Discussion
We found that our sample of heliconiian butterflies and poison frogs all reflected detectable quantities of ultraviolet light. When comparing between VIS and UV sensitive visual models, this UV reflectance had a negligible effect on achromatic contrast but did affect chromatic contrast, to varying degrees. The visual signals of heliconiian butterflies and poison frogs have both evolved under the influence of UV sensitive predators for the purpose of mitigating predation risk via aposematism. However, we found that UV reflectance from butterfly color patterns had a much greater effect on enhancing chromatic contrast, both in terms of absolute (change in JNDs) and proportional (percent increase due to the addition of UV) change, than was recorded from the color patterns of either of the poison frog species.
Maximizing visual contrast is not necessarily the goal of aposematic signals, and two patterns can be visually distinct (e.g., different combinations of colors) while being equally contrasting. However, by quantifying the contribution of UV to achromatic and chromatic contrast, our approach allows us to estimate the relative importance of these wavelengths to signal design. That said, it is important to note that the presence of UV reflectance within a color pattern does not equate to UV serving an explicit function; and depending on context UV reflectance could act as aposematism, camouflage, sexual signaling, thermoregulation, or protection from solar radiation (Umbers 2013). Moreover, pigments and structural colors will interact with light beyond the wavelengths visible to observers, and reflectance characteristics outside of the visible range may evolve without direct selection. For example, selection for very high reflectance across 400-700 nm would very plausibly, as a byproduct, also produce significant reflectance in the near ultraviolet (350-400 nm) and perhaps near infrared (700-750 nm).
We cannot, therefore, definitively state a function for ultraviolet reflectance in poison frogs, if indeed there is a function. However, by directly comparing the characteristics of frog coloring to the well-known UV signals of heliconiian butterflies we can provide guidance for future investigations.
Many heliconiian butterflies have evolved highly contrasting signals that contain a significant amount of UV light. However, despite high contrast, and likely being visible to potential predators, UV reflectance does not appear to play an important role in predator aversion (Dell’Aglio et al. 2018; Finkbeiner et al. 2017). The most compelling selection-based explanation for the evolution of UV+ signals comes from their potential use(s) for sexual selection in the genusHeliconius, where UV+ 3-hydroxy-DL-kynurenine (3-OHK) yellow pigments coincide with the duplication of UVS opsin genes (Briscoe et al. 2010). These signals have, therefore, co-evolved with complex UV sensitive visual systems that allow heliconiian butterflies to tune into UV reflectance for both mate choice and species recognition (Briscoe et al. 2010; Bybee et al. 2012; Finkbeiner et al. 2014; Finkbeiner et al. 2017). Visual discrimination that potentially plays an important role in preventing intergeneric hybridization between mimetic Heliconiusand Eueides butterflies (Finkbeiner et al. 2017). Whereas allHeliconius species have duplicated UV coding opsin genes, it is only likely truly influencing vision in certain clades ofHeliconius (such as H. erato, UV contrast shown in Figure 1E), yet even in these clades, expression is sex-specific and benefits apparently restricted to females (Finkbeiner and Briscoe 2020; McCulloch et al. 2016).
In comparison, UV reflectance in poison frogs appears to only have a small effect on pattern contrast, and its utility, if any, remains unknown (Yeager and Barnett 2020). Preliminary findings do not suggest UV in A. bilinguis inguinal flash marks enhances detection with model human predators (McEwan personal communication ). Although color can be an important intra-specific signal for poison frogs (Maan and Cummings 2009; Yang et al. 2019) the lack of UV sensitive photoreceptors in the dendrobatid visual system means that it is unlikely that ultraviolet reflectance has evolved in response to mating preferences or intraspecific recognition. Importantly, however, visual perception has only be characterized for O. pumilio (Siddiqi et al. 2004), a species that lacks UV reflectance (Chaves-Acuña et al. 2020; Maan and Cummings 2009; Siddiqi et al. 2004; Summers et al. 1999). Therefore, although it may be unlikely, we cannot conclusively rule out the presence of UV sensitive vision in other dendrobatid species. Moreover, as strong UV reflectance does not appear to affect predation risk in artificial targets (Lyytinen et al. 2001) or heliconiian butterflies (Finkbeiner et al. 2017), it also seems improbable that the comparatively weak UV reflectance observed in poison frogs would be an important contribution to aposematic signals. Indeed, where UV reflectance has been reported in poison frogs other color pattern combinations that lack UV have actually been found to result in greater visual contrast (Yeager and Barnett 2020), and are more likely to be avoided by avian predators (Lawrence and Noonan 2018).
To fully understand UV coloring in poison frogs we believe further study is required. Firstly, we believe that it is important to characterize the visual systems of a greater diversity of dendrobatid species considering the impressive diversity of intra- and inter-specific color patterns, where some species reflect UV and many well-studied species apparently do not. Secondly, behavioral trials both with potential predators and conspecifics are needed to examine whether observers respond differently to UV+ and UV- signals under natural lighting conditions. More widely, we believe that the role of UV reflectance in aposematic signaling deserves more attention, or perhaps publication bias against non-significant findings needs to be addressed. Finally, we caution about the over interpretation of function in animal coloration, such as in UV signals, and specifically suggest that neutral evolutionary processes may be more common in shaping animal color patterns than currently acknowledged.
Here we investigated the degree to which UV reflectance affected the visual contrast of conspicuous signals. UV is known to play an important role in intra-specific communication in heliconiian butterflies, and we found that UV had a correspondingly large effect on increasing the chromatic contrast of butterfly coloration. Conversely, poison frogs are not known to perceive UV light, and UV reflectance had a comparatively small effect on signal contrast. These data support the notion that UV reflectance does not necessarily have a special role in aposematic signal design and has likely evolved neutrally in many poison frogs (Yeager and Barnett 2020), however, much remains unknown.