DISCUSSION
We tested whether pigment concentrations are associated with environmental gradients indicative of carotenoid availability among agamid lizards, using a large interspecific dataset of pigment concentrations in coloured skin tissue. We found that species in more arid environments with high summer temperatures and radiation and lower vegetation productivity had lower concentrations of total carotenoids, higher concentrations of total pteridines and consequently, a lower ratio of carotenoids to pteridines. Across all species, the concentrations of carotenoid and pteridine pigments with similar hue (red ketocarotenoids and drosopterin, yellow dietary carotenoids and xanthopterin), were uncorrelated, indicating that carotenoids are not simply replaced with pteridine pigments of a similar hue (carotenoid mimicry). Although the concentration of dietary carotenoids was of similar magnitude to the concentration of drosopterin or xanthopterin, only pteridine concentrations predicted colour variation among species: redder hues were associated with higher concentrations of drosopterin, and more saturated colours were associated with higher concentration of pteridines (xanthopterin, other and total). We found no relationship between carotenoid or pteridine concentrations and indices of sexual selection (sexual dichromatism and sexual size dimorphism), which is consistent with the lack of association between carotenoid concentration and skin colour. Taken together, these results suggest that environmental carotenoid availability may alter the relative cost of acquiring and sequestering carotenoids vs synthesising pteridines to generate yellow-red skin colours.
In a series of pioneering studies, Grether and colleagues showed that genetically determined pteridine synthesis can compensate for environmental carotenoid availability among populations of guppies (Grether et al. 1999; Grether et al. 2001; Gretheret al. 2005). In this species, carotenoid and pteridine concentrations are positively correlated among populations to maintain a consistent ratio of tunaxanthin (yellow carotenoid) to drosopterin (red pteridine). This ratio produces the specific orange hue preferred by females (Grether et al. 1999; Grether et al. 2001; Deereet al. 2012). Stabilising selection acting on hue within guppies can explain the positive correlation between pigment types across streams. Across lizard species, however, hue varies greatly, and stabilising selection would not be expected as different species use different signalling colours. Our results instead suggest that pteridine synthesis balances geographic variation in carotenoid availability, irrespective of hue. In carotenoid-scarce environments, it may be less costly to synthesise pteridines than to acquire and metabolise carotenoids and vice versa in carotenoid-rich environments. However, the specific combination of pigments and skin colour pattern of each species is likely to depend on local selective pressures.
The strongest drivers of the association between total carotenoid concentrations in coloured skin and environmental PC1 were aridity and summer radiation. Most species of agamid lizards occupy semi-arid to arid environments, often with very little vegetation. All species of agamid lizard in this study are insectivorous, though some occasionally eat plant material including flowers (Cogger 2018; Melville 2019). Insects sequester carotenoids in proportion to their dietary availability (Heath et al. 2013); thus carotenoid availability may well be limited for both insects and their predators in arid environments. Limited dietary availability of carotenoids, however, does not necessarily mean that carotenoid availability is limiting for integument coloration. Available carotenoids may be sufficient to meet physiological and colour signalling requirements (Koch & Hill 2018). Furthermore, environmental availability can be compensated by more efficient carotenoid metabolism (e.g. assimilation and transport; Craig & Foote 2001; Koch & Hill 2018). Indeed, the prevailing view is that carotenoid limitation, where it exists, is due more to physiology (internal factors) than environmental availability (McGraw et al.2003; Hadfield & Owens 2006; Simons et al. 2014; Koch & Hill 2018). This view is primarily derived from the literature on birds, in which there is limited and inconsistent evidence for an association between feather carotenoid concentrations and diet (Mahler et al.2003; McGraw et al. 2003; Olson & Owens 2005). However, selection on carotenoid metabolism may differ greatly for birds compared to poikilothermic vertebrates (fish, amphibians, reptiles) because birds have different colour producing mechanisms and do not use pteridines to colour feathers. Thus, carotenoid availability may well be limiting for skin coloration in lizard species occupying arid environments.
We found that in agamid lizards, concentrations of ketocarotenoids were generally low (particularly astaxanthin) relative to other carotenoids. Astaxanthin is produced by a number of bacteria, fungi and algae, and can also be found in large quantities in some red flower petals (Ohmiya 2011). Agamid lizards are known to seek out and eat flower petals so could potentially obtain astaxanthin from the diet; however astaxanthin and other ketocarotenoids are generally rare in the diet of terrestrial animals (Svensson & Wong 2011; Heath et al. 2013; Koch & Hill 2018). Nevertheless, in some species, ketocarotenoids from dietary sources can accumulate when enzymes responsible for carotenoid breakdown, such as the β-carotene oxygenase enzymes BCMO1 and BCO2, are disrupted or deactivated (Twomeyet al. 2020a). More commonly, ketocarotenoids are metabolically converted from dietary yellow xanthophylls through oxidation reactions catalysed by ketolation enzymes (ketolases; Lopes et al. 2016; Mundy et al. 2016; Twyman et al. 2016). Metabolic conversion of dietary yellow xanthophylls to red ketocarotenoids has not been demonstrated in lizards, and the CYP2J19 gene that encodes the primary ketolase in birds and turtles is absent in squamates, tuataras and crocodilians (Twyman et al. 2016). A similar P450 enzyme (encoded by the gene CYP3A80) may act as a ketolase in the dendrobatid poison frog Ranitomeya sirensis and possibly other amphibians (Twomey et al. 2020a) but whether this may be the case in reptiles is not currently known. In this species of frog, the carotenoid cleavage enzyme BCO2 is also disrupted, possibly facilitating accumulation of ketocarotenoids and their dietary precursors (Twomeyet al. 2020a). BCO2 is associated with yellow coloration in the wall lizard, but not other polymorphic lacertids (Andrade et al.2019). Therefore, it is unclear whether agamid lizards have evolved mechanisms to enhance assimilation or enable conversion of dietary carotenoids to ketocarotenoids. The positive association we identified between the concentration of dietary carotenoids and ketocarotenoids could indicate increased ketocarotenoid conversion when dietary carotenoid availability is high, or that ketocarotenoids are similarly more available through diet. An absence of a mechanisms for ketocarotenoid conversion may explain the prevalence of drosopterin to produce orange and red hues in lizards and some other groups of poikilothermic vertebrates.
Among the 28 taxa in our dataset, skin colour was associated with the concentration of pteridines rather than carotenoids and there was no correlation between the two. In most other lizards, yellow is produced by high relative concentrations of dietary carotenoids and orange-red is produced by a high relative proportion of red pteridines (usually drosopterin; Ortiz et al. 1963; Ortiz & Maldonado 1966; Macedonia et al. 2000; Steffen & McGraw 2009; Weiss et al. 2012; Haisten et al. 2015; McLean et al. 2017; Andrade et al. 2019). Although carotenoids contribute to skin coloration, carotenoid concentrations are often uncorrelated with hue, saturation or luminance (Steffen et al. 2010; Weiss et al.2012). Instead, hue frequently corresponds to the concentration of red pteridines, particularly drosopterin (Steffen et al. 2010; Weisset al. 2012; Andrade et al. 2019). Our data is consistent with these studies and suggests that yellow-red coloration is seldom a reliable indicator of carotenoid content in lizards. This suggests in turn that expression of yellow-red signalling colours in lizards is unlikely to convey information on individual quality through mechanisms of honest carotenoid signaling such as resource trade-offs or indicator mechanisms (Koch et al. 2017; Koch & Hill 2018). Instead, the honesty of these colour signals may be maintained by other costs such as predation risk associated with conspicuous coloration (Stuart-Foxet al. 2003; Amdekar & Thaker 2019). More generally, honest carotenoid signalling may not apply to the many species of poikilothermic vertebrates that use a combination of pteridine and carotenoid pigments to generate yellow-red hues and have complex colour generation mechanisms.
Our comparative analysis uncovered broad patterns in pigment concentration; however, mechanisms underlying skin colour in reptiles are complex and influenced by structural components. In ectothermic vertebrates, colour is produced by the combination of chromatophore cells containing different pigment types or crystalline structures and structural components of the dermis (e.g. collagen and connective tissue). Xanthophores containing yellow to red carotenoid and/or pteridine pigments comprise the upper layer of chromatophores and may be underlain by iridophores containing periodically arranged guanine crystals, and melanophores containing melanin pigments (reviewed in Grether et al. 2004; Bagnara & Matsumoto 2006; Olsson et al. 2013; Ligon & McCartney 2016). The extraordinary diversity of integument colours in reptiles and other animals is produced by the interaction of pigments and structural components (Kemp et al.2012). For example, within a mimicry complex of poison frogs (Dendrobatidae), drosopterin contributes to orange coloration but variation in hue across the group is predominantly associated with the thickness of crystalline platelets within iridophores (i.e. structural; Twomey et al. 2020b). Furthermore, skin tissue commonly contains high concentrations of crystalline pteridines such as isoxanthopterin and pterin (Bagnara & Matsumoto 2006; McLean et al. 2017; McLeanet al. 2019; Twomey et al. 2020b). We found an association between the concentration of these pteridines and skin colour saturation and luminance. This represents novel evidence that pteridines that are often assumed to be colourless are associated with variation in integument coloration and may contribute to skin colour due to their crystallinity rather than spectral absorption (Oliphant & Hudon 1993; Palmer et al. 2018).
Overall, our results support a scenario where environmental carotenoid availability influences the relative concentrations of carotenoid and pteridine pigments used to generate yellow to red skin colours in lizards. Environmental gradients may shape the ecology and evolution of animal coloration by altering the relative cost of environmentally acquired and self-synthesised pigments.