Illumination geometry, detector position and the objective determination of animal signal colours in natural light

The study of colour patterns used in animal signals often requires objective measurement of spectral quality and intensity under natural field conditions. In practice this can be difficult because animals are mobile and may display dynamically at particular locations or times of day (Endler 1992; Endler & Thery 1996). The most common solution is to measure the spectral reflectance of the coloured surface under controlled laboratory conditions, measure the light typically striking the surface in the field (habitat irradiance) and multiply the two values together to produce an estimate of the intensity and/or spectral quality (e.g. Andersson 2000; Macedonia 2001; Cummings et al. 2003; Stuart-Fox et al. 2003). In general, however, the geometrical arrangement of light striking the surface (i.e. the density distribution and directional characteristics of light rays) and the position of the viewer are not the same in laboratory and field (Endler 1990). If so, this procedure will produce an accurate estimate of signal appearance only if two critical assumptions are met. First, the measured surface must reflect light in a highly diffuse manner for all visible wavelengths (Fig. 1). Second, reflected light must dominate the spectrum, with little contribution from transmitted light. In this paper, we illustrate a method for quantifying the effects of illumination and detection geometry on the spectrum and intensity of natural surfaces to test these critical assumptions. We use the body and dewlap (coloured throat fan) of the lizard Anolis cristatellus as examples. We then explore the consequences of failing to take deviations from these assumptions into account, by