and intraspecific gene exchange within a ... - Wiley Online Library

Molecular Ecology (2015) 24, 3107–3121

doi: 10.1111/mec.13233

Persistent inter- and intraspecific gene exchange within a parallel radiation of caterpillar hunter beetles (Calosoma sp.) from the Gal apagos FREDERIK HENDRICKX,*†1 THIERRY BACKELJAU,*‡ WOUTER DEKONINCK,* STEVEN M. VAN B E L L E G H E M , * † V I K I V A N D O M M E † and C A R L V A N G E S T E L * † 1 *Royal Belgian Institute of Natural Sciences, Vautierstraat 29, Brussels 1000, Belgium, †Terrestrial Ecology Unit (TEREC), Biology Department, Ghent University, K.L. Ledeganckstraat 35, Gent 9000, Belgium, ‡Evolutionary Ecology Group, Department of Biology, University of Antwerp, Groenenborgerlaan 171, Antwerp 2020, Belgium

Abstract When environmental gradients are repeated on different islands within an archipelago, similar selection pressures may act within each island, resulting in the repeated occurrence of ecologically similar species on each island. The evolution of ecotypes within such radiations may either result from dispersal, that is each ecotype evolved once and dispersed to different islands where it colonized its habitat, or through repeated and parallel speciation within each island. However, it remains poorly understood how gene flow during the divergence process may shape such patterns. In the Galapagos islands, three phenotypically similar species of the beetle genus Calosoma occur at higher elevations of different islands, while lowlands are occupied by a fourth species. By genotyping all major populations within this radiation for two nuclear and three mitochondrial gene fragments and seven microsatellite markers, we found strong support that the oldest divergence separates the highland species of the oldest island from the remaining species. Despite their morphological distinctness, highland species of the remaining islands were genetically closely related to the lowland population on each island and within the same magnitude as lowland populations sampled at different islands. Repeated evolution of highland ecotypes out of the lowland species appears the most likely scenario and estimates of geneflow rates revealed extensive admixture among ecotypes within islands, as well as between islands. These findings indicate that gene exchange among the different populations and species may have shaped the phylogenetic relationships and the repeated evolution of these ecotypes. Keywords: adaptive radiation, admixture, ecological speciation, introgression, islands, phylogeography Received 6 February 2015; revision received 4 May 2015; accepted 6 May 2015

Introduction A frequent observation of species radiations on island archipelagos is that phenotypically similar species live in comparable habitats on different islands (Gillespie & Roderick 2002; Losos & Ricklefs 2009). The genesis of such replicated species assemblages within an archipelCorrespondence: Frederik Hendrickx, Fax: +32 2 627 41 32; E-mail: [email protected] 1 Contributed equally to this work. © 2015 John Wiley & Sons Ltd

ago is expected to depend on the interplay between the rate of among-island colonization vs. within-island diversification (Fukami et al. 2007; Urban et al. 2008; Gillespie & Baldwin 2010; Warren et al. 2015) for which two, nonmutually exclusive, scenarios can be proposed. First, the occupation of a particular habitat or niche evolved only once in a species or population, and subsequently, this species or population colonizes comparable habitats on the different islands, that is a ‘species sorting’ mechanism (Leibold et al. 2004). Second, the occupancy of a particular habitat evolves repeatedly,

3108 F . H E N D R I C K X E T A L . that is parallel evolution (Schluter 2000; Gillespie 2004; Ryan et al. 2007; Gavrilets & Losos 2009; De Busschere et al. 2010; Mahler et al. 2013; Pascoal et al. 2014). Both scenarios can generally be distinguished by comparing the degree of genetic relatedness among species, with within-island radiations, resulting in a close genetic relatedness of species living on the same island (Losos & Ricklefs 2009). Irrespective of which colonization scenario is involved, it remains poorly understood to what extent genetic structuring of species and ecologically divergent populations within-island radiations is shaped by historic and ongoing levels of (interspecific) gene flow (Wang et al. 2013). It becomes increasingly clear that species boundaries can be permeable to interspecific gene exchange (Barton 2001; Seehausen 2004; Mallet 2005; Niemiller et al. 2008; Abbott et al. 2013; Martin et al. 2013). This is particularly expected within island archipelagos as (i) ecologically divergent species living on the same island have geographically restricted and potentially overlapping distributions and (ii) climatic and geological dynamics may translate into repeated episodes of fission and fusion of diverging populations (Ali & Aitchison 2014; Emerson & Faria 2014; Garrick et al. 2014). Indeed, cases of hybridization and introgression are frequently reported between species that radiated within island archipelagos (Shaw 2002; Grant et al. 2005; Jordal et al. 2006; Grant & Grant 2010b; Farrington et al. 2014; De Busschere et al. 2015). Alternatively, ecologically similar species residing on different islands may exhibit high neutral differentiation because of their strong geographic isolation (Wang et al. 2013). To explore patterns of admixture, multiple loci with variable evolutionary rates should be screened. Incongruent phylogenetic patterns, mostly among nuclear and mitochondrial genes, may then provide insights

into ancient colonization, radiation and hybridization patterns (Funk & Omland 2003; Chan & Levin 2005; Hein et al. 2010). Such in-depth knowledge on dispersal patterns and introgression rates is pivotal to assess the relative importance of local in situ diversification vs. interisland colonization in shaping replicated species assemblages in island archipelagos. In this study, we unravel the evolutionary history and patterns of gene flow within a radiation of caterpillar hunters of the genus Calosoma (Coleoptera: Carabidae) in the Gal apagos islands where the genus comprises four morphologically well-defined species (Fig. 1) (Desender et al. 1992; Peck 2006). Calosoma granatense is a large and generally long-winged and highly dispersive species that occurs chiefly in shrubby vegetation, often with sparse trees in the arid and transition zones on all major islands in the archipelago (Desender & De Dijn 1989; Peck 2006). The other three species have reduced wings and are restricted to the open grassy pampa vegetation zone at higher altitudes (>500 m) of the older and eastern islands San Cristobal (Calosoma linelli), Santa Cruz (Calosoma leleuporum) and Santiago (Calosoma galapageium) (Fig. 1) (Desender & De Dijn 1989). On these islands, small numbers of the lowland species are also found at higher elevations, resulting in a partial overlap in the distribution of highland and lowland species. Highland species share phenotypic similarities, particularly with respect to traits related to the loss of dispersal by flight (e.g. wing and metepisternal traits) as inferred from a morphometric comparison based on 26 external traits (Fig. 2A) (Desender & De Dijn 1990). The lowland species C. granatense is also abundant in pampa vegetation at higher elevations on the largest, western and younger islands Isabela and Fernandina, and just as the highland species, these populations show a gradual reduction in

Pinta (PIN) (0.7–?)

C. galapageium

Marchena (MAR) (0.8–?) Genovesa (GEN) (0.3–?) Volcan Wolf (IVW)

C. leleuporum

Santiago (SAN) (0.8–1.4)

Volcan Darwin

Fernandina (0.035–0.07)

Volcan Alcedo (IVA)

Sierra Cerro Negra (ISN) Azul (ICA)

C. granatense N

Santa Cruz (SCZ) (1.1–2.3)

San Cristobal (SCB) (2.4–4.0)

Isabela (0.5–0.8)

100 km

Santa Fe (SFE)

Floreana (FLO) (1.5–2.3)

C. linelli

Fig. 1 Distribution of Calosoma species from the Gal apagos (light yellow: Calosoma granatense—long-winged lowland populations; blue: Calosoma linelli; red: Calosoma leleuporum; green: Calosoma galapageium; dark yellow: C. granatense— short-winged highland populations) based on Desender et al. (1991) and Peck (2006). Codes between brackets are the abbreviations of the island names. Estimated minimum geological age and maximum geological age in MY for each island are depicted below the island names between brackets (Geist et al. 2013).

Española (ESP) (3.0–3.5)

© 2015 John Wiley & Sons Ltd

P A G O S B E E T L E R A D I A T I O N 3109 ADMIXTURE PATTERNS WITHIN A GALA wing size with the centre of the cline at approximately 500 m altitude (Fig. 2B) (Desender et al. 1991). All species are predators that feed exclusively on caterpillars. They are nocturnal and hide during daytime under stones and in crevices in the ground. While discrete wing size dimorphism is a common phenomenon in insects to counteract local extinctions (Roff 1994) or stochastic fluctuations in population size in heterogeneous landscapes (Hendrickx et al. 2013), only few species, such as the Calosoma species from Galapagos, show a gradual variation in wing size with strong differences in mean wing size among populations (Desender 1989; Zera & Denno 1997). As shown for the wing polymorphic populations of the carabid species Pogonus chalceus from Atlantic Europe, interdemic variation in this polygenic trait probably involves adaptation to local differences in habitat stability (Dhuyvetter et al. 2007; Van Belleghem & Hendrickx 2014). Because the morphological differences between C. granatense and the highland species increase with island age (Fig. 2) and because repeated loss of flight ability has previously been documented in island radiations (Olivieri 2009), a separate and repeated adaptive divergence on each island is considered the most likely evolutionary scenario (Desender et al. 1992). Based on sequence variation in the mitochondrial genes cox1, nadh1 and cytb and the nuclear genes cad and enolase, and allele frequencies at seven microsatellite loci, we here attempt to (i) reconstruct the evolutionary history of the repeated habitat occupancy of highland habitats within this Galapagos radiation and (ii) infer to what extent gene exchange among the different populations and species may have shaped the phylogenetic relationship and the repeated evolution of these ecotypes.

Materials and methods Taxon and data sampling Specimens were sampled by hand in the period 1996– 2010 from almost all known populations of the four species (see Tables S1 and S2, Supporting information, for detailed information of individuals used for genetic analysis). For Calosoma galapageium, only four individuals were available for DNA analysis and insufficient genetic information could be obtained to reliably reconstruct its evolutionary history. Although the recovered haplotypes from this species are included in the phylogenetic trees, the species was excluded for estimating introgression rates with the other species. Phylogenetic relationships and rates of introgression were estimated based on partial sequences of two nuclear (cad and enolase) and three mitochondrial (cox1, © 2015 John Wiley & Sons Ltd

cytb and nadh1) gene fragments, yielding a total sequencing length of 1729 and 1419 bp for the nuclear and mitochondrial sequences, respectively (see Appendix S1, Supporting information, for detailed information for primers and conditions for PCR amplification). It remains at present not known which continental Calosoma species is most closely related to the Gal apagos clade. The phylogenetic relationship among taxa within the genus Calosoma, which has a worldwide distribution and occurs on all continents, is currently unknown, and the classification into subgenera is based on morphological grounds only (Jeannel 1940). To root the trees and estimate relative substitution rates among the different genes (see Sequence data analysis), we included five New World Calosoma species. Calosoma sayi, Calosoma marginale, Calosoma macrum and Calosoma wilcoxi were provided by Martin Husemann and sampled at Baylor Camous, Waco, McLennan County, Texas, USA (31.547946°N, 97.103654°E). From those species, C. sayi is situated within the same subgenus (Castrida) as the species from Gal apagos (Jeannel 1940). We also included published sequences of cad (EU860105) and enolase (EU677562) of the species Calosoma scrutator collected from Pena Blanca, Arizona, USA (Wild & Maddison 2008). The specimen was provided by D. Maddison and further sequenced for the three mitochondrial genes. The total number of individuals that yielded unambiguous sequence reads was 56 (enolase), 103 (cad) and 118 (mitochondrial genes) (see Tables S1 and S2, Supporting information). We further genotyped 134 individuals at seven microsatellite loci that cross-amplify in the different species (see Tables S1 and S2, Supporting information). Primer pairs to amplify the microsatellites are reported by Dhuyvetter et al. (2002) (CALO1, CALO3 and CALO4). We further added four extra loci (CALO16, CALO21, CALO23 and CALO24) that were developed following the same protocol outlined by Dhuyvetter et al. (2002) (see Appendix S1, Supporting information).

Sequence data analysis Haplotypes were reconstructed by means of the PHASE 2.1 algorithm (Stephens et al. 2001). The sequences of the three mtDNA genes were concatenated into composite haplotypes, because their patterns of sequence divergence were congruent [coefficient of concordance (Campbell et al. 2011): W = 0.88; P < 0.001]. For phylogenetic inference, the best fitting nucleotide substitution model was selected based on the Bayesian information criterion as implemented in JMODELTEST 0.1.1 (Posada 2008). No evidence for recombination was found in the mitochondrial and two nuclear sequences based on GARD analysis (Pond et al. 2006).

3110 F . H E N D R I C K X E T A L . (A) San Cristobal, Santa Cruz and Santiago 8

C. linelli C. leleuporum C. galapageium C. granatense (SCB) C. granatense (SCZ) C. granatense (SAN) Other populations

6

PC 2

4

Head width Profemur width Thorax width (frontal) Eye size Head length Metafemur width

2

Total length

0 –2

Metepisternal width Metepisternal length

–4

Wing width Wing length

–6 –30

–20

–10

0

10

20

30

PC1

(B)

C. granatense, Isabela 8

IVA lowland IVA highland ICA lowland ICA highland ISN lowland ISN highland Other populations

6

PC 2

4


2

Total length

0 –2


–4


–6 –30

–20

–10

0

10

20

30

PC 1

(C)

C. granatense, smaller islands 8 6

PC 2

4

SFE SEY ESP MAR FLO PIN Other populations


2

Total length

0 –2


–4 –6 –30


–20

–10

0

10

20

30

PC 1

Fig. 2 Morphological variation in the different Calosoma species and populations from the Galapagos. Principal component analysis, based on length measurements of 26 morphological traits, displaying morphological variation among the species. Only the eigenvectors of traits that are most important in distinguishing the species are annotated. Individual annotation was divided over three different plots to avoid overlap of the symbols: (A) individuals from the major islands San Cristobal, Santa Cruz and Santiago, (B) individuals from Isabela, with individuals sampled at >500 m altitude depicted in yellow and (C) individuals from the remaining smaller islands. Island abbreviations in the legends and colour codes of panel (A) are as in Fig. 1. See Desender & De Dijn (1990) for a more detailed analysis on the morphological variation.

Phylogenetic relationships were reconstructed for the concatenated mtDNA and for each nuclear gene fragment separately by Bayesian methods. The different gene trees were rooted with sequences of the five outgroup species as we are, besides the topology, equally interested in the relative timing of the different branching events recovered from the different genes and relative substitution rates of the different genes. As no statistical support was found for the use of relaxed clock models, we opted for a strict molecular clock analysis using BEAST 1.7.1 (Drummond & Rambaut 2007) and used the split between the most closely related outgroup species (C. sayi) and the Gal apagos Calosoma species as a reference (see Appendix S1, Supporting information, for details on the BEAST analysis and argumentation for the use of strict clock model).

Microsatellite analysis Deviations from Hardy–Weinberg equilibrium were tested with an exact test as implemented in GENEPOP 4.2 (Rousset 2008). Expected heterozygosity and number of private alleles per species and population were calculated with GENALEX 6.5 (Peakall & Smouse 2012) for populations with a minimum of eight genotyped individuals. The degree and significance of genetic differentiation among species and populations were estimated by calculating D (Jost 2008) using the DEMETICS 0.8.5 package (Gerlach et al. 2010) implemented in R 2.14.2. We used an individual-based Bayesian analysis implemented in STRUCTURE 2.3.4. (Pritchard et al. 2000) to infer clusters (K) of individuals based on their multilocus genotypes. We applied an admixture model with five independent runs of K = 2–8, 100 000 Markov chain Monte Carlo repetitions with a burn-in period of 30 000, correlated allele frequencies and no prior information on the population of origin. We used the modal value of the statistic DK as a criterion to delineate the true number of genetic clusters (K), thereby maximizing the rate of change of the likelihood function (ln Pr (X| K)), rather than the likelihood per se (Evanno et al. 2005). Replicate analyses were aligned using CLUMPP 1.1.2 (Jakobsson & Rosenberg 2007). To allow a balanced sampling design, that is a between-ecotype comparison on each island, we excluded individuals of Espa~ nola, Floreana and Pinta in this latter analysis as only the lowland species Calosoma granatense is present on those islands.

Estimating introgression rates A coalescent approach was used to test the significance of admixture among the current and ancestral populations and species, and to estimate their rates. It assumes © 2015 John Wiley & Sons Ltd

P A G O S B E E T L E R A D I A T I O N 3111 ADMIXTURE PATTERNS WITHIN A GALA that populations diverged t generations ago and subsequently were subject to gene flow, that is an ‘isolation with migration model’ sensu (Hey & Nielsen 2004). This allows us to assess the relative importance of incomplete lineage sorting vs. gene exchange in generating gene genealogies. The model estimates the marginal posterior probability densities for the following demographic parameters: mutation-scaled population size i (hi = 4Nil, with l the geometric mean of mutation rate across loci), mutation-scaled divergence times between populations i and j (tij = tijl with tij the divergence time expressed as number of generations) and mutation-scaled migration (i.e. gene flow) rates (mij = Mij/l with Mij being the probability of migration per gene copy per generation from population j into population i). Based on these estimates, one can obtain the more useful estimate of the population migration rate, being the effective number of gene copies from population j received by population i per generation 2NiMij = hi.mij/2. The model has been extended to analyse gene exchange among more than two populations and implemented in the program IMA2 (Hey 2010). However, given that the number of estimated parameters increases exponentially with the number of populations, we restricted the analyses to a set of four populations, that is the highland and lowland species of Santa Cruz (Calosoma leleuporum and C. granatense, respectively) and San Cristobal (Calosoma linelli and C. granatense, respectively). Parameters were estimated based on the sequence data of cad, enolase, the concatenated mtDNA sequences and four microsatellites with perfect repeats (CALO1, CALO16, CALO21, CALO23) (see Appendix S1, Supporting information, for detailed information on the MCMC conditions used for the IM analysis). As the substitution rates of the different genes are expressed in terms of the number of substitutions since divergence from the continental species C. sayi, they only provide information on the relative rate and cannot be used to estimate demographic parameters.

Results Phylogenetic relationships and relative divergence times among the species Phylogenetic relationships among the 51 cad haplotypes yielded two highly supported clades, that is one comprising all haplotypes of the San Cristobal highland species Calosoma linelli and one comprising all haplotypes of both the lowland species Calosoma granatense and those from the highland species Calosoma leleuporum and Calosoma galapageium. The relationships among the cad haplotypes within this latter clade could, however, not be resolved (Fig. 3). © 2015 John Wiley & Sons Ltd

A fairly similar phylogenetic pattern was found for the 30 enolase haplotypes, showing the same well-supported clade uniting only haplotypes of the San Cristobal highland species C. linelli (Fig. 3). Clustering of the remaining haplotypes was, in contrast, only weakly supported (PP = 0.81). The most abundant haplotype was shared among C. leleuporum, C. galapageium and C. granatense throughout the archipelago (see Table S1). The tree of the 118 concatenated mtDNA haplotypes yielded a very different topology wherein all C. linelli haplotypes clustered with high support with six of the 16 C. leleuporum haplotypes (Figs 3 and 4). The 10 remaining haplotypes of C. leleuporum clustered within a clade that mainly comprised haplotypes of the lowland species C. granatense from Santa Cruz and Santiago and all haplotypes of the highland species C. galapageium from Santiago. In few cases, these C. leleuporum haplotypes were very closely related to those of C. granatense from Santa Cruz (Fig. 4). In contrast to the nuclear genes, the mitochondrial haplotypes showed a clear clustering according to island groups within C. granatense (Fig. 4). At the deepest level, haplotypes from the two oldest islands (San Cristobal, Espa~ nola) form a clade with those from the northern islands Marchena and Genovesa and the southern island Floreana. A second well-supported clade mainly comprises haplotypes from Santa Cruz, Santiago and Isabela. This clade also contains haplotypes from C. leleuporum and the three haplotypes of C. galapageium. The third clade contains the majority of haplotypes from Isabela, Fernandina, the northeastern island Pinta and a few haplotypes from Santiago (Fig. 4). Relative timing of the branching events of the different genes revealed that the oldest supported divergence events were recovered with cad and enolase (Fig. 3). These events separate the haplotypes of C. linelli from those of the other species. The mtDNA chronogram shows the next well-supported divergence, that is the separation between the C. granatense populations from the easternmost and old islands (San Cristobal, Espa~ nola, Floreana, Marchena and Genovesa) and the remaining populations (including C. linelli, C. leleuporum and C. galapageium).

Genetic divergence based on microsatellites With the exception of the highland species C. galapageium, at least eight individuals from each species could be genotyped for seven microsatellites (Table 1). No significant deviations from Hardy–Weinberg equilibrium were detected, and expected heterozygosities did not differ substantially between species and populations (data not shown). Yet, the number of private alleles was higher in the highland species C. linelli, followed by the

3112 F . H E N D R I C K X E T A L . Fig. 3 Phylogenetic relationship among the haplotypes of the nuclear genes cad and enolase and mitochondrial DNA. Chronograms (BEAST) summarizing the relative divergence times of the major supported clades of the cad, enolase and concatenated mtDNA haplotypes based on a strict molecular clock analysis. Divergence times are scaled relative to the divergence from the outgroup species Calosoma sayi. Vertical bars right of the trees depict the frequency of haplotypes of each species that was found in each clade with colour codes as in Fig. 1. For the lowland species Calosoma granatense, the origin of the individuals is further indicated by the outline colour of the boxes and corresponds to the colour code of the highland species, that is blue = San Cristobal, red = Santa Cruz, green = Santiago and light yellow = other islands. Node values represent posterior probability values. Horizontal grey bars indicate 95% credibility intervals of the relative divergence times.

1.00

1.00

0.33

cad 1.00

1.00

0.78 0.99

0.99 0.98

C. sayi C. marginale C. wilcoxi C. macrum C. scrutator 0.004

1.00

enolase

1.00

0.81

1.00

1.00 0.97

C. sayi C. marginale C. wilcoxi C. scrutator 0.004

1.00 1.00

1.00

1.00 1.00

mtDNA

1.00 1.00

1.00

1.00

1.00

0.33

C. sayi C. marginale C. macrum C. scrutator C. wilcoxi

0.42 1.00

0.03

highland species C. leleuporum, compared to the remaining species and populations (Table 1). Although the overall degree of genetic differentiation among species and populations was substantially higher than zero (D = 0.113; 95% CI: 0.100–0.129), pairwise D values revealed that this was mainly due to the differentiation between the San Cristobal highland species C. linelli and all other species, with D values ranging between 0.490 and 0.620 (Table 1). For the remaining species– population comparisons, the degree of differentiation did not exceed 0.073. As such, no significant differentiation was observed between the highland (C. leleuporum) and lowland (C. granatense) species on Santa Cruz (D = 0.019; 95%CI = 0.007 to 0.059). Similarly, there was no significant differentiation between the high(C. galapageium) and lowland (C. granatense) species of Santiago, or between the morphologically differentiated high- and lowland populations of C. granatense on Isabela (Table 1). To infer the optimal cluster number, we did not consider likelihood values of K = 8 as replicate analyses for this K value appeared highly inconsistent (high between replicate variability). Examination of DK statistics indicated that optimal clustering was obtained at K = 4 (DKmax = 8.99; see Appendix S2, Supporting information). The results of the cluster analysis are in line with the estimated levels of genetic differentiation (D), as they showed the same noticeable genetic divergence between C. linelli and the other species (Fig. 5). The low level of genetic differentiation between the other species and populations was confirmed by the STRUCTURE analysis, which showed no strong assignments of populations/species to specific clusters (Fig. 5).

Estimating direction and rates of introgression To assess whether the incongruent gene genealogies are due to admixture, rather than to incomplete lineage


P A G O S B E E T L E R A D I A T I O N 3113 ADMIXTURE PATTERNS WITHIN A GALA sorting, an ‘isolation with migration’ (IM) analysis was performed (Hey 2010). As the number of parameters to be estimated increases exponentially with the number of populations included, we restricted the IM analysis to the species of the islands San Cristobal and Santa Cruz. These four populations also showed the most pronounced inconsistencies for the different gene trees (Fig. 3). IM analysis with multiple populations requires a prespecified tree topology for the different populations and species. Based on the relative timing of the split events (Fig. 3), we recovered that the most ancient split separates the species C. linelli from the other species and populations in the archipelago. The next well-supported split was recovered from the mtDNA data and divides the majority of C. granatense haplotypes from the oldest islands, including San Cristobal, from the C. granatense populations of the more recent islands, including Santa Cruz and C. leleuporum. Within this latter mtDNA clade, Calosoma grantense and C. leleuporum haplotypes clustered together with high support and we therefore assumed this to be the most recent separation among those species and populations. Based on this topology, we thus assumed the following ancestral populations (from most recent to oldest): (i) an ancestral population on Santa Cruz that gave rise to the Santa Cruz C. granatense population and C. leleuporum; (ii) an ancestral population that gave rise to C. granatense from both San Cristobal and Santa Cruz as well as C. leleuporum; and (iii) an common ancestor to all populations (Fig. 6). In general, IMa analysis revealed multiple directions of gene exchange among the different contemporary and ancestral populations (Fig. 6). High introgression rates were inferred on Santa Cruz from the C. granatense population into the highland species C. leleuporum with a most probable estimate of 9.3 C. granatense gene copies that introgressed into C. leleuporum per generation (P < 0.001). Although less significant (0.01 < P < 0.05), an estimated 23.0 C. leleuporum gene copies introgressed per generation into the opposite direction on Santa Cruz. Significant signatures of gene exchange among the contemporary populations were further observed between the two C. granatense populations that reside on the different islands, with the direction being lower from Santa Cruz towards San Cristobal compared to the opposite direction (3.3 vs. 14.6 gene copies/generation) (Appendix S3, Supporting information). For the ancestral populations, significant gene flow could be detected between the C. granatense population of San Cristobal and the ancestral Santa Cruz population. Apparently, this ancestral Santa Cruz population was also exposed to low (2NM = 0.05), though highly significant (P < 0.001), gene flow from the population that gave rise to C. linelli before C. leleuporum and © 2015 John Wiley & Sons Ltd

C. granatense diverged on Santa Cruz. Gene flow was further observed between the ancestral population that gave rise to C. granatense and C. leleuporum and the ancestral population giving rise to C. linelli (9.7 gene copies/generation). However, these estimates should be treated with caution as the size of the former ancestral population could not be estimated with sufficient accuracy. In sum, these analyses revealed that the high levels of gene exchange took place during the divergence process both between ecologically divergent species residing on the same island and between the winged, geographically separated lowland populations.

Discussion Genetic relationships A general pattern emerging from the nuclear gene fragments is that the oldest split separates the lineage that gave rise to the highland (Calosoma linelli) and lowland (Calosoma granatense) species from San Cristobal and shows that the ecotypic divergence on this most ancient island is the oldest recovered split within this radiation. Consistent with our expectation of a repeated and parallel evolution of highland species out of the lowland species C. granatense (Desender & De Dijn 1990), the pairwise genetic divergence between the species on the younger islands is much smaller. Despite their morphological divergence and taxonomic status as clearly distinct species, no consistent clustering of nuclear haplotypes was observed for C. granatense and either Calosoma leleuporum or Calosoma galapageium. Similarly, nuclear and mtDNA haplotypes obtained from longwinged lowland and short-winged highland populations on the recent island Isabela were strongly admixed. This close genetic relatedness between highland and lowland species points into the direction of a very recent divergence or rampant introgression between the ecologically divergent species on the more recent islands Santa Cruz, Santiago and Isabela. Given this close genetic relationship between highland and lowland species on the younger islands, it may be not surprising that the microsatellite allele frequency differences between these high- and lowland species are very small and not statistically significant. This degree of differentiation is within the same range compared to the degree of differentiation among geographically separated C. granatense populations. As microsatellite allele frequencies are expected to reflect the patterns of contemporary gene flow (Avise 2004), it is likely that gene flow is currently ongoing between C. granatense and C. leleuporum while maintaining the strong morphological differentiation in wing size, head

3114 F . H E N D R I C K X E T A L .

0.98 1 1

1

1 1 1 C230_SAN_900 m

0.98

1

0.99

0.99

0.99 0.99

C239_SAN_900

0.99

1 0.95

1

C228_SAN_700 m

0.99 1

C116_SAN_700 m C117_SAN_700 m

0.99 0.99 0.99

C233_SAN_700 m

0.99 1

1 0.93 0.99 C232_SAN_700 m

1 0.94

0.94

0.90

C231_SAN_700 m C236_SAN_700 m

1 0.97 1 1

0.99 1 1

0.99 1

0.99 0.96 1

1

0.94 0.98 0.97 1 0.97 1 0.97 1

0.03


P A G O S B E E T L E R A D I A T I O N 3115 ADMIXTURE PATTERNS WITHIN A GALA Fig. 4 Detailed phylogenetic relationship of the mtDNA haplotypes. Chronograms (BEAST) summarizing the relative divergence times of different haplotypes of the concatenated mtDNA (cox1, nadh1 and cytb) haplotypes based on a strict molecular clock analysis. Outgroup taxa have been excluded from this figure. Node values represent posterior probabilities of the clades. Horizontal grey bars indicate 95% credibility intervals of the relative divergence times. Taxon labels at the tips of the tree represent individual ID and island code as specified in Fig. 1. Colour codes are as in Fig. 1, with individuals of Isabela sampled at >500 m altitude depicted in yellow. Grey intensities of the islands depicted right from the tree are relative to the proportion of haplotypes from a particular island contained within each clade, with darker values corresponding to higher proportions.

shape and even genital traits (Desender & De Dijn 1990). However, given that the number of individuals that could be used for microsatellite genotyping was small, more subtle patterns of genetic differentiation cannot be inferred. Increased levels of differentiation in relation to island age may result from both the time available to allow adaptation to different niches and the time to allow reproductive barriers to evolve between species occupying different niches (Emerson & Gillespie 2008; Roderick et al. 2012). The close genetic relationship between the morphologically divergent high- and lowland populations on the recent island Isabela (0.5–0.8 MY) suggests that adaptation to different niches evolved within this time frame. Complete reproductive isolation between high- and lowland populations did, however, not yet evolve. Only for the oldest island San Cristobal, with an estimated age of 2.4–4 MY (Geist et al. 2013), clear genetic clustering of haplotypes within each species, high differentiation in microsatellite allele frequencies and lack of any signatures of recent gene flow based on the IM analyses suggest that reproductive isolation between high- and lowland species is more complete, in particular as low numbers of the lowland species can be found in sympatry with the highland species C. linelli (F. Hendrickx, personal observation). Despite the consistent pattern observed for the nDNA, part of the mtDNA haplotypes produced a contrasting phylogenetic pattern in two respects. First, the oldest mtDNA divergence does not include the ancient split between C. linelli and the other Calosoma populations as recovered from the nuclear gene fragments. The oldest divergence shown by the mtDNA data separated the C. granatense populations from the more eastern and older islands San Cristobal, Espa~ nola, Marchena and Genovesa from the remaining Calosoma populations (including C. linelli, C. leleuporum and C. galapageium). Furthermore, this divergence occurred far more recently than the split between C. linelli and the remaining species that was recovered from nDNA (Fig. 3). This suggests that the old divergence between C. linelli and the remaining Calosoma populations could not be recovered by the mtDNA haplotypes, most likely due to the substitution of the ancestral C. linelli mtDNA by ancestral C. leleuporum haplotypes. Second, the mtDNA haplotypes of C. leleuporum formed two distinct © 2015 John Wiley & Sons Ltd

groups. One group was placed among haplotypes of the lowland species of Santa Cruz, which is in line with the nDNA. The other haplotype group of C. leleuporum is the sister clade of the highland species of San Cristobal (C. linelli). Posterior densities of the timing of these two divergence events (Fig. 4) indicated that this apparent shared ancestry between the mtDNA clades of the two highland species C. leleuporum and C. linelli is older than the mtDNA divergence between C. leleuporum and C. granatense on Santa Cruz. This suggests that the close genetic relationship between high- and lowland species on Santa Cruz is the result of either incomplete lineage sorting or introgression of C. granatense haplotypes into C. leleuporum. Because not all individuals were sequenced for all gene fragments (Table S1, Supporting information), incongruences in the mtDNA and nDNA gene topologies could potentially be caused by sampling different gene fragments in different individuals. However, for C. leleuporum, mtDNA and cad haplotypes originated from the same set of individuals (Table S1, Supporting information). Also for the other populations and species, 78% of the individuals were sequenced for both mtDNA and cad, and therefore, it renders less likely that the observed incongruence is caused by sampling different individuals.

Introgression patterns and ecotypic differentiation Gene genealogies can exhibit substantial stochastic variation for a given species divergence scenario, rendering it difficult to investigate the role of incomplete lineage sorting vs. admixture in generating incongruent tree topologies. IM analyses can be a powerful tool to infer such rates of past and contemporary gene exchange between species, but the number of estimated parameters grows exponentially with the number of populations. Combined with the large stochasticity of gene genealogical patterns for a given divergence scenario, it is necessary to include a large number of genetic markers to obtain reliable estimates. Here, six nuclear markers (cad, enolase and four microsatellites) and one mitochondrial marker were used for the IM analysis and we therefore restricted our analysis to the species of the older islands Santa Cruz and San Cristobal only. As this number of markers is rather limited for making

Colour codes are as in Fig. 1

Table 1 Number of genotyped individuals (N), average expected heterozygosity per locus (He), average number of private alleles per locus (PA) and genetic differentiation (Jost’ D) between the different Calosoma species and Calosoma granatense populations from the Gal apagos based on allele frequencies of seven microsatellites. D values whose 95% credibility interval did not contain zero are indicated with an asterisk

3116 F . H E N D R I C K X E T A L . accurate estimates of gene exchange, population sizes and divergence times (Hey 2010), confidence intervals around the estimates were large and estimates could be potentially biased towards single markers. The IM analysis suggests substantial introgression between the high- and lowland species on Santa Cruz. Hence, the close genetic relationship between both species on Santa Cruz is, at least partly, due to introgression rather than to a very recent divergence. Ecotypic differentiation on Santa Cruz thus most likely took place under high levels of gene flow. Whether this divergence occurred under continuous gene flow, or rather under different episodes of allopatric divergence interspersed with phases of high gene flow, is difficult to infer with the data at hand. For the two species on the oldest island San Cristobal, both IM analysis and the high number of private alleles in the highland species indicate that contemporary gene flow between the species on this island is most probably absent. However, the initial divergence between C. linelli and the ancestral C. granatense population probably took place under gene exchange as indicated by the relatively high estimates of gene flow into C. linelli. Whether the close genetic relationship between C. galapageium and C. granatense on Santiago and between short- and longwinged C. granatense populations on Isabela is also the result of admixture remains difficult to infer. Despite the low number of individuals of C. galapageium, no clear signature of genetic differentiation with the lowland population of C. granatense on Santiago could be detected and at least suggests that also for those species, interspecific gene flow is not unlikely. Besides these signals of within-island gene exchange, isolation with migration further revealed the significant levels of gene exchange among islands after their initial divergence, in particular for the lowland species C. granatense. Hence, our results suggest that the occupation of Santa Cruz is not the result of a single colonization event, but involves a continuous flux of Calosoma individuals from San Cristobal towards Santa Cruz and vice versa. Given that San Cristobal and Santa Cruz have not been connected by land (Geist et al. 2013; Ali & Aitchison 2014), it is not surprising that migration between both islands was highest for the winged lowland species compared to the flightless highland species. Nevertheless, low but significant signals of admixture involving between-island migration of highland species was observed, that is from the ancestor of C. linelli towards the Santa Cruz clade. Although it is not known whether this ancestral species that gave rise to C. linelli was flightless during these among-island colonization events, many flightless beetle genera have colonized the different islands within the archipelago (e.g. Galapaganus, Pterostichus) (Sequeira et al. 2000; Peck © 2015 John Wiley & Sons Ltd

P A G O S B E E T L E R A D I A T I O N 3117 ADMIXTURE PATTERNS WITHIN A GALA

Isabela (Cerro Azul)

Santiago

Santa Cruz

2006). Such among-island migration events of flightless species that are currently restricted to highland habitats could have occurred during colder Pleistocene periods when the typical highland habitat probably extended to much lower altitudes (Collinvaux 1984; Grant 1999). Both the probably larger population sizes of highland species during colder periods and their closer proximity to coastal habitats could have facilitated dispersal across the ocean by floating material. Importantly, migration events of high- and lowland ecotypes across islands may have important implications for the evolution of both ecotypes on Santa Cruz and within the archipelago in general. First, sporadic migration events may have resulted in the introgression of adaptive alleles that facilitate evolution in similar directions (Jones et al. 2012; Pardo-Diaz et al. 2012; Stern 2013; Faria et al. 2014; De Busschere et al. 2015). Indeed, alleles that are differentially selected in sympatric ecotype pairs often show a singular mutational origin, suggesting that reuse of these adaptive alleles might play an important role in repeated adaptation (Colosimo et al. 2005; Van Belleghem et al. 2015). More generally, admixture of different gene pools can provide strong fitness benefits and may as such create opportunities for the successful colonization of new habitats (Rius & Darling 2014). Alternatively, both ecotypes may independently colonize the same islands and subsequently exchange neutral alleles, which can produce a signal of in situ divergence, such as observed in Darwin’s finches (Grant & Grant 2010b; Farrington et al. 2014). On the Galapagos, several cases of within-island radiation have been reported (Parent et al. 2008) including Galapaganus weevils (Sequeira et al. 2008), Stomion beetles (Finston & Peck 2004), Bulimulus snails (Parent & Crespi 2006) and Hogna wolf spiders (De Busschere © 2015 John Wiley & Sons Ltd

C. linelli

C. granatense

C. leleuporum

C. granatense

C. galapageium

C. granatense

C. granatense (short-winged - highland)

C. granatense (long-winged/ lowland)

Fig. 5 Assignment of 113 individuals of the highland–lowland species pairs from San Cristobal, Santa Cruz, Santiago and Isabela, genotyped for seven microsatellites, into four distinct clusters as obtained by STRUCTURE v 2.4.3. Species colours are as in Fig. 1.

San Cristobal

et al. 2010). These within-island radiations were all inferred from the close morphological or genetic similarity of species from the same island. Yet, extensive molecular multilocus data sampled from many individuals per population are still scarce. Only for the Hogna wolf spiders, molecular data detected signatures of introgression and revealed that within- as well as among-island introgression took place in this parallel radiation (De Busschere et al. 2015). Mating experiments further suggested that distantly related species within this radiation are still capable of interspecific mating (De Busschere & Hendrickx 2013). Island archipelagos can be expected to show a pattern of relatively high genetic relationships of ecotypic differentiated species residing on the same island compared to the genetic relationship between conspecific populations from different islands. First, populations of the same species on different islands could be strongly isolated from each other. Also in this study, populations of C. granatense from different islands (e.g. San Cristobal, Espa~ nola and Floreana vs. Santa Cruz, Santiago and Isabela) often showed a more clear genetic structuring with respect to mtDNA compared to different ecotypes living on the same island. Although based on a relatively low number of individuals, significant differentiation in microsatellite alleles was observed between populations Espa~ nola and Floreana vs. Santiago and Santa Cruz. Second, substantial gene flow can be expected between species from the same island that exploit different habitats. Evidence is indeed accumulating that both speciation in the face of gene flow and hybridization and introgression are much more prevalent than previously thought (Shaw 2002; Grant et al. 2005; Jordal et al. 2006; Nosil 2008; Grant & Grant 2010a; Rius & Darling 2014; De Busschere et al. 2015). In

3118 F . H E N D R I C K X E T A L . C. linelli (SCB)

C. granatense C. leleuporum (SCB) (SCZ)

C. granatense (SCZ) 9.3***

3.3*** 0.39 tu

23.0* 14.6***

0.05** 3.4*** 1.6 tu 0.33**

9.7**

5.8 tu

addition, population sizes of insular species can be small and subjected to strong fluctuations, for example in response to changes in climatic conditions (Grant 1999; Gillespie & Roderick 2014), but in particular during the initial stages of colonization events. This may on its turn strongly enhance the fixation of genes introgressed from the native population into the invading population (Currat et al. 2008). Estimates of gene flow based on our IM analysis indeed show that gene flow between C. leleuporum and Calosoma grantense on Santa Cruz is of comparable magnitude compared to geneflow estimates between the C. granatense populations of Santa Cruz and San Cristobal. However, given that only a small subset of

Fig. 6 Estimated introgression rates among the highland and lowland species from the islands Santa Cruz and San Cristobal. Estimated posterior densities of the population introgression rates (2NM) between the current species of Santa Cruz (i.e. Calosoma leleuporum and Calosoma granatense (SCZ)) and San Cristobal (i.e. Calosoma linelli and C. granatense (SCB)) as obtained from an ‘isolation with migration’ (IMA2) analysis (Hey 2010). Boxes represent contemporary and historic populations, with the width of each box being proportional to the estimated population size. Boxes with grey borders represent ancestral populations whose population size could not be estimated with sufficient accuracy. Arrows depict significant directions of gene introgression, forward in time. Asterisks above migration rates (2NM) represent the level of significance based on a likelihood ratio test (*0.05 > P 0.01; ***P < 0.001). Time (tu) is expressed in mutation-scaled generation times.

the genome was sampled in our study and that only a small number of individuals could be analysed in some populations, which both result in wide confidence intervals around these estimates, it remains to be investigated whether this is a general pattern that is also apparent at a genomewide level.

Conclusions This study suggests that contemporary and historic inter- and intraspecific gene flow strongly drives genetic structure within the Calosoma radiation in Gal apagos. While repeated differentiation of highland species out of the common lowland species Calosoma granatense © 2015 John Wiley & Sons Ltd

P A G O S B E E T L E R A D I A T I O N 3119 ADMIXTURE PATTERNS WITHIN A GALA appears the most plausible evolutionary scenario, repeated colonization events of both high- and lowland species on different islands likely played an important role in this parallel radiation. High levels of admixture within islands increase the genetic relationship between morphologically and ecologically highly distinct species, resulting in genetic relationships that are of comparable magnitude compared to intraspecific differentiation of populations residing on different islands. Given that accurate phylogenetic analyses are indispensable to reconstruct the evolutionary history of insular radiations, future detailed studies on patterns of past and contemporary gene exchange between the different species and islands could strongly enhance our understanding on the evolution of island biota.

Acknowledgements This study could not have been conducted without access to the extensive collection of the late Konjev Desender, who collected the majority of Calosoma individuals used in this study. Leon Baert, Charlotte De Busschere and Jean-Pierre Maelfait (†) provided help in collecting additional specimens. Excellent cooperation and field logistic support were provided by the Charles Darwin Research Station (CDRS; Isla Santa Cruz, Galapagos, Ecuador); the Galapagos National Park Service and the Department of Forestry, Ministry of Agriculture of Ecuador. Three anonymous referees and the handling editor Prof. R. Gillespie provided some constructive comments on an earlier version of this manuscript. Outgroup specimens were kindly provided by Martin Husemann, and muscle tissue of the C. scrutator specimen was kindly provided by David R. Maddison. Financial support for the expeditions was achieved from the Royal Belgian Institute of Natural Sciences and the King Leopold III Fund. IM analyses were carried out using the STEVIN Supercomputer Infrastructure at Ghent University, funded by Ghent University, the Flemish Supercomputer Center (VSC), the Hercules Foundation and the Flemish Governement—department EWI. This work was conducted within the framework of BRAIN-Be BR/121/PI/GENESORT, the Interuniversity Attraction Poles programme IAP (SPEEDY), Action1—MO/36/025 of the Belgian Science Policy Office and FWO research network BeBoL (W0.009.11N).

References Abbott R, Albach D, Ansell S et al. (2013) Hybridization and speciation. Journal of Evolutionary Biology, 26, 229–246. Ali JR, Aitchison JC (2014) Exploring the combined role of eustasy and oceanic island thermal subsidence in shaping biodiversity on the Galapagos. Journal of Biogeography, 41, 1227–1241. Avise JC (2004) Molecular Markers, Natural History, and Evolution. Sinauer Associates, Sunderland, Massachusetts. Barton NH (2001) The role of hybridization in evolution. Molecular Ecology, 10, 551–568. Campbell V, Legendre P, Lapointe F-J (2011) The performance of the Congruence Among Distance Matrices (CADM) test in phylogenetic analysis. BMC Evolutionary Biology, 11, 64.


Chan KMA, Levin SA (2005) Leaky prezygotic isolation and porous genomes: rapid introgression of maternally inherited DNA. Evolution, 59, 720–729. Collinvaux PA (1984) The Gal apagos climate: present and past. In: Key Environments Galapagos (ed. Perry R), pp. 55–69. Pergamont Press Ltd., Oxford. Colosimo PF, Hosemann KE, Balabhadra S et al. (2005) Widespread parallel evolution in sticklebacks by repeated fixation of ectodysplasin alleles. Science, 307, 1928–1933. Currat M, Ruedi M, Petit RJ, Excoffier L (2008) The hidden side of invasions: massive introgression by local genes. Evolution, 62, 1908–1920. De Busschere C, Hendrickx F (2013) Does behavioural isolation prevent interspecific mating within a parallel ecotypic wolf spider radiation from the Gal apagos? Journal of Arachnology, 41, 25–30. De Busschere C, Hendrickx F, Van Belleghem SM et al. (2010) Parallel habitat specialization within the wolf spider genus Hogna from the Gal apagos. Molecular Ecology, 19, 4029–4045. De Busschere C, Van Belleghem SM, Hendrickx F (2015) Inter and intra island introgression in a wolf spider radiation from apagos and its implications for parallel evolution. MolecuGal lar Phylogenetics and Evolution, 84, 73–84. Desender K (1989) Dispersal and ecology of carabid beetles in Belgium: an evolutionary approach. Bulletin of the Royal Belgian Institute of Natural Sciences 54, 1–36. Desender K, De Dijn B (1989) The Calosoma species of the Galapagos archipelago. I. Redescription and distribution of the species. Bulletin of Royal Belgian Institute of Natural Sciences: Entomology, 59, 131–144. Desender K, De Dijn B (1990) The Calosoma species (Coleoptera, Carabidae) of the Gal apagos archipelago. II. Discriminant analyses and species identification key. Bulletin of Royal Belgian Institute of Natural Sciences: Entomology, 60, 55–68. Desender K, Baert L, Maelfait JP (1991) Evolutionary systematics of Calosoma WEBER carabid beetles of the Galapagos archipelago, Ecuador (Coleoptera: Carabidae). In: Advances in Coleopterology (eds Zunino M, Belles M, Blas M), pp. 193–200. AEC, Barcelona, Spain. Desender K, Baert L, Maelfait J-P (1992) Distribution and speciation of carabid beetles in the Gal apagos archipelago (Ecuador). Bulletin of Royal Belgian Institute of Natural Sciences: Entomology, 62, 57–65. Dhuyvetter H, Gaublomme E, Verdyck P et al. (2002) Isolation and characterisation of microsatellite loci in Gal apagos caterpillar hunters (Coleoptera, Carabidae, Calosoma). Molecular Ecology Notes, 2, 284–286. Dhuyvetter H, Hendrickx F, Gaublomme E, Desender K (2007) Differentiation between two salt marsh beetle ecotypes: evidence for ongoing speciation. Evolution, 61, 184–193. Drummond AJ, Rambaut A (2007) BEAST: bayesian evolutionary analysis by sampling trees. BMC Evolutionary Biology, 7, 214. Emerson BC, Faria R (2014) Fission and fusion in island taxa— serendipity, or something to be expected? Molecular Ecology, 23, 5132–5134. Emerson BC, Gillespie RG (2008) Phylogenetic analysis of community assembly and structure over space and time. Trends in Ecology & Evolution, 23, 619–630. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology, 14, 2611–2620.

3120 F . H E N D R I C K X E T A L . Faria R, Renaut S, Galindo J et al. (2014) Advances in ecological speciation: an integrative approach. Molecular Ecology, 23, 513–521. Farrington HL, Lawson LP, Clark CM, Petren K (2014) The evolutionary history of Darwin’s finches: speciation, gene flow, and introgression in a fragmented landscape. Evolution, 68, 2932–2944. Finston TL, Peck SB (2004) Speciation in Darwin’s darklings: taxonomy and evolution of Stomion beetles in the Gal apagos Islands, Ecuador (Insecta: Coleoptera: Tenebrionidae). Zoological Journal of the Linnean Society, 141, 135–152. Fukami T, Beaumont HJE, Zhang X-X, Rainey PB (2007) Immigration history controls diversification in experimental adaptive radiation. Nature, 446, 436–439. Funk DJ, Omland KE (2003) Species-level paraphyly and polyphyly: frequency, causes, and consequences, with insight from animal mitochondrial DNA. Annual Review of Ecology, Evolution and Systematics, 34, 397–423. Garrick RC, Benavides E, Russello MA et al. (2014) Lineage fusion in Galapagos giant tortoises. Molecular Ecology, 23, 5276–5290. Gavrilets S, Losos JB (2009) Adaptive radiation: contrasting theory with data. Science, 323, 732–737. Geist D, Snell H, Snell H, Goddard C, Kurz M (2013) A paleogeographic model of the Galapagos Islands and biogeographical and evolutionary implications. In: The Galapagos: A Natural Laboratory for the Earth Sciences (eds Harpp K, d’Ozouville N, Mittelstaedt E, Graham D). American Geophysical Union Monograph, Washington, District of Columbia. Gerlach G, Jueterbock A, Kraemer P, Deppermann J, Harmand P (2010) Calculations of population differentiation based on G(ST) and D: forget G(ST) but not all of statistics!. Molecular Ecology, 19, 3845–3852. Gillespie R (2004) Community assembly through adaptive radiation in Hawaiian spiders. Science, 303, 356–359. Gillespie R, Baldwin BG (2010) Island biogeography of remote archipelagoes: interplay between ecological and evolutionary processes. In: The Theory of Island Biogeography Revisited (eds Losos JB, Ricklefs E), pp. 358–387. Princeton University Press, Princeton, New Jersey. Gillespie RG, Roderick GK (2002) Arthropods on islands: colonization, speciation and conservation. Annual Review of Entomology, 47, 595–632. Gillespie R, Roderick GK (2014) Geology and climate drive diversification. Nature, 509, 297–298. Grant PR (1999) Ecology and Evolution of Darwin’s Finches. Princeton University Press, Princeton, New Jersey. Grant PR, Grant BR (2010a) Conspecific versus heterospecific gene exchange between populations of Darwin’s finches. Philosophical Transactions of the Royal Society B-Biological Sciences, 365, 1065–1076. Grant PR, Grant BR (2010b) Sympatric speciation, immigration, and hybridization in island birds. In: The Theory of Island Biogeography Revisited (eds Losos JB, Ricklefs E), pp. 326–357. Princeton University Press, Princeton, New Jersey. Grant PR, Grant BR, Petren K (2005) Hybridization in the recent past. American Naturalist, 166, 56–67. Hein J, Schierup MH, Wiuf C (2010) Gene Genealogies, Variation and Evolution: a Primer in Coalescent Theory. Oxford University Press, New York, New York.

Hendrickx F, Palmer SCF, Travis JMJ (2013) Ideal Free distribution of fixed dispersal phenotypes in a wing dimorphic beetle in heterogeneous landscapes. Ecology, 94, 2487–2497. Hey J (2010) Isolation with migration models for more than two populations. Molecular Biology and Evolution, 27, 905–920. Hey J, Nielsen R (2004) Multilocus method for estimating population sizes, migration rates and divergence time, with applications to the divergence of Drosophila pseudoobscura and D. persimilis. Genetics, 167, 747–760. Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics, 23, 1801–1806. Jeannel R (1940) Les Calosomes. Memoires du Musee National d’Histoire Naturelle, Editions du Musee, Paris. Jones FC, Grabherr MG, Chan YF et al. (2012) The genomic basis of adaptive evolution in threespine sticklebacks. Nature, 484, 55–61. Jordal BH, Emerson BC, Hewitt GM (2006) Apparent ‘sympatric’ speciation in ecologically similar herbivorous beetles facilitated by multiple colonizations of an island. Molecular Ecology, 15, 2935–2947. Jost L (2008) GST and its relatives do not measure differentiation. Molecular Ecology, 17, 4015–4026. Leibold MA, Holyoak M, Mouquet N et al. (2004) The metacommunity concept: a framework for multi-scale community ecology. Ecology Letters, 7, 601–613. Losos JB, Ricklefs R (2009) Adaptation and diversification on islands. Nature, 457, 830–836. Mahler DL, Ingram T, Revell LJ, Losos JB (2013) Exceptional convergence on the macroevolutionary landscape in an island lizard radiations. Science, 341, 293–295. Mallet J (2005) Hybridization as an invasion of the genome. Trends in Ecology and Evolution, 20, 229–237. Martin SH, Dasmahapatra KK, Nadeau NJ et al. (2013) Genome-wide evidence for speciation with gene flow in Heliconius butterflies. Genome Research, 23, 1817–1828. Niemiller ML, Fitzpatrick BM, Miller BT (2008) Recent divergence with gene flow in Tennessee cave salamanders (Plethodontidae: Gyrinophilus) inferred from gene genealogies. Molecular Ecology, 17, 2258–2275. Nosil P (2008) Speciation with gene flow could be common. Molecular Ecology, 17, 2103–2106. Olivieri I (2009) Dispersal. In: Encyclopedia of Islands (eds Gillespie R, Clague DA), pp. 224–228. University of California Press, Berkeley, California. Pardo-Diaz C, Salazar C, Baxter SW et al. (2012) Adaptive introgression across species boundaries in Heliconius butterflies. PLOS Genetics, 8, e1002752. Parent CE, Crespi BJ (2006) Sequential colonization and diversification of Galapagos endemic land snail genus Bulimulus (Gastropoda, Stylommatophora). Evolution, 60, 2311–2328. Parent CE, Caccone A, Petren K (2008) Colonization and diversification of Galapagos terrestrial fauna: a phylogenetic and biogeographical synthesis. Philosophical Transactions of the Royal Society B-Biological Sciences, 363, 3347–3361. Pascoal S, Cezard T, Eik-Nes A et al. (2014) Rapid convergent evolution in wild crickets. Current Biology, 24, 1369–1374. Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research –an update. Bioinformatics, 28, 2537–2539.


P A G O S B E E T L E R A D I A T I O N 3121 ADMIXTURE PATTERNS WITHIN A GALA Peck SB (2006) The Beetles of the Galapagos Islands, Ecuador: Evolution, Ecology, and Diversity (Insecta: Coleoptera). NRC Research Press, Ottawa, Ontario. Pond SLK, Posada D, Gravenor MB, Woelk CH, Frost SDW (2006) Automated phylogenetic detection of recombination using a genetic algorithm. Molecular Biology and Evolution, 23, 1891–1901. Posada D (2008) jModelTest: phylogenetic model averaging. Molecular Biology and Evolution, 25, 1253–1256. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959. Rius M, Darling JA (2014) How important is intraspecific genetic admixture to the success of colonising populations? Trends in Ecology & Evolution, 29, 233–242. Roderick GK, Croucher PJP, Vandergast AG, Gillespie RG (2012) Species differentiation on a dynamic landscape: shifts in metapopulation genetic structure using the chronology of the Hawaiian Archipelago. Evolutionary Biology, 39, 192–206. Roff DA (1994) Habitat persistence and the evolution of wing dimorphism in insects. American Naturalist, 144, 772–798. Rousset F (2008) GENEPOP ‘007: a complete re-implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources, 8, 103–106. Ryan PG, Bloomer P, Moloney CL, Grant TJ, Delport W (2007) Ecological speciation in South Atlantic island finches. Science, 315, 1420–1423. Schluter D (2000) The Ecology of Adaptive Radiation. Oxford University Press, Oxford. Seehausen O (2004) Hybridization and adaptive radiation. Trends in Ecology and Evolution, 19, 198–207. Sequeira AS, Lanteri AA, Scataglini MA, Confalonieri VA, Farrell BD (2000) Are flightless Galapaganus weevils older than the Galapagos islands they inhabit? Heredity, 85, 20–29. Sequeira AS, Lanteri AA, Albelo RL, Bhattacharya S, Sijapati M (2008) Colonization history, ecological shifts and diversification in the evolution of endemic Galapagos weevils. Molecular Ecology, 17, 1089–1107. Shaw K (2002) Conflict between nuclear and mitochondrial DNA phylogenies of a recent species radiation: what mtDNA reveals and conceals about modes of speciation in Hawaiian crickets. Proceedings of the National Academy of Sciences of the USA, 99, 16122–16127. Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for haplotype reconstruction from population data. American Journal of Human Genetics, 68, 978–989. Stern DL (2013) The genetic causes of convergent evolution. Nature Reviews Genetics, 14, 751–764. Urban MC, Leibold MA, Amarasekare P et al. (2008) The evolutionary ecology of metacommunities. Trends in Ecology & Evolution, 23, 311–317. Van Belleghem SM, Hendrickx F (2014) A tight association in genetically unlinked traits in sympatric and allopatric populations of a saltmarsh beetle. Genetica, 149, 1–9. Van Belleghem SM, Roelofs D, Hendrickx F (2015) Evolutionary history of a dispersal-associated locus in allopatric and sympatric divergent populations of a wing polymorphic beetle. Molecular Ecology, 24, 890–908.


Wang IJ, Glor RE, Losos JB (2013) Quantifying the roles of ecology and geography in spatial genetic divergence. Ecology Letters, 16, 175–182. Warren BH, Simberloff D, Ricklefs E et al. (2015) Islands as model systems in ecology and evolution: prospects fifty years after MacArthur-Wilson. Ecology Letters, 18, 200– 217. Wild AL, Maddison DR (2008) Evaluating nuclear protein-coding genes for phylogenetic utility in beetles. Molecular Phylogenetics and Evolution, 48, 877–891. Zera AJ, Denno RF (1997) Physiology and ecology of dispersal polymorphism in insects. Annual Review of Entomology, 42, 207–230.

F.H. designed research; F.H., S.V.B., W.D. and C.V.G. collected specimens, F.H., C.V.G. and V.V. performed research; F.H. and C.V.G. analysed data; F.H., C.V.G. and T.B. wrote the manuscript.

Data accessibility DNA sequences: GenBank accessions KC245496– KC245690 and KR132532–KR132553. Specimen information, sampling locations, microsatellite genotypes, haplotypes as Supporting information. IMA2 input files, raw BEAST tree files, haplotype information and microsatellite data: Dryad doi:10.5061/ dryad.b56n0.

Supporting information Additional supporting information may be found in the online version of this article. Table S1 Specimen information, sampling locations, haplotype information and microsatellite genotypes of the individuals used in this study. Table S2 Number of analysed individuals per gene fragment and population. Appendix S1 Method details. Appendix S2 Plot of the natural logarithm of the likelihood of the data (left) and deltaK (right) for different numbers of putative populations, K, as obtained from STRUCTURE v2.3. (Pritchard et al. 2000, Pritchard et al. 2010(1)). Appendix S3 Summary of the posterior distribution of population mutation rate and mutation scaled and population migration rates of all current and ancestral populations.