Journal of Biogeography

Insight into an island radiation: the of the archipelago

ForJournal: Journal Peer of Biogeography Review

Manuscript ID: draft

Manuscript Type: Special Issue Date Submitted by the Author:

Complete List of Authors: Vasconcelos, Raquel; CIBIO-UP Centro de Investigação em Biodiversidade e Recursos Genéticos, ICETA Carranza, Salvador; Institue of Evolutionary Biology (CSIC-UPF), Phylogeny and Systematics Harris, D.J.; CIBIO-UP Centro de Investigação em Biodiversidade e Recursos Genéticos, ICETA

Tarentola, Cape Verde Islands, cytochrome b, 12S rRNA, Key Words: Phylogeny, Biogeography, Island radiation

Page 1 of 30 Journal of Biogeography

1 1 2 3 4 Original article 5 6 7 8 Insight into an island radiation: the Tarentola geckos of the 9 10 Cape Verde archipelago. 11 12 13 R. Vasconcelos, CIBIO-UP, Centro de Investigação em Biodiversidade e Recursos Genéticos, Campus Agrário de Vairão, R. 14 Padre Armando Quintas, 4485-661 Vairão, Portugal; Faculdade de Ciências da Universidade do Porto Pr. Gomes Teixeira, 15 4099-002 Porto, Portugal. E-mail: [email protected] 16 17 S. Carranza, Institute of Evolutionary Biology (CSIC-UPF) CMIMA - Passeig Marítim de la Barceloneta, 37-49, E-08003 18 Barcelona, España. E-mail: [email protected] 19 D.J. Harris, CIBIO-UP, Centro de Investigação em Biodiversidade e Recursos Genéticos, Campus Agrário de Vairão, R. Padre 20 Armando Quintas, 4485-661For Vairão, Portugal. Peer Faculdade de CiênciasReview da Universidade do Porto Pr. Gomes Teixeira, 4099- 21 002 Porto, Portugal. E-mail: [email protected] 22 23 24 Corresponding author: 25 Salvador Carranza, Institute of Evolutionary Biology (CSIC-UPF) CMIMA - Passeig Marítim de la Barceloneta, 37-49, E-08003 26 Barcelona, España. E-mail: [email protected] 27

28 29 Running head: Cape Verdian Tarentola island radiation 30 31 ABSTRACT 32 Aim To reassess relationships of Tarentola from the Cape Verde Islands by including specimens from all 33 34 islands, including distinct subspecies never previously analysed genetically. To adequately determine 35 for the first time variation within forms by sequencing over 400 new specimens, decreasing the 36 37 possibility of any cryptic form being overlooked and resolving some of the issues raised in earlier works. 38 Extensive within-island sampling was also used to shed light on distributions and to make comparisons 39 with age and geological features of the islands including size, altitude and availability to explain 40 41 genetic diversity. 42 Location An oceanic archipelago situated in the Atlantic Ocean belonging to the Macaronesian 43 44 biogeographic region and located around 500 km off the coast of Senegal: the Cape Verde Islands 45 (latitudes 14° – 17° N and longitudes 22° – 25° W). 46 47 Methods A total of 405 specimens of Tarentola were collected in 276 localities from nine islands and 48 three islets within three groups with very different geological histories, topography, climate and 49 . Mitochondrial cytochrome b gene and 12S rRNA partial sequences were obtained and analysed 50 51 using Bayesian and Maximum Likelihood methods and networks to infer molecular diversity, 52 demographic features and phylogeographic patterns. Correlations with geological and ecological 53 54 features of the islands were performed to relate these variables with genetic variability. 55 Results The total picture of the phylogenetic relationships of the Cape Verdian Tarentola was presented 56 57 for the first time, the positions of new forms revealed and some rearrangements of previously 58 hypothesised relationships. Contrary to what was expected and despite the large sample size, low 59 60 intraspecific diversity was found using the cyt b gene fragment. Star-like haplotype networks and statistical tests suggested the past occurrence of a rapid demographic and geographical expansion over Journal of Biogeography Page 2 of 30

2 1 2 3 most of the islands examined. Size, altitude and habitat availability of the islands are positively 4 5 correlated with genetic variability . 6 Main Conclusions Biogeographic patterns have in general a high concordance with phylogenetic breaks 7 and with the three eco-geographical island groups. Volcanism, habitat availability, intimately linked with 8 9 orography and present and historical size of the islands appear to be the main factors explaining the 10 genetic diversity of Tarentola from the Cape Verdes. 11 12 Keywords 13 Tarentola , Cape Verde Islands, cyt b, 12S, phylogeny, biogeography, island radiation 14 15 16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 3 of 30 Journal of Biogeography

3 1 2 3 INTRODUCTION 4 Islands are ideal model systems to study evolution and phylogeography, and through time have become 5 6 known as “natural laboratories” for this purpose (Williamson, 1981). Oceanic archipelagos, such as the 7 Galapagos, Hawaii, Madeira and , permit testing of various evolutionary hypotheses (Losos & 8 9 Ricklefs, 2009). Gene flow between the island and the mainland or among islands may be practically non- 10 existent, allowing fixation of genetic variation and so differentiation of populations can occur by geographical 11 isolation. The sea depth isolates each island into a discrete evolutionary unit (Emerson, 2002). Once the ages 12 13 of the islands are known, phylogeography of taxa in archipelagos can be analysed within a known timeframe. 14 Additionally, island populations have higher risk of extinction than mainland populations (Frankham, 1997); 15 even though they generally present a depauperate number of relative to similar mainland areas, the 16 17 number of endemic species is usually very high (Whittaker, 1998), as is the risk of introduction of exotic 18 species (World Conservation Monitoring Centre, 1992). As a result, investigation and protection of endemic 19 island species is particularly important. 20 For Peer Review 21 The Cape Verde Islands are an oceanic archipelago located approximately 500-800 km from the West 22 African coast. The archipelago is volcanic (the last eruption occurred on Fogo in 1995), and has never been 23 connected to the mainland (Mitchell-Thomé, 1976). It has ten main islands, plus several islets, that are 24 25 politically separated in Windward and Leeward Islands but eco- and topologically divided in north-western, 26 eastern and southern islands (Fig. 1). They are arranged in a horse-shoe shape with its concavity facing 27 28 westwards. The islands are between 6 to 26 million years old (My) and the youngest ones are on the tips of 29 the arc (see Griffiths et al. , 1975; Grunau et al. , 1975; Stillman et al. , 1982; Mitchell et al. , 1983; Carracedo, 30 1999; Torres et al. , 2002; Plesner et al. , 2002). During sea level fluctuations in the Pleistocene (1.6 My) some 31 32 of the islands of the north-western group (S. Vicente, Sta. Luzia, Branco and Raso islets) were most likely 33 linked, and possibly also Boavista with Maio. Elsewhere the water channels are very deep so it is highly 34 improbable that the other islands were ever connected by land (Morris, 1998). Their sizes and topographies 35 36 vary dramatically; Santiago is the largest (around 1,000 km 2) and Raso islet (<6 km 2) among the smallest, 37 Fogo is the highest (near 2,800 m) and Santa Maria islet the flattest. 38 The fossil record on volcanic islands is generally scarce, thus in most cases no direct evidence of past 39 40 evolutionary events is available. However, organisms may show evidence of past population decline and 41 growth in their genetic variation (Beheregaray et al. , 2003). In this way can be considered model 42 organisms as they are non-volant, decreasing migration importance in the evolutionary process, are easy to 43 44 handle in the field and are present in all the Cape Verde islands. 45 Although Darwin, during his voyage on the Beagle, considered the Cape Verde islands to be “utterly 46 47 sterile” (Darwin, 1845), he noticed the in the more humid valleys. There are approximately 27 48 currently recognized native taxa, all endemic to the Cape Verdes (100% endemicity), which are divided 49 into three genera: the skinks Mabuya and the geckos Tarentola and Hemidactylus . The genus Tarentola is 50 51 biogeographically interesting because it arrived to the Cape Verde islands approximately seven million years 52 ago, from a propagule that dispersed from the western Canary Islands situated 1500 Km to the north 53 (Carranza et al. , 2000). The endemic Tarentola have been previously studied phylogenetically (Carranza et al. , 54 55 2000, 2002; Jesus et al. , 2002) and estimates of relationships indicated possible cryptic species and paraphyly 56 of some species. However, in all these studies not all the islands of the archipelago where included and 57 therefore not all taxa. Moreover, those studies were based on a very small number of samples per taxa and 58 59 island, studying only the deep divergent lineages and not assessing intraspecific variation. Such information is 60 valuable, as Tarentola on the Canary Islands show considerable intraspecific variation, possibly associated with island sizes, volcanic activity, ecological niche availability, or a combination of these factors (González et Journal of Biogeography Page 4 of 30

4 1 2 3 al. , 1996; Juan et al ., 2000; Gübitz et al. , 2005). Assessing intraspecific variation may uncover additional 4 cryptic lineages, and is highly relevant for any conservation assessment (Schwartz et al. , 2006; Kahindo et al. , 5 6 2007). 7 8 9 The aim of this study was to reassess relationships of Tarentola from the Cape Verde islands by including 10 specimens from all islands, including distinct subspecies never previously analysed genetically. At the same 11 time by sequencing over 400 new specimens, variation within forms can for the first time be adequately 12 13 determined. Our expectation is that this more complete sampling will resolve some of the issues raised in 14 earlier phylogenetic works, while it decreases the possibility that any cryptic forms have been overlooked. 15 Extensive within island sampling will also shed light on possible barriers to gene flow within species and its 16 17 population genetics and these can be compared to the age, geological and ecological features of the islands 18 including size, altitude and habitat availability. 19

20 For Peer Review 21 MATERIALS AND METHODS 22 23 Sampling and gathering of molecular data set 24 25 The ten islands of the Cape Verde archipelago were prospected between 2006 and 2008, during mid May to 26 mid July. The sampling stations were chosen randomly and stratified according to the different habitats 27 28 existing in each island, based on the agro-ecological and vegetation zoning maps (Diniz & Matos, 1986, 1987, 29 1988 a, b, 1993, 1994, 1999 a, b, c) and the number of sites per habitat according to its area. In this way, most 30 of the variability between and within each habitat – altitude, vegetation, climate and geographic position – 31 32 was contained in the different sampling stations. The sampled area, 440 stations of 1x1 squared kilometres, 33 corresponds to around 11% of the country area. Each station was sampled along transects for 35 minutes on 34 average, depending on the difficulty of the terrain, by two observers walking parallel to each other, totalling 35 36 more than 263 hours of sampling. 37 A total of 405 new specimens of Tarentola were included in the genetic analyses. Specimens were 38 identified in the field using discriminant characters published by Joger (1984, 1993) and Schleich (1987), 39 40 digital photographs taken, and a piece of tail was removed and stored in 100% ethanol. Sampled were 41 released immediately afterwards. Identification codes, localities and GenBank accession numbers of the 42 samples used are listed in Table 1. 43 44 Total genomic DNA was extracted from small pieces of tail using standard methods, following Harris et al. 45 (1998). Polymerase Chain Reaction (PCR) primers used in amplification and sequencing were 12Sa and 12Sb 46 47 for the 12S rRNA, and cyt b 1 and cyt b 2 (Kocher et al. , 1989) and cyt b 2F and CB3-3’ (Palumbi, 1996) for the 48 cytochrome (cyt b ) gene. Thermocycling consisted of an initial 3 min at 95 ºC followed by 35 cycles of 30 s at 49 95 ºC, 50 ºC and 72 ºC and then a single cycle of 7 min at 72 ºC. Amplified mitochondrial ( mt DNA) fragments 50 51 were sequenced from both strands on a 3100 Applied Biosystems DNA Sequencing Apparatus. 52 The two fragments of the cyt b gene (684 base pairs, bp) plus the 12S rRNA (403 bp) were used for the 53 phylogenetic analyses, totalling 1087 bp. Once all major lineages had been identified through the phylogenetic 54 55 analysis, intraspecific variation was assessed using the first fragment of the cyt b gene (cyt b 1 and cyt b 2 56 primers – 303 bp) for the network analysis. 57

58 59 Data analyses 60 Phylogenetic analyses Page 5 of 30 Journal of Biogeography

5 1 2 3 DNA sequences were aligned using ClustalX (Thompson et al., 1997) with default parameters (gap opening = 4 10; gap extension = 0.2). All the cyt b sequences had the same length and therefore no gaps were postulated. 5 6 These sequences were translated into amino acids using the vertebrate mitochondrial code and no stop 7 codons were observed, suggesting that they were probably all functional. Although some gaps were 8 9 postulated in order to resolve length differences in the 12S rRNA gene fragment, all positions could be 10 unambiguously aligned and were therefore included in the analyses. 11 Two methods of phylogenetic analysis were employed for the two independent partitions (cyt b and 12S) 12 13 and the combined data set and their results compared. These were: Maximum likelihood (ML) and Bayesian 14 analysis (BI). Modeltest v. 0.1.1 (Posada, 2008) was used to select the most appropriate model of sequence 15 evolution for the ML and BI of the independent partitions and the combined data sets, under the Akaike 16 17 Information Criterion. The models selected were: GTR+I+G for the cyt b partition; GTR+G for the 12S rRNA 18 partition; and GTR+I+G for the cyt b+12S combined data set. 19 Bayesian analyses were performed with MrBayes v. 3.0b4 (Huelsenbeck & Ronquist, 2001). For the 20 For Peer Review 21 combined data set, each gene had its own independent model of evolution (see above) and model parameters. 22 Four incrementally heated Markov chains with the default heating values were used. All analyses started with 23 randomly generated trees and ran for 2x10 6 generations, with sampling occurring at intervals of 100 24 25 generations producing 20,000 trees. For each independent analysis, the log-likelihood values of all trees saved 26 were plotted against the generation time. After verifying that stationarity had been reached both in terms of 27 28 likelihood scores and parameter estimation, the first 4,000 trees in the cyt b+12S data set were discarded and 29 independent majority rule consensus trees were generated from the remaining (post-“burn-in”) trees. The 30 frequency of any particular clade of the consensus tree represents the posterior probability of that node 31 32 (Huelsenbeck & Ronquist, 2001); only values above 95% were considered to indicate that nodes were 33 significantly supported (Wilcox et al ., 2002). 34 ML analyses were performed in PHYML (Guindon & Gascuel, 2003) with model parameters fitted to the 35 36 data by likelihood maximization. Reliability of the ML trees was assessed by bootstrap analysis (Felstenstein, 37 1985), involving 1000 replications. 38 Topological incongruence among partitions was tested using the incongruence length difference (ILD) 39 40 test (Michkevich & Farris, 1981; Farris et al ., 1994). In this test, 10,000 heuristic searches were carried out 41 after removing all invariable characters from the data set (Cunningham, 1997). To test for incongruence 42 among data sets a reciprocal 70% bootstrap proportion was also used (Mason-Gamer & Kellogg, 1996) or a 43 44 95% posterior probability threshold. Topological conflicts were considered significant if two different 45 relationships for the same set of taxa were both supported with bootstrap values ≥ 70% or posterior 46 47 probability values ≥ 95%. 48 Topological constraints to test alternative topologies were constructed using MacClade v. 4.0 (Maddison 49 & Maddison, 1992) and compared to optimal topologies using the Shimodaira-Hassegawa (SH) (Shimodaira & 50 51 Hasegawa, 1999) test implemented in PAUP*4.0b10 (Swofford, 1998). 52 53 Population genetics, demographic analyses and correlations 54 55 Network approaches are more effective than classical phylogenetic ones for representing intraspecific 56 evolution (Posada & Crandall, 2001). Therefore, the genealogical relationships among groups were assessed 57 with haplotypes networks constructed using statistical parsimony (Templeton et al ., 1992), as implemented in 58 59 the program TCS v1.21 (Clement et al. 2000) with a connection limit of 95%. Genetic differentiation between 60 populations belonging to the same network was calculated using the Snn statistic (Hudson, 2000) implemented in the program DnaSP v. 5 (Rozas et al ., 2003) (Appendix S1 in Supporting Information). Journal of Biogeography Page 6 of 30

6 1 2 3 Independent networks and those island populations which were part of a network but presented significant 4 Snn values were considered distinct Evolutionarily Significant Units (ESU), following Fraser & Bernatchez 5 6 (2001). 7 Haplotye (Hd) and nucleotide diversity (Pi) values, number of haplotypes (h) and segregation sites (S) 8 9 were also calculated with DnaSP v. 5 (Table 2). To test for the hypothesis of a rapid expansion and to estimate 10 the time since expansion a series of analyses were carried out. First the Fu’s Fs statistic (Fu, 1997) was 11 calculated to test for deviations from the neutral Wright-Fisher model consistent with a population expansion 12 13 under neutrality hypothesis, using coalescent simulations in DnaSP (based on the segregating sites and 14 assuming no recombination, with 10,000 replicates and 0.95% as confidence interval). To further characterize 15 expansion ARLEQUIN version 3.1 (Excoffier et al ., 2005) was used to determine the historical demography of 16 17 the population using mismatch distributions with the models of Rogers & Harpending (1992) and Rogers 18 (1995). Recent growth is expected to generate a unimodal distribution of pairwise differences between 19 sequences (Rogers & Harpending, 1992). The distribution is compared with that expected under a model of 20 For Peer Review 21 population expansion (Rogers, 1995), calculating the estimator expansion time (τ) and the initial and final θ 22 (θ0 and θ1, respectively), according to Schneider & Excoffier (1999). The fit of the mismatch distribution to 23 the theoretical distribution under an expansion scenario was assessed by Montecarlo simulations of 1,000 24 25 random samples. The sum of squared deviation between observed and expected mismatch distributions was 26 used as a test statistic and its p value represents the probability of obtaining a simulated sum of squared 27 28 deviations (SSD) larger than or equal to the observed one (Table 2). The τ parameter is an estimate of time 29 after expansion ( t) in mutational units. If the divergence rate per nucleotide and year (τ = 2 , were is the 30 substitution rate per lineage) and the number of nucleotides of the fragment analysed (l) are known, it is 31 32 possible then to calculate the age when the expansion occurred using the expression τ = l t, modified from 33 Harpending et al ., (1993). 34 For establishing comparisons between the number of sequences (N), number of haplotypes (h), ESU and 35 36 geographical and ecological characteristics of the islands: size (area, perimeter), altitudes (maxima, medium 37 and median), location (latitude and longitude of the centroid of the island) and habitat availability (number of 38 39 habitats) (Appendix S2 in Supporting Information), correlations were calculated with the Spearman rank 40 correlation method (Table 3), which is less sensitive to outliers. Correlations between these variables were 41 considered if Spearman rank correlation coefficient (ρ) was ≥0.60 and p<0.05 and calculated using the JMP 42 43 package (SAS Institute, Cary, North Carolina, USA). The geographical variables were obtained using a 44 Geographic Information System (GIS) ArcMap 9.0 (ESRI, 2004) with altitudes being derived from a digital 45 elevation model (Jarvis et al ., 2006) and ecological adapting the information available on the agro-ecological 46 47 and vegetation zoning maps (Diniz & Matos, 1986, 1987, 1988 a, b, 1993, 1994, 1999 a, b, c). 48 49 Estimation of divergence times and mutation rate ( ) 50 51 In order to estimate divergence times between lineages, the computer program r8sb v1.6.4 was used 52 (Sanderson, 1997, 2002). The outgroup sequence ( Tarentola americana (Gray, 1831)) was eliminated from 53 the tree previous to the analysis. Smoothing of rate variation along the tree was performed with the Langley & 54 55 Fitch (1974) and penalized likelihood (Sanderson, 2002) methods. Sixteen smoothing factors with log 10 from - 56 2 to 5.5 were used for the penalized likelihood method. The lowest χ2 cross-validation score, as calculated by 57 58 r8s, was used to select the best method. In order to account for the error involved in the calibration of the 59 Tarentola phylogeny a parametric bootstrap analysis was performed in which 1,000 Monte Carlo simulations 60 of alignments of the same length than the complete data sets were generated with Seq-Gen (Rambaut & Grassly, 1997) using the phylogenetic tree and model parameters previously obtained. This allowed Page 7 of 30 Journal of Biogeography

7 1 2 3 evaluating the stochastic errors of date estimates associated to sampling a finite number of base pairs 4 (Sanderson & Doyle 2001; Lalueza-Fox et al . 2005). 5 6 To estimate absolute rates, two calibration points were used. The first one was based on the assumption 7 that divergence between Tarentola boettgeri hierrensis Joger and Bischoff, 1983 (endemic to the island of El 8 9 Hierro) and Tarentola boettgeri bischoffi Joger, 1984 (endemic to the Selvagens Islands) began approximately 10 1 Ma, soon after El Hierro was formed, and rapid colonisation from Selvagens by the ancestor of Tarentola 11 boettgeri hierrensis occurred (see Carranza et al . 2000). The second calibration point was based in the 12 13 assumption that Tarentola delalandii Duméril & Bibron, 1836 from north Tenerife colonised the island of La 14 Palma 2 Ma, soon after this island was formed (Ancochea et al ., 1994; Gübitz et al ., 2000). 15 To test if the calibrations were appropriate, the inferred arrival dates of endemic Cape Verde Tarentola to 16 17 particular islands were checked to see if these were congruent with the origins of the islands themselves, 18 whenever these were known. In order to infer the average mutation rate ( ) for the genus Tarentola the ML 19 phylogenetic tree from Fig. 2 was also used with the same calibration points as stated above but using the 20 For Peer Review 21 Langley-Fitch algorithm. 22 23 24 RESULTS 25 26 Phylogenetic analyses 27 28 Independent ML and BI analyses of the two gene fragments (cyt b and 12S) produced trees that differed in 29 some minor arrangements of taxa or individual samples. In all cases, these differences had low bootstrap and 30 posterior-probability support, the respective values being less than 70% and less than 95%, respectively. 31 32 Consequently, it was considered that there were no major topological conflicts between the two gene- 33 partitions (Mason-Gamer & Kellogg, 1996). The ILD test ( P > 0.60) showed the two independent data sets 34 were not incongruent, and therefore it was decided to combine them for further analyses. In total, the 35 36 combined data set included 1087 bp (684 bp of cyt b and 403 bp of 12S rRNA). Of these, 674 positions were 37 variable and 637 parsimony-informative, the respective numbers for each gene partition being as follows: 522 38 and 515 for cyt b and 152 and 122 for 12S rRNA. 39 40 The results of the ML and BI phylogenetic analyses of the combined cyt b + 12S rRNA data sets are shown 41 in Fig. 2 and support the hypothesis that Tarentola from the Cape Verde archipelago is a monophyletic group 42 43 that originated by a single transoceanic dispersal event from the Canary Islands. 44 The combined tree of the ML and BI analyses shows four major groups (see Fig. 2): a) the T. darwini Joger, 45 1984 – T. “rudis” boavistensis Joger, 1993 group, not well supported; b) the T. caboverdiana Schleich, 1984 46 47 group with subspecies from S. Vicente ( T. c. sustituta Joger, 1984), Santa Luzia and Raso islet ( T. c. raziana 48 Schleich, 1984) and Santo Antão ( T. c. caboverdiana Schleich, 1984); c) the T. caboverdiana nicolauensis 49 Schleich, 1984 form group and d) the T. gigas (Bocage, 1896) – T. rudis Boulenger, 1906 group; all the last 50 51 three very well supported. 52 The phylogeny indicates that T. rudis is polyphyletic, with T. “rudis” boavistensis , endemic to the island of 53 Boavista, being more closely related to T. darwini from S. Nicolau, Fogo and Santiago than to the remaining 54 55 species of T. rudis . To test this result further, the log likelihood of the ML tree presented in Fig. 2 (-6468.896) 56 was compared with the log likelihood of a ML tree constrained so that T. rudis was monophyletic (- 57 6501.70948). The results of the SH test showed that the constrained tree had a significantly worse log 58 59 likelihood value than the unconstrained solution (Dif –ln L = 32.81265; p<0.005) and hence the tree from Fig. 60 2, where T. rudis is polyphyletic, is consequently preferred. Journal of Biogeography Page 8 of 30

8 1 2 3 Within group ‘a’, the three lineages of T. darwini are very deep, indicating that populations from Fogo, S. 4 Nicolau and Santiago have been evolving in isolation for several My. Thus, the bootstrap and pp values that 5 6 support the monophyletic status of T. darwini are very low (Fig. 2). 7 Tarentola caboverdiana , T. gigas and T. rudis form a very well supported clade, which is sister to group ‘a’ 8 9 formed by T. darwini and T. “rudis” boavistensis. The phylogenetic analyses show that T. caboverdiana is 10 paraphyletic, with the subspecies from S. Nicolau (T. c. nicolauensis ) (group ‘c’) being more closely related to 11 T. rudis and T. gigas (group ‘d’) than to the remaining subspecies of T. caboverdiana ( T. c. caboverdiana, T. c. 12 13 raziana and T. c. substituta – group ‘b’). However, the results of the SH test showed that the log likelihood of 14 the constrained tree in which T. caboverdiana was forced monophyletic (-6474.343) was not significantly 15 worse than the “best” tree presented in Fig. 2 (Dif –ln L = 5.446; P > 0.40). The three subspecies of T. 16 17 caboverdiana from group ‘b’ form a robust monophyletic assemblage that is further subdivided into the 18 population from Santo Antão ( T. c. caboverdiana ) and the populations from S. Vicente ( T. c. substituta ), S. 19 Luzia, Raso and Branco ( T. c. raziana ). 20 For Peer Review 21 Within group ‘d’, T. gigas and T. rudis from the southern islands form a very well supported clade. 22 Tarentola gigas appears in the phylogeny as sister to T. r. rudis from Santiago, although the support for this 23 assemblage is extremely low. A constraint analysis in which T. rudis from the southern islands was forced 24 25 monophyletic produced a tree with a log likelihood almost identical to the log likelihood of the unconstrained 26 tree presented in Fig. 2 (Dif –ln L = 0.314; P > 0.79), indicating that the apparent paraphyletic status of T. rudis 27 28 recovered in Fig. 2 is not well supported by our data. It is also shown that T. rudis protogigas Joger, 1984 29 populations from the southern islands of Fogo, Brava and Rombos islets form a clade apart from T. r. 30 maioensis Schleich, 1984 from Maio. 31 32 33 Population genetics and demographic analyses 34 A fragment of 303 bp of the cyt b gene was analysed for a total of 459 sequences (405 new plus 54 from 35 36 GenBank) of Tarentola , corresponding to 276 localities from nine islands and three islets across the Cape 37 Verde archipelago. Over the whole data set 105 polymorphic sites and 120 haplotypes were identified. Based 38 on the connection limit of 95%, eight independent networks could be inferred. The phylogenetic lineages 39 40 leading to those independent networks have been highlighted in Fig. 2 and the networks are shown in Fig. 3. 41 The significant Snn comparisons tests (Appendix S1) indicate that populations from Santiago (a4), the three 42 island populations of T. caboverdiana from network b (b1, b2 and b3), and four populations from network d 43 44 (d1, d2, d3 and d4) are genetically differentiated and should be considered as independent units in the 45 demographic analyses (see below). Therefore, the mt DNA analyses highlighted the existence of 14 46 47 independent genetic units in Tarentola from the Cape Verde archipelago, which are considered here as ESU. 48 The number of individuals sampled (N), number of haplotypes (h), nucleotide diversity (Pi), haplotypes 49 diversity (Hd), segregation sites (S) and other relevant data for each of the 14 differentiated ESU are given in 50 51 Table 2. 52 As expected from the star-like topologies of some of the networks, in all the 14 differentiated populations 53 identified in Fig. 3, seven cases (a2, a3, a4 South, b1, b2, b3 and c) were detected in which the Fu’s test was 54 55 significantly negative indicating, under a neutral model, that these populations could have experienced a 56 demographic expansion event. To characterize the expansion pattern further, a model of sudden demographic 57 growth was fitted to the pairwise sequence mismatch distribution of the seven populations. In six of these 58 59 cases, the mismatch distributions were not significantly different from the sudden expansion model of Rogers 60 and Harpending (1992). The results of the Fu’s test ( Fs ), the squared deviation statistic (SSD) and other relevant demographic parameters are given in Table 2. The mutation rate inferred from the ML tree using r8s Page 9 of 30 Journal of Biogeography

9 1 2 3 (see materials and methods) was 3.7 x 10 -8 per site and, assuming a generation time of 1 year for Cape Verdian 4 Tarentola (pers. obs.), we estimated the approximate onset of expansion for the six populations (Table 2). 5 6 Correlations between the genetic variability parameters and the geographical characteristics of the 7 islands showed that the number of haplotypes (h) present in each island and the habitat availability are 8 9 strongly and positively correlated with the island area and maximum, mean and median altitudes but not with 10 the latitudinal or longitudinal location of the island (Table 3). The analyses also showed that the number of 11 samples was strongly and positively correlated with the same variables and also with the island perimeter 12 13 and habitat availability, highlighting that the sampling effort was proportional to island size and ecological 14 heterogeneity. The number of habitats was also positively and significantly correlated with the number of 15 haplotypes present in the islands. 16 17 18 19 DISCUSSION 20 For Peer Review 21 22 Phylogeography of Tarentola from the Capeverdes 23 Mitochondrial markers are widely used for phylogeographic studies for various reasons including laboratory 24 25 technique simplicity, the general fast rate of evolution and lack of recombination when compared with nuclear 26 markers (Hare, 2001). Moreover, the effective population size for mitochondrial genes is four times lower 27 28 than for nuclear markers, so any genetic drift effects are emphasized and isolation by distance patterns should 29 be easier to detect (Díaz-Almela et al. , 2004). Additionally, as it is often used, comparative analyses can be 30 carried out, such as in the present work. 31 32 In Cape Verdian Tarentola , a well supported monophyletic group, more than one species is found in some 33 of the islands (Schleich, 1987) and some species have paraphyletic assemblages. For example, Carranza et al. 34 (2000) indicated that the ‘ T. darwini ’ individual from S. Nicolau produced an unsolved trichotomy between 35 36 the mtDNA lineage of T. darwini from Fogo, Santiago and T. “rudis” boavistensis from Boavista. Also that this 37 latter subspecies was most closely related to populations assigned to T. darwini than T. rudis , making T. rudis 38 polyphyletic. Both examples pointed to the need of revising the systematics of the group by increasing the 39 40 sampling range and size. In this study, three new forms were included in the analyses: two new taxa, T. rudis 41 maioensis and T. caboverdiana caboverdiana , endemic subspecies from Maio and Santo Antão, respectively and 42 T. rudis protogigas , which also occurs on Brava and Rombos islets, from a previously unsampled island, Fogo. 43 44 With the addition of this new data, the phylogenetic tree presented in Fig. 2 includes now representatives of 45 all known taxa from all the islands where this genus occurs. The estimate of phylogeny of the Cape Verdian 46 47 Tarentola was slightly different from previous studies. Most branches are now better supported and the 48 positions of the three new forms revealed, making it possible to assess for the first time the complete picture 49 of this group. Knowledge of a complete and robust phylogeny for the Cape Verdian Tarentola is essential for 50 51 future conservation of these endemic geckos as it defines the ESU to be protected in the projected protected 52 areas. 53 It is clear that the T. caboverdiana nicolauensis form that before appeared as sister taxa to T. c. raziana 54 55 and T. c. substituta , T. rudis and T. gigas is unrelated with the other T. caboverdiana present on the north- 56 western group in Santo Antão ( T. c. caboverdiana ), S. Vicente ( T. c. substituta ), Santa Luzia and Raso islet ( T. c. 57 raziana ) which form a well supported group. Currently T. c. nicolauensis appears to be most closed related 58 59 with the T. rudis – T. gigas complex. It is here referred to as a ‘complex’ as T. gigas does not appear as basal 60 taxa of T. rudis as in Carranza et al. (2000), but as a sister group of T. r. rudis from Santiago, although the bootstrap and posterior probability values are very low and the constraint analyses does not reject the Journal of Biogeography Page 10 of 30

10 1 2 3 alternative hypothesis of monophyly of southern T. rudis (see results). It is also apparent that T. rudis 4 protogigas populations from the southern islands of Fogo, Brava and Rombos islets form a monophyletic 5 6 group with a good support and that T. r. maioensis from Maio, belonging to the southern group but ecologically 7 and geologically closer to the eastern group, forms another clade that is weekly supported as sister taxa of this 8 9 last. 10 In addition, it is revealed that the detection by Jesus et al. (2002) of “ T. gigas” in S. Nicolau, was a 11 misinterpretation due to the previous lack of samples from Maio Island. As new samples from Maio were 12 13 included in the present analysis, it became clear that it is in fact a T. r. maioensis specimen. This highlights the 14 importance of including all known lineages of a group and samples from the entire range of species in 15 phylogenetic analyses. 16 17 In this group, it is also strongly evident the agreement of the phylogenetic structure within the different 18 clades with the three eco-geographical regions of the archipelago. In this way, the ‘a’ group is subdivided into 19 three units, each one assigned to one of the eastern, north-western and southern regions (see Fig. 1). It is also 20 For Peer Review 21 important to notice that some Tarentola species are exclusive of one of these regions, such as T. caboverdiana 22 that only appears in the north-western islands group and T. rudis that is present in all southern islands, as 23 what was also found in endemic Hemidactylus and Mabuya , two other another radiations of endemic reptiles 24 25 of the archipelago (Carranza et al. , 2001; Brown et al ., 2001; Arnold et al. , 2008). 26 According to the new phylogenetic hypothesis and inferred dates (see Fig. 2), Tarentola colonised the 27 28 Cape Verde archipelago from the western Canary Islands between 7.73 – 5.99 Ma. The most parsimonious 29 explanation is that the first island to be colonised was S. Nicolau, which is part of the north-western island 30 group (Fig. 1). As S. Nicolau consisted of two independent units up to 4.7 – 2.6 Ma, when these were finally 31 32 united by volcanic activity (Dupart et al ., 2007), it is hypothesized here that the first speciation event that 33 separated the ancestor of group ‘a’ ( T. darwini + T. “rudis” boavistensis ) and the ancestor of groups ‘b’, ‘c’ and 34 ‘d’ ( T. caboverdiana + T. rudis from the southern islands, plus T. gigas from Raso and Branco) approximately 35 36 5.99 ± 1.6 Ma took place by allopatric speciation in this island. From there, the ancestor of T. caboverdiana 37 colonised all the remaining north-western islands, while the ancestor of group ‘a’ colonised the eastern island 38 of Boavista and the southern islands of Santiago and Fogo. The topology presented in Fig. 2 also suggest that 39 40 another colonisation event took place from S. Nicolau to Branco, Raso or Santa Luzia between 3.49 ± 1.2 and 41 2.53 ± 0.9 Ma, giving rise to T. gigas , which nowadays only survives in the islets of Branco and Raso, where it 42 43 coexists with the much smaller T. caboverdiana . Also from the north, arrived at approximately 2.53 ± 0.9 Ma 44 the ancestor of the three subspecies of T. rudis present in all the southern islands. Other movements between 45 islands have happened in more recent times but most probably are the result of human introductions and are 46 47 discussed below. 48 49 Distribution of the genetic diversity 50 51 As the network analyses from the group ‘a’ showed that not even the populations considered belonging to the 52 same species, T. darwini , could be linked together, it can be concluded following Hart & Sunday (2007) that 53 54 cryptic species probably have been overlooked. Geckos are thought to be morphologically more conservative 55 than most other reptiles (Harris et al. , 2004). Supporting this, Jesus et al. (2002) when assessing mt DNA 56 variation in Tarentola from the Cape Verde islands had already concluded that it was high between species 57 58 relative to other reptiles from the same islands, such as the endemic Mabuya skinks (Brehm et al. , 2001). The 59 networks also highlighted the presence of low haplotypic diversity within Tarentola from Boavista. 60 The extensive sampling allowed confirmation of the occurrence of several individuals of the very distinct ‘T. darwini ’ form in S. Nicolau reassuring that a population is indeed present after a unique individual was Page 11 of 30 Journal of Biogeography

11 1 2 3 reported by Carranza et al. (2000). Joger (1984) first reported the finding of T. darwini on S. Nicolau but this 4 was considered doubtful by Schleich (1987) as Joger himself considered two of three animals found as 5 6 doubtfully assigned. The present analysis confirms that indeed a different form exists on the island, but that it 7 is very distinct to other T. darwini , and probably represents a new species. Extensive sampling also identified 8 9 its distribution, for the first time, which is restricted to the eastern part of the island (see Fig. 1). Additionally, 10 based on the present results, several forms considered as subspecies should be raised to species status, 11 however, further genetic studies (nuclear markers) and ongoing detailed morphological analyses will be 12 13 addressed in a near future to confirm these data. 14 As explained above, the presence of the two allopatric Tarentola species in S. Nicolau, T. darwini and T. 15 caboverdiana , can be explained by allopatric speciation. However, the presence of two Tarentola species in 16 17 Santiago, T. rudis and T. darwini , can be explained by two independent colonisation events from the North, 18 following the direction of the main currents and trade winds. Future GIS modelling of the species distribution 19 might shed light on which factors may explain the current range of different species on the same island. These 20 For Peer Review 21 two species from Santiago are both morphologically and genetically distinct and occurr in sympatry in the 22 south of the island. This fact was first noticed by Schleich (1987) and only confirmed here, refuting parapatry 23 referred by Joger (1984). It is important to notice that of the 149 individuals sequenced, nine of them, 24 25 assigned to T. rudis based on morphology, presented T. darwini type mt DNA. This implies that limited 26 hybridization may be occurring, and that the movement of mtDNA across the species boundary may be 27 28 unidirectional. However, more detailed analyses of nuclear markers and morphological characters will be 29 needed to confirm this, especially since these two taxa diverged approximately 6 Ma (see Fig. 2). 30 Within T. darwini , examination of the networks shows little evidence of structuring within islands, except 31 32 on Santiago. Here there are two well geographically delimited subgroups, one in the North and another one in 33 the South of the Island (a4 north and a4 south in Fig. 2 and 3) that appear genetically differentiated according 34 to the Snn test (see Appendix S1) and are considered here independent ESU. 35 36 Based on the network analysis, the presence of T. r. maioensis in S. Nicolau is possibly due to an 37 introduction as it presents a haplotype only one mutational step away from the more common one found in 38 Maio (Fig. 3d2). Besides that, after an extensive sampling no other individual of that taxon was found on that 39 40 island and additionally it was found on the coast, Ponta Cachorro, and not inland. Analogously, the two 41 individuals of T. c. nicolauensis from S. Vicente cited by Jesus et al. (2002) also seem to be the result of recent 42 introductions, as they present common haplotypes with the ones found in S. Nicolau (Fig. 3c) and because 43 44 they were found in , which has a major port from where boats periodically reach S. Nicolau. This 45 approach also pointed to another possible introduction of T. c. substituta (endemic of S. Vicente) in S. Antão, 46 47 , a village of fishermen that regularly fish in other islands (Fig. 3b). However, in this later case, the 48 presence of a common haplotype between T. c. substituta and T. c. caboverdiana (Fig. 3b) can also be explained 49 by the fact that S. Antão and its neighbour islands (S. Vicente, S. Luzia, Raso and Branco islets), which were 50 51 connected during the Pleistocene see level falls, were only some few kilometres apart at that time. This could 52 have allowed gene flow between the populations of the north-western group. This is further shown in the 53 higher number of connections between the star-like networks of each island population and the low number 54 55 of mutational steps between them, contrasting with the other Tarentola forms. Geckos are often introduced 56 from one island to another, as for instance the introduced T. mauritanica from Madeira Island to Porto Santo, 57 belonging to same archipelago (Jesus et al. , 2008) or the two independent introductions of Hemidactulus 58 59 angulatus in Cape Verde from two different African sources (Arnold et al ., 2008). Even islands endemics can 60 be introduced to the mainland, as in the case of T. delalandii , from the Canaries to Cantabria (Gómez, 2006). Journal of Biogeography Page 12 of 30

12 1 2 3 The network analyses further show that T. rudis from the southern islands group, together with T. gigas 4 from Branco and Raso islets, form a well supported unit, as previously reported by Carranza et al. (2000) with 5 6 well separated monophyletic lineages in each island (Fig. 3d1). The only populations with shared haplotypes 7 in group ‘d’ are the ones belonging to islets of the same islet group, as between Branco and Raso, or between 8 9 these and the main island to which they belong, as between Rombos islets and Brava and Santa Maria islet and 10 Santiago. This is explained by the fact that indeed these southern islands were never connected and that gene 11 flow between them, if it occurs, seems to be extremely low. 12 13 14 Biogeographic patterns 15 In Tenerife and Gran Canaria, from the Canary Islands, deep molecular divergences between lineages 16 17 of the same island have been reported (e.g. Chalcides sexlineatus and C. viridanus : Pestano & Brown, 1999, 18 Brown et al ., 2000, Carranza et al ., 2008; Tarentola delalandii and T. boettgeri : Nogales et al ., 1998, Gübitz et 19 al ., 2005; Gallotia galloti and G. intermedia/G. goliath : Thorpe et al ., 1996; González et al ., 1996; Maca-Meyer, 20 For Peer Review 21 2003). The main reasons suggested to explain this pattern are geographical or ecological isolation, i.e., 22 multiple geological origins and marked ecological differences between regions of the islands that enhanced 23 opportunities to evolve allopatrically (Juan et al ., 2000, Thorpe & Malhotra, 1996 ). In the same way, 24 25 homogeneity at the molecular level on the smaller islands of Fuerteventura, Lobos and Lanzarote was 26 explained by the absence of geographical barriers and ecological similarity within these islands (Nogales et al. , 27 28 1996). So, since the Cape Verde archipelago belongs to the same biogeographical region and presents islands 29 of different sizes, a similar pattern of divergent new lineages following the extensive sampling would be 30 expected for the larger and more mountainous islands, as it has been demonstrated that both area and 31 32 altitude positively affect speciation rates (Rosenzweig, 1995; Hobohm, 2000, respectively) Moreover, this 33 archipelago is even older that the Canaries, in which the oldest island is thought to have formed 22 Ma (Abdel- 34 Monem, 1971), compared to the inferred age of 26 My for the eastern islands of Cape Verde. However, within 35 36 the same form, the presence of two and not very deep mitochondrial lineages was noticed only on Santiago. 37 Furthermore, half of the median-joining networks revealed a ‘star-like’ haplotype network (Fig. 3) and 38 presented strongly negative Fs (Fu, 1997) values, indicating that rapid recent expansions (Slatkin & Hudson, 39 40 1991) previewed by strong bottlenecks occurred all over the archipelago (see Table 2). Demographic analyses 41 presented in Table 2 further demonstrate that five out of the six expansion events inferred from our data set 42 43 occurred between 114,000 and 168,000 years ago, although the poor knowledge of the past geological history 44 of the Cape Verdes does not allow us to know the exact event that might have produced the expansions. One 45 possible explanation is that these expansions occurred after volcanic eruptions that decimated the faunas. In 46 47 fact volcanism younger than 1.1 Ma occurred on several of the islands: S. Vicente, Fogo (in 1995, with 26 48 volcanic eruptions since the 15th century), Santiago, Sal (0.4Ma), Santo Antão (0.09Ma) and S. Nicolau (0.1Ma) 49 (see Plesner et al. , 2002; Torres et al. , 2002; Knudsen et al. 2003; Schlüter, 2006; Duprat et al. , 2007). 50 51 However, it did not occur recently in Maio or Boavista, for example, (Stillman et al. , 1982; Mitchell et al. , 1983) 52 and these also present low, if not lower, intraspecific mitochondrial divergences. Thus, recent volcanism can 53 be a factor but not the only factor involved. As both archipelagos present islands with abrupt topography, with 54 55 half of the Cape Verde islands (Santiago, Fogo, Brava, S. Antão and S. Nicolau) having steep mountain areas, 56 one reaching almost 3,000 meters, the presence of geographical barriers is unquestionable. Even more if 57 considering that Tarentola species are typically found in dry areas at altitudes below 1500 m (Barbadillo et al. , 58 59 1999; R. Vasconcelos, pers. observ). On the other hand, reduced ecological differences within an island were 60 shown by the low number of “floristic altitudinal zones” (used by many authors as an indicator of the macro habitats diversity) that is always lower than two in all islands (Duarte et al. 2007). This could explain why Page 13 of 30 Journal of Biogeography

13 1 2 3 only in Santiago, the biggest islands of the archipelago, two different mtDNA lineages with geographic 4 structure were observed. Careful GIS modelling is now needed to understand which eco-topographic factors 5 6 explain this distribution. Santiago is also the island presenting the highest number of ESU and haplotypes, 7 followed by S. Antão, which is the second biggest island (see Appendix S2), and that is also one of the 8 9 youngest, having around 7.6 My (Knudsen et al. , 2003; Plesner et al. 2002), Fogo and S. Nicolau. Those islands 10 exhibit a strong orography, owing to the fact that erosion processes did not have enough time to flatten and 11 aridify them as markedly as the oldest eastern islands (Sal, Boavista) and Maio, allowing an altitudinal 12 13 ecological gradient. Moreover, based on the agro-ecological-vegetation maps (Diniz & Matos, 1986, 1987, 14 1988 a, b, 1993, 1994, 1999 a, b, c) and observations on the terrain (see Appendix S2) these islands presents a 15 relatively high habitat diversity (both climatic and topographic) that contrasts with the arid and semi-arid low 16 17 eastern counterparts, where severe pluriannual droughts occur periodically and are recorded since the XVI 18 century (Langworthy & Finan, 1997). This is also probably why in Boavista, the third biggest island, Tarentola 19 was found in low densities (R. Vasconcelos, A. Perera pers. observ.) and why this specie is apparently not in 20 For Peer Review 21 Sal (Carranza, 2000), even though an undetermined Tarentola species was reported from there by Angel 22 (1935, 1937) and Mertens (1955). If those records are correct, we could hypothesize an extinction scenario on 23 that last extremely arid island. If the records are incorrect probably Tarentola never reached that island, as it 24 25 is located further north, against the main currents and trade winds which have a northeast-southwest 26 direction. So, presenting the Cape Verde Islands, relatively to the Canaries possibly less ecological niches and 27 28 high ecological pressure which produces strong bottlenecks, Tarentola presents shallow mt DNA networks 29 with recent coalescent times. 30 31 32 In this way, the Tarentola radiation was clarified and clearly associated with historical islands sizes, oceanic 33 currents and trade winds and distances between the three island groups, although more morphological and 34 genetic studies are needed to clarify the , and GIS modelling to understand the barriers to gene flow 35 36 between ESU existing in the same island. Two factors were accounted for the low specific and intraspecific 37 variation observed in each island of the Cape Verdes: 1) the recent volcanic activity and high ecological stress 38 that could lead to population extinctions; and 2) the poor habitat diversity within some islands. Some geological 39 40 and ecological features of the islands were positively correlated to its genetic diversity such as area, altitude and 41 number of habitats. 42

43 44 ACKNOWLEDGEMENTS 45 R.V. is grateful to S. Rocha, M. Fonseca, J. C. Brito and A. Perera from CIBIO, J. Motta, H. Abella and A. Nevsky for 46 47 help during fieldwork; to Eng. J. César, Dr. Domingos, Eng. Orlando, Eng. J. Gonçalves, Eng. Lenine, Dr. C. Dias, 48 and staff from MAA and to Dr I. Gomes and all staff from INIDA for logistical aid and to J. Roca for lab 49 assistance. Research was supported by grants from Fundação para a Ciência e Tecnologia (FCT): 50 51 SFRH/BD/25012/2005 (to R.V.), PTDC/BIA-BDE/74288/2006 (to D.J.H.); from the Ministerio de Educación y 52 Ciencia, Spain: CGL2005-06876/BOS, Grup de Recerca Emergent of the Generalitat de Catalunya: 53 2005SGR00045, and an Intramural Grant from the Consejo Superior de Investigaciones Científicas, Spain: 54 55 2008301031 (to S.C.). Samples were obtained according to license nr. 07/2008 by DGA, MAA, Cape Verdian 56 Government. 57

58 59 REFERENCES 60 Angel, F. (1935) Lézards des Îles du Cap Vert, rapportés par M. le Professeur Chevalier. Description d’espèces nouvelles. Bulletin Du Museum d’Histoire Naturelle, Paris , 2, 165–169. Journal of Biogeography Page 14 of 30

14 1 2 3 Angel, F. (1937) Sur la Faune Herpétologique de l’Archipel du Cap Vert. XII. Congrès International Zoologie, Lisbonne , 1935 , 9, 4 1693–1700. 5 6 Ancochea, E., Hernan, F., Cendrero, A., Cantagrel, J.M., Fuster, J.M., Ibarolla, E. & Coello, J., (1994) Constructive and 7 destructive episodes in the building of a young oceanic island, La Palma, Canary islands, and genesis of the Caldera de 8 Taburiente. Journal of Volcanology and Geothermal Research , 60 , 243–262. 9 Abdel-Monem, A., Watkins, N.D. & Gast, P.W. (1971) Potassium-Argon ages, volcanic stratigraphy and geomagnetic polarity 10 11 history of the Canary Islands: Lanzarote, Fuerteventura, Gran Canaria and La Gomera. American Journal of Science , 271, 12 490–521. 13 Arnold, E.N. Vasconcelos, R., Harris, D.J. Mateo, J. A. & Carranza, S. (2008) Systematics, biogeography and evolution of the 14 15 endemic Hemidactylus geckos (Reptilia, , Gekkonidae) of the Cape Verde Islands: based on morphology and 16 mitochondrial and nuclear DNA sequences. Zoologica Scripta , 37 , 619–636. 17 Barbadillo, L.J., Lacomba, J.I., Pérez-Mellado, V., Sancho, V. & López-Jurado, L.F. (1999) Anfibios Y Reptiles de la Península 18 Ibérica, Baleares Y Canarias . Editorial Planeta, Barcelona, Spain. 19 20 Beheregaray, L.B., Ciofi, C.,For Geist, D., Gibbs, Peer J.P., Caccone, A. &Review Powell, J.R. (2003) Genes record a prehistoric volcano eruption 21 in the Galapagos. Science , 302 , 75. 22 Brehm, A., Jesus, J., Pinheiro, M. & Harris, D.J. (2001) Relationships of scincid lizards Mabuya (Reptilia: Scincidae) from the 23 24 Cape Verde Islands based on mitochondrial and nuclear DNA sequences. Molecular Phylogenetics and Evolution , 19 , 25 311– 316. 26 Brown, R.P., Campos-Delgado, R. & Pestano, J. (2000). Mitochondrial DNA evolution and population history of the Tenerife 27 skink Chalcides viridanus . Molecular Ecology 9, 1061–1067. 28 29 Brown, R.P., Suarez, N.M., Smith, A. & Pestano, J. (2001). Phylogeography of Cape Verde Islands skinks ( Mabuya ). Molecular 30 Ecology , 10 , 1593–1597. 31 Carracedo, J.C. (1999). Growth, structure, instability and collapse of Canarian volcanoes and comparisons with the Hawaiian 32 33 volcanoes. Journal of Volcanology and Geothermal Research , 94 , 1–19. 34 Carranza, S., Arnold, E.N., Mateo, J.A. & Geniez, P. (2002) Relationships and evolution of the North African geckos Geckonia 35 and Tarentola (Reptilia: Gekkonidae), based on mitochondrial and nuclear DNA sequences. Molecular Phylogenetics and 36 Evolution , 23 , 244–256. 37 38 Carranza, S., Arnold, E.N., Mateo, J.A. & López-Jurado, L.F. (2001) Parallel gigantism and complex colonization patterns in the 39 Cape Verde scincid lizards Mabuya and Macroscincus (Reptilia: Scincidae) revealed by mitochondrial DNA sequences. 40 Proceedings of the Royal Society of London b , 268 , 1595– 1603. 41 42 Carranza, S., Arnold, E.N., Mateo, J.A. & López-Jurado, L.F. (2000) Long-distance colonization and radiation in gekkonid 43 lizards, Tarentola (Reptilia: Gekkonidae), revealed by mitochondrial DNA sequences. Proceedings of the Royal Society of 44 London B, 267 , 637–649. 45 Carranza, S., Arnold, E.N., Geniez, P., Roca, J. & Mateo, J.A. (2008) Radiation, multiple dispersal and parallelism in the skinks, 46 47 Chalcides and Sphenops (Squamata: Scincidae), with comments on Scincus and Scincopus and the age of the Sahara 48 Desert. Molecular Phylogenetics and Evolution , 46 1071–1094. 49 Clement, M., Posada, D. & Crandall, K.A. (2000) TCS: a computer program to estimate gene genealogies. Molecular Ecology , 9, 50 1657–1660. 51 52 Cunningham, C.W. (1997) Is congruence between data partitions a reliable predictor of phylogenetic accuracy? Empirically 53 testing an iterative procedure for choosing among phylogenetic methods. Systematic Biology, 46 , 464-478. 54 Darwin, C. (1845) Journal of researches into the natural history and geology of the countries visited during the voyage of H.M.S. 55 56 Beagle round the world: Under the command of Capt. Fitz Roy . John Murray, London. 57 Díaz-Almela, E., Boudry, P., Launey, S., Bonhomme, F. & Lapègue, S. (2004) Reduced Female Gene Flow in the European Flat 58 Oyster Ostrea edulis . Journal of Heredity , 95, 510–516. 59 Diniz, A. C. & Matos, G. C. (1986) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde I – Ilha de Santiago. Garcia 60 da Horta, Série de Botânica. IICT, 8, 39–82. Page 15 of 30 Journal of Biogeography

15 1 2 3 Diniz, A. C. & Matos, G. C. (1987) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde II – Ilha do Fogo. Garcia da 4 Horta, Série de Botânica. IICT , 9, 35–69. 5 6 Diniz, A. C. & Matos, G. C. (1988a) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde III – Ilha do Maio. Garcia 7 da Horta, Série de Botânica. IICT , 10 , 19–48. 8 Diniz, A. C. & Matos, G. C. (1988b) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde IV – Ilha da Boavista. 9 Garcia da Horta, Série de Botânica. IICT , 10 , 49–72. 10 11 Diniz, A. C. & Matos, G. C. (1993) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde V – Ilha do Sal. Garcia da 12 Horta, Série de Botânica. IICT , 11 , 9–30. 13 Diniz, A. C. & Matos, G. C. (1994) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde VI e VII – Ilha de S. Vicente – 14 15 Ilha Sta. Luzia. Garcia da Horta, Série de Botânica. IICT, 12 , 69-100. 16 Diniz, A. C. & Matos, G. C. (1999a) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde VIII – Ilha de S. Nicolau. 17 Garcia da Horta, Série de Botânica. IICT , 14 , 1–54. 18 Diniz, A. C. & Matos, G. C. (1999b) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde IX – Ilha Brava. Garcia da 19 20 Horta, Série de Botânica.For IICT , 14 , 55–82.Peer Review 21 Diniz, A. C. & Matos, G. C. (1999c) Carta de zonação Agro-Ecológica e da Vegetação de Cabo Verde X – Ilha de Santo Antão. 22 Garcia da Horta, Série de Botânica. IICT , 14 , 1–34. 23 24 Duarte, M. C., Rego, F. Romeiras, M. M. & Moreira, I. (2007) Plant species richness in the Cape Verde Islands—eco- 25 geographical determinants. Biodiversity and Conservation , 17 : 453–466. 26 Duprat , H.I., Friis, J., Holm, P.M., G randvuinet, T. & Sørensen, R.V. (2007) The volcanic and geochemical development of São 27 Nicolau, Cape Verde Islands: Constraints from field and 40Ar/39Ar evidence. Journal of Volcanology and Geothermal 28 29 Research , 162 , 1–19. 30 Emerson, B.C. (2002) Evolution on oceanic islands: molecular phylogenetic approaches to understanding pattern and 31 process. Molecular Ecology, 11 , 951–966. 32 33 ESRI (2004) ArcMap 9.0. Environmental Systems Research Institute, Inc. 34 Excoffier, L. Laval, G. & Schneider, S. (2005) Arlequin ver. 3.0: An integrated software package for population genetics data 35 analysis. Evolutionary Bioinformatics Online , 1, 47–50. 36 Farris, J.S., Kallersjo, M., Kluge, A.G., & Bult, C. (1994) Testing significance of incongruence. Cladistics, 10 , 315-319. 37 38 Felsenstein, J. (1985) Confidence-limits on phylogenies - an approach using the bootstrap. Evolution , 39 , 783-791. 39 Frankham, R. (1997). Do island populations have less genetic variation than the mainlaind populations? Heredity , 78 , 281– 40 307. 41 42 Fraser, D.J & Bernatchez, L. (2001) Adaptive evolutionary conservation: towards a unified concept for defining conservation 43 units. Molecular Ecology , 10 , 2741–2752. 44 Fu, Y.X. (1997) Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. 45 Genetics , 147, 915–925. 46 47 Gómez, J.M. (2006) Salamanquesas canarias ( Tarentola delalandii ) en Cantabria. Boletín de la Asociación Herpetológica 48 Española, 17 , 80–81. 49 González, P., Pinto, F., Nogales, M., Jiménez-Asensio, J., Hernández, M. & Cabrera, V.M. (1996) Phylogenetic relationships of 50 the Canary Islands endemic lizards genus Gallotia (Sauria: Lacertidae), inferred from mitocondrial DNA sequences. 51 52 Molecular Phylogenetics and Evolution , 6, 63–71. 53 Griffiths, J., Cantagrel, J.M., Alves, C.A., Mendes, F., Serralheiro, A. & Macedo, J.R. (1975) Donnés radiométriques 54 potassiumargon sur quelques formations magnetiques dês îles de l’árchipel du Cap-Vert. Compte Rendu de l’Academie 55 56 de Sciences, Paris , Sér. D. , 280 , 2429–2432. 57 Grunau, H.R., Lehner, P., Cleintuar, M.R., Allenbach, P. & Bakker, G. (1975) New Radiometric Data from Fuerteventura 58 (Canary Islands), Maio (Cape Verde) and São Tomé (Golf of Guinea). In G. J. Borradaille (Ed.) Progress in Geodynamics 59 (pp. 90–118). Amsterdam: Royal Society of Netherlands Academy of Arts and Sciences. 60 Guindon, S. & Gascuel, O. (2003) A simple, fast and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology, 52 , 696-704. Journal of Biogeography Page 16 of 30

16 1 2 3 Gübitz, T., Thorpe, R.S. & Malhotra, A. (2005) The dynamics of genetic and morphological variation on volcanic islands. 4 Proceedings of the Royal Society of London, Series B, 272 , 751–757. 5 6 Gübitz, T, Thorpe, R.S & Malhotra, A. (2000) Phylogeography and natural selection in the Tenerife Tarentola 7 delalandii : testing historical and adaptive hypotheses. Molecular Ecology , 9, 1213–1221. 8 Hare, M.P. (2001) Prospects for nuclear gene phylogeography. Trends in Ecology and Evolution , 16 , 700–706. 9 Harpending H.C., Sherry, S.T., Rogers, A.R., Stoneking, M. (1993) Genetic structure of ancient human populations Current 10 11 Anthropology , 34 ,483–496. 12 Harris, D.J., Arnold, E.N. & Thomas, R.H. (1998) Relationships of the lacertid lizards (Reptilia: Lacertidae) estimated from 13 mitochondrial DNA sequences and morphology. Proceedings of the Royal Society of London, Series B, 265 , 1939–1948. 14 15 Harris, D.J., Batista, V., Lymberakis, P. & Carretero, M.A. (2004) Complex estimates of evolutionary relationships in Tarentola 16 mauritanica ( Reptilia: Gekkonidae) derived from mitochondrial DNA sequences. Molecular Phylogenetics and Evolution , 17 30 , 855–859. 18 Hart, M.W. & Sunday, J. (2007) Things fall apart: biological species form unconnected parsimony networks. Biological 19 20 Letters , 3, 509–512. For Peer Review 21 Hobohm, C. (2000) Plant species diversity and endemism on islands and archipelagos, with special reference to the 22 Macaronesian Islands. Flora 195 , 9–24. 23 24 Huelsenbeck, J.P. & Ronquist, F. (2001) MRBAYES: Bayesian inference of phylogeny. Bioinformatics, 17 , 754-755. 25 Hudson, R.R. (2000) A new statistic for detecting genetic differentiation. Genetics , 155 , 2011–2014. 26 Jarvis A., H.I. Reuter, A. Nelson & E. Guevara (2006) Hole-filled seamless SRTM data V3 . International Centre for Tropical 27 Agriculture (CIAT), available from http://srtm.csi.cgiar.org. 28 29 Jesus, J., Brehm, A., Harris & D.J. (2002) Relationships of Tarentola Reptilia: Gekkonidae) from the Cape Verde islands 30 estimated from DNA sequence data. Amphibia-Reptilia , 22 , 235-242. 31 Joger, U. (1984) Taxonomische revision der Gattung Tarentola (Reptilia, Gekkonidae). Bonner zoologische Beiträge , 35 , 129– 32 33 174. 34 Joger, U. (1993) On two collections of reptiles and amphibians from the Cape Verde Islands, with descriptions of three new 35 taxa. Courier Forschungsinstitut Senckenberg , 159 , 437–444. 36 Juan, C., Emerson, B.C., Orom, P. & Hewitt, G.M. (2000) Colonization and diversification: towards a phylogeographic 37 38 synthesis for the Canary Islands. Trends in Ecology & Evolution 15 , 104–109. 39 Kahindo, C., Bowie, R.C.K. & Bates, J.M. (2007) The relevance of data on genetic diversity for the conservation of Afro- 40 montane regions. Biological Conservation , 134, 262–270. 41 42 Kocher, T.D., Thomas R.K., Meyer, A., Edwards, S.V., Pääbo, S., Villablanca, F.X. & Wilson, A.C. (1989) Dynamics of 43 mitochondrial DNA evolution in animals: amplification and sequencing with conserved primers. Proceedings of the 44 National Academy of Sciences USA , 86 , 6196–6200. 45 Knudsen, M.F., Abrahamsen, N. & P. Riisager (2003) Paleomagnetic evidence from Cape Verde Islands basalts for fully 46 47 reversed excursions in the Brunhes Chron. Earth and Planetary Science Letters , 206 , 199–214. 48 Lalueza-Fox, C., Castresana, J., Samprieto, L., Marquez-Bonet, T., Alcover, J.A. & Bertranpetit, J. (2005) Molecular dating of 49 caprines using ancient DNA sequences of Myotragus balearicus , an extinct endemic Balearic mammal. BMC Evolutionary 50 Biology , 5, 70. 51 52 Langley, C. H. & Fitch, W. M. (1974) An examination of the constancy of the rate of molecular evolution . Journal of Molecular 53 Evolution , 3: 161-177. 54 Langworthy, M. & Finan, T.J. (1997) Waiting for rain - agriculture and ecological imbalance in Cape Verde. Lynne Rienner 55 56 Publishers, 212pp. 57 Losos, J.B. & Ricklefs, R.E. (2009) Adaptation and diversification on islands. Nature , 457, 830–836. 58 Maddison, W. P. & Maddison, D. R. (1992). MacClade, version 3: Analysis of phylogeny and character evolution. Sinauer, 59 Sunderland, Massachusetts. 60 Mertens, R. (1955) Die Eidechsen der Kapverden. Commentationes Biologicae , 15 , 1–17. Page 17 of 30 Journal of Biogeography

17 1 2 3 Maca-Meyer, N., Carranza, S., Rando, J.C., Arnold E.N. & Cabrera, V.M. (2003) Status and relationships of the extinct giant 4 Canary Island lizard Gallotia goliath (Reptilia: Lacertidae), assessed using ancient mtDNA from its mummified remains, 5 6 Biological Journal of the Linnean Society , 80 , 659–670. 7 Mason-Gamer, R.J. & Kellogg, E.A. (1996) Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae 8 (Gramineae). Systematic Biology, 45 , 524-545. 9 Michkevich, M.F. & Farris, J.S. (1981) The implications of congruence in Menidia . Systematic Zoology, 30 , 351-370. 10 11 Mitchell, J.G., LeBas, M.J., Zielonka, J. & Furnes, H. (1983) On dating the magmatism of Maio, Cape Verde Islands. Earth and 12 Planetary Science Letters , 64 , 66–76. 13 Mitchell-Thomé, R.C. (1976) Geology of the Middle Atlantic Islands. Stuttgart: Science Publishers. 14 15 Morris, R.O. (1989). Navigational Chart Number 366 ‘Arquipélago de Cabo Verde’ . UK: Taunton, UK: Hydrographic Society. 16 Nogales, M., López, M., Jiménez-Asensio, J., Larruga, J.M., Hernández, M. & González, P. (1998) Evolution and biogeography 17 of the genus Tarentola (Sauria: Gekkonidae) in the Canary Islands, inferred from mitochondrial DNA sequences. Journal 18 of Evolutionary Biology , 11 , 481-494. 19 20 Palumbi, S.R. (1996). NucleicFor acids, II: thePeer polymerase chain Review reaction. Molecular Systematic s. Hillis, D.M., Moritz, C. & Mable, 21 B.K. (eds.). Sunderland, Mass.: Sinauer Associates, Inc., 205–247. 22 Pestano, J & Brown, R.P. (1999) Geographical structuring of mtDNA in Chalcides sexlineatus within the island of Gran 23 24 Canaria. Proceedings of the Royal Society of London, Series B, 266 , 815–823. 25 Plesner, S., Holm, P.M. & Wilson, J.R. (2002). 40Ar-39Ar geochronology of Santo Antão, Cape Verde Islands. Journal of 26 Volcanology and Geothermal Research , 120 , 103–121. 27 Posada, D. (2008) ModelTest: Phylogenetic Model Averaging. Molecular Biology and Evolution , 25 , 1253-1256. 28 29 Posada, D. & Crandall, K.A. (2001) Evaluation of methods for detecting recombination from DNA sequences: Computer 30 simulations. Proceedings of the National Academy of Sciences USA , 98 , 13757–13762. 31 Rambaut, A. & Grassly, N.C. (1997) Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along 32 33 phylogenetic trees. Computer Applications in the Biosciences, 13, 235-238. 34 Rogers, A.R. & Harpending, H. (1992) Population growth makes waves in the distribution of pairwise genetic differences. 35 Molecular Biology and Evolution, 9, 552–569. 36 Rogers, A.R. (1995) Genetic evidence for a Pleistocene population expansion. Evolution , 494 , 608–615. 37 38 Rosenzweig, M.L. (1995). Species Diversity in Space and Time . Cambridge University Press, Cambridge, UK. 39 Rozas, J., Sánchez-Delbarrio, J.C., Messeguer, X. & Rozas, R. (2003) DnaSP, DNA polymorphism analyses by the coalescent 40 and other methods. Bioinformatics 19 , 2496–2497. 41 42 Sanderson, M.J. & Doyle, J.A. (2001) Sources of error and confidence intervals in estimating the age of the angiosperm from 43 rbcL and 18S rDNA data. American Journal of Botany , 88 , 1499–1516. 44 Sanderson, M.J. (1997) A nonparametric approach to estimating divergence times in the absence of rate constancy. 45 Molecular Biology and Evolution, 14 , 1218-1231. 46 47 Sanderson, M.J. (2002) Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood 48 approach. Molecular Biology and Evolution , 19 , 101-109. 49 Slatkin, M. & Hudson, R.R. (1991) Pairwise comparisons of mitochondrial DNA sequences in stable and exponentially 50 growing populations. Genetics , 129 , 555–562. 51 52 Schleich, H.H. (1987) Herpetofauna Caboverdiana. Spixiana , 12 , 1–75. 53 Schneider, S. & Excoffier, L. (1999) Estimation of demographic parameters from the distribution of pairwise differences 54 when the mutation rates vary among sites: application to human mitochondrial DNA. Genetics , 152 , 1079–1089. 55 56 Schlüter, T. (2006) Geological Atlas of Africa with Notes on Stratigraphy, Tectonics, Economic Geology, Geohazards and 57 Geosites of Each Country . Springer Berlin, Heidelberg Chapter4, pp 64. 58 Schwartz, M.K., Luikart, G. & Waples, R.S. (2006) Genetic monitoring as a promising tool for conservation and management. 59 Trends in Ecology and Evolution , 22 , 25–33. 60 Shimodaira, H. & Hasegawa, M. (1999) H. Shimodaira and M. Hasegawa, Multiple comparisons of Log-likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution , 16 , 1114–1116. Journal of Biogeography Page 18 of 30

18 1 2 3 Stillman, C.J., Furnes, H., LeBas, M.J., Robertson, A.H.F. & Zielonka, J. (1982) The geological history of Maio, CapeVerde 4 Islands. Journal of the Geological Society of London , 139 , 347–361. 5 6 Swofford, D.L. (1998) PAUP*: Phylogenetic Analysis using Parsimony (and other methods), v 4.0 . Sinauer Associates, 7 Sunderland, MA. 8 Templeton, A.R., Crandall, K.A. & Sing, C.F. (1992) A cladistic analysis of phenotypic associations with haplotypes inferred 9 from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics , 132 , 619–633. 10 11 Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmourgin, F. & Higgins, D.G. (1997) The clustalX windows interface: flexible 12 strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research, 24 , 4876-4882. 13 Thorpe, R.S, Black, H. & Malhotra, A. (1996) Matrix correspondence tests on the DNA phylogeny of the Tenerife Lacertid 14 15 elucidate both historical causes and morphological adaptation. Systematic Biology , 45 , 335–343. 16 Thorpe, R.S & Malhotra, A. (1996) Molecular and morphological evolution within small islands. Philosphical Transactions of 17 the Royal Society of London, Series B , 351 , 815–822. 18 Torres, P.C., Silva, L.C., Serralheiro, A., Tassinari, C. & Munhá, J. (2002) Enquadramento geocronológico pelo método K/Ar 19 20 das principais sequênciasFor vulcano-estratigráficas Peer da ilhaReview do Sal — CaboVerde. Garcia de Orta, Série Geológica, 18 , 9–13. 21 Whittaker, R.J. (1998) Island Biogeography – Ecology, Evolution and Conservation . Oxford University Press, Oxford. 22 Wilcox, T.P., Zwickl, D.J., Heath, T.A. & Hillis, D.M. (2002) Phylogenetic relationships of the dwarf boas and comparison of 23 24 Bayesian and bootstrap measures of phylogenetic support. Molecular Phylogenetics and Evolution, 25 , 361-371. 25 Williamson, M. (1981) Island populations . Oxford University Press, Oxford. 26 World Conservation Monitoring Centre (1992) Global Biodiversity: Status of the Earth’s living resources . Chapman & Hall, 27 London. 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 19 of 30 Journal of Biogeography

19 1 2 3 4 Supporting Information 5 6 7 Additional Supporting Information may be found in the online version of this article: 8 9 Appendix S1 Genetic differentiation between populations belonging to the same network : Snn values. 10 Appendix S2 Variables used in the correlation analyses. 11 12 13 14 15 16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Biogeography Page 20 of 30

20 1 2 3 4 5 BIOSKETCHES 6 7 8 Raquel Vasconcelos investigates the herpetofauna of the Cape Verde Islands combining GIS and molecular 9 tools, as a PhD student at Centro de Investigação em Biodiversidade e Recursos Genéticos, Oporto University 10 (CIBIO/UP) and Institute of Evolutionary Biology, Pompeu-Fabra University (CSIC-UPF). 11 12 13 Salvador Carranza is a Researcher at CSIC-UPF, Spain, with a long-standing interest in the herpetofauna of 14 the Mediterranean Basin, Old World deserts and oceanic islands. 15 16 17 D. James Harris is an invited Assistant Professor in CIBIO/UP, Portugal, interested in phylogenetics, 18 especially of reptiles. He leads several projects on determining the genetic and morphological variation of 19 20 reptiles of the MediterraneanFor Basin andPeer other parts of theReview world. http://cibio.up.pt 21 22 23 Commissioning editors: Niels Raes & Menno Schilthuizen 24 25 26 27 The papers in this Special Issue arose from the symposium ‘Evolutionary islands 150 years 28 after Darwin’ (http://science.naturalis.nl/darwin2009), held from 11 to 13 February 2009 29 at the Museum Naturalis, Leiden, the Netherlands. The theme of the symposium was to 30 explore the contribution of islands to our understanding of evolutionary biology and to 31 analyse the role of island biological processes in a world in which the insularity of island 32 and mainland ecosystems is being drastically altered. 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 21 of 30 Journal of Biogeography

21 1 2 3 4 Figure 1 Map of the Cape Verde Islands showing the geographic location and altitudes of the archipelago islands, 5 6 and the origins of the Tarentola samples included in the analyses (Geographic Coordinate System, Datum WGS 84). 7 Dashed line divides the T. darwini southern and northern haplotypes from Santiago. 8 9 10 Figure 2 Maximum likelihood (ML) tree inferred using the GTR+I+G model of sequence evolution (Log Likeliohood = 11 -6468.896) showing relationships and estimated times of divergence of endemic Cape Verde Tarentola and their 12 relatives from the Canary Islands. The tree is rooted using Tarentola americana . Bootstrap support values above 13 60% for the ML analysis are shown below nodes. Posterior probability values (pp) higher than 95% for the Bayesian 14 15 analysis are represented by an asterisk (*) and are shown above nodes. Italic numbers in some selected nodes 16 (highlighted with a filled circle), indicate the estimated age of the speciation event of that node in millions of years, 17 followed by the standard deviation obtained with parametric bootstrap using the original topology (see materials 18 19 and methods). Sequences downloaded from GenBank are shown in the figure with their respective GenBank 20 accession numbers for theFor cyt b and 12S Peer rRNA genes separated Review by a dash. For locality data and GenBank accession 21 numbers of the new sequences see Table 1. Letters immediately to the right of island names correspond to the 14 22 Evolutionarily Significant Units (ESU) recognised in the present work and shown in Fig.3. 23 24 25 Figure 3 Networks corresponding to cyt b sequence variation. Lines represent a mutational step, dots missing 26 haplotypes and circles haplotypes. The circle area is proportional to the number of individuals. Dashed lines 27 represent probable ancestral haplotype. For correspondences of sample and location codes see Table 1. a) Tarentola 28 29 “rudis” boavistensis from Boavista ( 1), T. darwini from S. Nicolau ( 2), Fogo ( 3) and Santiago ( 4). b) T. caboverdiana 30 from S. Vicente, Santo Antão, Santa Luzia and Branco and Raso islets. c) Tarentola c. nicolauensis from S. Nicolau. d) 31 T. rudis from Santiago, Fogo, Brava and Rombos islets, T. gigas from Branco and Raso ( 1) and Tarentola r. maioensis 32 33 from Maio ( 2). 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Biogeography Page 22 of 30

22 1 2 3 Fig. 1 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30

31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 23 of 30 Journal of Biogeography

23 1 2 3 Fig. 2 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Biogeography Page 24 of 30

24 1 2 3 Fig. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 25 of 30 Journal of Biogeography

25 1 2 3 Table 1 Details of material and sequences used in the present study. Tcs stands for T. caboverdiana substituta , Tcr 4 for T. caboverdiana raziana , Tcc for T. caboverdiana caboverdiana , Tcn for T. caboverdiana nicolauensis , Td for T. 5 6 darwini , Trb for T. “rudis” boavistensis , Tgg for T. gigas gigas , Trr for T. rudis rudis , Trp for T. rudis protogigas and Trm 7 for T. rudis maioensis . SV stands for S. Vicente, SL for Sta. Luzia, ra for Raso, br for Branco, SA for Santo Antão, SN for 8 S. Nicolau, ST for Santiago, F for Fogo, B for Brava, M for Maio and BV for Boavista. Individuals marked with * have 9 ambiguous identification. 10 Code Taxa Island Locality Genbank codes Code Taxa Islan d Locality Genbank codes 11 T001 Tcs SV Mindelo To be added T256 Tcc SA Ponta Aguadinha To be added 12 T003 Tcs SV Mindelo To be added T259 Tcc SA Tarrafal de To be added T010 Tcs SV To be added cv105 Tcc SA Dogoi To be add ed 13 T014 Tcs SV Calhau To be added cv107 Tcc SA Lagoa To be added T017 Tcs SV Calhau To be added cv108 Tcc SA Lagoa To be added 14 T021 Tcs SV Calhau -Madeiral road To be added cv110 Tcc SA Lagoa To be added T026 Tcs SV Pico Alves Martinho To be added cv111 Tcc SA Lagoa To be added 15 T032 Tcs SV To be added cv113 Tcc SA Porto Novo To be added T033 Tcs SV Monte Verde To be added cv114 Tcc SA Rb.ra da Cruz To be added T035 Tcs SV To be added T261 Tcn SN Tarrafal To be added 16 T037 Tcs SV Salamansa To be added T264 Tcn SN Branca To be added T038 Tcs SV Salamansa To be added T267 Tcn SN Praia Branca graveyard To be added 17 T043 Tcs SV Salamansa To be added T271 Tcn SN Praia Branca graveyard To be added T045 Tcs SV Salamansa To be added T273 Tcn SN Monte Furado To be added 18 T046 Tcs SV Salamansa To be added T278 Tcn SN Chã do Curral To be added T047 Tcs SV Calhau To be added T280 Tcn SN Rb.ra da Prata To be added 19 T048 Tcs SV S. Pedro To be added T281 Tcn SN Rb.ra da Prat a To be added T051 Tcs SV S. Pedro To be added T283 Tcn SN Cabeçalinho To be added 20 T055 Tcs SV S. Pedro For PeerTo be added T285Review Tcn SN Cabeçalinho To be added T059 Tcs SV S. Pedro To be added T286 Tcn SN Cabeçalinho To be added 21 T064 Tcs SV Mindelo To be ad ded T287 Tcn SN Cabeçalinho To be added T070 Tcs SV Mindelo To be added T288 Tcn SN Lombo de Morro To be added 22 T077 Tcs SV road to Madeiral To be added T294 Tcn SN Tarrafal To be added T084 Tcs SV To be added T295 Tcn SN Luís Afonso To b e added 23 T092 Tcs SV Monte Verde To be added T298 Tcn SN Luís Afonso To be added T094 Tcs SV Monte Verde To be added T306 Tcn SN Rb.ra Brava To be added 24 T095 Tcs SV Mato Inglês To be added T307 Tcn SN Rb.ra Brava To be added T096 Tcs SV Mato Inglês To be added T308 Tcn SN Rb.ra Brava To be added 25 T097 Tcs SV Mato Inglês To be added T309 Tcn SN Rb.ra Brava To be added T102 Tcs SV Mato Inglês To be added T311 Tcn SN Campinho To be added 26 T105 Tcs SV Lazareto To be added T313 Tcn SN Fajã de Baixo To be added T109 Tcs SV Lazareto To be added T315 Tcn SN Estância Brás cross To be added 27 T110 Tcs SV To be added T316 Tcn SN Estância Brás cross To be added T116 Tcs SV Monte Cara To be added T318 Tcn SN Cabeçalinho To be added 28 T122 Tcs SV road to S. Pedro To be added T321 Tcn SN Praia de Baixo To be added T124 Tcs SV road to S. Pedro To be added T326 Tcn SN Campo do Porto To be added 29 T125 Tcs SV road to S. Pedro To be added T327 Tcn SN Fajã de Baixo To be added T126 Tcs SV road to S. Pedro To be added T328 Tcn SN da Covada To be added 30 T128 Tcs SV road to S. Pedro To be added T331 Tcn SN Preguiça Airport To be added T129 Tcs SV To be added T334 Tcn SN Caldeira da Preguiça To be added T132 Tcs SV Pico do Vent o To be added T337 Tcn SN Caldeira da Preguiça To be added 31 T133 Tcs SV Pico do Vento To be added T338 Tcn SN Caldeira da Preguiça To be added T134 Tcs SV Pico do Vento To be added T339 Tcn SN Hortelão To be added 32 T135 Tcs SV Flamengos To be added T3 42 Tcn SN Ponta da Praia do Garfo To be added T136 Tcs SV Flamengos To be added T346 Tcn SN Ponta Pataca To be added 33 T137 Tcs SV Flamengos To be added T353 Tcn SN Ponta Coruja To be added T140 Tcs SV Flamengos To be added T355 Tcn SN Chã de Norte To be added 34 T142 Tcs SV Flamengos To be added T356 Tcn SN Chã de Norte To be added T144 Tcr SL Água Doce To be added T366 Tcn SN Mombaixa To be added 35 T145 Tcr SL Água Doce To be added T367 Tcn SN Mombaixa To be added T150 Tcr SL Água Doce To be added T301 Td SN Carriçal To be added 36 T151 Tcr SL Água Doce To be added T302 Td SN Carriçal To be added T154 Tcr SL Ponta Salina To be added T303 Td SN Carriçal -Juncalinho road To be added 37 T158 Tcr SL Ponta de Praia To be added T304 Td SN Carriçal -Juncali nho road To be added T162 Tcr SL Praia de Palmo a Tostão To be added T348 Td SN Juncalinho To be added 38 T164 Tcr SL Monte Espia To be added T349 Td SN Juncalinho To be added T165 Tcr SL Topinho de Nhô Lopes To be added T351 Td SN Ponta Larga To be ad ded 39 T166 Tcr SL Rib.ra de Casa To be added T358 Td SN Aguada de Falcão To be added T170 Tcr SL Morro da Prainha Branca To be added T359 Td SN Aguada de Falcão To be added 40 T172 Tcr SL Monte Creoulo To be added T360 Td SN Aguada de Falcão To be added T174 Tcr SL Rib.ra de Freira To be added T361 Td SN Monte Vermelho To be added 41 T178 Tcr SL Rib.ra de Freira To be added T362 Td SN Monte Vermelho To be added raT1 Tcr ra Ponta de Casa To be added T363 Td SN Ponta Mota To be added 42 raT2 Tcr ra Ponta de Casa To be added T364 Td SN Ponta Mota To be added raT3 Tcr ra Ponta de Casa To be added T365 Td SN Ponta Mota To be added 43 raT5 Tcr ra Ponta de Casa To be added T370 Td ST S. Lourenço dos Órgãos To be added raT6 Tcr ra Ponta de Casa To be added T37 2 Td * ST To be added 44 raT7 Tcr ra Ponta de Casa To be added T373 Td ST Cidade Velha To be added T181 Tcc SA To be added T374 Td ST Cidade Velha To be added 45 T182 Tcc SA To be added T375 Td ST Cidade Velha To be added T183 Tcc SA Cova To be added T376 Td ST Achada Pedra To be added T187 Tcc SA da Garça To be added T378 Td ST Ribeirão Chiqueiro To be added 46 T188 Tcc SA To be added T379 Td ST Ribeirão Chiqueiro To be added T189 Tcc SA Lagoa To be added T380 Td * ST Ribeirão Chiqueiro To be added 47 T192 Tcc SA Espongueiro cross To be added T381 Td * ST Ribeirão Chiqueiro To be added T193 Tcc SA Sinagoga To be added T383 Td * ST Ribeirão Chiqueiro To be added 48 T194 Tcc SA Lombo To be added T384 Td * ST S. Nicolau Tolentino To be added T196 Tcc SA Morro de Passagem To be added T389 Td ST To be added 49 T198 Tcc SA Morro de Passagem To be added T390 Td ST Achada Fazenda To be added T199 Tcc SA Chã de Norte To be added T391 Td ST Achada Fazenda To be added 50 T201 Tcc SA Chã de Norte village To be added T392 Td ST To be added T203 Tcc SA Aldeia To be added T394 Td ST Porto Gouveia To be added 51 T204 Tcc SA Porto Novo To be added T395 Td ST Rb.ra Grande de Santiago To be added T206 Tcc SA Porto Novo To be added T397 Td ST Rb.ra Grande de Santiago To be added 52 T207 Tcc SA Chã de Norte To be added T398 Td ST Ponta Bombardeiro To be added T210 Tcc SA Chã de Lagoinha To be added T399 Td ST Ponta Bombardeir o To be added 53 T211 Tcc SA S.Tomé To be added T400 Td ST Ponta Bombardeiro To be added T214 Tcc SA Porto Novo To be added T402 Td ST S. Martinho Pequeno To be added 54 T215 Tcc SA Rib de Bodes To be added T403 Td ST João Varela To be added T216 Tcc SA Chã do Brejo To be added T404 Td ST João Varela To be added 55 T219 Tcc SA Chã de Banca To be added T405 Td ST João Varela To be added T222 Tcc SA Chã de Nhã Nica To be added T406 Td ST João Varela To be added 56 T225 Tcc SA Chã de Nhã Nica To be added T4 07 Td ST João Varela To be added T228 Tcc SA Lombo do Meio To be added T409 Td ST S. Martinho Pequeno To be added 57 T230 Tcc SA Rabo de Gamboeza To be added T411 Td ST S. Martinho Pequeno To be added T234 Tcc SA Curralete To be added T412 Td ST Praia B aixo To be added 58 T238 Tcc SA Curralete To be added T413 Td ST To be added T240 Tcc SA Curralete To be added T414 Td ST Praia Baixo To be added 59 T241 Tcc SA Ponte Sul To be added T415 Td ST Praia Baixo To be added T248 Tcc SA Monte Trigo To be added T416 Td ST Praia Baixo To be added 60 T254 Tcc SA Covão To be added T417 Td ST Praia Baixo To be added

Journal of Biogeography Page 26 of 30

26 1 2 3 4 Code Taxa Island Locality Genbank code Code Taxa Island Locality Genbank code 5 T418 Td ST Praia Baixo To be added T577 Td F Campanas de Baixo To be added T420 Td ST Nossa Sra da Luz To be added T578 Td F Campanas de Baixo To be added 6 T421 Td ST Nossa Sra da Luz To be added T579 Td F Campanas de Baixo To be added T422 Td ST To be added T581 Td F Velho Manuel To be added T423 Td ST Cancelo To be added T583 Td F Lomba To be added 7 T424 Td ST Cancelo To be added T584 Td F Mosteiros To be added T425 Td ST Cancelo To be added T585 Td F Mosteiros To be added 8 T426 Td ST Cancelo To be added T586 Td F Fonsaco To be ad ded T428 Td ST Cancelo To be added T587 Td F Fonsaco To be added 9 T429 Td ST S. Lourenço dos Órgãos To be added T588 Td F Mosteiros To be added T430 Td ST S. Lourenço dos Órgãos To be added T589 Td F Mosteiros To be added 10 T431 Td ST Calheta S. Miguel To be added T590 Td F Mosteiros To be added T432 Td ST Calheta S. Miguel To be added T591 Td F Santa Catarina do Fogo To be added 11 T433 Td ST Calheta S. Miguel To be added T592 Td F Santa Catarina do Fogo To be added T434 Td ST Calheta S. Miguel To b e added T593 Td F Santa Catarina do Fogo To be added 12 T435 Td * ST S. Lourenço dos Órgãos To be added T594 Td F S. Filipe To be added T436 Td * ST S. Filipe de Cima To be added T595 Td F S. Filipe To be added 13 T437 Td ST Rb.ra da Barca To be added T59 6 Td F S. Filipe To be added T438 Td ST Rb.ra da Barca To be added T597 Td F Monte Vermelho To be added 14 T439 Td ST Rb.ra da Barca To be added T599 Td F S. Filipe To be added T440 Td ST Rb.ra da Barca To be added T600 Td F Cova Figueira To be added 15 T441 Td * ST Curral Grande To be added T601 Td F Cova Figueira To be added T443 Td * ST Curral Grande To be added T602 Td F Cova Figueira To be added 16 T444 Td ST Pedra Barro To be added T603 Td F Monte Verde To be added T445 Td ST Pedra Barro To be add ed T604 Td F Monte Verde To be added 17 T446 Td ST Calheta S. Miguel To be added T605 Td F Monte Verde To be added T447 Td ST Calheta S. Miguel To be added T606 Td F S. Filipe To be added 18 T448 Td * ST Ribeirão Galinha To be added T661 Trb BV Caminho cr uz João Santo To be added T449 Td ST Achada Além To be added T662 Trb BV Caminho cruz João Santo To be added 19 T451 Td ST Santa Ana To be added T663 Trb BV Caminho cruz João Santo To be added T452 Td ST Santa Ana To be added T664 Trb BV Monte Estância To be added 20 T453 Td ST Santa AnaFor ToPeer be added Review T665 Trb BV Ervatão To be added T454 Td * ST Santa Ana To be added T666 Trb BV Ervatão To be added 21 T456 Td ST Santa Ana To be added T667 Trb BV Chão de Palhal To be added T457 Td ST Santa Ana To be added T668 Trb BV Salamansa To be added 22 T458 Td ST Santa Ana To be added T669 Trb BV Salamansa To be added T459 Td ST Santa Ana To be added T671 Trb BV Lomba de Malva To be added T460 Td ST Santa Ana To be added T672 Trb BV Cabeça de Cachorro To be added 23 T462 Td ST Praia To be added T675 Trb BV Chã de Calheta To be added T464 Td ST To be added raTg2 Tgg ra Ponta de Casa To be added 24 T465 Td ST Chão Bom To be added raTg3 Tgg ra Ponta de Casa To be added T468 Td ST S. Lourenço dos Órgãos To be added raTg4 Tgg ra Ponta de Casa To be added 25 T469 Td ST Montanhinha To be added T371 Trr ST Cidade Velha To be added T470 Td ST Palha Carga To be added T377 Trr ST Ribeirão Chiqueiro To be added 26 T471 Td ST S. Lourenço dos Órgãos To be added T382 Trr ST Ribeirão Chiqueiro To be added T472 Td ST S. Lourenço dos Órgãos To be added T385 Trr ST S. Nicolau Tolentino To be added 27 T473 Td ST S. Lourenço dos Órgãos To be added T386 Trr ST S. Nicolau Tolentino To be added T474 Td ST Chão de Tanque To be added T387 Trr ST S. Nicolau Tolentino To be added 28 T475 Td ST Santa Catarina To be added T401 Trr ST S. Martinho Pequeno To be added T476 Td ST Santa Catarina To be added T408 Trr ST S. Martinho Pequeno To be added 29 T477 Td ST Santa Catarina To be add ed T410 Trr ST S. Martinho Pequeno To be added T478 Td ST Santa Catarina To be added T419 Trr ST Nossa Sra da Luz To be added 30 T479 Td ST Santa Catarina To be added T442 Trr ST Curral Grande To be added T480 Td ST Santa Catarina To be added T455 Trr ST Santa Ana To be added 31 T482 Td ST Chão Bom To be added T461 Trr ST Praiaaia To be added T483 Td ST Chão Bom To be added T463 Trr ST Praiaaia To be added 32 T484 Td ST Chão Bom To be added T517 Trr ST Barnabé To be added T486 Td ST Chão Bom To be adde d T518 Trr ST Barnabé To be added 33 T487 Td ST Chão Bom To be added T519 Trr ST Barnabé To be added T488 Td ST Ponta do Lobrão To be added T521 Trr ST Barnabé To be added 34 T489 Td ST Ponta do Lobrão To be added T532 Trp F Lagariça To be added T490 Td ST Ponta do Lobrão To be added T535 Trp F Monte Calhau To be added 35 T491 Td ST Trás os Montes To be added T598 Trp F Monte Vermelho To be added T492 Td ST Trás os Montes To be added T541 Trp B Favatal To be added 36 T493 Td ST Trás os Montes To be added T544 Trp B Lima Doce To be added T494 Td ST Trás os Montes To be added T545 Trp B EsPraiaadinha To be added 37 T495 Td ST Trás os Montes To be added T548 Trp B Fajã de Água To be added T496 Td ST Trás os Montes To be added T549 Trp B Cova Rodela To be added T497 Td ST Tarrafal To be added T550 Trp B Porto de Ferreiros To be added 38 T499 Td ST Tarrafal To be added T556 Trp B Palhal To be added T500 Td ST Tarrafal To be added T561 Trp B Chão de Sousa To be added 39 T501 Td ST Flamengos To be added T565 Trp B Chão de Aguada To be added T502 Td ST Flamengos To be added T569 Trp B To be added 40 T503 Td ST Flamengos To be added T570 Trp B Cachaço To be added T504 Td ST Flamengos To be added T571 Trp B Morro Largo To be added 41 T505 Td ST Jalalo Ra mos To be added T575 Trp B Campo da Porca To be added T507 Td ST To be added T576 Trp B Chão Queimado To be added 42 T508 Td ST Serra Malagueta To be added T607 Trm M Calheta de Cima To be added T509 Td ST Serra Malagueta To be added T610 Trm M Monte Batalha To be added 43 T510 Td ST Serra Malagueta To be added T613 Trm M Rocha Albarda To be added T511 Td ST Serra Malagueta To be added T616 Trm M Volta Grande To be added 44 T512 Td ST Porto Madeira To be added T619 Trm M Morro To b e added T513 Td ST Porto Madeira To be added T621 Trm M To be added 45 T514 Td ST Barragem To be added T625 Trm M Casas Velhas To be added T515 Td ST Barragem To be added T628 Trm M Fig. da Horta - Pilão Cão To be added 46 T516 Td ST Ba rragem To be added T631 Trm M Rb.ra D. João To be added T522 Td ST Praia Baixo To be added T634 Trm M To be added 47 T523 Td ST Praia Baixo To be added T637 Trm M Laje Branca To be added T524 Td ST Praia Baixo To be added T640 Trm M Mont e Branco To be added 48 T526 Td ST Porto Rincão To be added T643 Trm M Pilão Cão de Cima To be added T527 Td ST Porto Rincão To be added T645 Trm M Pêro Vaz To be added 49 T528 Td ST Porto Rincão To be added T647 Trm M Pêro Vaz To be added T529 Td ST Entre Picos de Rede To be added T650 Trm M Ponta Rabil To be added 50 T530 Td ST Entre Picos de Rede To be added T651 Trm M Monte Batalha To be added T531 Td ST Entre Picos de Rede To be added T653 Trm M To be added 51 T533 Td F Campanas de Baixo To be added T657 Trm M Monte Vermelho To be added T534 Td F Monte Calhau To be added cv90 Trm M To be added 52 T536 Td F Monte Calhau To be added cv91 Trm M Pêro Vaz To be added T537 Td F Monte Calhau To be added cv92 Trm M Pilão Cão de Cima To be added 53 T538 Td F Luzia Nunes To be added cv93 Trm M Rb.ra D. João To be added 54 55 56 57 58 59 60 Page 27 of 30 Journal of Biogeography

27 1 2 3 Table 2 Mitochondrial cyt b diversity, neutrality tests and demographic parameters in the 14 Evolutionarily 4 Significant Units (ESU) of Tarentola from Cape Verde Islands. (N), sample size; (Pi), nucleotide diversity; (Hd), 5 6 haplotype diversity; (h), number of haplotypes; (S), segregation sites; (Fs) Fu's statistics; (SDD), Sum of Squared 7 deviation statistics; (τ), Tau; (θ0), initial Omega; (θ1), final Omega and (t), expansion time for the six populations for 8 which the tests suggested expansion. Statistical significant p values (* p< 0.05, ** p< 0.01). Taxa and island 9 abbreviation as in Table 1. 10 11 Group ESU N Pi h Hd S Fs SSD τ θ0 θ1 t (years) 1 – Trb BV 17 0.00451 4 0.654 4 0.53603 0.03746 2.461 0.00200 2.750 12 2 – Td SN 16 0.00289 6 0.542 7 -3.10275** 0.00347 1.537 0.00000 1.487 139.728 13 a 3 – Td F 39 0.00373 12 0.75 12 -8.36432** 0.01009 1.256 0.00352 99999 114.169 4 North – Td ST 66 0.00574 11 0.717 11 -2.86202 0.01050 2.908 0.00000 3.32168 14 4 South – Td ST 72 0.00867 18 0.815 17 -6.62733** 0.02075 3.383 0.00000 7.461 307.546 15 b 1 – Ts SV 52 0.00342 12 0.632 11 -7.96204** 0.00325 1.367 0.00000 3.470 124.273 2 – Tcr SL+br+ra 24 0.00377 8 0.764 7 -3.92264* 0.01285 1.316 0.00000 99999 119.637 16 - 3 – Tcc SA 44 0.01241 23 0.942 25 0.10558** 1.309 0.00000 99999 17 13.00275** c Tcn SN 49 0.00576 11 0.844 10 -3.53715* 0.00142 1.85 0.00000 99999 168.182 18 1 – Tg br+ra 6 0.00198 2 0.6 1 0.79518 0.05428 0.947 0.00000 99999 19 d 2 – Trr ST 23 0.00172 3 0.379 2 -0.03308 0.00421 0.887 0.00000 0.900 20 3 & 4 – Trp B+ro+F For 26 0.00660 6Peer 0.702 9 0.17212 Review 0.03556* 0.998 0.00000 99999 21 4 – Trp B+ro 23 0.00269 5 0.628 4 -1.47199 0.01347 0.932 0.00200 99999 22 5 – Trm M 25 0.00341 6 0.577 6 -1.80635 0.00850 0.242 0.84199 99999 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Biogeography Page 28 of 30

28 1 2 3 Table 3 Correlation ρ values between (N), sample size, genetic variability parameters: (h), number of haplotypes; 4 number of (ESU), Evolutionarily Significant Units and geographical and ecological characteristics of the islands: size 5 6 (Area, Perimeter), altitude (maximum, mean and media), location (latitude and longitude of the centroid of the 7 island) and habitat availability (number of habitats). Statistical significant p values (* p< 0.05, ** p< 0.01). 8 9 Habitat 10 Size Altitude Location availability 11 Area Perimeter maximum mean media longitude latitude habitats N 0.7802** 0.6967** 0.8022** 0.7503** 0.8471** -0.2220 0.0681 0.8912** 12 h 0.6785** 0.5414 0.7624** 0.7246** 0.8020** -0.3492 0.1724 0.7967** 13 ESU 0.1949 0.1283 0.2591 0.2952 0.3298 -0.0469 -0.1826 0.2957 habitats 0.8578** 0.8667** 0.9289** 0.8577** 0.9111** -0.0867 0.1245 14 15

16 17 18 19 20 For Peer Review 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 29 of 30 Journal of Biogeography

1 2 3 4 5 SUPPORTING INFORMATION FOR 6

7 8 Insight into an island radiation: the Tarentola geckos 9 10 of the Cape Verde archipelago. 11 12 13 By R. Vasconcelos, S. Carranza & D. J. Harris 14 15 16 17 18 Appendix S1 Genetic differentiation between populations belonging to the same network : Snn values. Statistical 19 20 significant p values (*p<0.05,For ** p<0.01). Peer Review 21 Populations Snn 22 23 a4 North a4 South 1.00000** d2 d4 1.00000** 24 d2 d3 1.00000** 25 d2 d1 1.00000** d4 d3 1.00000** 26 d4 d1 1.00000** 27 d3 d1 1.00000** 28 b3 b1 0.96716* 29 b3 b2 0.97386** 30 b1 b2 0.97368** 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Journal of Biogeography Page 30 of 30

1 2 3 4 SUPPORTING INFORMATION FOR 5 6 7 Insight into an island radiation: the Tarentola geckos 8 9 of the Cape Verde archipelago. 10 11 12 By R. Vasconcelos, S. Carranza & D. J. Harris 13 14 15 16 17 Appendix S2 Variables used in the correlation analyses between genetic variability and geological and ecological 18 features of the Cape Verde Islands. (N), number of samples; (h) number of haplotypes; (ESU), number of 19 20 Evolutionarily SignificantFor Units, size (areaPeer and perimeter), Review altitude (maximum, max, mean and media), Location 21 (latitude and longitude of the centroid of the island) and number of habitats for each island. For islands codes see fig. 22 1 or Table 1. 23 Genetic Habitat 24 variability Size (Km2/ Km) Altitude (m) Location (DD) availability 25 Island N h ESU Area Perimeter maximum mean media Longitude Latitude habitats

26 M 24 5 1 273.55 109.15 392 46 159 -23.1614 15.21709 7 27 ST 159 32 3 1003.96 326.65 1339 275 586 -23.6247 15.08353 13 28 sm 2 2 1 0.07 1.44 24 4 8 -23.5075 14.9076 1 29 F 42 13 2 471.42 121.56 2781 865 1339 -24.3844 14.92816 12 30 B 19 5 1 62.87 71.16 959 382 459 -24.7056 14.85122 9 31 ro 4 1 1 3.03 35.91 97 3 34 -24.6610 14.96909 2 32 BV 17 3 1 630.95 134.93 360 57 164 -22.8144 16.09726 7 33 S 0 0 0 220.88 124.01 381 31 137 -22.9315 16.73702 7 SN 63 17 2 345.82 209.42 1282 269 563 -24.2569 16.59846 13 34 SV 52 12 1 225.40 113.75 711 119 318 -24.9679 16.84547 12 35 SL 15 5 1 34.72 34.90 351 21 144 -24.7452 16.76634 6 36 ra 12 6 2 5.79 9.60 135 22 55 -24.5877 16.61791 3 37 br 3 3 2 2.77 10.85 322 31 137 -24.6702 16.65844 2 38 SA 43 22 1 785.11 203.55 1971 654 969 -25.1699 17.05633 12 39

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60