Conspicuousness
As a proxy for larvae conspicuousness to aerial and terrestrial predators, we estimated the contrast between the tadpole body coloration and the water background. We estimated conspicuousness from above because phyllomedusinae larvae are known to congregate at the water surface during the daytime (Branch 1983; Caramaschi and Jim 1983). Our experiments (see the previous section) had confirmed this presumption, at least for large-sized larvae (see Results). We are aware that aquatic predators, such as arthropod and fish, also prey on phyllomedusinae larvae (Magnusson & Hero 1991) yet the visibility of these ponds was below 5 cm, as estimated from the “disappearance” of larvae when they swim deeper from the surface. Moreover, the metallic and putatively conspicuous patch was located on the dorsal surface of the tadpole head and is arguably easier to spot from above the water.
The body-background contrast was measured on lateral and dorsal photographs of 193 larvae of P. vaillanti , 113 of P. bicolor , and 30 of C. tomopterna at variable developmental stages. A Kodak color card was included in each photograph as a control for white balance. We then used the software ImageJ (Schneider et al. 2017) to measure image red, green, and blue (RGB) brightness on six areas of the larvae’s body: the interocular patch, the loreal area, the middle dorsal area, the lateral area, the dorsolateral area, and the ventral zone (Figure 1-B). For background measurements, we took photographs of the water surface under daytime natural light and at the very ponds where we found the tadpoles.
To quantify and compare coloration we used the RGB decomposition method on the digital images (Endler 1990; Stevens et al. 2007; Amézquita et al. 2009). Briefly, we estimated relative brightness on each color channel: red (R* = r/ (r + g + b), green G* = g/ (r + g + b), and blue B* = b/ (r + g + b). Then, to map those measurements onto a two-dimensional space, we calculated two scores: LM = R* ̵̶ G*, and B* = B (Endler 1990; Amézquita et al. 2009). This approach provides a reasonable estimate (long-medium and short wavelengths) of the photoreceptor’s response in animals with trichromatic vision (Kebler et al. 2003). Finally, we used the multivariate distances between each tadpole area and the background coloration as our proxy for contrast or conspicuousness (Endler 1990).