In trying to understand the evolution of visual signal diversity we are often faced with problem of asking which of two possible signals is most visible against a particular natural background. Alternatively we may want to know how easily different signals (for example signals of sympatric species) can be told apart.
When animals use colorful signals to communicate with conspecifics, selection will tend to favor colors that are highly visible against the natural background and colors that can easily and unambiguously be distinguished from signals utilized by other similar-looking species. Thus, in order to understand the evolutionary pressures on signal properties, we often want to know how distinct two different colors appear to an animal viewer. Perception of color by most animals differs from human perception because of differences in visual-system anatomy and physiology. How, then, can we quantify the difference in appearance of colors to an animal viewer?
A useful approach is to quantify the perceived difference between color stimuli as a distance in the animal’s “perceptual color space.” This refers to some measure of the difference in sensory stimulation elicited by different colors. At present most studies of visual ecology and communication use one of two methods to measure perceptual color space distance between pairs of colors. The first involves determining the relative stimulation of the different classes of cone photoreceptors by each color stimulus in units of relative photon capture, plotting colors in n-dimensional space (where n is the number of photoreceptor classes), and then measuring a simple Euclidian distance between the points. The second approach involves estimating the noise within each photoreceptor channel, determining the detection threshold – defined as the just-noticeable difference, or JND, between the two colors – determined by photoreceptor noise, and then estimating the distance between two colors in units of multiples of this threshold, or in units of JNDs. The photoreceptor noise method has been shown to be effective for estimating color discrimination thresholds, but the idea that differences between supra-threshold colors can be measured in multiples of the JND is more controversial. Each of these methods has some theoretical support, and each has advantages and disadvantages in terms of assumptions required and the information needed to do the calculations. However, empirical tests of which method works best are lacking, and there is not much agreement on the best way to measure perceived differences between colors that are well beyond the threshold for discrimination.
What we need here is a model system – and that usually means anoles! Anoles have an excellent, well-understood color vision system. Moreover a method has been developed that uses visual attention as an assay of signal visibility. Fleishman et al. (2016) used this method to quantify the visibility of different colors viewed against a gray background by Anolis sagrei. They then used these behavioral results to test the accuracy of the two different popular methods of measuring distance in visual color space. In these experiments anoles viewed a checkerboard of gray squares in their visual periphery (Fig. 1). One square was then abruptly shifted to some other color. The distance in color space from the initial gray to the newly appearing color was determined using the two different metrics described above. The probability of a gaze shift towards the novel colored stimulus was quantified, and colors throughout the anoline color space were tested. The probability of detection was then plotted against the different measures of perceptual color distance.
Figure 1. (a) Lizard sits on the perch in a small cage. The experiment waits until its gaze is directed straight outward towards the video camera (t1). At t2 a small colored stimulus square is presented in the middle of the checkerboard visual field. The camera records whether or not the lizard notices the newly introduced color-square and shifts its gaze towards it. See the video above. The video at the top of the page illustrates the appearance of the stimulus to a lizard and shows a positive response.
Which method best predicted the results? It turns out both methods gave excellent predictions of the behavioral data! In this case anoles are the peacemakers. Basically, the information available to the color vision system is how much difference there is in the photon capture by different cone classes exposed to different colors. These two different ways of quantifying this change turn out to give rather similar predictions, and both sets of predictions accurately predict detection probability. Thus the anoles give a firm go ahead for both methods of modeling distance in color space. The ability to easily model distances between colors in anoline perceptual color space should greatly facilitate studies of the evolution of dewlap color diversity because we can easily figure out which color is most visible under each set of natural habitat light conditions.
Now we just have to figure out why Anolis dewlaps come in so many different colors!