Conserved visual sensitivities across divergent lizard lineages that differ in an ultraviolet sexual signal

Abstract The sensory drive hypothesis predicts the correlated evolution of signaling traits and sensory perception in differing environments. For visual signals, adaptive divergence in both color signals and visual sensitivities between populations may contribute to reproductive isolation and promote speciation, but this has rarely been tested or shown in terrestrial species. We tested whether opsin protein expression differs between divergent lineages of the tawny dragon (Ctenophorus decresii) that differ in the presence/absence of an ultraviolet sexual signal. We measured the expression of four retinal cone opsin genes (SWS1, SWS2, RH2, and LWS) using droplet digital PCR. We show that gene expression between lineages does not differ significantly, including the UV wavelength sensitive SWS1. We discuss these results in the context of mounting evidence that visual sensitivities are highly conserved in terrestrial systems. Multiple competing requirements may constrain divergence of visual sensitivities in response to sexual signals. Instead, signal contrast could be increased via alternative mechanisms, such as background selection. Our results contribute to a growing understanding of the roles of visual ecology, phylogeny, and behavior on visual system evolution in reptiles.

However, visual sensitivities can be influenced by the relative proportion of different photoreceptor types (and therefore cone opsin expression) in the retina (reviewed in Carleton, 2014). For example, in New World warblers (Parulidae), relative opsin expression is associated with plumage dichromatism and light environment (Bloch, 2015). Similarly, a high abundance of ultraviolet (UV) sensitive cones have been associated with the presence of a UV signal in a lizard (Fleishman, Loew, & Whiting, 2011). Opsin gene expression may evolve more readily in response to varying selection than opsin spectral tuning (i.e., changing the wavelength of peak photoreceptor sensitivity, λ max ). However, evidence for an association between visual signals and color vision achieved by modifying the relative gene expression of cone opsins is currently limited in terrestrial species (Bloch, Morrow, Chang, & Price, 2015;Coyle, Hart, Carleton, & Borgia, 2012;Tseng et al., 2018;Yewers et al., 2015).
The tawny dragon lizard (Ctenophorus decresii; Duméril & Bibron, 1837) is a good candidate for examining changes in visual sensitivity associated with divergence of a sexual signal and the signaling environment. The species comprises two genetically and phenotypically distinct lineages which differ markedly in a sexual color signal, male throat coloration (Figure 1). Northern lineage males are polymorphic with four discrete throat morphs: orange, yellow, yellow with an orange center, and gray (Teasdale, Stevens, & Stuart-Fox, 2013), all of which lack significant UV reflectance. By contrast, southern lineage males are monomorphic with UV-blue throats with a consistent UV reflectance peak . This throat coloration is prominently displayed during territorial and courtship displays involving head-bobbing and push-ups (Gibbons, 1979(Gibbons, , 1977Osborne, Umbers, Backwell, & Keogh, 2012;Stuart-Fox & Johnston, 2005) and is locally adapted to increase conspicuousness against the predominant background colors of native lichen in their respective ranges . The northern lineage is primarily found in semi-arid sparsely vegetated habitats, whereas the southern lineage occurs in wetter, temperate, more vegetated habitats (Houston, 1974).
Characteristic of diurnal lizards, C. decresii has tetrachromatic vision with UV sensitive (SWS1; 364-383 nm), short-wavelength sensitive (SWS2; 440-467 nm), medium-wavelength sensitive (rod-like cone opsin RH2; 483-501 nm), and long-wavelength sensitive (LWS; 560-625 nm) cone opsins and one rod opsin (RH1; Yewers et al., 2015). There are no significant differences in the absorption spectra of visual pigments between lineages of C. decresii, nor in amino acid sequences of opsin genes (Yewers et al., 2015). However, color discrimination may be fine-tuned by differences in the relative proportion of photoreceptor types rather than shifts in their peak wavelength sensitivities. Given the occurrence of locally adapted throat F I G U R E 1 Sampling localities and male throat colors of the tawny dragon (Ctenophorus decresii). (a) Map showing localities in the northern and southern lineages. Elevated rocky ranges are shaded in gray. Average reflectance of male throat colors found in the (b) northern lineage: orange, yellow, gray, and in the (c) southern lineage: blue. Ultraviolet wavelengths are highlighted in gray; (d) Representative male throats for each color coloration in C. decresii, we hypothesized that the lineages may differ in visual sensitivities via the relative expression of cone opsin genes.
Specifically, we predicted that the southern lineage would exhibit higher expression of the UV sensitive SWS1 opsin gene due to the UV reflectance peak found on male throats.

| Animals
We analyzed the visual sensitivities of seven northern lineage (six males and one female) and nine southern lineage (seven males and two females) individuals (Table 1). We focused primarily on males because in C. decresii, males compete for access to females and opportunities for female male choice appear to be limited, as is generally the case in lizards (Lailvaux & Irschick, 2006;Lebas & Marshall, 2001;Olsson & Madsen, 1995;Smith & Zucker, 1997;Tokarz, 1995). Male-male interactions are therefore the strongest determinants of mating success in territorial lizards (Gullberg, Olsson, & Tegelstrom, 1997;Simon, 2011;Stamps & Krishnan, 1997;Tokarz, 1998); however, we included a subset of females for comparison.  (Slater & Birney, 2005). A single mix of primers and probes was used for each opsin gene for the length of the experiment.

| Statistical analyses
We used a linear mixed-effects model to examine the effects of lineage on opsin gene expression. Specifically, we had lineage, gene, and their interaction as fixed terms and included months in captivity, lizard ID, and age as random-effect terms (opsin expression~lineag e + gene + lineage*gene + (1|months) + (1|ID)) + (1|age). We used our model to examine the normality of residuals and found significant departure (Shapiro-Wilk normality test, p < .0001). We log-transformed the data and confirmed normality (p = .30). The random-effect terms each accounted for a negligible amount of variability in the model (<0.05 total variance). Further, we repeated this analysis on a dataset comprising only males (excluding females; opsin expr ession~lineage + gene + lineage*gene + (1|months) + (1|ID) + (1|age) ). Statistical tests were performed in R v3.3 (R Core Development Team, 2017) with the packages effects (Fox, 2003), lme4 (Bates, Mächler, Bolker, & Walker, 2014), and lmerTest (Kuznetsova, 2017).
We conducted a post hoc power analysis in the program G*Power v3.1.9.4 (Faul, Erdfelder, Buchner, & Lang, 2009;Faul, Erdfelder, Lang, & Buchner, 2007), and determined that the datasets comprising all samples and adult males had powers of 0.94 and 0.86, respectively. A power of 0.80 is generally regarded as an appropriate level of statistical power (Cohen, 1988).

| RE SULTS
There were no significant differences in gene expression of individual cone opsins between the northern and southern lineages (p > .05; Table 3). For the full dataset, the mean relative gene expression ± SE for the northern lineage was 0.011 ± 0.007, 0.044 ± 0.022, 0.373 ± 0.224, 0.572 ± 0.201 (SWS1, SWS2, RH2, LWS, respectively). Likewise for the southern lineage, mean relative gene expression ± SE was 0.007 ± 0.002, 0.032 ± 0.013, 0.286 ± 0.129, 0.675 ± 0.124 (Figure 2a). We found significant differences between opsin genes (p < .0001; Table 3). The relative expression patterns for the four opsin genes were similar between lineages with SWS1 as the lowest expressed gene, followed by SWS2, RH2, and TA B L E 2 Forward primer, probe, and reverse primer sequences for each of the four cone opsins

| D ISCUSS I ON
The coevolution of color signals and vision is central to the sensory drive hypothesis, a key proposed mechanism of adaptive divergence and speciation. We examined visual sensitivities of C. decresii and found no evidence that relative opsin gene expression levels correspond to divergence between lineages which differ in the presence/ absence of a UV sexual signal. We found differences in expression between the four opsin genes; this was similar to the relative ex-  (Barbour et al., 2002;Bowmaker, Loew, & Ott, 2005;Tseng et al., 2018). Our results support the hypothesis that diurnal lizards share a highly conserved ancestral pattern of tetrachromatic vision extending into the ultraviolet spectrum as characterized in the families Agamidae (Barbour et al., 2002;Yewers et al., 2015), Chamaeleonidae , Cordylidae (Fleishman et al., 2011), Dactyloidae (Kawamura & Yokoyama, 1998;Loew, Fleishman, Foster, & Provencio, 2002;  TA B L E 3 Results of linear mixedeffects models on all individuals (N = 16) and a subset of only males (N = 13), statistically significant values are italicized F I G U R E 2 Mean relative expressions (copies per µl) of the four cone opsin genes in the northern and southern lineages for (a) all individuals and (b) a subset of only males. 95% confidence bounds were calculated using parameters estimated from the fitted model inferred to have been present in the ancestral vertebrate (Cronin, Johnsen, Marshall, & Warrant, 2014).
These findings add to the growing body of evidence suggesting that the coevolution of color signals and visual systems is rare in terrestrial systems (Lind, Henze, Kelber, & Osorio, 2017). The strength of selection on visual sensitivities to optimize signal perception is likely to be weaker in terrestrial than aquatic systems because terrestrial light environments are less distinct and more variable over time and space whereas aquatic systems vary more steeply and consistently in background radiance (Chiao, Vorobyev, Cronin, & Osorio, 2000;Endler, 1993b;Goldsmith, 1990;Levine & MacNichol, 1979). ). This suggests that individuals could select a background substrate to maximize contrast or crypsis. There is evidence from various terrestrial taxa of local adaptation and/or that individuals can select or modify their environment to alter their conspicuousness (Bortolotti, Stoffel, & Galvan, 2011;Endler & Day, 2006;Endler & Thery, 1996;Gunderson, Fleishman, & Leal, 2018;Heindl & Winkler, 2003;Klomp, Stuart-Fox, Das, & Ord, 2017;Leal & Fleishman, 2002;Macedonia, 2001;Marshall, Philpot, & Stevens, 2016;Nafus et al., 2015;Sicsú, Manica, Maia, & Macedo, 2013;Uy & Endler, 2004). For example, two closely related species of Anolis lizards, A. cooki, and A. cristatellus, differ markedly in dewlap UV reflectance and have adaptively diverged in microhabitat preference to select light conditions that maximize signal contrast (Leal & Fleishman, 2002). However, similar to our findings, there are no significant differences in spectral sensitivity between ecologically diverse species of Anolis with varying dewlap colorations (Fleishman et al., 1997;Loew et al., 2002).
Other ecological factors may generate selection pressures on aspects of the visual system that constrain divergence in response to sexual signals. Visual systems have evolved to accommodate a broad gauntlet of activities critical to survival and fitness. Lizards rely primarily on visual signals at longer distances (López & Martín, 2001;López, Martín, & Cuadrado, 2002;Whiting, Webb, & Keogh, 2009), and individuals must detect and interpret many objects in their environment including suitable shelter, potential rivals and mates, predators, and prey. The lineages of C. decresii share similar predominantly avian predators and a generalist diet of insects (Gibbons, 1977), which often have color patterns that reflect or absorb selectively in the UV spectrum (Théry & Gomez, 2010 (Tseng et al., 2018). However, the great majority of evidence for environmental regulation of opsin expression comes from fish, which experience drastic changes in ambient light environment with habitat changes during ontogeny (Bowmaker & Kunz, 1987;Cheng, 2004;Cottrill et al., 2009;Shand, Archer, & Collin, 1999;Shand et al., 2008;Shand, Hart, Thomas, & Partridge, 2002) and even across diurnal cycles (Johnson, Stanis, & Fuller, 2013). It is possible that environmental factors contributed variation to our data; however, there were no differences in opsin gene expression between sexes and these groups do not differ in diet or habitat preferences (Gibbons, 1977).

| CON CLUS IONS
In summary, we found no evidence for divergence in visual sensitivities between lineages of C. decresii that differ in the presence of a UV sexual signal. Our findings are consistent with weaker divergent selection on visual sensitivities within and between closely related species in terrestrial systems. The lack of divergence in visual sensitivities between lineages of C. decresii can likely be attributed to similar selection on color vision imposed by the abiotic and biotic environment. Instead, male C. decresii may increase conspicuousness of their throat coloration to conspecifics by selecting contrasting backgrounds. By testing for opsin expression divergence in a terrestrial reptile, our study contributes to a growing understanding of broad-scale patterns in the coevolution of signals and sensory systems, and how they differ between taxonomic groups and environments.

ACK N OWLED G M ENTS
We are grateful to Madeleine Yewers for access to tissue samples and fieldwork; Katrina Rankin for animal husbandry assistance; Jessica Hacking and Adam Elliott for fieldwork; Franca Casagranda for ddPCR guidance; and the landowners who granted access to their property for data collection. This work was supported by the Australian Research Council (DP1092908 to D.S.-F.).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

AUTH O R CO NTR I B UTI O N S
All authors contributed to study design and statistical analysis. CAM contributed to fieldwork. CMD designed primers, conducted ddPCR data collection, and wrote the manuscript. All authors edited and approved the final manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset and R code are available from Dryad: https ://doi.