Computational protocol: Fear no colors? Observer clothing color influences lizard escape behavior

Similar protocols

Protocol publication

[…] We selected two blue T-shirts that appeared to match the color of the lizards’ ventral patches, but differed in brightness to the human eye (dark blue and light blue). We also included a red and a gray T-shirt because both have reflectance spectra distinct from blue. Gray is also a background color on the lizards, but probably not a sexually-selected signaling color and should be less conspicuous than red. We used reflectance spectroscopy to measure the peak wavelengths and reflectance spectra of each T-shirt and of the blue ventral patches of adult male S. occidentalis. We caught and measured multiple body regions of four adult male S. occidentalis at Stunt Ranch: dorsal blue, dorsal background (brown/gray), center of the blue throat patch, throat background, black border of the blue abdominal patch on the left side, center of the blue abdominal patch on the left side, and abdomen background. For each area of interest, we took three separate measurements, which we averaged together. We used an Ocean Optics spectrometer (USB 2000) with a fiber optic reflectance probe (Ocean Optics R200-7-UV-VIS) and a pulsed xenon light source (Ocean Optics PX-2) to measure reflectance in a 1.3 mm diameter patch. To reduce glare, the probe was placed at a 45° angle relative to the surface being measured. We measured reflectance relative to a Labsphere certified reflectance standard in Ocean Optics’ software OOIBase32. We processed and plotted reflectance spectra using the R package pavo [] (, ).To determine quantitatively whether our blue T-shirts resembled the lizards’ ventral blue patches, we used Vorobyev and Osorio's [] receptor noise model to calculate the just noticeable difference values (JNDs) for both chromatic and luminance (achromatic) contrasts between all T-shirts and the lizards’ blue patches. A JND value less than 1 indicates that two regions cannot be discriminated while a JND greater than 1 indicates that the regions are visually distinct, with higher values indicating greater contrast. We used the vismodel function in pavo to calculate the quantal catch from the visual system of the collared lizard Crotaphytus dickersonae (from []), using the D65 illuminant, and calculating the illuminance catch based on the sum of all cones. We then calculated JNDs from data generated from this visual model. Because published information on the retinal cone sensitivities of Sceloporus lizards is not available, we used the receptor spectral sensitivities of the collared lizard, another diurnal tetrachromatic iguanid lizard species, which has been used previously as a proxy for Sceloporus vision []. We used the default values for Weber fractions and cone ratios in the pavo package.To determine the conspicuousness of the T-shirts in the environment, we took photographs of one person (BJP) wearing each T-shirt at 10 randomly selected locations in each study site. Photos were taken at a distance of 2 m from the observer who was wearing the T-shirt along with an 18% gray standard that was centered in the camera’s field of view. Pictures were taken with a Canon Rebel XSi fitted with an EF-S 18–55 mm lens (Canon U.S.A., Inc; Melville, NY); we used manual settings for shutter speed and lens aperture, the white balance was set to daylight, and images were saved in RAW file format. All photos were taken with the observer in the open, in direct sunlight, and facing the sun to minimize differences in lighting among images. Photos were taken between 0900–1300 h, the same timeframe when behavioral trials occurred.We used the Multispectral Image Calibration and Analysis Toolbox plugin for ImageJ [] to analyze our images. Prior to analysis, images were screened for exposure issues and we found that several had been overexposed. Thus only 4 from Stunt and 6 from Griffith Park were used. Suitable images were linearized and equalized using the gray standard. We transformed the camera’s red, green, and blue channel pixel values to a lizard-specific color space, converting to animal cone-catch values using the camera spectral sensitivity functions (under the D65 illuminant), database of natural spectra (400–700 nm), and the visual system of the collared lizard. In short, simulations predict the photoreceptor responses of the lizard to the natural spectra, and the camera's responses to the same spectra under the D65 standard illuminant. A model is then generated that can predict animal photoreceptor cone-catch values from camera values.For each pixel, we calculated long wave sensitive (LWS), medium wave sensitive (MWS), short wave sensitive (SWS), and double cone photon catches (LUM) of the model species. Although fence lizards likely are able to detect UV, we regarded that the UV cone type would play a negligible role in the relative differences in detectability among the shirts because the shirts reflected minimally in the UV range (). We divided each image into 2 regions: T-shirt (area just below the gray standard) and background (all area surrounding the observer from the waist up). We manually selected these regions using the ‘rectangular selection’ and ‘polygon selection’ tools in ImageJ respectively, and we calculated JNDs between these two regions for color and luminance, thus accounting for chromatic and achromatic contrasts respectively [,]. We used a Weber fraction of 0.05, and we assumed equal cone ratios (following [,]). With these comparisons, we could determine whether lizards were responding to color per se or conspicuousness. […]

Pipeline specifications

Software tools pavo, ImageJ
Applications Miscellaneous, Microscopic phenotype analysis
Organisms Homo sapiens, Sceloporus occidentalis