Opsin variation in a poison frog that shows color-associated behavioral biases

doi:10.21203/rs.3.rs-3129154/v1

Download PDF

Research Article

Opsin variation in a poison frog that shows color-associated behavioral biases

https://doi.org/10.21203/rs.3.rs-3129154/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Animals send and receive signals, which are honed by evolution for detection by their intended receiver, for a variety of purposes. For signals used in mate choice, selection should favor those that are easy to detect by the receiving sex’s sensory system(s). Here we test for genetic differences affecting color vision among populations and color morphs of the polytypic poison frog, Oophaga pumilio, a species where both males and females show color-assortative behavioral biases. Opsin genes, expressed in rod and cone cells in the retina, play an important role in the tuning of color vision. We therefore hypothesized that the frog’s behavioral biases may be underlain by differences in the expression or sequences of these genes. To test this, we compared closely related populations of O. pumilio containing red and green morphs to test for convergence in opsin expression and sequence by color. We also compared opsins in sympatric morphs from two polymorphic regions with those of neighboring monomorphic populations to test for differences that would confer improved color discrimination in the polymorphic area. We found no significant differences in opsin expression among red and green populations nor among frogs from one polymorphic (red/blue) zone. However, consistent with our predictions for greater color discrimination, we found greater RH1 expression in frogs from a second polymorphic (red/yellow) region. We found several opsin sequence variants that showed no clear association with color morph, though some mutations were only found in polymorphic populations and some mutations should have significant impacts on color vision in affected individuals. Overall, we found limited evidence that differences in color vision have evolved among distinctly colored O. pumilio populations, though the variation we observed within populations does suggest that the raw material needed for selection to act on color vision is present.

retina

cone cell

color perception

Dendrobatidae

aposematic

mate choice

The ability to send and receive signals is critical to animal communication and successful signaling often results in direct fitness benefits (e.g., Kiriazis and Slobodchikoff 2006; Mappes et al. 2005). Oftentimes obtaining a mate also depends on successful signaling to the opposite sex (e.g., Loyau et al. 2005; Vidya and Sukumar 2005). For a signal to be effective, three things must happen; the signaler must produce the signal, the signal must travel to the receiver, and the signal must be perceivable by the receiver, standing out from background noise (Endler 1992). For example, for a peacock to signal to a female, the male must produce long, brightly pigmented feathers and display them to a female, the female must have the visual capability to perceive feathers’ details, and this signal must stand out from the green of the forest habitat. Thus, the sensory system(s) of the receiver is/are an important component of signal evolution, and selection favors signals that are tuned to match the receiver’s sensory capabilities. Variation in perceivability of signals by receivers can thus act as a selective force shaping sexual traits.

Sexual selection as an agent of evolutionary change relies on members of one sex, typically males, successfully signaling their fitness to members of the choosier sex, typically females. The ability of the chooser to perceive differences in the signals of potential mates is thought to influence which mate is chosen (rev. in Kokko et al. 2002; Ryan and Cummings 2013) and males with a non-optimal signal with respect to female perception may have their signal lost against background noise (Endler 1992). Animals generally have better perception in some portion(s) of their total sensory range and males are predicted to be more successful in attracting a female’s attention and interest when their sexual signals fall within the range that the female can best perceive (Endler 1992). If signal perception varies among females, this could lead to males “specializing” on different ranges of the sexual signal, producing variation in the signal driven by variation in the female sensory system (Endler and Basolo 1998). For visual signals, this has been seen in aquatic species like sticklebacks (Boughman 2001) and African cichlids (Maan et al. 2006). However, for terrestrial species such sensory divergence is more commonly found in other modalities (e.g., auditory: Patricelli and Blickley 2006).

Vision in vertebrates is made possible by light sensing photoreceptors (rods and cones) in the retina of the eye, which possess the photopigment opsin. Color vision results when multiple cone types, each containing an opsin protein that absorbs light at a different range of wavelengths, are present in the retina (rev. in Cronin et al. 2014). These opsins absorb incoming light, which produces a conformational change in the chromophore, a vitamin-A based ligand bound to the opsin protein (rev. in Bloch 2016). This reaction sets off the phototransduction cascade which, among other events, results in the disassociation of the opsin and the chromophore (known as photoreceptor bleaching) and causes a signal to be sent through the nervous system to the brain (Fain et al. 2010). The relative signal strengths from the different cone types in the same section of the retina are compared by the brain to determine an object’s color (rev. in Gegenfurtner and Kiper 2003; Cronin et al. 2014). This complex sequence of events allows for numerous ways that color vision can be modified by evolution.

Changes to the opsin protein sequences and their expression are two such ways. The amino acid sequence of the opsin protein, particularly at certain key sites, is important in determining the range of wavelengths that the opsin absorbs (Yokoyama and Radlwimmer 1998; rev in Yokoyama 2008; Bowmaker 2008). Over time, evolution can select for changes in the opsin sequence that improve color vision in areas of the spectrum where selection favors better sensitivity and/or discrimination (rev. in Price 2017). Changes to the expression of the opsin mRNA transcripts is another way that vertebrate visual systems can evolve to detect and discriminate signals more effectively. Greater opsin expression is generally thought to indicate a greater density of the opsin protein in the retina, either due to higher density of photoreceptors, or due to a greater density of opsin protein per photoreceptor (rev. in Price 2017). Since each photoreceptor has a unique opsin protein, expression levels of the opsin genes can be used as a measure of visual sensitivity across different regions of the visible spectrum and variation in opsin expression has been shown to correlate with behavioral sensitivity to that opsin’s color range (Li et al 2005). Thus, variation in the expression of the different opsin proteins can be used to infer differences in color vision among individuals or populations.

Sexual selection can cause behavioral reproductive isolation, a step on the road to speciation, when divergence in mating traits and preferences for them prevent mating between individuals or populations bearing different mating trait variants (Ritchie 2007; Servedio and Boughman 2017; Coyne and Orr 2004; Panhuis et al. 2001). A few studies have shown convincing evidence for color vision differences occurring in association with differential female preferences in aquatic species, creating the sort of disruptive selection that can drive speciation (rev. in Carleton 2009). However, to understand more fully the role that intraspecific differences in sensory systems play in speciation we need to study how this mechanism works in a wider range of taxa. The optical properties of air and water differ in how they affect light attenuation at various wavelengths (Endler 1993; Cronin et al. 2014) and thus, the properties of the eyes of animals in each type of habitat have evolved in different ways. To better understand whether the likelihood of intraspecific divergence in color vision differs among these habitats more studies in terrestrial systems are needed. To date, most comparative opsin work in terrestrial systems has tested for phylogenetic patterns in opsin presence, for example among bird species (Bloch 2015; Bloch et al. 2015). The studies we are aware of that have tested for intraspecific differences in color vision associated with sexual selection in terrestrial animals have thus far not found a clear link between color vision and sexual selection (e.g., lizard, Yewers et al. 2015; damselflies; Bybee et al. 2012; Huang et al. 2014), though they remain few in number.

Oophaga pumilio is a terrestrial poison frog that shows extreme color polymorphism and polytypism across the Bocas del Toro archipelago of Panama (Myers and Daly 1976; Wang and Summers et al. 2010; Rudh et al. 2010). This archipelago formed recently (apx. 1000–9000 years ago, Anderson and Handley 2002), and recent work suggests that sexual selection played a role in shaping the evolution of color variation in this species. Female O. pumilio tend to prefer males of the local color (Reynolds and Fitzpatrick 2007; Maan and Cummings 2008; Richards-Zawacki and Cummings 2010; Maan and Cummings 2012) and territorial males are more aggressive toward encroaching males of the local color (Yang et al 2016) though these behavioral biases appear to be stronger in monomorphic populations than in polymorphic ones (Yang et al. 2016; Richards-Zawacki et al. 2012). Thus, it appears that both sexes have the capability to distinguish differences between color morphs and differences in color vision could have contributed to the evolution of these differential color-based biases. Recent evidence suggests that female mate preference and male aggression biases are learned as tadpoles through imprinting on the mother’s color (Yang et al. 2019b). Thus, the ability to accurately distinguish colors appears to be important from an early age.

Our knowledge of the visual system of this (and most other) frogs is far from complete. What we do know about O. pumilio comes from microspectrophotometry (MSP) data from a single captive individual and suggests that these frogs have trichromatic color vision owing to three cone types being present in the retina: a red cone with wavelength of maximum absorption, (λ_max) of 561nm, a green cone with λ_max of 489nm, and a blue cone with λ_max of 466nm (Siddiqi et al. 2004). While the typical red (LWS) and blue (SWS1) cone opsins are present in the O. pumilio genome, it appears that the typical green cone opsin, RH2, has been lost from Dendrobatid frogs, including O. pumilio (Wan et al. in review). However, based on the MSP data from Siddiqui et al. (2004), the pigment in the green cones of O. pumilio has a similar absorbance spectrum to the pigment in the rod, suggesting that the green cones are using the same opsin protein (RH1) as rods.

Here we test the hypothesis that two aspects of color vision, opsin sequence and opsin mRNA expression, differ between morphs of O. pumilio in ways that putatively improve discrimination of the local frog color. We tested for convergence in color vision characteristics among populations of a similar color, taking advantage of the existence of multiple morphs that appear to have converged on red or green coloration. If color vision has been important in shaping mate preference, the red morphs should have more similarity to each other in their opsin sequence and expression than they do to the green populations (and vice versa). Additionally, we examined these same color vision components in two polymorphic populations, where frogs must be able to accurately discriminate the different colors present in order to show color associated behavioral biases. Here we predicted that frogs in the polymorphic populations would have visual systems that are better able to discriminate between the local colors than neighboring monomorphic populations would. The polymorphic nature of this species offers a unique chance to test for differences in the visual system within a species that is putatively diverging due to sexual selection; it allows a test of the idea that divergence in visual systems can contribute to the early stages of speciation. Additionally, this study adds to our understanding of how intraspecific variation in the visual system of a terrestrial species may evolve.

To test for differences in opsin sequence and expression between O. pumilio morphs, we collected five frogs from each of 10 monomorphic populations as well as five frogs per morph from two polymorphic populations. Red is the ancestral color of this species, though it seems to have evolved more than once (Freeborn 2020). Three monomorphic green and three monomorphic red populations were used for our red/green comparisons (Fig. 1). To minimize the potential for similarities due to common ancestry, we chose same-colored populations from across the phylogeny for our comparisons so as to maximize the genetic distances among them (Fig. 2). The remaining four monomorphic populations we used were the closest (geographically, and often genetically, see Figs. 1 and 2) monomorphic populations bearing similar coloration to one of the morphs from each polymorphic zone. That is, for the Dolphin Bay polymorphic population (which contains red, blue, and intermediately-colored brown frogs), we used a nearby red monomorphic (SCBH) and a blue monomorphic (NRA) for comparison (the intermediate brown phenotype is not found elsewhere). Likewise, for the Cemetery polymorphic population, which contains red and yellow-green frogs, we used nearby yellow-green (ICN) and red (LL) monomorphic populations for comparison (Fig. 1). We used the monomorphic/polymorphic population sets to test for differences in color vision between frogs that encounter a wider variety of conspecific coloration in their local population (polymorphic) and their monomorphic neighbors (Fig. 1).

Frogs were hand captured in the field, housed in plastic terraria for transport to the Bocas del Toro field station at the Smithsonian Tropical Research Institute (STRI), and processed at approximately the same time of day. Processing involved euthanasia by sharp decapitation followed by double pithing, after which the eyes were quickly removed then punctured (to allow penetration of RNAlater). Eyes were then placed in RNAlater (Ambion) and stored overnight at 4^oC then moved to -20^oC. The samples were shipped on dry ice to the University of Pittsburgh where they were again stored at -20°C until extraction.

RNA extraction

We extracted total RNA from the retina of one eye per animal and reverse transcribed the mRNA into cDNA. To do this, immediately prior to extraction, we dissected out the retina of one haphazardly chosen eye and homogenized it in Trizol (Invitrogen) with a mini pestle in a 1.5mL tube. We extracted total RNA using the DirectZol RNA mini-prep kit (Zymo) following the manufacturer’s instructions and including the optional 15min on-column DNase incubation. The resulting RNA was then quantified using an Epoch Microplate Spectrophotometer (Biotek) and 250ng of total RNA was converted into cDNA using the SuperScript IV VILO kit (Invitrogen), which includes both oligo(dt)18 and random hexamers to prime the cDNA synthesis reaction. Additionally, for each sample, we ran a no reverse transcriptase reaction (NRT) that had all the same components as the reverse transcribed (RT) samples, except for the reverse transcriptase. We used these samples as controls to confirm the lack of contamination in the sample during qPCR reactions. cDNA samples were diluted 1:10 in molecular grade water prior to qPCR to overcome PCR inhibition.

Sequencing

We designed primers to span the entirety of each O. pumilio opsin gene transcript. Since the opsin transcripts are longer than a typical Sanger sequence (LWS-1223, RH1-1406, SWS1-1309 bp), we designed two overlapping primer sets (Table S1) from the transcripts of each O. pumilio opsin gene (provided by L. O’Connell) using Primer3 (Untergasser et al. 2012; Koressaar and Remm 2007). Initially we used GoTaqGreen (Promega) for PCR reactions (see Supplementary Information for reaction conditions). However, due to concerns about the error rate of Taq polymerase we switched to a hot-start, high-fidelity Q5 polymerase (New England Biolabs, Ipswich, MA) part way through the study. We re-sequenced with Q5 to verify SNPs that were first identified in a GoTaqGreen-produced sequence. We visualized PCR products on a 0.75% 50mL agarose gel, cut out bands of the expected size, and cleaned these using the Monarch DNA Gel Extraction Kit (New England Biolabs, Ipswich, MA) before having thenm Sanger sequenced on an ABI 3730xl (Azenta Life Sciences, Chelmsford, MA). We trimmed sequence ends and inspected trace files visually to assess sequence quality. We manually checked base pairs called as “N” for possible heterozygosity using Geneious v 9.1.8 (www.geneious.com). We then aligned the sequence halves for each gene with the full mRNA transcript (provided by L. O’Connell) in MEGA v 7.026 (Kumar et al. 2016), manually combined sequences to create contigs, and translated contigs using ExPASy (web version, Swiss Institute of Bioinformatics). In cases where the two sequence parts failed to overlap, we inserted Xs to fill the gap. mRNA and protein sequences were aligned in MEGA v 7.026 (Kumar et al. 2016) and examined for any differences between sequences. For every SNP and amino acid change, we manually examined the trace file at the relevant base pair(s) to confirm the sequence variant.

Gene Expression Assay

We developed qPCR primers for the O. pumilio opsin proteins (Supplementary Information: methods and Table S2). After choosing primers, we ran qPCR on the cDNA samples using a QuantStudio 3 (Applied Biosystems) to measure gene expression of the opsins relative to our reference gene. Samples were run in triplicate with all genes from a given sample run on the same plate. If one of the opsin genes required a redo of the qPCR (see Supplemental Information), the reference gene was rerun as well to avoid variation due to differences between plate set-ups and qPCR runs. For each sample, we included a no-reverse transcription control well for each opsin gene (i.e., four no RT controls per sample) on the qPCR run. These controls included all qPCR components, but the NRT sample was used for the template instead of the RT. Each run also had a no template control well for each gene to ensure there were no contaminants introduced during qPCR setup. We calculated the relative quantity of transcripts for each sample using the delta Ct method (Pfaffl 2001), where delta Ct is calculated as the Ct value of the reference gene (PDE6g) minus the Ct value of the opsin gene and used this value for all downstream analyses.

Statistical Analysis

All analyses were run in R v. 3.6.2 (R Core Team 2021) except where otherwise noted. We assessed that model assumptions were met by 1) checking residual vs fitted value plots for linearity, 2) checking the Q-Q plots of the residuals for normality, and 3) checking residuals vs leverage plots for outliers. We used the lm function in package “lme4” (Bates et al. 2015) to specify linear models comparing opsin expression between monomorphic populations in the red and green frog comparison (red: GLB, HOSP, OLP; green: NIPA, PUNT, CAN; Fig. 1). We used separate linear models for each gene with main effects of frog color (red vs. green), population, and sex (male vs. female), and included a color by population interaction term. We used Tukey tests in GraphPad Prism v 9.3.1 for Windows (GraphPad Software, San Diego, CA) for post-hoc comparisons of expression for each population pair.

We tested each of the two polymorphic populations for differences in opsin expression among its constituent color morphs using separate linear models. Each sympatric morph was included as a separate category in these models, as were the nearest (geographically) monomorphic populations that shared the same colorations. These included frog color, population, sex, and a color by population interaction term as main effects. Post-hoc comparisons for each population set (for Dolphin Bay: DPB (three morphs), SCBH, and RA; for Cemetery: CEM (two morphs), ICN and LL; Fig. 1) were done using Tukey tests in GraphPad Prism.

If opsin expression differs among populations or morphs, those of the same color should have similar overall (i.e., across all three genes) opsin expression profiles. To test this, we used a principal components analysis (PCA) that included the delta Ct values from all three opsin genes. Separate PCA’s were run for each comparison (green/red, Cemetery, and Dolphin Bay populations, as described above) using the prcomp function in R. We then visually examined graphs of PC1 vs PC2 for clustering by population and/or morph.

Sequencing

We sequenced LWS, RH1, and SWS1 for 49 individuals from 12 populations. For most populations we sequenced two individuals but for a few populations (NIPA, NRA) we sequenced a third individual because one individual was amplified using GoTaqGreen (Promega) before we made the shift to using high fidelity Q5 polymerase (New England Biolabs) and the resequencing attempt of the GoTaqGreen sample was unsuccessful with Q5. All but two of the SNPs we detected in this study found were in samples that were sequenced with Q5, and one of the two SNPs found using GoTaqGreen was also found in two other individuals that were sequenced with Q5, suggesting that the inclusion of the GoTaqGreen samples did not influence the conclusions of our study. Our initial sequencing pass resulted in the discovery of a nonsense mutation in one of the polymorphic populations. Seeing this, we increased our sequencing effort to include all samples collected for both polymorphic sites (five per morph, totals: n = 10 from CEM and n = 15from DBP). We were able to sequence all three opsin genes for most individuals. However, we were unable to obtain sequences for parts of RH1 and SWS1 for one individual from the GLB population. Additionally, we were unable to obtain overlapping sequences from our two primer sets per gene for some individuals, which necessitated filling gap in 13 sequences from 11 individuals. This was more common in RH1 and SWS1 (5 and 6 samples, respectively) than in LWS (2 samples) and the gaps ranged from 3-172 bp in size (mean 58.3bp). Gaps were longer in SWS1 sequences (mean = 90.4bp) than in LWS (mean = 62.5bp) and RH1 (mean = 19.3bp).

We found several SNPs resulting in changes to the amino acid sequences. LWS and RH1 each had four amino acid residues with missense mutations, one of which was shared by individuals from different populations (shared by two populations in LWS, three in RH1). One of the missense mutations in LWS was in the K296 site (bovine rhodopsin numbering) critical for chromophore binding. Another in LWS was shared by two differently-colored frogs from the polymorphic DBP population. SWS1 had 11 amino acid residues with missense mutations and one with a nonsense mutation (residue 108 of O. pumilio, residue 111 in bovine rhodopsin, Hargrave et al. 1983). One residue had a missense mutation that was shared by two differently-colored frogs from the polymorphic DBP population. Most frogs with a SNP were heterozygous for the wild type sequence at that base pair, but this was not the case for the K296 mutation or the nonsense mutation (Fig. 3). Additionally, both the nonsense mutation and the K296 mutation were found in frogs sequenced using Q5.

Gene Expression Assay

RNA extractions were successful for all 74 retina samples. Most samples resulted in extractions that permitted us to use 250ng of RNA as template in the cDNA reaction. Some samples produced nearly 250ng of RNA, and for these we investigated the potential impact of a smaller template amount using our qPCR data. The average Ct values for samples that contained < 250ng of RNA were generally within 1 cycle of the average of samples with the full 250ng (range of Ct differences between groups = 0.32–1.029). However, this difference in starting template values should have been adequately controlled by our use of a reference gene, as concentrations of opsins and reference gene should have been equally impacted.

There were no significant expression differences between red and green frogs for any opsin (Fig. 4; linear model, main effect of color: |t| ≤ 1.40, p ≥ 0.173, Table S4). However, there was a marginally significant color by population interaction for RH1 (linear model: |t| = 1.97, p = 0.06) where HOSP (red) had lower expression and GLB (red) had higher expression (Fig. 5). When comparing among the overall opsin expression means for the red/green frogs, RH1 expression was generally the greatest, followed by LWS, then SWS1 (Table S5). In the comparison of Dolphin Bay polymorphic zone frogs (red, blue, and brown morphs) with their nearest monomorphic neighbors, we found no significant differences in expression among morphs or populations and no significant morph by population interactions for any of the opsin genes (Fig. 6; linear model: |t| ≤ 1.74, p ≥ 0.10, Table S4). Here again, RH1 had the highest expression, followed by LWS then SWS1(Table S5). For the Cemetery red/yellow polymorphic zone, there was a significant color by population interaction for RH1 (linear model: |t| = 2.33, p = 0.03, Table S4) with both Cemetery morphs having greater RH1 expression than either of the nearby monomorphic morphs. However, in post-hoc pairwise comparisons the RH1 expression of polymorphic Cemetery morphs were only significantly found to be higher than the yellow-green (ICN) morph (Tukey, p ≤ 0.01). There were no other significant effects of color or population, or significant color by population interactions (linear models: |t| ≤ 1.437, p ≥ 0.17) for any of the three opsins. Similar to the other comparisons, RH1 had the highest expression in the Cemetery polymorphic zone, followed by LWS, and SWS1 (Table S5).

The PCA including expression for all three opsins showed significant eigenvalues for PC1 (1.495) and PC2 (1.001), which explained 49.848% and 33.367% of the variation, respectively. RH1 and SWS1 both aligned strongly and negatively with PC1, while LWS aligned strongly and positively with PC2 (Table S3). When PC1 was plotted against PC2, frogs did not visually cluster together by population, nor did they cluster together by dorsal color category (i.e., red, green, etc.; Figures S1-S3).

Our hypothesis that opsin sequences would differ among morphs was largely unsupported by the data. There were no mutations found in residues at any of the key sites for spectral tuning that have been identified in other taxa (Yokoyama and Radlwimmer 1998). Additionally, most of the mutations in the sequences that we found were only found in one individual frog and almost all of the mutations that were found in multiple individuals were not found in individuals from the same population, nor did they typically occur in frogs of the same color, suggesting the mutations did not coevolve with color. It should be noted, however, that by sampling only two-five frogs per morph we likely missed some of the variation that is present in opsin gene sequences within and among populations. Deeper sampling across a smaller number of populations might reveal spatial patterns in some of the mutations we found.

While our main hypothesis for opsin sequence variation was not supported, a couple of the mutations we identified are likely to have a major impact on color vision for the individual frogs in which they were found. For example, we found a nonsense mutation in the SWS1 opsin of a red frog from the polymorphic Cemetery (CEM) population that occurs early in the protein. Since the individual is homozygous for the mutation, it is likely that this frog has dichromatic color vision achieved using red (LWS) and green (RH1) cones and therefore sees quite differently than its trichromatic conspecifics. Moreover, this frog probably has a reduced ability to distinguish between the red and yellow-green morphs present within its polymorphic population, which would appear to it as different shades of red (Jefferson and Harvey 2006). Another variant we found that is likely to have an impact on color vision is the mutation of the lysine residue at position 311 in the O. pumilio LWS transcript (296 by bovine rhodopsin numbering) as it is a mutation in the site of chromophore binding (rev. in Pepe 1999). This mutation was also found in the homozygous state, so there is no wild-type protein to rescue the mutant phenotype. Since the mutation results in a change in the amino acid residue (Table 1), chromophore binding is likely to be disrupted and result in limited functionality of the LWS opsin for that frog. That suggests that this frog, too has dichromatic color vision, but using green (RH1) and blue (SWS1) cones. Since the O. pumilio green cones have low absorption at wavelengths of light longer than 575nm (Siddiqui et al. 2004), this frog likely has difficulty detecting photons of light in the orange and red ranges of the color spectrum. This frog was from the Dolphin Bay polymorphic (DBP) population, which has red, blue, and intermediate (brownish) colored frogs (Yang et al. 2019a), suggesting that the individual would have difficulty discriminating between red frogs and intermediately colored brown frogs. Additionally, mutations in this site are linked with retinitis pigmentosa, a group of diseases characterized by retina degradation in humans (Robinson et al. 1992), suggesting that this mutation might have wider ranging effects on the frog than just impacting its color vision.

Table 1

Summary of amino acid residue mutations found with numbering according to *O. pumilio* opsin proteins. Letters in sample names indicate population, as in Fig. 1. Amino acid properties are coded as: N = nonpolar, P = polar, + = positive charge, - = negative charge. Additional variants found at these sites in other species are listed, along with the species used and their GenBank accession numbers.
Gene	Mutation	Number of frogs with mutation	Sample found in	Amino acid properties of change	Amino acid variants at residue	Other species used (GenBank accession number)
LWS	A95D	2	DBP08 DBP11	P→ X	invariable	Nanorana parkeri¹⁴ (XP_018416216), Ambystoma tigrinum¹ (AAC96070.1), Cynops pyrrhogaster² (BAB55453.1), Xenopus laevis³ (NP_001084114.1), X. tropicalis⁴(NP_001096331.1), Rana catesbeiana⁵ (PIO33359.1), human red⁶ (NP_064445.2), human green⁷ (NP_001041646.1)
	L220P	1	CAN04	N →N	M, I, L
	A240T	1	CEM10	N→ P	H, Q
	K311N	1	DBP04	+ → P	invariable
RH1	V137M	1	ICN03	N → N	L, I	Xenopus laevis⁸ (NP_001080517.1), X. tropicalis⁹ (NP_001090803.1), Nanorana parkeri¹⁴ (XP_018410729.1), Ornithorhynchus anatinus¹⁰ (NP_001121099.1), Crocodylus porosus (XP_019392098.1), Tachyglossus aculeatus¹¹ (AFO70161.1), Gavialis gangeticus¹⁴ XP_019360201.1), Geospiza fortis¹⁴ (XP_005426698.1)
	I218V	3	CAN05 GLB04 LL02	N → N	I, V
	I319N	1	DBP08	N → P	I, L
	E343K	1	CEM09	- → +	invariable
SWS1	G101A	1	CEM04	N → N	invariable	Xenopus laevis¹² (NP_001079121.1), X. tropicalis¹² (NP_001119548.1), Rana catesbeiana⁵ (PIO33958.1), Nanorana parkeri¹⁴ (XP_018416245.1), Haliaeetus leucocephalus¹⁴ (XP_010567082.1), Picoides pubescens¹⁴ (XP_009898521.1), Mus pahari¹⁴ (XP_021045511.1), Condylura cristata¹⁴ XP_004677073.1), Caunus lupis familiaris¹⁴ (XP_539386.2), Uta stansburiana¹³ (AAZ79909.1)
	E108Stop	1	CEM03	- → stop	E, D
	E129K	1	CEM03	- → +	invariable
	R130K	1	CEM03	+ → +	invariable
	V151 M/G	2	CEM04 DBP10	N → N	invariable
	L157F	1	ICN02	N → N	invariable
	Q179P	1	DBP10	P → N	invariable
	T191I	2	DBP11 DPB15	P → N	invariable
	F205L	1	NIPA02	N → N	invariable
	S235F	1	NIPA03	P → N	invariable
	S255C	1	DBP15	P → P	invariable
References: ¹Xu et al. (1998), ²Sakakibara et al. (2002), ³Babu et al. (2002), ⁴Sato et al. (2011)., ⁵Warren et al. Unpub, ⁶Wang et al. (2023), ⁷Hoffman et al. (2017)., ⁸Kroeger et al. (2014), ⁹Lodowski et al. (2013)., ¹⁰Davies et al. (2007), ¹¹Bickelmann et al. (2012), ¹²Klein et al. (2002), ¹³Su et al. (2006)., ¹⁴NCBI Genbank predicted protein sequence.

It is hard to be certain what impact the other mutations we found could have on the frogs’ color vision as they are not at any of the known key sites based on the previously studied (non-amphibian) species and it is unclear whether or how mutations that are merely close to key sites, as ours often were, might affect the λ_max. Several of the mutations we found are in sites that are invariable across a wide range of species (see Table 1), suggesting that these are important sites for the protein’s function (e.g., changes could impact λ_max or disrupt protein folding). Additionally, since many of the mutations we found are in the trans-membrane domains, it is possible that they could cause minor disruption to the protein structure. Most known key sites from studies of other taxa occur in the transmembrane domains, and it makes sense that mutations in this part of the protein could change interactions between residues and potentially affect λ_max in O. pumilio, particularly since many of the changes we found result in a change of charge or polarity at that residue (Table 1). Additionally, our findings of more mutations in SWS1 compared to LWS and RH1 is consistent with the findings of a survey of several frog clades, including the poison frog family Dendrobatidae, which found more amino acid residues under positive selection in SWS1 than in LWS and RH1 (Wan et al. in review). This suggests stronger selection on SWS1, perhaps due to the apparent loss of SWS2 in Dendrobatid frogs (Wan et al. in review) and the shift in SWS1 opsin from the ancestral state of having a λ_max in the UV range to having a λ_max the blue range in O. pumilio. Interestingly, none of these sites under positive selection in the broader amphibian opsin study (Wan et al. in review) or in a study of frog species from Texas, USA (Schott et al. 2022) were found in sites known to be key for spectral tuning in other, previously studied taxa. This suggests the potential for amphibian key sites to be different than those found in other non-amphibian taxa. Or, perhaps amphibians share some of key sites with other vertebrate groups while other key sites are unique to amphibians?

Most of what is known about key sites for spectral tuning comes from studies done in a handful of species. The typical approach taken to identify key sites has been to compare the protein sequence of the species of interest with that of the ancestral opsin and investigate the effect of residues that differ using MSP on proteins that have been altered via site-directed mutagenesis (rev. in Hauser et al. 2014). At least 35 species have been studied for spectral tuning sites in LWS (26 mammals, 3 birds, 2 reptiles, 1 amphibian, 3 fish; Yokoyama and Radlwimmer 1998; Yokoyama et al. 2002; Yokoyama and Radlwimmer 2001 and refs within), with five sites having been considered “key” because of their influence on spectral tuning (Yokoyama and Radlwimmer 1998). The range of wavelength shift sizes reported in LWS is 2-16nm and the average wavelength shift size for all key site mutations in LWS is 9.5nm (rev. in Yokoyama et al. 2008). For RH1, at least 45 species have been studied for spectral tuning sites (17 mammals and 28 cichlid fishes; Sakmar et al. 1989; Hunt et al. 2001; Yokoyama et al. 2005; Sugawara et al. 2005; Levenson et al. 2006 and refs within), with 10 key sites having been identified. Mutations in these residues cause shifts of 3–21 nm (mean 10.8 nm) in λ_max (rev. in Yokoyama et al. 2008). For SWS1, at least 19 species have been studied for spectral tuning sites, with five key sites that cause an average shift of 28nm from wildtype, with a range from 0-72nm shift sizes having been found (8 mammals, 6 birds, 2 reptiles, 1 amphibian, 2 fish; rev. in Hauser et al. 2014). Thus, while multiple species have been tested for spectral tuning sites in each opsin, these have been mainly limited to mammals and fish. Therefore, it is possible that amphibian opsins have different or additional spectral tuning sites that have yet to be detected. Studies that specifically test for key sites that result in shifts in λ_max in amphibian opsins are necessary to better understand opsin evolution in this group.

While we hypothesized that opsin expression would differ among morphs, in most cases we found opsin expression patterns to be similar across O. pumilio color morphs; RH1 had the highest expression and SWS1 had the lowest expression, with LWS expression falling between that of the reference gene (PDE6g) and SWS1. Comparisons of opsin gene expression for red and green frogs from multiple populations also revealed no consistent expression differences among color categories. While we did find a marginally significant color by population interaction, the differences were not consistent with our hypothesis that populations of the same color have more similar opsin expression profiles to one another than they do populations of a different color. Instead, our marginally significant interaction seems to show that there was more variation in opsin expression among populations of red frogs than among populations of green frogs. Overall, we interpret these results as suggesting that opsin expression has not evolved in concert with coloration for these populations.

We also saw a significant color by population interaction for RH1 when we compared frogs from one of our polymorphic populations with frogs from nearby monomorphic populations. Both red and yellow-green morphs from the polymorphic Cemetery (CEM) population of Bastimentos Island had greater expression of RH1 than frogs from the monomorphic yellow-green population from Colón Island (ICN). The greater expression of the RH1 opsin for the Cemetery morphs suggests that these frogs have better color discrimination in the yellow to orange-red range compared to the monomorphic green frogs. An increase in RH1 expression would increase the overlap in wavelengths absorbed by the green and red cones, thus allowing for better color discrimination in the overlap region which covers the green, yellow, and orange part of the color spectrum (Siddiqui et al. 2004, Cronin et al. 2014). Increased opsin expression has been linked to increased opsin density in the photoreceptors, as well as the density of photoreceptors within the retina, and to increased sensitivity to light (rev. in Price 2017). Therefore, it is conceivable that this increase in expression for the RH1 is due to selection favoring frogs being able to better distinguish between the red and yellow frogs present in the Cemetery polymorphic population. However, because RH1 appears to be used in both the rods and green cones (Wan et al. in review), this increase in expression of RH1 could also be explained by increased rod usage. Most studies that have tested for changes in rhodopsin (RH1) mRNA transcript levels throughout the day have found that numbers of mRNA transcripts increase throughout the day and decrease overnight (Hartman et al. 2001; Kamphuis et al. 2005; Yu et al. 2007; but see McGinnis et al. 1992). Thus, it appears that rod expression is generally highest in daylight, and therefore our measure of increased RH1 expression in Cemetery morphs of O. pumilio could be due to greater rod use, cone use, or both. All samples for this study were taken at approximately the same time of day, suggesting that variation associated with circadian variation of opsin expression are unlikely to explain the observed differences. This, coupled with the fact that lighting conditions are similar among O. pumilio populations (Yeager 2015), makes the explanation of greater rod use in just this population seem less plausible.

Most of the differences in opsin sequences and expression levels we found in O. pumilio that have the potential to impact color vision were detected in the polymorphic Cemetery and Dolphin Bay populations. It seems plausible that color discrimination, which could be achieved by changes in opsin expression levels, opsin sequences, or both, could be selected for more strongly (or perhaps exclusively) in polymorphic populations where multiple color morphs co-occur and choices to fight or court with differently colored individuals may affect individual fitness (Yang et al. 2016, Richards-Zawacki et al. 2012, Yang et al. 2019b). Considering the two polymorphic populations we studied, it makes sense that we might see evidence of selection for increased color discrimination, particularly in the middle- to long-wavelength opsins, in the Cemetery (yellow-green vs. red) population because these two sympatric morphs are closer in color than the two pure Dolphin Bay morphs (blue vs. red), which are at opposite ends of the visual spectrum. In our sequence data we did find several residues with mutations shared by two Dolphin Bay frogs, which we did not see in any other population. While this could potentially be due to evolution of spectral tuning to the different color morphs present in the polymorphic zone this seems unlikely as the mutations were not found at sites known to affect the λ_max. Instead, the greater number of mutations discovered in Dolphin Bay frogs may reflect sampling bias, as 15 individuals (5 red, 5 blue, 5 intermediate) from this population were sampled, whereas only 10 (5 red, 5 yellow-green) were sampled from the Cemetery population, and two or three were sampled from every other population in this study. The apparent correlation between number of frogs sampled and number of opsin mutations found suggests that sampling a larger number of individuals from each population would likely yield more mutations and enable a more rigorous comparison of opsin variation among populations.

In summary, differences in color vision due to opsin expression or sequence differences do not appear to have been a driving force behind the development of the color-based behavioral biases seen in different color morphs of O. pumilio as neither opsin expression nor opsin sequence showed consistent differences among morphs. However, variation in both opsin sequence and expression are present in O. pumilio, and we detected such variation more commonly in the polymorphic populations where color discrimination may be important in the defense of territories by males and in female choice. The fact that differences in opsin expression and sequence exist among individuals, and that such variants are not particularly rare, suggests the potential for color vision to evolve in response to ongoing sexual selection. It is important to note that the small and uneven sample sizes we were able to achieve with this, the first published study of opsin variation in O. pumilio, would preclude us from detecting all but the largest effect sizes at the morph and population levels. Follow up studies with deeper sampling within morphs and populations and studies that test whether changes in opsin expression and sequence result in meaningful changes in color vision would go a long way toward understanding the role that opsin evolution may have played in the species' striking diversification.

The general lack of differences in spectral tuning among O. pumilio color morphs appears to stand in stark contrast with results from aquatic systems (Guppy: Smith et al. 2011; Sandkam et al. 2015a; Sandkam et al. 2015b; Cichlid fishes: rev. in Carleton 2009). This could be indicative of a more general pattern, where the properties of water as a medium for visual communication produce stronger selection on visual systems compared to air. However, it is important to remember that O. pumilio’s recent radiation into distinct color morphs, which is hypothesized to have occurred as the islands formed ~ 1,000–9,000 years ago (Anderson and Handley 2002), is orders of magnitude more recent than the cichlid radiation (~ 300,000 years ago, rev. in Seehausen 2006). Thus, even if selection were strong in air, O. pumilio morphs may simply not have had sufficient time to diverge in their opsin expression and/or sequences in response to it.

Additionally, it is possible that while opsin expression and sequences do not generally vary among morphs, that other aspects of color vision, which were not the focus of this study, do. For instance, variation in chromophore usage between morphs could be used for spectral tuning, as frogs have been found to use two different chromophore forms and the form of the chromophore that is used affects λ_max.. However, what little we know about chromophore usage in amphibians comes primarily from differences found between tadpoles and adult frogs in a single species (Rana pipiens) and this pattern appears not to have been investigated any further in frogs (Liebman and Entine 1968). Additionally, O. pumilio photoreceptors contain an oil droplet, which could be used to tune color vision via pigmentation in the oil (Cronin et al. 2014). Thus far what little we know about oil droplets in this species has come from work on captive animals where the droplets were found to be clear and unpigmented (Hailman 1976, Siddiqui et al. 2004). This suggests that the oil droplets may not be a tuning mechanism in this species, though this is another area that would benefit from further study as it is unclear how the oil droplets of captive frogs, which had not been eating the species typical diet for some time, might differ from those of wild frogs. While this study adds to our knowledge of the potential for color vision evolution within a color polymorphic terrestrial species, there is still much more to learn about color vision’s role in shaping color-based biases and their role sexual selection.

Acknowledgements

We thank L. Freeborn and G. Freeborn for their help collecting frogs in the field and processing them at Smithsonian Tropical Research Institute’s Bocas de Toro research station. Thanks also to J. Boyle for countless discussions when troubleshooting the qPCR assay and the use of his lab space for setting up qPCR plates. Thanks to M. Lee for help in troubleshooting the RNA extraction and to J. Stephenson for helpful comments on earlier drafts of this manuscript.

Funding

Support for this research came from start-up funds provided by the University of Pittsburgh to CLR-Z.

Conflicts of Interest/Competing Interest

The authors declare no conflicts of interest or competing interests

Ethics Approval

The work was approved by The University of Pittsburgh Institutional Animal Care and Use Committee (IACUC) protocol #21089903.

Consent for Participation

N/A

Consent for Publication

All authors consent to publication of this work

Availability of Data and Material/Code availability

Raw data will be deposited into Dryad upon acceptance.

Authors’ Contributions

KAH and CLR-Z conceived of and planned experiments, KAH conducted experiments, KAH and CLR-Z wrote the paper.

Anderson RP, and Handley Jr CO (2002) Dwarfism in insular sloths: biogeography, selection, and evolutionary rate. Evol 56(5):1045-1058
Babu S, McIlvain V, Whitaker SL, et al. (2006) Conserved cis-elements in the Xenopus red opsin promoter necessary for cone-specific expression. FEBS letters 580(5): 1479-1484.
Bates D, Kliegl R, Vasishth S, and Baayen H (2015) Parsimonious mixed models. arXiv preprint arXiv:1506.04967
Bickelmann C, Morrow JM, Mueller J, et al. (2012) Functional characterization of the rod visual pigment of the echidna (Tachyglossus aculeatus), a basal mammal. Visual Neuroscience, 29(4-5): 211-217.
Boughman JW (2001) Divergent sexual selection enhances reproductive isolation in sticklebacks. Letters to Nature 411:944-948
Bowmaker JK (2008) Evolution of vertebrate visual pigments. Vision Research 48(20):2022-2041
Bloch NI (2015) Evolution of opsin expression in birds driven by sexual selection and habitat. Proc of the Royal Society B 282:20142321
Bloch NI, Price TD, and Chang BSW (2015) Evolutionary dynamics of RH2 opsins in birds domonstrate an episode fo accelerated evolution in the New World warblers (Setophaga). Mol Ecol 24(10):2449-2462
Bloch NI (2016) The evolution of opsins and color vision: connecting genotype to a complex phenotype. Acta Biológica Colombiana 21(3):481-494
Bybee SM, Johnson KK, Gehring EJ et al (2012) All the better to see you with: a review of odonate color vision with transcriptomic insight into the odonate eye. Organisms, Diversity, and Evol 12:241-250
Carleton K (2009) Cichlid fish visual systems: mechanisms of spectral tuning. Integrative Zool 4(1):75-86
Coyne JA. And Orr HA (2004) Speciation. Sinauer Associates Inc., Sunderland, MA
Cronin, TW, Johnson S, Marshall NJ, and Warrant EJ (2014) Visual Ecology. Princeton University Press, Princeton, NJ
Davies WL, Carvalho LS, Cowing JA, et al. (2007) Visual pigments of the platypus: a novel route to mammalian colour vision. Current Biology 17(5): R161-R163.
Endler JA (1992) Signals, signal conditions and the direction of evolution. The Am Nat 139:S125-S153
Endler JA (1993) The color of light in forests and its implications. Ecol Monographs 63(1):1-27
Endler JA, and Basolo AL (1997) Sensory ecology, receiver biases and sexual selection. TREE 13: 415-420.
Fain GL, Matthews HR, Cornwall MC et al (2001) Adaptation in vertebrate photoreceptors. Phys Rev 81(1):117-151
Freeborn L (2020) The genetic, cellular, and evolutionary basis of skin coloration in the highly polymorphic poison frog, Oophaga pumilio. Dissertation, University of Pittsburgh
Gegenfurtner KR, and Kiper DC (2003) Color vision. Annua Rev of Neuro 26:181-206
Hailman JP (1976) Oildroplets in the eyes of adult anuran amphibians: a comparative survey. J of Morph 148(4):453-468
Hartman SJ, Menon I, Haug-Collet K et al (2001) Expression of rhodopsin and arrestin during the light-dark cycle in Drosophila. Mol Vis 7:95-100
Hauser FE., van Hazel I, and Chang BS (2014) Spectral tuning in vertebrate short wavelength‐sensitive 1 (SWS1) visual pigments: Can wavelength sensitivity be inferred from sequence data?. J of Exp Zool Part B: Mol and Develop Evol 322(7):529-539
Hofmann L, Alexander NS, Sun W, et al. (2017) Hydrogen/deuterium exchange mass spectrometry of human green opsin reveals a conserved pro-pro motif in extracellular loop 2 of monostable visual G protein-coupled receptors. Biochemistry 56(17):2338-2348.
Huang S-C, Chiou T-H, Marshall J et al (2014) Spectral sensitivities and color signals in a polymorphic damselfly. Plos One 9(1):e87972
Hunt DM, Dulai KS, Partridge JC, Cottrill, P, et al. (2001) The molecular basis for spectral tuning of rod visual pigments in deep-sea fish. J of Exp Bio 204(19):3333-3344
Jefferson L, and Harvey R (2006) Accommodating color blind computer users. Proc of the 8th Intl ACM SIGACCESS Conf on Comp and Acc pp 40-47
Kamphuis, W, Cailotto C, Dijk F et al (2005) Circadian expression of clock genes and clock-controlled genes in the rat retina. Biochem and Biophysi Reseach Comm 330(1):18-26
Kiriazis, J, and Slobodchikoff, CN (2006) Perceptual specificity in the alarm calls of Gunnison's prairie dogs. Behav Proc 73(1):29-35
Klein SL, Strausberg RL, Wagner L, et al. (2002) Genetic and genomic tools for Xenopus research: the NIH Xenopus initiative: a peer reviewed forum. Developmental dynamics: an official publication of the American Association of Anatomists 225(4): 384-391.
Kokko, H., Brooks R, Jennions MD et al (2003) The evolution of mate choice and mating biases. Proc of the Royal Society of London. Series B: Biol Sci 270(1515):653-664
Koressaar, T, and Remm M (2007) Enhancements and modifications of primer design program Primer3. Bioinformatics 23:1289–1291.
Kroeger H, LaVail MM, and Lin JH (2014) Endoplasmic reticulum stress in vertebrate mutant rhodopsin models of retinal degeneration (pp. 585-592). Springer New York.
Kumar S, Stecher G, and. Tamura K (2016) MEGA7: Molecular Evolultionary Genetics Analysis version 7.0 for bigger datasets. Mol Biol and Evol 33:1870-1874
Levenson DH, Ponganis PJ, Crognale MA, et al (2006) Visual pigments of marine carnivores: pinnipeds, polar bear, and sea otter. J of Comp Physiol A 192:833-843
Li P, Temple S, Gao Y, Haimberger TJ et al (2005) Circadian rhythms of behavioral cone sensitivity and long wavelength opsin mRNA expression: a correlation study in zebrafish. The J of Exp Biol 208:497-504
Liebman PA, and Entine G (1968) Visual pigments of frog and tadpole (Rana pipiens). Vision Research 8(7):761-IN7
Lodowski KH, Lee R, Ropelewski P, et al. (2013) Signals governing the trafficking and mistrafficking of a ciliary GPCR, rhodopsin. Journal of Neuroscience 33(34):13621-13638
Loyau A, Saint Jalme M, Cagniant C et al (2005) Multiple sexual advertisements honestly reflect health status in peacocks (Pavo cristatus). Behav Ecol and Socio 58(6):552-557
Maan ME and Cummings ME (2008) Female preferences for aposematic signal components in a polymorphic poison frog. Evol 62:2334-2345.
Maan ME and Cummings ME (2012) Poison frog colors are honest signals of toxicity, particularly for bird predators. The Am Nat 179:E1–E14
Maan ME, Hofker KD, van Alphen JJM et al (2006) Sensory drive in cichlid speciation. The Am Nat 167:947-954
Mappes J, Marples N and Endler JA (2005) The complex business of survival by aposematism. Trends in Ecol & Evol 20(11):598-603
McGinnis JF, Whelan JP, and Donoso LA 1992. Transient, cyclic changes in mouse visual cell gene products during the light‐dark cycle. J of Neuro Research 31(3):584-590
Myers CW, and Daly JW (1976) Preliminary evaluation of skin toxins and vocalizations in taxonomic and evolutionary studies of poison-dart frogs (Dendrobatidae). Bulletin of the Am Museum of Nat History 157:174-262
Panhuis TM., Butlin R, Zuk M et al (2000) Sexual selection and speciation. Trends in Ecol and Evol16:364-371
Patricelli GL, and Blickley JL (2006) Avian communication in urban noise: Causes and consequences of vocal adjustment. The Auk 123(3):639-649
Pepe IM (1999) Rhodopsin and phototransduction. J of Photochemistry and Photobiology B: Biology 48(1):1-10
Pfaffl MW (2001) A new mathematical model for relative quantitation in real-time RT-PCR. Nucleic Acids Research 29(9)
Price TD (2017) Sensory drive, color, and color vision. The Am Nat 190(2):157-170
R Core Team (2021) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
Reynolds RG and Fitzpatrick BM (2007) Assortative mating in poison-dart frogs based on an ecologically important trait. Evol 61:2253-2259
Ritchie MG (2007) Sexual selection and speciation. Ann Rev of Ecol Evol and Syst 38:79-102
Richards-Zawacki CL, Wang IJ, and Summers K (2012) Mate choice and the genetic basis for colour variation in a polymorphic dart frog: inferences from a wild pedigree Mol Ecol 21:3879-3892
Richards Zawacki CL and Cummings ME (2010) Intraspecific reproductive character displacement in a polymorphic poison dart frog, Dendrobates pumilio. Evol 65:259-267
Robinson PR, Cohen GB, Zhukovsky EA et al (1992) Constitutively active mutants of rhodopsin. Neuron 9(4):719-725
Rudh A, Rogell B, Hastad O et al (2010) Rapid population divergence linken with co-variation between coloration and sexual display in strawberry poison frogs. Evol 65(5):1271-1282
Ryan MJ, and Cummings ME (2013) Perceptual biases and mate choice. Ann Rev of Ecol, Evol, and Syst 44:437-459
Sakakibara S, Hiramatsu H, Takahashi Y, et al. (2002) Opsin expression in adult, developing, and regenerating newt retinas. Mol Brain Res 103(1-2):28-35
Sakmar TP, Franke RR. and Khorana HG (1989) Glutamic acid-113 serves as the retinylidene Schiff base counterion in bovine rhodopsin. Proc of the National Academy of Sci 86(21):8309-8313
Sandkam B, Young CM, and Breden F (2015a). Beauty in the eyes of the beholders: colour vision is tuned to mate preference in the Trinidadian guppy (Poecilia reticulata). Mol Ecol 24:596-609
Sandkam, BA, Young CM, Breden FMW et al (2015b) Color vision varies more among populations than among species of live-bearing fish from South America. BMC Evol Biolc15:225
Sato K, Yamashita T, Ohuchi H, et al. (2011). Vertebrate ancient-long opsin has molecular properties intermediate between those of vertebrate and invertebrate visual pigments. Biochem 50(48):10484-10490
Schott, R. K., Perez, L., Kwiatkowski, M. A., Imhoff, V., & Gumm, J. M. (2022). Evolutionary analyses of visual opsin genes in frogs and toads: Diversity, duplication, and positive selection. Ecol and Evol 12(2):e8595
Seehausen O (2006) African cichlid fish: a model system in adaptive radiation research. Proc of the Royal Society B: Biol Sci 273:1987-1998
Servedio MR, and Boughman JW (2017) The role of sexual selection in local adaptation and speciation. Ann Rev of Ecol, Evol, and Syst 48:84-109
Siddiqi A, Cronin TW, Loew ER et al (2004) Interspecific and intraspecific views of color signals in the strawberry poison frog Dendrobates pumilio. J of Exp Biol 207:2471-2485
Smith AR, D’Annunzio L,.Smith AE et al (2011) Intraspecific cone opsin expression variation in the cichlids of Lake Malawi. Mol Ecol 20:299-310
Su CY, Luo DG, Terakita A, et al. (2006) Parietal-eye phototransduction components and their potential evolutionary implications. Science 311(5767):1617-1621.
Sugawara T, Terai Y, Imai H, Turner GF, et al. (2005) Parallelism of amino acid changes at the RH1 affecting spectral sensitivity among deep-water cichlids from Lakes Tanganyika and Malawi. Proc of the National Academy of Sci102(15):5448-5453
Untergasser A, Cutcutache I, Koressaar T, et al (2012) Primer3-new capabilities and interfaces. Nuc Acids Res 40:1–12
Vidya TNC, and Sukumar R (2005) Social and reproductive behaviour in elephants. Curr Sci 1200-1207
Wan Y, Navarrete MJ, O’Connell LA et al In review. Diurnality shapes the visual opsin genes of colorful Neotropical frogs.
Wang IJ and Summers KJ (2010) Genetic structure is correlated with phenotypic divergence rather than geographic isolation in the highly polymorphic strawberry poison-dart frog. Mol Ecol 19:447-458
Warren RL, Vandervalk BP, Kucuk E, et al. Unpublished .Assembly of the North American Bullfrog Genome Unpublished
Xu L, Hazard III ES, Lockman DK, et al. (1998) Molecular cloning of the salamander red and blue cone visual pigments. Mol Vis 4(10)
Yang Y, Richards-Zawacki CL, Devar A et al (2016) Poison frog color morphs express assortative mate preferences in allopatry but not sympatry. Evol 70:2778-2788
Yang Yang, Blomenkamp YS, Dugas MB, et al (2019a) Mate choice versus mate preference: Inferences about color-assortative mating differ between field and lab assays of poison frog behavior. The American Naturalist 193(4): 589-607
Yang Y, Servedio MR, Richards-Zawacki CL (2019b) Imprinting sets the stage for speciation. Nature 574 (7776): 99-102
Yeager J (2015) Causes and consequences of warning color variation in a polytypic poison frog. Dissertation, Tulane University.
Yewers MS, McLean CA, Moussalli A et al (2015) Spectral sensitivity of cone photoreceptors and opsin expression in two colour-divergent lineages of the lizard Ctenophorus decresii. The J of Exp Biol 218:1556-1563
Yokoyama S (2008) Evolution of dim-light and color vison pigments. Ann Rev of Genomics and Human Genetics 9:259-282
Yokoyama S and Radlwimmer FB (1998) The “five sites” rule and the evolution of red and green color vision in mammals. Mol Biol and Evol 15(5):560-567
Yokoyama S, Yang H, and Starmer WT (2008) Molecular basis of spectral tuning in the red-and green-sensitive (M/LWS) pigments in vertebrates. Genetics 179(4):2037-2043
Yokoyama S. (2002) Molecular evolution of color vision in vertebrates. Gene 300(1-2):69-78
Yokoyama S, Takenaka N, Agnew DW, et al. (2005) Elephants and human color-blind deuteranopes have identical sets of visual pigments. Genetics 170(1):335-344
Yu C, Gao Y, Li P et al (2007) Synchronizing multiphasic circadian rhythms of rhodopsin promoter expression in rod photoreceptor cells. J of Exp Biol 210(4):676-684

No competing interests reported.

OpsinMSsupplement6302023.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Opsin variation in a poison frog that shows color-associated behavioral biases

Status:

Version 1

Abstract

Figures

Introduction

Methods

RNA extraction

Sequencing

Gene Expression Assay

Statistical Analysis

Results

Sequencing

Gene Expression Assay

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1