Genetic effects on life-history traits in the Glanville fritillary butterfly

Introduction

Human land use has led to natural habitat loss and fragmentation such that numerous species and populations are now at risk of extinction (Hanski, 2005; Hughes, Daily & Ehrlich, 1997). The impacts of the remaining patch matrix structure or metapopulations, of the size and the number of remaining suitable habitat patches, of the level of connectivity or the isolation degree between the patches on the persistence and evolutionary dynamics of species and populations have been intensely investigated (Hanski & Mononen, 2011; Hanski & Ovaskainen, 2000). Life-history traits such as body size, reproductive output or dispersal capacity are known to often vary accordingly to the degree of fragmentation experienced by the different populations (Hanski, 2011; Saastamoinen, 2007b; Wheat et al., 2011). Although the type and the strength of natural selection in fragmented habitat might vary locally, given enough time and strong enough selection pressures, persisting populations might adapt to fragmentation and experience genetic differentiation. To date, the evolution of the genetic basis of many life-history traits according to habitat fragmentation remain however poorly understood. This also remains true in the case of the Glanville fritillary butterfly, Melitaea cinxia (L.), despite that the ecology of the species has been intensively investigated over the last 20 years in the Åland Islands, and the species is now considered as a model organism in the dynamics and evolution of natural (meta)population.

Allelic variation in metabolic genes is often associated with life-history traits in natural populations (Marden, 2013). A much-studied example is the gene Phosphoglucose isomerase (Pgi) in butterflies but also in other insects (Dahlhoff & Rank, 2000) and plants (Mitchell-Olds & Pedersen, 1998). In the Orange Sulfur butterfly (Colias eurytheme, Pieridae), Pgi allozyme alleles are associated with survival in the field (Watt, 1977) and reproductive success (Watt, Carter & Blower, 1985), while in the Glanville fritillary butterfly (M. cinxia, Nymphalidae), a SNP in Pgi is associated with flight metabolic rate (Niitepõld, 2010; Niitepõld et al., 2009) and dispersal rate in the field (Hanski & Mononen, 2011; Niitepõld et al., 2009), population growth rate (Hanski & Saccheri, 2006), female fecundity (Saastamoinen, 2007b), body temperature at flight (Saastamoinen & Hanski, 2008), lifespan (Orsini et al., 2009; Saastamoinen, Ikonen & Hanski, 2009) and larval development (Saastamoinen et al., 2013). Considering insect studies on other candidate genes, Saastamoinen et al. (2013) found genetic associations between the incidence of an extra larval instar and allelic variation in Pgi, Serpin-1, and Vitellin-degrading protease precursor genes. Importantly, many of these associations involve a significant interaction between genotype and ambient (Niitepõld, 2010) or acclimation temperature (Wong et al., 2016), consistent with the hypothesis of temperature-dependent enzyme kinetics.

Apart from association studies with candidate genes, genetic effects on life-history traits in the Glanville fritillary are demonstrated by significant heritabilities of many traits (de Jong et al., 2014; Klemme & Hanski, 2009; Kvist et al., 2015; Mattila & Hanski, 2014). For instance, recent studies have shown high heritabilities for both post-diapause larval development time (h² = 0.37) and lifetime offspring production (h² = 0.62) (de Jong et al., 2014; Kvist et al., 2013). These studies do not discriminate between single gene and multigene effects, though de Jong et al. (2014) found that genetic polymorphism in the cytochrome P450 gene CYP337 accounted for 14% of the estimated heritability in egg production.

A generic problem with many candidate gene studies on wild populations is relatively small sample sizes and lack of replication, which cast doubt on the conclusions even in the case of significant results. Exceptions that demonstrate convincing single-locus associations include the study of the Sdhd gene polymorphism (Sdhd is a subunit of a Succinate dehydrogenase, SDH, gene), which is associated with a reduction in SDH enzyme activity in flight muscles and an increase of metabolic rate during flight in the Glanville fritillary (Marden et al., 2013; Wheat et al., 2011). Variation in the same locus is associated with longevity in hyperoxic conditions in Drosophila (Walker et al., 2006).

Here, we aimed to test whether habitat fragmentation might drive genetic differentiation between five natural populations of the Glanville fritillary butterfly, using numerous SNP marker-phenotypic trait associations. We show evidence of associations at various loci, through a candidate gene study involving a total of 49 single nucleotide polymorphisms (SNPs) from 26 candidate genes (metabolic genes). All SNPs were picked mostly based on previous similar, but more limited studies in terms of total number of traits, populations, or markers included. We investigated 21 different life-history traits, selected from traits previously recorded by a study by Duplouy, Ikonen & Hanski (2013) investigating phenotypic differentiation of butterfly populations under landscape fragmentation, and that we found to have high importance in three principal components (PCs) analyses. Importantly, the present study is replicated by including butterfly samples from five different regional populations in the Baltic area, one population in Finland, three in Sweden, and one in Estonia. Additionally, we also provide data from a pilot study, providing a temporal semi-replication to the main study (only on larval and pupal development traits). Several studies have shown that some differences in life-history traits (Duplouy, Ikonen & Hanski, 2013), allelic frequencies (Fountain et al., 2016) and gene expression profiles (Somervuo et al., 2014) occur between the populations inhabiting fragmented versus continuous landscapes. As both northern regional populations inhabit a highly fragmented landscape, including the well-studied metapopulation in the Åland Islands (Hanski, 2011), while the three remaining southern regional populations inhabit a more continuous habitat, we incorporated an environment factor confounding both the degree of habitat fragmentation and the latitude (or climatic region) of the populations, to our analyses to test if and how such factor may affect the Glanville fritillary butterfly at the genetic level. Although other major population characteristics, such as the different degree of relatedness between the populations, might play a role, our results present various potential genetic associations that despite being non-significant after accounting for all tests included in the study (N = 1,086) are consistent across the populations, and the two environments.

Materials and Methods

Study material

We used the material reported in the phenotypic life-history study of Duplouy, Ikonen & Hanski (2013). We included the results from a pilot study that took place in 2007, as they provide independent replicate to the findings from the main 2010 study. Sample sizes for each year in given in Table S1.

Pre-diapause larvae of the Glanville fritillary were sampled from five regional populations in the Baltic Sea region (Fig. 1) in September 2006 and September 2009. Both the Åland Islands (ÅL) metapopulation in Finland (60°07′N, 19°54′E) and the Swedish east-coast Uppland (UP, 59°90′N, 17°50′E) population occur in highly fragmented landscapes, where the meadow habitat occurs in small fragments (average area: 0.19 ha in ÅL, Ojanen et al., 2013), which cover only about 1% of the total landscape area (Hanski, 2011). In contrast, in Öland (ÖL, 56°39′N, 16°38′E) and Gotland (GO, 57°47′N, 18°49′E) in Sweden, and in Saaremaa (SA, 58°05′N, 22°15′E) in Estonia, there are extensive continuous calcareous meadows (alvars), often exceeding several hundred hectares in size. The larval host plants Plantago lanceolata and Veronica spicata are the same in all five populations (Duplouy, Ikonen & Hanski, 2013; Nieminen, Siljander & Hanski, 2004). Females lay large clutches of ca 150 eggs and the larvae live in groups of full sibs (Nieminen, Siljander & Hanski, 2004). Despite generally colder monthly temperature minima recorded across the years for both northern fragmented populations (Fig. S1), the butterfly phenologies are generally considered similar between the five populations of this study (Duplouy, Ikonen & Hanski, 2013). In any case, we confounded both the degree of fragmentation and the latitude of the population into one environment factor: northern fragmented environment versus southern continuous environment. For a more detailed description of the species’ ecology and eco-evolutionary dynamics see Nieminen, Siljander & Hanski (2004) and Hanski (2011), respectively.

For the pilot study ran in the summer 2007, our samples originated from different larval family groups in each population (N_ÅL = 9, N_UP = 13, N_GO = 2, and N_SA = 12), with several individuals from the same family included. In contrast, for the experiment ran in the summer 2010, our samples originated from 50 larval family groups from each population, widely distributed across the respective study areas, and for which the degree of relatedness was previously estimated to be low (estimate of genetic similarity by Duplouy, Ikonen & Hanski (2013)). Most butterflies belong to the eastern mitochondrial clade, originating from the Asian glacial refuge area, with the minor exception of a few butterflies from ÖL belonging to the southeastern clade (Duplouy, Ikonen & Hanski, 2013; Wahlberg & Saccheri, 2007). According to the IUCN Red List, the butterfly is not classified as a threatened species in the five study regions, hence no permits were required for sampling.

For a detailed description of the rearing conditions and measuring of the phenotypic traits, see Duplouy, Ikonen & Hanski (2013). In brief, the caterpillars diapaused in the constant temperature of 3 °C in incubators, the post-diapause larvae were reared indoors under 12:12 L/D light and 25/15 °C day/night temperature and were fed with the leaves of P. lanceolata, cultivated in the laboratory. Adult butterflies were individually marked and released into an outdoor population cage to collect data on behavioral and reproductive traits as described below.

Results

After correction, no significant results were conserved (α = 0.1, all p > 0.12). Due to the large number of tests conducted and corrected for in this study, and to the relatively small sample size, our analyses might however have generated false negative results. Thus, we present below the results from the most significant associations before correction (Table 1; Table S7). In order to achieve 80% power, our sample size for testing a large effect SNP should be of 84 or 59, for dominant or additive model, respectively, when LD is of 0.5. However, as our SNPs are more likely to be small effect SNPs, to achieve 80% power, our sample size should be of 439 and 310 for dominant and additive models, respectively, with LD of 0.5. In all case, the sample size would increase to N = 11,670 for dominance model, and N = 7,845 for additive model, when the LD is small (LD = 0.1).

Associations involving larval and pupal traits

The traits of larval and pupal development were summarized by the first four PCs of PCA₁ (Table S3) (Duplouy, Ikonen & Hanski, 2013). Of the four PCs, PC_1–1 is positively correlated with pupal weight, but negatively correlated with pupal period. The other PC’s show strongest correlations with the weights and durations of specific larval instars. The sex of the larvae had a strong significant effect on PC_1–1 and PC_1–3 (p < 5.13e⁻⁶), with females showing lowest PC_1–1 and PC_1–3 values. Sex had no effect on neither PC_1–2 nor PC_1–4 (Table S7, Duplouy, Ikonen & Hanski, 2013).

One association was found between Pgi: c.331A > C and PC_1–3 (p = 3.84e⁻⁴, dominant effect). Individuals with one or two copies of the C allele had higher values of PC_1–3 than the AA homozygotes (Fig. 2). PC_1–3 is primarily negatively correlated with the weights of the fifth instar post-diapause larvae (Table S3), thus suggesting that AA homozygotes tend to be heavier after diapause than the other genotypes (CC 4.24 mg, CA 4.04 mg, and AA 5.27 mg, Fig. S3A). This is true for all four populations, even if the frequency of the C allele is significantly higher in the southern continuous than in northern fragmented environments (p < 10⁻⁶, Table S8). Similarly, the pupae with one or two copies of the C allele were heavier than the AA homozygotes in the 2007 pilot material (CC 192 mg, CA 193 mg, and AA 176 mg, p = 0.0032, dominant effect, Fig. S3B).

Both SNPs c480_est:926G>A and c480_est:1051G>A (125 bp’s apart) in the Endocuticle structural glycoprotein gene (SgAbd-8) showed significant interactions with environment type in affecting PC_1–2 (p = 0.0023 and 0.002, respectively, dominant effects, Fig. 3). There were no significant differences between the genotypes in the southern continuous environments (p = 0.28), but the individuals from the northern fragmented environments carrying at least one G allele had lower values of PC_1–2 than the AA homozygotes (p = 0.019, Fig. 3). PC_1–2 is mostly negatively correlated with the weight of the seventh larval instar (Table S3).

Finally, there was a significant effect of two SNPs in the Hemolymph proteinase-5 gene (89 bp’s apart) on PC_1–2 (p = 1.15e⁻³, and p = 1.21e⁻³, over-dominant effects, Fig. S4) and a SNP in the Heat shock protein gene on PC_1–1 (p = 2.19e⁻⁴, over-dominant effect, Fig. S5). In each case, the homozygotes show higher PCs values than heterozygote individuals.

Associations involving adult traits

The male adult-traits were summarized by the first three PCs of PCA_M (Table S4) to reflect on male quality. PC_M1 is negatively correlated with both survival and average distance flown per day, indicating that males of short survival do fly less. PC_M2 is negatively correlated with pupal weight but positively correlated with age at first mating, indicating that smaller males mate at older age. Finally, PC_M3 positively correlated flight activity (probability to fly) with mating success (total number of copulations and age at first mating), thus male flying more mate more, despite starting to mate at older age. There were no direct effects of environment on any of the PCs.

We found an association between PC_M1 and the SNP c3917_est:386A>C in the Serine proteinase-like protein gene (p = 1.64e⁻⁴, recessive effect, Fig. S6). Butterflies with at least one copy of the A allele showed lower PC_M1 values (AA = −3.07, AC = 0.13, and CC = 15.09).

We found an association between SNPs in the SgAbd-8 gene and PC_M2 (SNP c480_est:1003G>C, p = 2.15e⁻⁴, dominant effect of G allele, Fig. 4A), and PC_M3 (SNP c480_est:926G>A and SNP c480_est:1051G>A, 125 bp’s apart, p = 8.56e⁻⁴, dominant effects of G allele). Additionally, the direct effects of the two SNPs in the SgAbd-8 gene were coupled with the effect of environment by genotype interactions (p = 1.87e⁻⁴, over-dominant effects, Fig. 4B). There were no significant differences between the genotypes in the northern fragmented environments, but the heterozygote individuals from the southern continuous environments had higher PC_M3 values than the homozygote individuals (Fig. 4B).

There was also a significant association between PC_M3 and the SNP hsp_1:206T>G in the Heat shock 70 kDa protein coding gene (p = 1.33e⁻⁴, recessive effect, Fig. S7A). In general, individuals that carry at least one T allele show lower PC_M3 values (GG = 0.45, GT = −0.14, and TT = −0.07). Finally, there were significant associations between PC_M3 and the SNP c50_est:824A>G in the Hemolymph proteinase-5 gene. Homozygote CC males show highest PC_M3 values (CC = 0.18, AC = −0.02, and AA = −0.16, p = 2.35e⁻⁴, additive effect, Fig. S7B).

The female adult and oviposition related-traits were summarized by the first three PCs (PC_F1 to PC_F3) of a female PCA_F (Table S5). PC_F1 is negatively correlated with female offspring production (both total eggs and total caterpillars production), suggesting that females laying small clutches produce less caterpillars. PC_F2 shows positive correlation with first clutch size and hatch rate (ArcSin corrected), suggesting that larger clutches show higher hatch rate success. Finally, PC_F3 is strongly positively correlated to female age at first oviposition, and to a lower degree to the size of the first clutch laid, suggesting that the first clutch of older females are larger. There were no effects of the environment nor the genotype on neither of the three PCs from PCA_F.

Discussion

We highlight several associations between SNPs in candidate genes and some of the PCs from PCAs on various life-history traits in the Glanville fritillary butterfly (Table 1), however these associations did not hold up following correction for multiple testing. One reason for the lack of significant results may remain the small sample size for each population, and the large amount of tests done, leading to false negative results. We therefore believe many of the positive results before correction, are robust and most likely driven by the fact that the selection of candidate genes was largely based on previous studies (discussed below), in which these genes had already shown significant or nearly significant effects. Furthermore, we used phenotypic traits, some of which have been highly heritable in previous studies (de Jong et al., 2014; Kvist et al., 2013) or have shown phenotypic variation among environments suggestive of local adaption (Duplouy, Ikonen & Hanski, 2013). Although functional validations are needed to confirm our findings, the present results provide a starting point for genome-wide association studies on the Glanville fritillary. Such studies have remained impractical so far, but they are becoming feasible in the future, with the help of the now available genome (Ahola et al., 2014), linkage map (Rastas et al., 2013) and other tools for the Glanville fritillary, as well as the decreasing cost of whole-genome sequencing.

The material for the present study originates from five distinct populations, over two different collection years. Consistent results across this material can be considered to be more robust than would be comparable results for a single population in the Glanville fritillary butterfly. However, because the phylogenetic relatedness varies between the five populations (see Somervuo et al., 2014, and Table 2 for estimates of population divergence; and Wahlberg & Saccheri, 2007, for description of the different mitochondrial clades), it is not possible to conclude with certitude that the genes with known or presumed function in life-history ecology show significant differences in allele frequencies due to variable natural selection in the two environments (northern fragmented versus southern continuous). Other non-investigated population characteristics, such as the historical relatedness or geographical closeness of the populations from similar environments, might be of similar or higher importance in this context. Although the pairwise genetic distances between populations, calculated based on the genotypes at 33 SNPs of 176 butterflies are not significantly grouping the populations in the two environments, future association studies in the Glanville fritillary butterfly should further address this issue by including additional neutral markers. In any case, the fact that out of the 49 SNPs from 26 genes investigated, 15 SNPs from eight genes seem to show differences in allele frequencies between the two environmental extremes suggests their potential role in local habitat adaptation.

In each case, there is support from previous studies on the Glanville fritillary, or from other insect studies, and to some extend from the present study, that these genes may influence life-history traits or other phenotypic traits. For example, while working on four of the five populations included in the present study (Åland, Uppland, Öland, and Saaremaa), Somervuo et al. (2014) identified almost 2,000 genes with dissimilar expression levels between the northern fragmented and southern continuous environments. This list of genes in Somervuo et al. (2014) includes the eight genes for which we found diverging allele frequencies (at one or several SNPs) between environmental types. The key published results supporting our findings regarding the association of these eight genes to variation in life-history traits or other important phenotypic traits include the following:

1. The glycolytic Phosphoglucose isomerase (Pgi) gene, for which genetic polymorphism has previously been linked to variation in a large number of traits (Dahlhoff & Rank, 2000; Hanski & Saccheri, 2006; Niitepõld et al., 2009; Orsini et al., 2009; Saastamoinen, Ikonen & Hanski, 2009) as further discussed below, and for which our results support for associations with larval development.

2. The Cytochrom p450 (CYP337) gene, which has been previously associated with life-time egg and larval production, and host plant preference in the Glanville fritillary (de Jong et al., 2014).

3. The Vitellin-degrading protease precursor gene, for which polymorphism has been associated with variation in larval development (Ahola et al., 2015, this study) and the incidence of an extra larval instar in the Glanville fritillary (Saastamoinen et al., 2013).

4. The Heat shock 70 kDa protein (Hsp70) gene, for which a previous study shown that gene expression is up-regulated during thermal stress (Luo et al., 2014; Mattila, 2015), as HSPs may provide cells with protection against damage from extreme temperature conditions (Sørensen, Kristensen & Loeschcke, 2003). Furthermore, in the Glanville fritillary, polymorphism in HSP genes have been previously associated with variation in fecundity traits, including egg hatch rate, age at first oviposition and first clucth size and in mobility (de Jong et al., 2014) and in male flight metabolic rate (Mattila, 2015). In the present study, polymorphism in the SNPs hsp_1:106G>A and hsp_1:206T>G are associated to variation in PC_1–1 and PC_M3 in males, respectively, thus suggesting association with larval development, or male flight activity and mating success.

5. The Hemolymph proteinase-5 and the Serine proteinase-like genes are two genes linked to immune and stress responses in Lepidoptera (An et al., 2009; Han et al., 2013) and other insects (Ashida & Brey, 1995), and to larval development traits in the Glanville fritillary (Ahola et al., 2015, this study). Again, the correlations we observe in larvae and in males between PC_1–2 (larval development) or PC_M3 (male mating success and flight activity) and polymorphism in the Hemolymph protease-5 gene might suggest variation in the quality, through for example variation in immune functions, between individuals in the Glanville fritillary.

6. The Endocuticle structural glycoprotein (SgAbd-8) gene, described as a structural gene associated with larval development in insects (e.g., in bumblebees, Colgan et al., 2011), showed genetic associations with male PCs, as well as a gene by environment type interaction on larval development in the present study.

7. The glucose-6-phosphatase dehydrogenase (G6PD) gene, in which polymorphism has been associated with mating success in Colias butterflies (Carter, 1988).

The results of the genetic association for the SNP (Pgi):c.331A>C are a case in point. We found significant associations with larval development. The AA homozygotes have heavier larvae following diapause, but smaller pupae than the other genotypes. These results were consistent across the four populations, even though there are great differences in the frequency of the C allele in the populations: the C allele is much more common in the two northern populations from continuous environment (around 50%) than in the southern populations from fragmented environment (around 20%).

Several previous studies conducted in the Åland Islands have demonstrated that the (Pgi):c.331A>C AC heterozygotes have superior flight metabolic rate and dispersal rate in the field compared with the AA homozygotes (Niitepõld et al., 2011, 2009). Both empirical and modeling studies have suggested that natural selection in highly fragmented landscapes favors heterozygosity because of associated superior dispersal and colonization abilities (Fountain et al., 2016; Hanski & Mononen, 2011), thus allowing the individuals to differently cope with spatial heterogeneity, and the species to best occupy the available habitat. Similarly, other phenotypic traits (e.g., larval and fecundity) may be affected by the prevailing environmental conditions (Saastamoinen, 2007b; Saastamoinen et al., 2013). Working on larval development, Saastamoinen et al. (2013) found that individuals with the C allele develop more often via an extra larval instar before pupation. As the extra instar is common in larvae that are small after diapause, this finding is consistent with our results showing that the AA larvae are heavy after diapause, as well as with results from Kallioniemi & Hanski (2011) showing that AA homozygotes produce heavier pupae than the other genotypes. Furthermore, the homozygote AA individuals have higher survival in low (stressful) temperature in the laboratory (Kallioniemi & Hanski, 2011), while the small CC individuals generally showed reduced survival (Orsini et al., 2009). These results are consistent with studies on other butterfly species. In the copper butterflies, Lycaena tityrus, Karl, Schmitt & Fischer (2008) found that Pgi genotypes are associated to several traits of larval and pupal development and cold stress resistance, while in C. eurytheme Pgi heterozygote individuals have higher survival than the homozygotes in natural populations (Watt, 1977).

Although, the current study did not highlight any genetic and environmental association with female adult traits (including survival, fecundity, and flight activity traits), previous studies have shown that Pgi-AC heterozygous females (SNP (Pgi):c.331A>C) lay larger clutches than AA homozygous females especially at low ambient temperatures (Saastamoinen, 2007a; Saastamoinen & Hanski, 2008). In Colias, Pgi heterozygous individuals in allozyme studies had similarly high male mating success and high female fecundity (Carter, 1988; Watt, 1992; Watt et al., 2003). The CC homozygous individuals are uncommon in the Åland Islands and hence previously little studied. We suggest that this lack of association with female traits may be due to the large number of traits included in the PCA, rather than the true lack of association with genetic polymorphism in the two environment types.

In summary, several studies on the Glanville fritillary examining associations between the SNP (Pgi):c.331A>C and life-history traits have found consistent results for several traits, including larval development. The individuals with the AA genotype do not appear to be as good dispersers as the heterozygotes (Niitepõld et al., 2009), but might re-allocate their resources towards other fitness traits, such as their own body size, making them competitive individuals within their habitat patches (e.g., in many butterflies, females select for larger males) and the environmental conditions (e.g., fluctuating temperatures). There may also be trade-offs between dispersal behavior and immune responses, as suggested by the significant associations observed between another gene, the Heat shock 70 kDa protein gene, and adult male activity (PC_M3). Active flight causes oxidative stress, which may become an important factor for dispersive butterflies. Somervuo et al. (2014) showed that butterflies from fragmented landscapes are more tolerant of hypoxia, which is suggested to be another likely genomic adaptation to dispersal.

Conclusion

The current literature shows multiple evidence pointing to the possibility that population-level variations in the environmental conditions, in the Baltic Sea region, may influence life-history evolution and thereby allele frequencies and gene expression in the Glanville fritillary butterfly. In the context of habitat fragmentation, traits such as dispersal and re-colonization, which are key processes facilitating the persistence of metapopulations in fragmented landscapes, are among the traits that have previously been suggested to be affected (Hanski, 2011). Equally, traits influencing performance in large, stable populations (in continuous landscapes) may be important. Studies on allelic variation in Pgi and sdhd genes have similarly demonstrated significant differences between new versus old local populations in the highly fragmented landscape in the Åland metapopulation (Hanski & Saccheri, 2006; Hanski, 2011; Wheat et al., 2011; Zheng, Ovaskainen & Hanski, 2009), supporting the notion that population turnover, extinctions and re-colonizations, impose selection in fragmented landscapes. Similarly, other gene expression studies demonstrated significant allelic differences in various genes between individuals experiencing different ambient temperature conditions (Kvist et al., 2013; Saastamoinen & Hanski, 2008; Saastamoinen et al., 2013). The differences between newly established versus old local populations in fragmented landscapes, and between individuals evolving in cold versus warmer environments reflect some of the selection processes that might operate at the population level. Although we could predict that the difference between environments should reveal the outcome of such processes during longer periods of time, further functional analyses are needed to fully tease apart the relative importance of the confounding and stochastic evolutionary processes related to the degree of habitat fragmentation and the geographical positions of the five populations used in our study.

Supplemental Information