Limits...
Changeable camouflage: how well can flounder resemble the colour and spatial scale of substrates in their natural habitats?

View Article: PubMed Central - PubMed

ABSTRACT

Flounder change colour and pattern for camouflage. We used a spectrometer to measure reflectance spectra and a digital camera to capture body patterns of two flounder species camouflaged on four natural backgrounds of different spatial scale (sand, small gravel, large gravel and rocks). We quantified the degree of spectral match between flounder and background relative to the situation of perfect camouflage in which flounder and background were assumed to have identical spectral distribution. Computations were carried out for three biologically relevant observers: monochromatic squid, dichromatic crab and trichromatic guitarfish. Our computations present a new approach to analysing datasets with multiple spectra that have large variance. Furthermore, to investigate the spatial match between flounder and background, images of flounder patterns were analysed using a custom program originally developed to study cuttlefish camouflage. Our results show that all flounder and background spectra fall within the same colour gamut and that, in terms of different observer visual systems, flounder matched most substrates in luminance and colour contrast. Flounder matched the spatial scales of all substrates except for rocks. We discuss findings in terms of flounder biology; furthermore, we discuss our methodology in light of hyperspectral technologies that combine high-resolution spectral and spatial imaging.

No MeSH data available.


Related in: MedlinePlus

Box plots showing the actual and ideal ΔS distributions calculated for two theoretical observers: dichromatic crab (a) and trichromatic ray (b). Results for each of these observers looking at P. dentatus are shown in the left panels; results for S. aquosus are shown in the right panels. The actual values for flounder, as determined by spectral measurements of flounder and substrates, are shown in dark grey; the theoretical ideal values (a calculation that considers the likelihood of detection if flounder were identical to background, i.e. a ‘perfect camouflage’ situation) are shown in light grey. Asterisks indicate that the ideal and actual distributions are statistically equivalent.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5383827&req=5

RSOS160824F7: Box plots showing the actual and ideal ΔS distributions calculated for two theoretical observers: dichromatic crab (a) and trichromatic ray (b). Results for each of these observers looking at P. dentatus are shown in the left panels; results for S. aquosus are shown in the right panels. The actual values for flounder, as determined by spectral measurements of flounder and substrates, are shown in dark grey; the theoretical ideal values (a calculation that considers the likelihood of detection if flounder were identical to background, i.e. a ‘perfect camouflage’ situation) are shown in light grey. Asterisks indicate that the ideal and actual distributions are statistically equivalent.

Mentions: What would perfect camouflage look like? Simplistically described, a flounder could be effectively concealed against a given background if it had the same spectral range and texture of the substrate (i.e. spatial patterning), assuming it did not cast any shadows. Here, we simplify this definition even further by omitting patterning and focusing only on perfect chromatic camouflage. We created a ‘perfect’ chromatic camouflage scenario by assuming that the fish and the background had the same spectral gamut. To obtain ΔL and ΔS distributions that we termed ‘ideal’, we combined all spectral measurements from each species of flounder taken on a given substrate, assumed that the background was composed of the same spectra, and calculated ΔL and ΔS for all possible pairwise combinations between them. Next, we calculated the same for the ‘actual’ spectral measurements for fish and substrate. Then we converted both the ideal and actual distributions to probability density functions. Finally, we used the Wilcoxon rank sum test [44] at the 5% significance level to test whether or not the ideal and actual scenarios came from distributions with equal medians, which gave us a mathematical tool to determine how good the flounders' camouflage was. For these comparisons, we identified two ways in which an actual distribution could be statistically equivalent to the ideal: (i) with the acceptance of the hypothesis for the Wilcoxon rank sum test at the 5% significance level or (ii) if every ΔL or ΔS element in the actual distribution was less than 1, implying that all pairs of spectra were indistinguishable from each other. Before applying the Wilcoxon rank sum test, we normalized each bin in each distribution by sample size in that bin, obtaining the probability of members in that bin in the range 0–1. While more mathematical than biological, comparing the distribution of JNDs between the theoretical ideal scenario and the actual measured scenario allows us to gain more insights into the detectability of the flounder, as seen through the eyes of relevant observers, when multiple colours are involved. For our dataset, this approach yielded 1200 JND values per flounder per substrate (total of 40 fish spectra; 30 substrate spectra, resulting in 1200 spectral pairs); these are presented as box plots in figure 6 (luminance contrast) and figure 7 (chromatic contrast). Note that since the distributions were normalized for probability, the amplitudes shown no longer correspond to ΔL or ΔS thresholds.


Changeable camouflage: how well can flounder resemble the colour and spatial scale of substrates in their natural habitats?
Box plots showing the actual and ideal ΔS distributions calculated for two theoretical observers: dichromatic crab (a) and trichromatic ray (b). Results for each of these observers looking at P. dentatus are shown in the left panels; results for S. aquosus are shown in the right panels. The actual values for flounder, as determined by spectral measurements of flounder and substrates, are shown in dark grey; the theoretical ideal values (a calculation that considers the likelihood of detection if flounder were identical to background, i.e. a ‘perfect camouflage’ situation) are shown in light grey. Asterisks indicate that the ideal and actual distributions are statistically equivalent.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5383827&req=5

RSOS160824F7: Box plots showing the actual and ideal ΔS distributions calculated for two theoretical observers: dichromatic crab (a) and trichromatic ray (b). Results for each of these observers looking at P. dentatus are shown in the left panels; results for S. aquosus are shown in the right panels. The actual values for flounder, as determined by spectral measurements of flounder and substrates, are shown in dark grey; the theoretical ideal values (a calculation that considers the likelihood of detection if flounder were identical to background, i.e. a ‘perfect camouflage’ situation) are shown in light grey. Asterisks indicate that the ideal and actual distributions are statistically equivalent.
Mentions: What would perfect camouflage look like? Simplistically described, a flounder could be effectively concealed against a given background if it had the same spectral range and texture of the substrate (i.e. spatial patterning), assuming it did not cast any shadows. Here, we simplify this definition even further by omitting patterning and focusing only on perfect chromatic camouflage. We created a ‘perfect’ chromatic camouflage scenario by assuming that the fish and the background had the same spectral gamut. To obtain ΔL and ΔS distributions that we termed ‘ideal’, we combined all spectral measurements from each species of flounder taken on a given substrate, assumed that the background was composed of the same spectra, and calculated ΔL and ΔS for all possible pairwise combinations between them. Next, we calculated the same for the ‘actual’ spectral measurements for fish and substrate. Then we converted both the ideal and actual distributions to probability density functions. Finally, we used the Wilcoxon rank sum test [44] at the 5% significance level to test whether or not the ideal and actual scenarios came from distributions with equal medians, which gave us a mathematical tool to determine how good the flounders' camouflage was. For these comparisons, we identified two ways in which an actual distribution could be statistically equivalent to the ideal: (i) with the acceptance of the hypothesis for the Wilcoxon rank sum test at the 5% significance level or (ii) if every ΔL or ΔS element in the actual distribution was less than 1, implying that all pairs of spectra were indistinguishable from each other. Before applying the Wilcoxon rank sum test, we normalized each bin in each distribution by sample size in that bin, obtaining the probability of members in that bin in the range 0–1. While more mathematical than biological, comparing the distribution of JNDs between the theoretical ideal scenario and the actual measured scenario allows us to gain more insights into the detectability of the flounder, as seen through the eyes of relevant observers, when multiple colours are involved. For our dataset, this approach yielded 1200 JND values per flounder per substrate (total of 40 fish spectra; 30 substrate spectra, resulting in 1200 spectral pairs); these are presented as box plots in figure 6 (luminance contrast) and figure 7 (chromatic contrast). Note that since the distributions were normalized for probability, the amplitudes shown no longer correspond to ΔL or ΔS thresholds.

View Article: PubMed Central - PubMed

ABSTRACT

Flounder change colour and pattern for camouflage. We used a spectrometer to measure reflectance spectra and a digital camera to capture body patterns of two flounder species camouflaged on four natural backgrounds of different spatial scale (sand, small gravel, large gravel and rocks). We quantified the degree of spectral match between flounder and background relative to the situation of perfect camouflage in which flounder and background were assumed to have identical spectral distribution. Computations were carried out for three biologically relevant observers: monochromatic squid, dichromatic crab and trichromatic guitarfish. Our computations present a new approach to analysing datasets with multiple spectra that have large variance. Furthermore, to investigate the spatial match between flounder and background, images of flounder patterns were analysed using a custom program originally developed to study cuttlefish camouflage. Our results show that all flounder and background spectra fall within the same colour gamut and that, in terms of different observer visual systems, flounder matched most substrates in luminance and colour contrast. Flounder matched the spatial scales of all substrates except for rocks. We discuss findings in terms of flounder biology; furthermore, we discuss our methodology in light of hyperspectral technologies that combine high-resolution spectral and spatial imaging.

No MeSH data available.


Related in: MedlinePlus