Discussion
Rapid adaptation is a common feature of biological invasions (e.g., Colautti & Barrett, 2013; Stern & Lee, 2020; Battlay et al., 2023). Counter to results for many invasions, previous studies of the brown anole in the southeastern United States have revealed a lack of evidence for adaptation to the novel environmental conditions experienced there (Kolbe et al., 2007a, Bock et al., 2021, 2023, Baeckens et al., 2023). This is surprising given that the colonization of the non-native range in this species likely coincided with substantial changes in ecological interactions, as well as a shift from a tropical to a temperate climate. Consistent with previous studies of phenotypic evolution in the brown anole invasion, our results show that variation in dewlap characteristics among non-native populations is primarily explained by genetic ancestry. Admixture of multiple native-range lineages strongly influences the current genetic composition of non-native populations (Kolbe et al., 2004, 2007b, 2008, Bock et al., 2021). GWAS results corroborated these findings and indicated that most dewlap traits appear to have a complex genetic architecture. While we find that multiple aspects of the dewlap are related to local environmental conditions, genomic analyses did not indicate that dewlap-associated SNPs retain a signature consistent with local adaptation. Whether natural selection plays a role in signal divergence during the brown anole invasion thus requires further investigation. This study provides insight into the ongoing evolutionary processes occurring in biological invasions, highlighting the importance of genetic ancestry in brown anole dewlap variation among populations. Below, we put our findings in the context of previous studies of the brown anoles to better understand how populations have evolved during this invasion.
Genetic ancestry and its role in shaping dewlap phenotypes
Genetic clustering analyses revealed that genetic ancestry of invasive brown anoles can be summarized at a coarse level as a combination of two genetically distinct native lineages: Western Cuba and Central-eastern Cuba. Genetic ancestry had strong effects on several aspects of the dewlap; lizards with increasing Western Cuba ancestry have larger, brighter red dewlaps with lower UV reflectance and less yellow (Table 1, Figure 1b, S8a-k). As well, the probability of lizards having a solid dewlap increased with the frequency of Western Cuba ancestry (Figure S9c). This result is consistent with findings from the native range which, while based on more limited sampling, also find that Western Cuba populations of A. sagrei have a greater proportion of solid dewlaps, whereas Central-eastern Cuba populations have a greater proportion of spotted dewlaps (Driessens et al., 2017). These results point to an important role of genetic ancestry in shaping multiple dimensions of dewlap variation across invasive populations in this system.
Our genome-wide association analyses further explored the effect of genetic ancestry on dewlap variation. We identified several SNPs significantly associated with red coloration and brightness of dewlaps, and dozens of SNPs with suggestive associations with all dewlap traits (Figure 3, Table S2). Our finding that genetic ancestry has strong and widespread effects on dewlap differentiation among non-native populations is consistent with previous studies of invasive species (e.g., Silene vulgaris , Keller & Taylor, 2010; sticklebacks, Lucek et al., 2010).
We previously used the same samples to identify a large-effect locus significantly associated with limb length in A. sagrei (Bock et al., 2021). Our results here indicate that the genetic architecture of the dewlap is likely different and more complex than that characterizing limb length, at least for A. sagrei of Cuban ancestry. This interpretation is further supported by among-trait correlations. While all limb length traits show strong and significant pairwise correlations (Bock et al., 2021), dewlap traits are characterized by more moderate among-trait correlations that are occasionally non-significant (Figure S11).
The associations identified here between SNPs and dewlap traits should be interpreted with caution for two reasons. First, aside from associations reported for red composition and total and mean dewlap brightness, the strength of the association signal is reduced for any one SNP. Second, in-depth analyses of standard and ancestry-specific associations (Figures 3 and 4) revealed that genotype classes with the largest difference in trait values are also the ones with the lowest sample size (e.g., genotype GG for Western Cuba ancestry, at Chr2: 166.587 Mb; Figure 4). Therefore, some of these results could be due to uneven sample sizes among genotype classes. With these caveats in mind, we note that the direction of the ancestry-specific effects reported here is consistent with our observations of the effect of Western Cuba ancestry on dewlap trait variation. Specifically, linear mixed effects models of dewlap traits based on genetic ancestry as well as the ancestry-specific GWAS indicated that Western Cuba ancestry is associated with increased brightness of dewlaps.
In many organisms, carotenoid pigments have been shown to be a source of variation in red, orange, and yellow ornamental coloration within and between species (reviewed in Toews et al., 2017). In most animal species, carotenoids cannot be synthesized and must be ingested. The extent to which dewlap color variation is influenced by nutritional condition has so far been investigated in a few anole species, A. distichus (Ng et al., 2013b) and A. sagrei (Steffen et al., 2010). Both studies found no difference in color between carotenoid and control treatments and dewlap coloration was heritable in A. distichus lizards. Moreover, biochemical investigations of dewlap pigment composition in several Anolis species have found that pterin pigments are important sources of coloration, although carotenoids were detected as well (Ortiz, 1962; Macedonia et al., 2000, Steffen & McGraw, 2007, 2009). In contrast to carotenoids, pterins are produced endogenously (reviewed in Andrade & Carneiro, 2021). Even so, identifying genes involved in pterin synthesis has been challenging because of the complexity of underlying biochemical pathways. For example, studies contrasting skin patches of different color in other reptiles have identified tens to hundreds of genes that are differentially expressed, and that are likely involved in the production of these pigments (e.g., McLean et al., 2017). Our finding that dewlap phenotypes are correlated with genome-wide estimates of genetic ancestry across non-native populations is in line with these previous studies and suggests a complex genetic architecture for dewlap traits.
Potential adaptation to light environments
We tested the hypothesis that dewlap phenotypes are correlated with environmental variation, which is expected under local adaptation. Supporting this prediction, we found significant relationships between multiple environmental variables and dewlap characteristics. Our strongest results show that dewlaps tend to be darker with relatively high UV reflectance as canopy openness (i.e., habitat light) increases. Also, we found that brown anoles occurring in habitats with greater precipitation had more UV reflectance. These findings are consistent with those obtained by previous studies in the native range of A. sagrei . Driessens et al. (2017) found that lizards occupying open forest habitats had higher UV reflectance than populations inhabiting environments with little light exposure.
Studies in other Anolis species show different correlations between UV reflectance and the environment, suggesting the relationship between aspects of the dewlap and the local environmental conditions might be species specific. For example, Ng et al. (2013a) found no difference in UV reflectance among A. distichus populations inhabiting environments that differed in light characteristics. Also, Leal and Fleishman (2002, 2004) reported that A. cristatelluspopulations in Puerto Rico occupying closed forest habitats with little light penetration exhibit highly reflective and transmissive dewlaps with more UV reflectance. As light exposure increased (i.e., greater canopy openness), dark dewlaps with low UV reflectance were favored (Leal & Fleishman, 2004).
Several studies on local adaptation of Anolis dewlaps have contrasted populations in xeric and mesic habitats. These habitats differ in characteristics such as habitat light, temperature, or precipitation. We sampled brown anole populations across a latitudinal gradient throughout Florida and into southern Georgia. The southeast region of the United States is known to have relatively uniform climatic conditions within seasons. As expected, we found less variation in annual mean temperature and precipitation patterns among our sampled populations (Figure S3b,c) as compared to previous studies of tropical mesic and xeric habitats in Puerto Rico and the Dominican Republic (Leal & Fleishman, 2004, Ng et al., 2013a). Despite this reduced variation in temperature and precipitation, we found that lizards inhabiting cooler environments tend to have dewlaps with greater orange composition, which is consistent with previous studies of A. distichus (Ng et al., 2013a). While this may indicate convergent evolution of dewlap orange composition in relation to temperature in A. distichus andA. sagrei , the underlying mechanism is unknown.
In contrast to analyses at the trait level discussed above, analyses at the SNP level did not provide evidence of adaptation. Specifically, SNPs associated with dewlap traits had similar F STvalues as random genome-wide SNPs (Figure S12a). As well, dewlap-associated SNPs were rarely classified asF ST outliers (Figure S12b). Moreover, the small number of dewlap-associated SNPs that do have extremeF ST values are highly genetically differentiated in one or two population pairs at most (Figure S12b). These results therefore indicate that, if some of the dewlap-associated SNPs reported here are involved in local adaptation, their contribution is likely specific to a small number of populations, with different SNPs being recruited by natural selection in different populations. The genotype-environment association analyses further supported these results, indicating that dewlap-associated SNPs are not also associated with the environmental variables predicted to shape dewlap traits under local adaptation (Figure S13). Given evidence presented here that points to a complex genetic architecture for dewlaps and considering that invasive brown anole populations are highly genetically diverse, it is possible that adaptation occurs via small changes in allele frequency at a large number of loci. Identifying the genetic signature of natural selection at loci involved in the control of dewlap is needed to confirm the occurrence of adaptation, as inferred from trait-environment correlations. Achieving this task will be challenging, and will require much denser genome-wide data, in line with observations for other highly polygenic traits (Lowry et al., 2017).
In conclusion, the brown anole invasion allowed us to study the evolution of a complex signaling phenotype during a biological invasion, revealing how genetic ancestry strongly influences among-population variation in dewlaps in the non-native range. Our study supports the importance of invasion history and admixture in determining patterns of phenotypic divergence during biological invasions (Kolbe et al., 2004; Keller & Taylor, 2010). Although we found some evidence that aspects of the dewlap are correlated with environmental variation among non-native populations, which is consistent with previous studies and suggestive of local adaptation, the loci underlying these dewlap characteristics did not show a genetic signature consistent with the action of natural selection. Future studies should consider denser sampling of SNPs along the genome, which may allow the signature of natural selection to be recovered for these polygenic traits. Additionally, to better understand dewlap evolution in non-native brown anole populations, future studies should consider other potential sources of selection, including species recognition (Losos, 1985; Vanhooydonck et al., 2009; Baeckens et al., 2018a), sexual selection (Vanhooydonck et al., 2009), and intrasexual selection (i.e., male-male competition) (Vanhooydonck et al., 2005; Lailvaux & Irschick 2007; Baeckens et al., 2018b).