ROCKS 90 MEAN 143 SIANDARD DEUIATION 2e 5 50- 25- OO 255 191 127 63 GRAY UALUE .0 WHOLE BOWL MEAN 127 SIANDARD DEUIATION S 25 50- 25 OU 191 255 63 GRAY UQLUE ONE ROCK MEAN 84 STANDARD DEUIATION IE H.25 50- 25- 00 191 127 63 GRAY UALUE 1.0- AREA NEXT TO ANIMAL MEAN 112 STANDARD DEUIATION S3 H25 50- 25 00 191 127 63 GRAY UQLUE FIGURE 13 ANIMAL *3 ON ROCKS 1.0 MEAN S7 SIANDAPC DEUIATICN 2 5 8.50- 25- 00 63 12. 191 GRAY UQLUE 1.0 AREAI NEXT TO ANIMAL MEAN 7 SIANDARD DEVIATION 34 75 .50- 25- 00 63 12. 191 255 GRAY UQLUE 255 1.O H75 50- 25- 00- 1.O- H.75 50- 25- 00 ONE ROCK MEAN STANDARD DEVIATION 15 63 127 191 2 GRAY UQLUE WHOLE BOWL MEAN 12 STANDARD DEUIATION as L 191 2 63 GROY UQLUE N J I + — —— — + J (sousnou) Whaas o FIGURE 14 Z J FIGURE 15 Whaltat am 8 — — Q Z I FIGURE 16 paaa- J N — I FIGURE 17 — snou J staa- 2 8 wha a FIGURE 18 0 N J FIGURE 19 —— waa — FICURE 20 PATTEFMIMIE OM GFAY SUESTEATE 100 + 90 - 80 - 60 - 50 40 50 l ER Animal +2 Animal #3 Animal fl EMING ON ROCK SUBSTRATE A 100 90 - 30 - 70 50 50 40 30 - L t Animal 13 Animal 2 NVEAK TEBRA SN DAR VSTPPE QVEAK OISAUPTNE N LIOET FIGURE 21 PAUIERNING ON ROCRS AND GEAT 100 - 90 80 - 70 - 50 - 40 - 30 20 101 4 Et Lt oEl GRA ROCKS NWEAK TBRA sroote JWEAK DISRUPTIVE HAUN LIGHT QUN. DARK