Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards

Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Alvarina Santos Couto Mestrado em Biodiversidade, Genética e Evolução Departamento Biologia Centro de Investigação em Biodiversidade e Recursos Genéticos (CIBIO) 2014 Orientadora Catarina Pinho, Post-Doc, Centro de Investigação em Biodiversidade e Recursos Genéticos (CIBIO), Universidade do Porto. Coorientadora Sara Rocha, Post-Doc, Centro de Investigação em Biodiversidade e Recursos Genéticos (CIBIO), Universidade do Porto e Departamento de Xenética, Bioquímica e Inmunoloxía, Facultad de Bioloxía, Universidade de Vigo. Todas as correções determinadas pelo júri, e só essas, foram efetuadas. O Presidente do Júri, Porto, ______/______/_________ Charles Darwin FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards i Acknowledgements A realização desta dissertação não teria sido possível sem a ajuda e apoio de várias pessoas. Em primeiro lugar gostaria de agradecer de forma especial à minha orientadora, Catarina Pinho, e co-orientadora, Sara Rocha, que despertaram o meu interesse pela genética e pelas Podarcis e me acolheram no projeto que deu origem a esta tese. Por todo o entusiasmo contagiante e ensinamentos, todo o interesse e disponibilidade e por terem sido dos melhores orientadores que poderia ter tido. Também gostaria de expressar um agradecimento especial à “equipa Podarcis” do CIBIO pelo input. Em especial à Antigoni Kaliontzopoulou, ao Miguel A. Carretero e ao James Harris pela partilha dos seus conhecimentos, discussões e comentários que ajudaram no meu trabalho. Agradeço a todos os que colaboraram nesta tese cedendo amostras e à Carolina Pereira e à Carla Luís pela contribuição na correção manual das sequências. Ao Pedro Tarroso pelos scripts e disponibilidade. Aos restantes colegas do CIBIO agradeço a boa-disposição, amizade e entreajuda. Agradeço a todos os meus colegas de mestrado, em especial ao Pedro Cardoso e à Mariana Ribeiro, por todas as conversas e apoio. Ao Bruno, por todo carinho e apoio, que sempre aturou as minhas mudanças de humor e me ouviu a falar de lagartixas mesmo sem entender o que eu dizia. Obrigada também por todos os scripts e ajudas em programas que ateimavam em não funcionar. Finalmente, à minha família por todo o apoio. Aos meus pais, irmã e às minhas tias que sempre me apoiaram e sem os quais aqui não tinha chegado. I would also like to express my gratitude to the Group of Phylogenomics at the University of Vigo that provided a computational cluster of extreme importance to perform the large-scale analyses of this thesis. Thank you David, for hosting me for two weeks and for helping me with guidance and comments regarding my work. I must also acknowledge Diego Mallo, Leonardo Martins and Ramon for their help in the important analyses of this thesis. ii FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards This work fits in the framework of the research project "On the road to speciation: an integrated analysis of the evolution of reproductive isolation in a cryptic species complex." - COMPETE and by national funds through FCT - PTDC/BIA-BEC/102179/2008; PI: Catarina Pinho. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards iii Resumo Avanços recentes no campo da filogenética permitem uma inferência abrangente dos processos evolutivos a partir de dados multilocus. Inferir relações evolutivas entre espécies, especialmente quando estas divergiram recentemente, ou muito rapidamente, oferece desafios significativos. Este é o caso das lagartixas do género Podarcis da Península Ibérica e do Norte de África, cuja taxonomia permaneceu controversa mesmo apesar dos vários estudos baseados em DNA mitocondrial (mtDNA), aloenzimas e morfologia. Neste trabalho foram usados 30 genes nucleares para 170 indivíduos representativos de todos morfotipos e linhagens de DNA mitocondrial conhecidas, a fim de avaliar os níveis de polimorfismo genético e inferir a árvore de espécies das lagartixas do género Podarcis da Península Ibérica e do Norte de África. Para isso, foi utilizada uma variedade de métodos para estimar a árvore das espécies abrangendo métodos estatísticos de distancia (NJst) e máxima pseudo-verosimilhança (MP-EST), métodos Bayesianos de supertree  (Guenomu) e probabilísticos Bayesianos (*BEAST). Todos estes têm em conta a persistência de polimorfismo ancestral, processo que se pensa ser a principal causa da incongruência entre árvores de genes e árvores de espécies no caso das Podarcis. As sequências nucleares apresentam altos níveis de partilha de haplótipos em praticamente todos os genes entre as linhagens de DNA mitocondrial previamente definidas. De acordo com estes genes e uma análise de agrupamento Bayesiana que atende à maximização do equilíbrio de Hardy-Weinberg dentro de cada grupo, as 17 linhagens de DNA foram subdivididas em 24 grupos geneticamente distintos. Além disso, também foi detectado fluxo génico entre grupos através da observação de genótipos miscigenados; alguns destes casos correspondem a exemplos previamente documentados (ex. P. bocagei/P. guadarramae lusitanica), mas outros correspondem à descrição de fluxo génico pela primeira vez (P. vaucheri “Spain”/P. carbonelli; P. vaucheri “Spain”/P. virescens; P. hispanica “Galera”/P. hispanica “Albacete/Murcia” e P. liolepis/P. hispanica sensu stricto/P. atrata). Algumas situações de clara discordância citonuclear foram reveladas. A fim de remover evidências claras de fluxo génico recente, indivíduos com uma proporção de menos de 95% do seu genoma atribuído a um único grupo foram excluídos para a inferência filogenética. As árvores das espécies obtidas corroboraram algumas das relações inferidas pelo mtDNA, mas também revelaram outras completamente diferentes. Um grupo de formas do Este da Península Ibérica (P. hispanica sensu stricto, P. liolepis e P. atrata) foi consistentemente recuperado pelos diferentes métodos, com altos valores de suporte e com apenas pequenas variações nas relações entre linhagens. O clado composto por formas de P. hispanica do Norte x FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure 3.3. Phylogeny of Iberian and North African Podarcis species as estimated with the NJst method using each gene ML tree topology from the “haplotypes” dataset; a) “best” NJ species-tree; b) NJst consensus (branch lengths are transformed and do not reflect any amount of evolution). Trees were inferred unrooted and rooted for visualization in the branch leading to P. erhardii, P. taurica, P. sicula and P. tiliguerta. The numbers on the nodes indicate multilocus bootstrap support values for branches calculated following Seo, 2008. .................... 34 Figure 3.5. Extended consensus of the posterior distribution of species-trees obtained with Guenomu. Values indicate the posterior probabilities. ................................................................ 36 Figure 3.7. Posterior density of species-trees (cloudogram) from *BEAST analyses for the Iberian and North African Podarcis species for all loci, for a chain length/tree prior of 341M/Yule. Each thin line corresponds to a sampled tree, so darker areas correspond to higher density of trees in agreement. Blue sets of trees represent those with the same topology as the most popular tree, the next most popular set appearing in red, and the third most popular green. Remaining trees are all dark green. Uncertainty in node heights is shown by smears around the mean node height. Maximum clade credibility tree and posterior probabilities of support above 50 are show in blue. ..................................................................................................................... 37 Figure 3.6. Posterior density of species-trees (cloudogram) from *BEAST analyses for the Iberian and North African Podarcis species for 21 loci, for a chain length/tree prior of a) 605M/Yule; b) 168M/coalescent. Each thin line corresponds to a sampled tree, so darker areas correspond to higher density of trees in agreement. Blue sets of trees represent those with the same topology as the most popular tree, the next most popular set appearing in red, and the third most popular green. Remaining trees are all dark green. Uncertainty in node heights is shown by smears around the mean node height. Maximum clade credibility tree and posterior probabilities of support above 50 are show in blue. .................................................................... 38 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards xi List of Tables Table 1.1. The most used programs for estimating species-tree and its characteristics. BCA, Bayesian concordance analysis; MSC, Multispecies coalescent model. ..................................... 12 Table 2.1. Number of sequences for each species and loci. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli, PGL, P. guadarramae lusitanica; PGG, P. guadarramae guadarramae; PV, P. virescens; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga"; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera”; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PE, P. erhardii; PS, P. sicula; PTA, P. taurica; and PM, P. muralis, PT, P. tiliguerta. ............................................................................................................ 18 Table 3.2. Number of admixed individuals and designation of the clusters to which they were assigned. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli; PGG, P. guadarramae guadarramae; PGL, P. guadarramae lusitanica; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga”; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PV, P. virescens; and PM, P. muralis. ..................................................................................................................... 28 Table 3.3. Number of sequences for each loci and unit defined with STRUCTURE. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli; PGG, P. guadarramae guadarramae; PGL, P. guadarramae lusitanica; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga”; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PV, P. virescens; PM, P. muralis; PE, P. erhardii; PS, P. sicula; and PTA, P. taurica. .............................................................................. 29 Table 3.1. Summary statistics and neutrally test for the 30 loci analysed in this study. Nseqs, number of sequences; NSites, total sequence lengths – () excluding sites with gaps / missing xii FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards data; h, number of haplotypes; Hd, haplotype diversity; Eta, total number of mutations; S, number of segregating sites, π, nucleotide diversity, θw, Theta-Waterson. ................................ 31 Table 3.4. Models of sequence evolution obtained with jModeltest2 and the models used in *BEAST for each gene. Nst, number of substitution rate categories. ......................................... 33! FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards xiii List of Abbreviations ACM4 - Acetylcholinergic Receptor M4 AICc - Akaike Information Criterion with correction BCA - Bayesian Concordance Analysis BI - Bayesian inference C-mos - Oocyte maturation factor Mos ESS - Effective Sample Sizes GDL - Gene Duplication and Loss GT - Gene-tree HGT - Horizontal Gene Transfer ILS - Incomplete Lineage Sorting IUCN - International Union for Conservation of Nature MC1R - Melanocortin Receptor 1 MCMC - Markov Chain Monte Carlo ML - Maximum Likelihood MtDNA - Mitochondrial DNA Mya - Million Years Ago NFYCint16 - Nuclear Transcription Factor Y, intron 16 Nst - number of Substitution rate PKM2int5Muscle Pyruvate Kinase 2, intron 5 RAG1 - Recombination Activating Protein 1 RAG2 - Recombination Activating Protein 2 SIC - Simple Indel Coding SNPs - Single Nucleotide Polymorphisms ST - Species-tree β-fibint7 - Beta-fibrinogen, intron 7 xiv FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 1 Chapter 1 Introduction Species definition is an important task in biology, but also one that already lead to more than half century of controversy. Determining what is a species has been difficult due the impossibility in observing some processes directly, and there is still no consensus on a single (and unifying) definition of species. The study of species-complexes, groups of closely related incipient species, such as the wall lizards of the genus Podarcis, is one of the best windows into the complexity of processes of diversification, and into the difficulty of defining and inferring the history of species differentiation. 1.1 The Iberian and North Africa clade of Podarcis Podarcis (Lacertidae, Squamata) are diurnal lizards with high morphological and ecological similarity between species. They are habitat generalists and use rocks, trunks, vegetation or bare ground for thermoregulation, foraging and shelter. Ecological modeling revealed that temperature is a key factor for some species (Carretero et al., 2006). These lizards evolved and diversified in the Mediterranean basin and they are widely distributed in Europe and North Africa (Arnold and Ovenden, 2002). Currently, this genus comprises 23 species according to international databases (Uetz and Hosek, 2014). The object of this thesis is a well-defined, monophyletic group, inside this genus: the Iberian and North Africa clade (Harris et al., 2005). Seven species are currently recognized within this group: P. bocagei, P. carbonelli, P. hispanica, P. vaucheri, P. liolepis, P. guadarramae and P. virescens (Uetz and Hosek, 2014; Geniez et al., 2014). In terms of conservation, these species are abundant and widely distributed and thus not considered threatened. The exception is P. carbonelli that is classified as “endangered” by the International Union for Conservation of Nature (IUCN) Red List of Threatened Species, due its fragmented distribution and loss of habitat (Sá-Sousa et al., 2009). 2 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 1.1.1 Taxonomy and species delimitation The taxonomy of Podarcis has for long been controversial. Species delimitation has been difficult and successive studies disagreed on distribution and taxonomic classifications. This is particularly true for the Iberian and North African clade, in which the debate continues even after extensive molecular, morphological, ecological and physiological studies, having been performed (reviewed in Carretero, 2008). After decades of controversy over morphological variation and classification attempts (e.g. Pérez-Mellado and Galindo, 1986; Geniez, 2001; Sá-Sousa, 2001) the first comprehensive studies regarding genetic variation in this group were based on mitochondrial DNA (mtDNA). These surveys evidenced the existence of several groups highly differentiated and with a strong association with geography and morphology (Harris and Sá-Sousa, 2002; Harris et al., 2002; Pinho et al., 2006). Phylogenetic and biogeographic hypotheses were then constructed, which have been constantly changing as new lineages are discovered (Harris and Sá-Sousa, 2002; Harris et al., 2002; Pinho et al., 2006; Kaliontzopoulou et al., 2011). Based on mtDNA (Kaliontzopoulou et al., 2011), the distribution of these species is as shown in Figure 1.1. P. bocagei inhabits the Iberian Northwest and P. carbonelli has a fragmented distribution along the Western Iberian coast, with an isolate in Southern Spain. P. guadarramae comprises P. guadarramae lusitanica from Northwest and central Iberia and P. guadarramae guadarramae from central Iberia (former P. hispanica type 1A and 1B, respectively; Geniez et al., 2014). P. hispanica type 2 (now called P. virescens; Geniez et al., 2014) is found in central and South-west Iberia. P. guadarramae and P. virescens are visibly complementary, with different ecological affinities (altitude and temperature) and the lack of range overlap suggesting mutual exclusion, at least in Portugal (Sá-Sousa et al., 2000). P. vaucheri inhabits the South of the Iberian Peninsula and throughout Morocco and Algeria. P. liolepis inhabits the Northeast of the Iberian Peninsula and Southern France and includes the form from the Columbretes islands (Spain) which was formerly known as P. atrata. Despite the synonymization of these two species (Renoult et al., 2010), for simplicity in this thesis we use the name P. atrata to designate individuals collected in the Columbretes. P. hispanica is now considered a paraphyletic complex of mtDNA lineages and, according to the terminology used for e.g. in Kaliontzopoulou et al., 2011 basically includes all forms which have not yet been elevated to the species status or assigned to one of the other species. This includes lineages from Morocco, Algeria and Tunisia in North Africa, and also from Southeastern Spain, where three divergent mtDNA lineages can be found (named “Galera”, “Albacete/Murcia” and “sensu stricto” in Kaliontzopoulou et al., 2011). FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 3 The distribution of these species is generally parapatric, with some exceptions. The pair which overlaps the most is P. guadarramae lusitanica and P. bocagei in the Northwestern Iberia (Pérez-Mellado, 1981a), but there are also cases of sympatry involving P. carbonelli and P. g. lusitanica and P.carbonelli and P. g. guadarramae in the Western-central system (PérezMellado, 1981b) P. virescens and P. carbonelli in the central zone of Portugal (Sá-Sousa, 2001) and P. hispanica “Galera” and P. hispanica sensu stricto lineages (Kaliontzopoulou et al., 2011). In Doñana, in Southern Spain, sympatric populations of P. vaucheri and P. carbonelli are found (Harris et al., 2002). Figure 1.1. Map of the Iberian Peninsula and North African showing the estimated distribution ranges for Podarcis mtDNA lineages. Image made according to Pinho et al., 2008 and Kaliontzopoulou et al., 2011. Nuclear markers, namely protein loci, confirm the evolutionary units inferred based on mtDNA, with some exceptions, giving support to the idea that these entities are really distinct species (Pinho et al., 2007a). Genealogies of two nuclear introns showed high degrees of haplotype sharing, by opposition to the monophyly of mtDNA (Pinho et al., 2008; Renoult et al., 2009; Kaliontzopoulou et al., 2011). This result was interpreted as mostly the outcome of incomplete lineage sorting, coupled with limited gene flow, suggesting recent speciation and incomplete barriers to gene flow. Indeed, variable degrees of introgression have been found, from rare hybridization between e.g. P. bocagei and P. guadarramae lusitanica (Pinto, 2013) to probably complete nuclear swamping resulting in a striking case of cytonuclear discordance P. liolepis P. carbonelli P. guadarramae lusitanica P. guadarramae guadarramae P. bocagei P. virescens P. atrata P. hispanica sensu stricto P. hispanica “Albacete/Murcia” P. hispanica “Galera” P. vaucheri “Southern-central Spain” P. vaucheri “Southern Spain” P. vaucheri “Morocco/Algeria” P. hispanica “Jebel Sirwah” P. hispanica “Azazga” P. hispanica “Batna” P. hispanica “Tunisia/Algeria” 4 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards between forms in Southeast Iberia (Pinho et al., 2007a, 2008; Renoult et al., 2009). A study in an Iberian contact zone (between P.bocagei and P.carbonelli) indicated that gene flow exists, but the hybridization observed is highly localized and bimodal (Pinho et al., 2009), pointing to the existence of strong reproductive barriers. One of the few comprehensive morphological surveys within the clade (Kaliontzopoulou et al., 2012) showed that there is morphological differentiation consistent with genetic variation, but also that the distinctiveness is subtle and only evident on a pairwise or one-against-all comparative basis. In short, i) the high differentiation suggested by mtDNA divergence between lineages, ii) the concordant structure inferred by nuclear markers, iii) the morphological differentiation patterns and iv) the existence of barriers to gene flow in contact zones suggest that the forms are most likely distinct species. However, the persistence of high levels of shared polymorphism and the permeability to gene exchange suggest that Podarcis species arose quite recently and have not concluded the process of speciation. Remarkably, although Podarcis are nowadays one of the best-studied groups of the Iberian fauna, the evolutionary history of this group remains poorly understood in many aspects. This is partly due to the high rate of cryptic lineage discovery in this system (Kaliontzopoulou et al., 2011), suggesting that there may still be undiscovered “species”. Also, recently detected mtDNA lineages such as those of P. hispanica from Batna and Azazga, in Algeria (Lima et al., 2009), P. vaucheri from South-central Spain or P. hispanica from the Albacete/Murcia area (Kaliontzopoulou et al., 2011) or forms with sampling difficulties such as P. liolepis from the Columbretes (former P. atrata) have not been characterized from a nuclear marker perspective, implying that they may or may not correspond to distinct evolutionary entities. Finally, there is discordant evidence from different studies concerning species distinction and gene flow, particularly in the Southeastern corner of the Iberian Peninsula (Pinho et al., 2007a, 2008; Renoult et al., 2009; Kaliontzopoulou et al., 2011). 1.1.2 Previous phylogenies and biogeographic hypotheses Previous assessments of the phylogeny of this group were done using allozymes (Pinho et al., 2003, 2007a) and mitochondrial DNA (Harris and Sá-Sousa, 2001; Harris and Sá-Sousa, 2002; Harris et al., 2002; Pinho et al., 2006; Kaliontzopoulou et al., 2011). Allozyme markers, despite their general utility in detecting species boundaries and hybridization in this system, showed a striking lack of resolution in the estimation of relationships between species (Pinho et al., 2007a). The proposed hypotheses recovered most lineages as monophyletic, but failed to suggest any relationship among species/lineages, exhibiting a star-like topology, which could be due to low resolution or be the result of rapid diversification. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 5 On the other hand, analyses using mtDNA recovered, in general, a well supported phylogeny. The most recent study (Kaliontzopoulou et al., 2011) recovered three main clades: one includes all forms from Western and Central Iberia (P. bocagei, P. guadarramae, P. carbonelli and P. virescens); another comprises the Southeastern Iberian and North African lineages (P. hispanica sensu stricto, P. hispanica “Albacete/Murcia”, P. vaucheri with three divergent lineages - one found in North Africa and two others in Southern Spain - and the P. hispanica forms from Tunisia, Algeria and Morocco); and a third clade, sister to the other two, that corresponds to P. liolepis and P.hispanica “Galera”, from Southern Spain (Figure 2). The status of P. atrata, endemic to the Columbretes archipelago, is considered doubtful. In a mitochondrial perspective, it is recognized as P. liolepis (Harris and Sá-Sousa, 2002). Figure 1.2. Estimate of relationships between Iberian and North African Podarcis based on maximum likelihood analyses of mtDNA. Above the node the Bayesian posterior probabilities and below the nodes the maximum likelihood/maximum parsimony bootstraps are given. Image modified from Kaliontzopoulou et al., 2011. The estimated divergence time between these mtDNA lineages (Kaliontzopoulou et al., 2011) predates the Messinian salinity crisis (~5Mya) and it was proposed that geological events taking place at the end of the Miocene in the area now corresponding to South-eastern Iberia, the rift mountain range in Morocco, and the North of Algeria. This was followed by the abrupt 12 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Most of these methods have been tested on simulated datasets but applications to empirical data are still relatively scarce, particularly for closely related species. In these cases, genetic distances between species can be small (or zero) and completely overlap with intraspecific distances, making it very hard to infer GT's without error, and harder to infer non-flat ST's distributions. Taking into account multiple individuals per "species" thus becomes necessary, but this feature makes GT (and ST) inference to become also exponentially harder. This type of data is incredible challenging and the relative performance of many methods is still not well understood in these cases. Table 1.1. The most used programs for estimating species-tree and its characteristics. BCA, Bayesian concordance analysis; MSC, Multispecies coalescent model. Name Method Input Output RF Distance Rooted gene-trees Rooted supertree consistent with maximum number of splits in the input trees, without branch lengths. MulRF Distance Unrooted gene-trees Unrooted supertree that minimizes the RF distance to the input multi-labeled trees, without branch lengths. MRP Parsimony Rooted gene-trees Unrooted consensus supertree of all multiple trees with the same maximum parsimony score. BUCKy Non-parametric BCA Unrooted distributions Unrooted supertree with the clades with the highest amount of genomic support, without branch lengths. ASTRAL Quartet compatibility Unrooted gene-trees Unrooted supertree that agrees with the largest number of quartet trees. iGTP Parsimony Un/rooted gene-trees Rooted or unrooted supertree without branch lengths or nodal support. Guenomu Bayesian supertree; Parsimony Unrooted gene-trees Rooted species-trees with posterior distributions. STAR Distance Rooted gene-trees Rooted species-tree without branch lengths and supported by bootstrap analysis. STEAC Distance Rooted gene-trees Rooted species-tree without branch lengths and supported by bootstrap analysis. iGLASS Distance Rooted gene-trees Rooted species-tree without branch lengths and supported by bootstrap analysis. NJst Distance Unrooted gene-trees Unrooted species-tree without branch lengths and supported by bootstrap analysis. STEM Maximum likelihood + Distance Rooted gene-trees Rooted species-tree with the highest likelihood and with branch lengths but without posterior probabilities or divergence times. MP-EST Maximum pseudolikelihood Rooted gene-trees Rooted species-tree with branch lengths and supported by bootstrap analysis. STELLS Maximum likelihood Rooted gene-trees Rooted species-tree with the highest likelihood and branch lengths. BEST Bayesian analysis; MSC DNA/protein sequences Rooted species-tree with branch lengths, posterior probabilities and divergence times. *BEAST Bayesian analysis; MSC DNA/protein sequences Rooted species-tree with branch lengths, posterior probabilities and divergence times. PHYLDOG Likelihood DNA/protein sequences Rooted species-tree with branch lengths, posterior probabilities but without divergence times. Supertree Methods Full Probabilistic Methods FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 13 1.3 Aims and organization of the thesis The main goal of this thesis is to explore the utility of several methods of inference of species-trees to infer the phylogeny of the Iberian and North Africa Podarcis wall lizards, based on 30 nuclear molecular markers. Several tools were applied to this DNA sequence dataset in order to perform alignment, alignment trimming, allele phasing, inference of gene genealogies and of "population" structure, gene flow estimates and, finally, species-tree inference. Specifically, it was aimed at: 1) Evaluating the levels of genetic polymorphism for the above-mentioned 30 loci dataset, and its distribution within and among Podarcis lineages; 2) Evaluating the evolutionary distinctiveness of mtDNA lineages, particularly of those which had not been studied before, using nuclear DNA markers; and 3) Determining the phylogeny of the Iberian and North Africa Podarcis species complex with different methodologies able to take into account disagreement between GT's and the ST due to incomplete lineage sorting. This thesis is organized in five chapters. The first chapter is a general introduction providing the necessary background information about the biology and controversial taxonomy of the study group and about the inference methodologies currently available to address our questions. In the next three chapters the methods used, the results and the discussion are presented. The 5th chapter corresponds to conclusions and future perspectives, with some further comments about what can be done in the future to address our main questions, what can be improved on current ongoing work, and new questions that have arisen. To finish, the literature references used along this work and Appendix data are presented. 14 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 15 Chapter 2 Material and Methods 2.1 Sample collection, DNA extraction and amplification The work leading to this report consisted essentially in alignment and statistical analysis of previously produced DNA sequences. Field and laboratory work were therefore not part of this work. DNA extraction, PCR conditions, amplification and sequencing were essentially carried out using the conditions described in Pinho et al., (2010) and Pereira et al., (2013). 2.2 Taxon and gene sampling For this project a set of 30 unlinked nuclear loci was used. This set included loci that were previously sequenced in Podarcis, such as β-fibint7 (Pinho et al., 2008) or the set of markers from Pinho et al. 2010; exons widely used in other Squamata, such as ACM4 (Gamble et al., 2008), PDC (Salvi et al., 2013), C-mos (Godinho et al., 2005), RAG1 (Pinho, unpublished), RAG2 (Hoegg et al., 2004) and MC1R, NFYCint16 e PKM2int5 (Pinho et al., 2010); plus a set of 21 anonymous markers developed for the genus Podarcis (Pereira et al., 2013). From a collection of tested markers, these loci were the ones which were successfully amplified and sequenced and for which good quality sequences were obtained. One hundred and seventy individuals, representative of all currently known species, morphotypes and mtDNA lineages of the Iberian and North African Podarcis clade, were Sanger sequenced. This data set included between five and 15 individuals per each of the 16 lineages described in Kaliontzopoulou et al., (2011) plus individuals of P. atrata. As outgroups, 20 individuals belonging to five species representative of all other clades of the genus (Salvi, personnal communication) were used: P. muralis (a species that exists throughout Europe, including the Iberian Peninsula, but that belongs to a different clade than our focal species), P. sicula, P. tiliguerta, P. taurica and P. erhardii. 16 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Sample codes, localities and the mtDNA correspondence for each individual are presented in Table A2 in the Appendix. The geographical origin of each sample is represented in Figure 2.1. Figure 2.1. Map of the Iberian Peninsula and North African showing the geographical origin of each sample used in this study and its respective mtDNA lineage/species. 2.3 Dataset assembly Sequences were examined and corrected by eye in Sequencher v.4.1.4. (Gene Codes Corporation). Several sequences were heterozygous for insertion/deletion polymorphisms, and the method outlined by Flot et al., (2006) was used to resolve them. For the majority of the loci, as they contained a considerable amount of indels, alignment was not trivial. We experimented a few alignment algorithms, such as the E/L/G-INS-I implemented on MAFFT v.7.122 (Katoh and Standley, 2013), as well as the automated method implemented in PRANK v.140110, that takes into account the evolutionary distances between sequences while also recognizing insertions and deletions as distinct evolutionary events (Löytynoja and Goldman, 2008). This method seems to outperform the other methods and thus was used for downstream analyses. A few final adjustments, when considered necessary, were made by hand. Because many of the alignments had large regions with indels, plus some highly variable regions, and this can be problematic for phylogenetic analyses, trimAL v.1.4 (Capella-Gutiérrez 0 P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei 0 40 40 10 10 20 20 Iberian Peninsula P. atrata P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. carbonelli P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. liolepis P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. guadarramae guadarramae P. guadarramae lusitanica P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. virescens P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. hispanica “Galera” P. hispanica sensu stricto P. hispanica “Albacete/Murcia” P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri “Southern-central Spain” P. vaucheri “Southern Spain” North African P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri “Morocco/Algeria” P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. hispanica “Jebel Sirwah” P. hispanica “Azazga” P. hispanica “Batna” P. hispanica “Tunisia/Algeria” Outgroups P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. vaucheri SC Spain P. hispanica Jebel Sirwah P. liolepis P. hispanica Tunisia/Algeria P. hispanica sensu stricto P. hispanica Galera P. hispanica Batna P. hispanica Azazga P. hispanica Albacete/Murcia P. hispanica type 2 P. hispanica type 1B P. hispanica type 1A P. carbonelli P. atrata P. vaucheri Morocco/Algeria P. vaucheri S Spain P. muralis P. erhardii P. sicula P. taurica P. tiliguerta P. bocagei P. muralis P. erhardii P. sicula P. tiliguerta P. taurica FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 17 et al., 2009) was used to remove large indels and poorly aligned regions. The heuristic “automated1” method was used to automatically decide the best method to trim each specific alignment, between “gappyout”, “strict” and “strictplus”, depending on the number of sequences, the average identity score among sequences and the average identity score for each most similar pair of sequences. In short, for each column of the alignment a gap-score is calculated and columns are sorted according to this score, producing a plot of gap-score thresholds versus percentage of the alignment below that threshold. The slope of this curve is then used to decide the optimal cut-off point for the “gappyout” option. The “strict” option removes the columns that would be deleted with “gappyout” plus blocks with at least five consecutive columns below a certain similarity cut-off. “Strictplus” is similar but selects automatically the block size to be eliminated, defined as 1% of the alignment size between a minimum size of 3 and a maximum size of 12. The main reason to use this approach instead of simply removing all indels was the fact that indels can be useful phylogenetic information (Freudenstein and Mark, 2001; Simmons et al., 2001; Young and John, 2003), and accountable for, especially in distance-related methods. The Bayesian algorithm implemented in the program PHASE v.2.1.1 (Stephens et al., 2001) was used to recover gametic phases, often assisted by the known haplotype phases determined using the Flot et al., (2006) method. The input files were prepared using DNAsp v.5.0 (Librado and Rozas, 2009) with minor modifications by hand. For each locus, the ingroup was analyzed separately from the outgroup species. Analyses were also performed independently for each of two well sampled outgroups (P. muralis and P. tiliguerta). Each dataset was analyzed using the general model for recombination rate (-MR) (Li and Stephens, 2003) with 1000 steps for burnin, one of thinning interval and 1000 main iterations. Each analysis was repeated five times with different random seeds and the consistency of the results was verified. Incongruences between runs were not accepted. The gametic phase of alleles for polymorphic sites was considered for base probabilities ≥ 0.75. This threshold limit was chosen based on the distribution of the probabilities of the inferred alleles for different runs for all loci; was chose a value that allowed us confidence in the results, but minimized the number of positions being discarded. Several inhouse perl or python scripts were used to process these data. They were used to produce the file describing the known phases, to summarize the results of different runs and then to replace the ambiguities by the inferred alleles. For P. erhardii, P. sicula and P. taurica, with fewer sequences, haplotypes were resolved by hand or left unresolved. BioEdit v.7.2.5 (Hall, 1999) was used to edit alignments when needed. The final number of sequences used for each species/loci is given in Table 2.1. 18 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Table 2.1. Number of sequences for each species and loci. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli, PGL, P. guadarramae lusitanica; PGG, P. guadarramae guadarramae; PV, P. virescens; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga"; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PE, P. erhardii; PS, P. sicula; PTA, P. taurica,; and PM, P. muralis, PT, P. tiliguerta. 2.4 Haplotype networks Median-joining networks (Bandelt et al., 1999) with MP posterior optimization (Polzin and Daneshmand, 2003) were constructed using Network v.4.612. To build these networks, all columns with indels were removed because each gap-character will count as one different mutation (5th state), thus increasing genetic distance between sequences when long indels occur. In parallel, we also codified all the indels with the simple indel coding (SIC) method (Simmons and Ochoterena, 2000) implemented in SeqState v.1.4.1 (Müller, 2005). SIC codes indels as separate characters taking into account the start/end of each indel and creates a Loci PA PB PC PH1A PH1B PH2 PHAM PHAZA PHBAT PHGAL PHJS PHSSN PHSSS PHTA PL PVMA PVSCS PVSS PE PS PTA PM PT sum ACM4 4 2 16 18 18 30 18 4 6 18 10 14 18 20 18 20 16 18 410 612 8308 β-fibint7 826 22 30 20 40 18 4 6 40 28 22 28 26 34 30 16 24 6 4 4 36 0472 C-mos 10 20 18 14 16 30 18 4 8 20 16 24 10 20 22 24 20 20 4 4 4 16 10 352 MC1R 820 20 18 18 28 18 4 8 18 16 24 10 18 24 20 20 18 410 416 10 354 NFYCint16 818 12 14 18 24 12 4 8 12 16 12 20 20 16 18 10 16 4 4 2 14 10 292 PDC 10 14 20 18 16 28 20 4 6 18 14 24 10 18 22 22 20 20 4 4 4 16 10 342 PKM2int5 10 18 22 18 20 30 20 4 8 20 16 16 18 12 20 22 16 22 4 4 4 16 10 350 RAG1 10 20 18 18 14 30 20 4 6 20 16 22 10 18 22 24 20 20 4 4 4 16 8348 RAG2 820 14 18 14 28 20 4 8 18 16 24 10 20 20 20 20 20 4 4 4 16 4334 Pod6b 10 14 10 12 14 16 10 4 6 16 14 12 18 18 22 22 612 2 2 0 16 0256 Pod7b 10 18 22 16 16 30 20 4 8 20 14 14 20 20 20 24 2 6 4 4 4 16 8320 Pod11 10 20 16 18 18 30 20 4 8 18 16 14 20 20 22 18 14 18 4 4 4 14 10 340 Pod12b 10 20 20 18 14 30 20 4 6 18 12 12 16 14 22 22 14 24 0 0 0 0 6 302 Pod13 614 14 16 14 26 14 4 8 16 12 16 20 16 22 24 10 22 4 4 4 8 2 296 Pod14 620 20 18 14 30 14 4 8 18 12 818 16 22 22 16 24 0 0 0 16 0306 Pod14b 10 20 22 16 20 20 12 4 8 12 16 14 16 18 20 20 10 18 4 4 4 14 10 312 Pod15 618 14 18 18 28 20 4 8 18 10 12 20 20 16 26 14 20 4 4 4 16 6324 Pod15b 10 20 18 18 16 28 20 4 8 18 16 16 20 20 20 20 16 22 410 416 4348 Pod16 8 4 16 18 18 28 8 4 8 18 10 16 20 18 22 24 10 14 4 4 0 14 10 296 Pod17 10 20 20 16 10 26 18 4 8 18 16 10 16 16 20 22 14 22 2 4 4 16 8320 Pod20 10 18 22 18 20 30 20 4 8 20 12 14 20 20 20 22 16 24 4 4 4 16 6352 Pod21 10 20 20 16 14 24 18 4 8 18 14 12 20 18 20 24 16 22 4 4 4 16 4330 Pod25 8 8 18 12 16 24 16 4 8 14 614 14 12 16 12 10 18 4 4 0 10 6254 Pod31 10 14 10 14 16 30 18 4 6 18 10 14 20 20 20 22 820 4 4 4 2 8 296 Pod33 10 18 16 16 14 26 20 4 8 20 16 14 18 20 16 24 824 4 4 0 16 6322 Pod38 10 16 16 14 14 30 20 4 0 16 016 20 022 24 14 14 0 0 0 16 0266 Pod43 820 16 14 10 20 16 4 8 14 12 16 18 16 18 18 14 18 4 4 2 12 4286 Pod55 10 20 20 20 16 30 20 4 8 20 14 14 20 18 22 22 16 22 4 4 0 16 10 350 Pod69 10 20 22 18 20 30 12 4 8 16 14 14 20 18 18 24 16 22 2 2 0 16 8334 Pod72 620 10 18 14 28 16 2 8 20 210 20 10 12 22 16 24 4 2 0 2 4 270 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 19 character matrix that in added to the alignment. The problem with this approach is that for indels completely comprised within other indels is impossible to determine their state. 2.5 Determining the units of analyses Although the mtDNA assignment of all samples is known a priori, this shouldn´t be used to delimit units given the possibility of mtDNA introgression among taxa. Consequently, we used STRUCTURE v.2.3.4 (Pritchard et al., 2000) to define units based on the 30 loci multilocus genotype of each individual. For these analyses, we considered each haplotype at each locus to be an independent allele, regardless of the genetic distance between haplotypes. A series of Python scripts were written to convert DNAsp haplotype distribution data files into the STRUCTURE input. P. muralis, one of the outgroups, also was included in this analysis since its geographical distribution encompasses the North of the Iberian Peninsula and there are some evidences of gene flow with species from our focal group (namely P. liolepis) (Pinho et al., 2008). Individuals with more than 60% of missing data were excluded. STRUCTURE implements a Bayesian model-based clustering algorithm to find clusters of individuals that minimize Hardy-Weinberg and linkage disequilibria without any a priori information about each individual’s origin. Given these underlying assumptions, STRUCTURE does not necessarily imply phylogenetic proximity among individuals recovered as belonging to the same cluster. Thus, including the most likely number of groups (K) as estimated by STRUCTURE, from the full data, as units of analyses, might introduce biases in estimated phylogenetic relationships. To minimize these biases, we opted by considering the highest possible subdivision in our sample supported by the data, whether the chosen groups correspond to species or phylogroups within them (the terms “units”, “lineages”, and “species” are herein used, sometimes interchangeably, and not necessarily with any taxonomic-ranking implication). In order to do so, we applied an iterative procedure to uncover structure in our data set by performing multiple separate analyses, first on the total data set and subsequently on each cluster recovered. Each analysis consisted of three steps: 1) a STRUCTURE run; 2) the choice of the appropriate number of populations present in each data set (K); 3) removal of admixed individuals, if any. All STRUCTURE runs were performed under an admixture model, for 200000 steps after 20000 steps discarded as burn-in, with 5 replicates for every possible K. K was allowed to vary from 1 to a variable number, depending on the dataset and chosen by inspecting trends in the outputs and in the likelihood values. The choice of the value of K more adequate to describe variation in each data set (X) was based on the run with the highest Ln Pr 20 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards (X|K) (following the recommendation by the developers of the program; see section 5 of STRUCTURE manual). In order to remove the influence of admixture in our data set, which would violate the assumptions of most phylogenetic methods, we chose to eliminate all individuals presenting less than 95% assignment to a single cluster. This value was chosen after inspection of the overall assignment proportions for the majority of individuals included in the analysis, typically around 99% (see Results). A threshold of 95% can be seen as perhaps too high; however, we assumed this value as a conservative cut-off in order to guarantee as little influence of gene flow as possible, even if pure individuals were inadvertently left out. This procedure was applied iteratively for each cluster defined until one of the three following possibilities were verified: a) the group contained only one individual; b) K=1 was the best K; c) (for K>1) the most part of the individuals appeared with less than 90% of the genome assigned to any of the groups (implying high gene flow or virtually no differentiation among those groups). 2.6 Conforming data to the phylogenetic methods´ assumptions Most of the methods used in this thesis make three important assumptions: 1) patterns of allele sharing are a result of ILS and not gene flow; 2) there is no recombination within loci; 3) there is free recombination among loci or knowledge about patterns of linkage among loci. 1) Gene flow As explained in the previous section above, we removed from the data set all the individuals presenting evidence of admixture between groups. Because this approach may fail to detected historical gene flow, was applied the coalescent model of divergence with gene flow, IMa2 (Hey and Nielsen, 2007) in a similar way as Pinho et al., (2008) but some preliminary results showed gene flow among most of the species and because this was a clear violation of the IM model assumptions (Strasburg and Rieseberg, 2009) we could not have confidence in our results and thus, this strategy was abandoned. 2) Recombination We used RDP3.44 (Martin et al., 2010) to assess whether sequence data were affected by recombination. For this, the possibility of recombination was investigated using three methods: RDP (Martin et al., 2010), GENECONV (Sawyer, 1989) and MaxChi (Smith and Smith, 1998); we used the option “automask” for an optimal recombination detection, setting the cut-off P- FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 21 value to 0.001. Following author recommendations (Martin, personnal communication), if recombination was not inferred for the three methods simultaneously, we assumed recombination-free alignments. 3) Patterns of linkage among loci We performed an exact test for genotypic disequilibrium using the program Genepop v.4.1.4 (Rousset, 2008) in order to evaluate whether any pair of loci were in physical linkage, thus sharing a common evolutionary tree. This analysis was performed for the five groups, among those recovered by STRUCTURE (see section 2.5), which had a sample size of at least 10 individuals (P. bocagei, P. carbonelli, P. virescens, P. vaucheri “Spain”, P. vaucheri “Morocco/Algeria 2”; see Results). 2.7 DNA sequence polymorphism To describe levels of genetic variation, summary diversity statistics for each gene were calculated for our group of interest (that is, after removing all outgroups). This was performed in DNAsp v.5.10.1 (Librado and Rozas, 2009). As this program does not accept ambiguity codes, we re-coded all unphased positions as Ns with an in-house python script. We calculated the number of haplotypes, haplotype and nucleotide diversity, the number of total mutations, the number of segregating sites and the population mutation parameter θ (Watterson, 1975). 2.8 Gene-trees inference Some of the species-trees inference methods used in this work take gene-trees or gene-trees distributions as input, thus creating a need for gene-tree estimation prior to analysis. This step was carried out under different methods. The best-fit model of evolution for each gene fragment was estimated with jModelTest2 (Guindon and Gascuel, 2003; Darriba et al., 2012), under the Akaike Information Criterion with correction (AICc). Gene-trees for each loci were estimated using maximum likelihood (ML) in RAxML v.8 (Stamatakis, 2014) and with MrBayes v.3.2.2 (Ronquist et al., 2012) for Bayesian inference (BI). For RAxML the GTR model was always used (as it is the only implemented), with or without G/I, depending on the highest ranked GTR model on the jModeltest2 weight table. The 28 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Table 3.1. Number of admixed individuals and designation of the clusters to which they were assigned. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli; PGG, P. guadarramae guadarramae; PGL, P. guadarramae lusitanica; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga”; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PV, P. virescens; and PM, P. muralis. Sample code mtDNA lineage Structure Region, Country 9.32 PHSS 0,009 (PA); 0,314 (PHSS); 0,677 (PL) Valencia, Spain B2 PHSS 0,052 (PA); 0,172 (PHSS); 0,777 (PL) Valencia, Spain Val1 PHSS 0,169 (PA); 0,309 (PHSS); 0,523 (PL) Valencia, Spain 1.15 PL 0,181 (PA); 0,151 (PHSS); 0,668 (PL) Aragón, Spain Cel1 PGL 0,062 (PB); 0,936 (PGL); 0,001 (PGG) Ourense, Spain CR1 PV 0,003 (PVMA); 0,002 (PC); 0,939 (PV); 0,055 (PVSCS/PVSS) Huelva, Spain MR16 PV 0,002 (PVMA); 0,004 (PC); 0,173 (PV); 0,821 (PVSCS/PVSS) Leiria, Portugal 8.26 PVSS 0,006 (PVMA); 0,908 (PC); 0,002 (PV); 0,084 (PVSCS/PVSS) Huelva, Spain 9.73 PHGAL 0,515 (PHGAL); 0,485 (PHAM) Murcia, Spain 9.67 PHGAL 0,585 (PHGAL); 0,415 (PHAM) Murcia, Spain DB4281 PM 0,917 (PM1); 0,083 (PM2) León, Spain DB4294 PM 0,936 (PM1); 0,064 (PM2) León, Spain ALB11=9.79 PHAM 0,222 (PHAM1); 0,778 (PHAM2) Albacete, Spain ALB2=9.77 PHAM 0,912 (PHAM1); 0,088 (PHAM2) Albacete, Spain DB1878 PHAM 0,834 (PHAM1); 0,166 (PHAM2) Albacete, Spain 7 300 PVMA 0,057 (PVMA1); 0,944 (PVMA2) Khenifra, Morocco DB1449 PVMA 0,920 (PVMA1); 0,080 (PVMA2) Ceuta, Spain DB8560 PL 0,013 (PL1); 0,811 (PL2); 0,176 (PL3) Soria, Spain DB1853 PHSS 0,066 (PHSS1); 0,934 (PHSS2) Jaén, Spain DB1748 PHSS 0,940 (PHSS1); 0,060 (PHSS2) Jaén, Spain Val1 PHSS 0,905 (PA/PL/PHSS); 0,037 (PHAM/PHGAL); 0,049 (PC/PV/PVMA/PVSCS/PVSS); 0,002 (PHAZA/PHBAT/PHJS/PHTA); 0,003 (PB/PGL/PGG); 0,001 (PM) Valencia, Spain Alb8 PC 0,003 (PA/PL/PHSS); 0,003 (PHAM/PHGAL); 0,927 (PC/PV/PVMA/PVSCS/PVSS); 0,060 (PHAZA/PHBAT/PHJS/PHTA); 0,003 (PB/PGL/PGG); 0,002 (PM) Salamanca, Spain FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 29 The number of sequences of each species and locus kept for further analyses after STRUCTURE is shown in Table 3.2. Table 3.2. Number of sequences for each loci and unit defined with STRUCTURE. PA, P. atrata; PB, P. bocagei; PC, P.carbonelli; PGG, P. guadarramae guadarramae; PGL, P. guadarramae lusitanica; PHAM, P. hispanica “Albacete/Murcia”; PHAZA, P. hispanica “Azazga”; PHBAT, P. hispanica “Batna”, PHGAL, P. hispanica “Galera; PHJS, P. hispanica “Jebel Sirwah”, PHSS, P. hispanica sensu stricto, PHTA, P. hispanica “Tunisia/Algeria”, PL, P. liolepis, PVMA, P. vaucheri “Morocco/Algeria”, PVSCS, P. vaucheri “Southern-central Spain”; PVSS, P. vaucheri “Southern Spain”; PV, P. virescens; PM, P. muralis; PE, P. erhardii; PS, P. sicula; and PTA, P. taurica. Loci PA1 PA2 PB PC PH1A PH1B PH2 PHAM1 PHAM2 PHAZA PHBAT PHGAL PHJS PHSS1 PHSS2 PHTA PL1 PL2 PL3 PVMA1 PVMA2 PVS PE PS PTA PM1 PM2 PT Sum ACM4 2 2 214 16 18 28 8 4 4 6 18 10 410 20 12 8 2 2 14 32 410 68 2 8274 β-fibint7 4 4 18 14 10 10 24 10 24 6 16 14 412 16 12 8 2 2 14 38 2448 2 0264 C-mos 6 4 20 14 12 16 28 8 4 4 8 20 16 412 20 14 8 2 2 18 38 4448 2 10 310 MC1R 6 2 20 18 16 18 28 8 4 4 8 18 16 412 18 14 8 2 2 14 36 410 48 2 10 314 NFYCint16 6 2 18 12 10 18 22 6 2 4 8 12 16 412 20 10 6 0 2 14 24 4428 0 10 256 PDC 6 4 14 16 16 16 26 10 44 6 18 14 412 18 14 8 2 2 18 38 4448 2 10 302 PKM2int5 6 4 18 18 16 18 28 10 44 8 18 16 410 12 14 6 2 2 16 36 4448 2 10 302 RAG1 6 4 20 16 16 14 28 10 44 6 18 16 412 18 14 8 2 2 18 38 4448 2 8308 RAG2 6 2 20 12 16 14 28 10 44 8 18 16 412 20 14 6 2 2 14 38 4448 2 4296 Pod6b 6 4 14 812 12 14 6 2 4 6 14 14 412 18 12 8 2 2 18 16 2208 2 0222 Pod7b 6 4 18 20 14 16 26 10 44 8 18 14 412 20 12 8 2 2 18 84448 2 8278 Pod11 6 4 20 16 16 16 26 10 44 8 16 16 412 20 14 8 2 2 16 34 4448 2 10 306 Pod12b 6 4 20 20 16 14 26 10 44 6 18 12 410 14 12 8 2 2 16 36 0000 0 6270 Pod13 2 4 14 12 14 14 22 6 4 4 8 16 12 412 16 12 8 2 2 18 30 4444 0 2254 Pod14 4 2 20 18 16 12 26 8 2 4 8 16 12 410 16 14 8 2 2 18 38 0008 2 0270 Pod14b 6 4 20 20 14 18 18 8 2 4 8 12 16 410 18 12 8 2 2 16 26 4446 2 10 278 Pod15 6 0 18 12 16 16 26 10 44 8 18 10 412 20 10 6 2 2 18 32 4448 2 6282 Pod15b 6 4 20 18 16 16 24 10 44 8 16 16 412 20 14 6 2 2 14 38 410 48 2 4306 Pod16 4 4 414 16 18 24 4 0 4 8 16 10 412 18 14 8 2 2 18 24 4408 2 10 256 Pod17 6 4 20 18 16 10 22 8 4 4 8 16 16 210 16 12 8 2 2 16 34 2448 2 8282 Pod20 6 4 18 20 16 18 26 10 44 8 18 12 412 20 14 6 2 2 18 38 4448 2 6308 Pod21 6 4 20 20 14 14 22 10 44 8 16 14 212 18 14 6 2 2 18 36 4448 2 4292 Pod25 4 4 816 12 14 22 10 24 8 10 6212 12 10 6 2 0 8 28 4404 2 6220 Pod31 6 4 14 812 14 26 8 4 4 6 16 10 412 20 12 8 2 2 18 26 4 4 4 0 0 8 256 Pod33 6 4 18 16 14 14 22 10 44 8 18 16 410 20 14 6 2 2 18 28 4408 2 6282 Pod38 6 4 16 14 12 12 26 10 44 0 16 0412 014 8 2 2 18 28 0006 2 0220 Pod43 4 4 20 16 12 10 18 4 4 4 8 14 12 412 16 10 8 2 2 14 32 4426 2 4252 Pod55 6 4 18 22 16 14 28 12 44 8 18 14 614 18 14 8 2 2 18 36 44010 210 316 Pod69 6 4 18 22 16 18 26 8 2 4 8 14 14 412 18 14 4 2 2 18 36 2208 2 8292 Pod72 2 4 20 616 14 26 10 22 8 18 2212 10 8 2 2 2 18 38 4200 2 4236 30 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure 3.2. Estimated probability of ancestry of the Iberian and North African Podarcis species complex and of the outgroup P.muralis, as calculated with STRUCTURE. Each horizontal bar represents one individual and is divided into K segments shown in different colours, with sizes proportional to the portion of the genome of each individual inferred to have originated from each of the K inferred clusters. Transparent boxes highlight admixed individuals, who were excluded from further analyses. The species assignment/mtDNA lineages of the individuals in question are shown in the vertical bars on the left, with the same colours used in Figure 2.1 and 3.1. The new units resulting from subdivisions by STRUCTURE are shown on right in bold. ! 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% P. hispanica sensu stricto P. bocagei P. carbonelli P. hispanica "Albacete/Murcia" P. hispanica "Azazga" P. muralis P. vaucheri "Spain" P. muralis 1 P. atrata 1 P. atrata 2 P. hispanica sensu stricto 1 P. liolepis 3 P. liolepis 1 P. vaucheri "Morocco/Algeria 2" P. vaucheri "Morocco/Algeria 1" P. hispanica "Albacete/Murcia 2" P. hispanica "Albacete/Murcia 1" P. liolepis 2 P. muralis 2 P. hispanica sensu stricto 2 P. atrata P. hispanica "Jebel Sirwah" P. hispanica "Tunisia/Algeria" P. hispanica "Batna" P. hispanica "Galera" P. vaucheri "Southern Spain" P. vaucheri "Southern-central Spain" P. vaucheri "Morocco/Algeria" P. virescens P. guadarramae guadarramae P. guadarramae lusitanica P. liolepis MtDNA& FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 31 3.3 Nuclear loci variability Seven out of the 30 newly sequenced loci had large indel regions, sometimes amounting to over 400bp. Initial alignment length varied between 320bp and 1392bp, and after trimming, final alignments vary between 306bp and 676bp. Summary diversity statistics for the 30 loci and considering all individuals of the Iberian and North African Podarcis group are given in Table 3.3. Table 3.3. Summary statistics and neutrally test for the 30 loci analysed in this study. Nseqs, number of sequences; NSites, total sequence lengths – () excluding sites with gaps / missing data; h, number of haplotypes; Hd, haplotype diversity; Eta, total number of mutations; S, number of segregating sites, π, nucleotide diversity, θw, Theta-Waterson. Polymorphism Loci Nseqs NSites h Hd Eta S π θw Indels ACM4 232 432 (385) 38 0,782 40 37 0,004670 0,01596 0 β -fibint7 244 519 (407) 102 0,981 110 101 0,011793 0,04087 11 C-mos 274 550 (537) 44 0,738 44 44 0,002672 0,01324 0 MC1R 272 694 (648) 45 0,795 38 38 0,002784 0,00949 0 NFYCint16 228 646 (405) 72 0,941 82 78 0,010893 0,03208 17 PDC 268 349 (327) 36 0,779 26 26 0,004037 0,01289 0 PKM2int5 270 459 (289) 42 0,807 41 36 0,004983 0,02018 9 RAG1 278 454 (431) 27 0,571 27 24 0,001690 0,00898 0 RAG2 268 676 (615) 48 0,896 51 51 0,003132 0,01345 0 Pod6b 208 489 (339) 53 0,845 81 73 0,007764 0,03642 11 Pod7b 248 385 (343) 39 0,931 43 41 0,011255 0,01963 4 Pod11 274 435 (264) 53 0,792 45 41 0,008175 0,02510 8 Pod12b 264 410 (241) 93 0,97 101 87 0,018868 0,05869 7 Pod13 236 360 (204) 44 0,644 45 40 0,006056 0,03247 2 Pod14 260 530 (402) 65 0,931 72 67 0,009209 0,02716 8 Pod14b 248 400 (242) 67 0,907 74 65 0,008356 0,04411 25 Pod15 254 420 (282) 48 0,895 47 43 0,006518 0,02495 3 Pod15b 274 504 (357) 78 0,891 71 66 0,006441 0,02987 15 Pod16 228 306 (192) 81 0,969 64 61 0,026945 0,05291 8 Pod17 254 336 (250) 51 0,681 55 50 0,004832 0,03272 3 Pod20 280 396 (212) 31 0,601 33 30 0,005253 0,02279 2 Pod21 266 316 (150) 55 0,873 54 44 0,013613 0,04763 7 Pod25 200 276 (241) 30 0,798 30 28 0,007636 0,01978 3 Pod31 236 507 (380) 89 0,968 78 75 0,010491 0,03268 8 Pod33 258 339 (158) 38 0,832 35 33 0,013199 0,03408 10 Pod38 212 461 (326) 17 0,566 21 20 0,004096 0,01034 0 Pod43 232 353 (320) 28 0,507 29 29 0,001952 0,01505 0 Pod55 286 421 (377) 30 0,493 33 32 0,001951 0,01362 0 Pod69 270 388 (294) 32 0,587 31 31 0,003009 0,01708 2 Pod72 224 462 (328) 67 0,936 63 59 0,015985 0,03005 7 32 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Nucleotide diversity (π) ranged between 0,00169 (RAG1) and 0,018868 (Pod12b) while haplotype diversity (Hd) varied between 0,571 (RAG1) and 0,981 ( β -fibint7). These levels of polymorphism are similar to the ones previously observed in the two nuclear introns (π β - fibint7=0.01269; π6-Pgdint7=0.01721) and, as expected, considerably lower than that detected in mtDNA (Pinho et al., 2008). The population mutation rate (θw) ranged between 0,00898 (RAG1) and 0,05869 (Pod12b). 3.4 Gene-trees Models of sequence evolution selected using the AICc in jModeltest2 for each locus are given in Table 3.4, as well as the respective models as implemented in *BEAST. All ML and BI gene-trees were very shallow, and usually poorly resolved, mostly comprised of polytomies. Often, the ingroup was not even recovered as monophyletic. 3.5 Species-tree Inference Species-trees inferred by the supertree summary statistics method NJst are shown in Figure 3.3 a) and b). We here present the topology of the “best” NJ species-tree (built using the ML trees for all loci) from the “haplotypes” dataset and the bootstrap NJst consensus, with multilocus bootstraps calculated following the 2-stage bootstrap procedure from Seo (2008). Remaining tree topologies (based on the dataset including all sequences (“full”) and bootstrap NJst consensus) can be found on Appendix. Well-supported groups are coincident in all cases. Tree topologies as estimated with NJst reveal overall concordance in many aspects. The ingroup (the Iberian and North African Podarcis excluding P. muralis) is in all cases monophyletic with great support, as well as P. muralis. A few groups are in all cases well supported: 1) the eastern Iberian group of P. atrata, P. liolepis and P. hispanica sensu stricto; 2) a clade with the three clusters of P. vaucheri; 3) all P. hispanica forms from North Africa; 4) the southeastern Iberian P. hispanica “Albacete/Murcia” and P. hispanica “Galera” and 5) P. bocagei plus P. guadarramae lusitanica. The sister-relationship between P. vaucheri and the African forms of P. hispanica was recovered by the consensus of bootstrap NJst estimates of both datasets (“full” and “haplotype”), and also by the “ML” NJst estimate of the “full” dataset, but not highly supported. In all cases except in the bootstrap consensus estimate using the “full” FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 33 dataset, the western Iberian species P. virescens, P. carbonelli, P. guadarramae and P. bocagei were also recovered as a clade, although with low support. Table 3.4. Models of sequence evolution obtained with jModeltest2 and the models used in *BEAST for each gene. Nst, number of substitution rate categories; AICc, Akaike Information Criterion with correction. Loci jModeltest2 *BEAST All individuals Haplotypes per species Model_AICc Nst Model_AICc Nst Model_AICc ACM4 K80+I+G 2 K80+I 2 HKY+I β-fibint7 JC 1 K80+G 2 HKY+G C-mos JC 1 K80+G 2 HKY+G MC1R JC 1 HKI+I+G 2 HKY+I+G NFYCint16 K80+G 2 HKI+G 2 HKY+G PDC TrNef+I+G 6 K80+I 2 HKY+I PKM2int5 K80+G 2 K80+G 2 HKY+G RAG1 K80+I 2 K80+I 2 HKY+I RAG2 K80 2 K80+I+G 2 HKY+I+G Pod6b JC 1 K80+G 2 HKY+G Pod7b K80 2 K80 2 HKY Pod11 TPM3+I+G 6 JC 1 JC69 Pod12b K80+G 2 K80+G 2 HKY+G Pod13 K80+I+G 2 K80+G 2 HKY+G Pod14 JC 1 K80+G 2 HKY+G Pod14b TPM2uf+G 6 JC 1 JC69 Pod15 K80+G 2 K80+G 2 HKY+G Pod15b TPM3uf+I+G 6 K80+G 2 HKY+G Pod16 TIM1+I+G 6 K80+G 2 HKY+G Pod17 HKY+G 2 HKI 2 HKY Pod20 HKY+I+G 2 JC 1 JC69 Pod21 TrNef+G 6 JC 1 JC69 Pod25 TrNef+G 6 TIM1ef+G 6 GTR+G Pod31 JC 1 TrNef+G 6 GTR+G Pod33 HKY+I+G 2 JC 1 JC69 Pod38 JC 1 K80+G 2 HKY+G Pod43 K80 2 K80 2 HKY Pod55 K80+G 2 K80 2 HKY Pod69 K80 2 K80 2 HKY Pod72 JC 1 K80+I+G 2 HKY+I+G 34 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure 3.3. Phylogeny of Iberian and North African Podarcis species as estimated with the NJst method using each gene ML tree topology from the “haplotypes” dataset; a) “best” NJ species-tree; b) NJst consensus (branch lengths are transformed and do not reflect any amount of evolution). Trees were inferred unrooted and rooted for visualization in the branch leading to P. erhardii, P. taurica, P. sicula and P. tiliguerta. The numbers on the nodes indicate multilocus bootstrap support values for branches calculated following Seo, 2008. 2.0 PH1B PE PL1 PA1 PVMA1 PHGAL PM2 PL2 PH1A PHBAT PC PHAM2 PM1 PL3 PT PHSS1 PS PHAZA PVS PHSS2 PVMA2 PHTA PA2 PHJS PB PTA PHAM1 PH2 99 97 38 83 12 64 63 17 99 78 28 100 98 20 34 89 99 99 14 15 30 100 64 99 23 46 P. taurica P. tiliguerta P. sicula P. erhardii P. muralis 1 P. muralis 2 P. liolepis 2 P. liolepis 1 P. atrata 2 P. liolepis 3 P. atrata 1 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. virescens P. carbonelli P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. hispanica “Jebel Sirwah” P. hispanica “Batna” P. hispanica “Tunisia/Algeria” P. hispanica “Azazga” Eastern Iberia Western and Central Iberia P. hispanica North Africa Southeastern Iberia P. vaucheri 700.0 639 628 1000 146 356 993 971 992 456 643 494,5 388 0 999 559 231 885 494,5 992 776 383 468 375 829 632 978 137 P. muralis 1 P. muralis 2 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. atrata 2 P. atrata 1 P. liolepis 2 P. liolepis 3 P. liolepis 1 P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. hispanica “Batna” P. hispanica “Azazga” P. vaucheri “Spain” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. virescens P. carbonelli P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. taurica P. tiliguerta P. sicula P. erhardii Eastern Iberia Southeastern Iberia Western and Central Iberia Western and Central Iberia P. hispanica North Africa P. vaucheri 700.0 639 628 1000 146 356 993 971 992 456 643 494,5 388 0 999 559 231 885 494,5 992 776 383 468 375 829 632 978 137 700.0 639 628 1000 146 356 993 971 992 456 643 494,5 388 0 999 559 231 885 494,5 992 776 383 468 375 829 632 978 137 a) b) FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 35 Species-trees inferred by MP-EST are shown in Figure 3.4. Figure 3.4. Phylogeny of Iberian and North African Podarcis species as estimated using the maximum pseudolikelihood coalescent method, MP-EST, using the 27 loci ML tree: a) maximum pseudo-likelihood estimate of 100 independent searches (branches are in coalescent units, T = 2τ/θ); b) extended consensus of the 100 searches (numbers on the nodes indicate the number of times a given bipartition was found in the 100 searches; branch lengths are transformed and do not reflect any amount of evolution). 2.0 PH1A PE PHAM2 PHSS1 PL3 PHBAT PHAM1 PVS PT PS PA2 PHSS2 PHAZA PL1 PHJS PTA PVMA2 PA1 PH2 PC PH1B PHTA PVMA1 PM2 PL2 PHGAL PB PM1 P. sicula P. taurica P. tiliguerta P. erhardii P. muralis 1 P. muralis 2 P. atrata 1 P. liolepis 2 P. virescens P. vaucheri “Morocco/Algeria 1” P. vaucheri “Morocco/Algeria 2” P. vaucheri “Spain” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. hispanica “Galera” P. bocagei P. guadarramae lusitanica P. liolepis 1 P. liolepis 3 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. atrata 1 P. carbonelli P. guadarramae guadarramae P. hispanica “Batna” P. hispanica “Azazga” P. hispanica “Tunisia/Algeria” P. hispanica “Jebel Sirwah” P. hispanica North Africa Southeastern Iberia Eastern Iberia Western and Central Iberia Western and Central Iberia P. vaucheri 90.0 PH2 PHAM1 PL2 PHGAL PA1 PTA PB PHSS2 PHAM2 PHAZA PHSS1 PVMA2 PL1 PH1B PA2 PE PM2 PH1A PHBAT PT PVS PVMA1 PC PHJS PS PL3 PM1 PHTA 81 100 100 10039 100 100 200 30 100 100 100 43 100 100 100 100 100 100 100 81 100 100 54 51 79 100 100 100 100 100 100 19 100 100 100 100 100 100 100 54 100 100 100 100 100 100 100 100 50 100 100 100 100 P. muralis 1 P. muralis 2 P. hispanica “Batna” P. hispanica “Azazga” P. hispanica “Tunisia/Algeria” P. hispanica “Jebel Sirwah” P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. virescens P. hispanica sensu stricto 1 P. hispanica sensu stricto 2 P. liolepis 3 P. liolepis 1 P. atrata 2 P. atrata 1 P. liolepis 2 P. guadarramae guadarramae P. carbonelli P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. guadarramae lusitanica P. bocagei P. taurica P. erhardii P. sicula P. tiliguerta Eastern Iberia South Iberia and North Africa Southeastern Iberia Western and Central Iberia Western and Central Iberia P. vaucheri P. hispanica North Africa 90.0 PH2 PHAM1 PL2 PHGAL PA1 PTA PB PHSS2 PHAM2 PHAZA PHSS1 PVMA2 PL1 PH1B PA2 PE PM2 PH1A PHBAT PT PVS PVMA1 PC PHJS PS PL3 PM1 PHTA 81 100 100 10039 100 100 200 30 100 100 100 43 100 100 100 100 100 100 100 81 100 100 54 51 79 100 100 100 100 100 100 19 100 100 100 100 100 100 100 54 100 100 100 100 100 100 100 100 50 100 100 100 100 P. muralis 1 P. muralis 2 P. hispanica “Batna” P. hispanica “Azazga” P. hispanica “Tunisia/Algeria” P. hispanica “Jebel Sirwah” P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. virescens P. hispanica sensu stricto 1 P. hispanica sensu stricto 2 P. liolepis 3 P. liolepis 1 P. atrata 2 P. atrata 1 P. liolepis 2 P. guadarramae guadarramae P. carbonelli P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. guadarramae lusitanica P. bocagei P. taurica P. erhardii P. sicula P. tiliguerta Eastern Iberia South Iberia and North Africa Southeastern Iberia Western and Central Iberia Western and Central Iberia 90.0 PH2 PHAM1 PL2 PHGAL PA1 PTA PB PHSS2 PHAM2 PHAZA PHSS1 PVMA2 PL1 PH1B PA2 PE PM2 PH1A PHBAT PT PVS PVMA1 PC PHJS PS PL3 PM1 PHTA 81 100 100 10039 100 100 200 30 100 100 100 43 100 100 100 100 100 100 100 81 100 100 54 51 79 100 100 100 100 100 100 19 100 100 100 100 100 100 100 54 100 100 100 100 100 100 100 100 50 100 100 100 100 P. muralis 1 P. muralis 2 P. hispanica “Batna” P. hispanica “Azazga” P. hispanica “Tunisia/Algeria” P. hispanica “Jebel Sirwah” P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. virescens P. hispanica sensu stricto 1 P. hispanica sensu stricto 2 P. liolepis 3 P. liolepis 1 P. atrata 2 P. atrata 1 P. liolepis 2 P. guadarramae guadarramae P. carbonelli P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. guadarramae lusitanica P. bocagei P. taurica P. erhardii P. sicula P. tiliguerta Eastern Iberia South Iberia and North Africa Southeastern Iberia Western and Central Iberia Western and Central Iberia a) b) 36 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards The MP-EST tree topology recovered largely coincides with previous estimates from NJst. Focusing on the consensus-tree, previously described groups 1, 3, 4 and 5 are again recovered with maximum “support”. The main differences are that P. vaucheri from Morocco (2) is not recovered as in the same clade as its remaining conspecifics, but instead groups with the African P. hispanica forms (in the consensus tree, with low support), and also that here a clade with P. carboneli and P. guadarramae guadarramae is recovered in all tree searches. This specific bipartition was never found in any of the NJst tree searches. The species-tree estimate (extended consensus) obtained with the Bayesian gene-trees distributions and Guenomu is presented in Figure 3.5. In this case, no clade is supported. The strict consensus tree (not shown) is a polytomy between all units, showing that in the 1000 species-trees used to construct the consensus no common clade exists in the big majority of the gene-trees. Figure 3.5. Extended consensus of the posterior distribution of species-trees obtained with Guenomu. Values indicate the posterior probabilities. 3.0 PHJS PH1A PL2 PH2 PS PA2 PVMA2 PTA PE PL1 PHSS1 PHBAT PVS PH1B PT PHAZA PC PHTA PL3 PM2 PB PHAM1 PA1 PHSS2 PHGAL PVMA1 PM1 PHAM2 P. liolepis 2 P. liolepis 1 P. atrata 2 P. atrata 1 P. bocagei P. guadarramae lusitanica P. hispanica “Azazga” P. hispanica “Batna” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. tiliguerta P. erhardii P. vaucheri “Morocco/Algeria 2” P. vaucheri “Spain” P. guadarramae guadarramae P. vaucheri “Morocco/Algeria 1” P. muralis 1 P. muralis 2 P. hispanica sensu stricto 1 P. sicula P. hispanica sensu stricto 2 P. liolepis 3 P. carbonelli P. taurica P. hispanica “Albacete/Murcia 1” P. hispanica “Albacete/Murcia 2” P. hispanica “Galera” P. virescens 7 1 0 0 0 0 0 7 6 14 7 4 2 3 1 5 2 8 3 4 3 9 5 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 37 Some results from *BEAST are shown in Figures 3.6 and 3.7. The distributions of tree topologies based only on species-trees consensus (consensus per topology within run) can be found on Appendix. Convergence revealed to be extremely hard to achieve for many of the parameters, especially for gene-tree likelihoods and gene-tree heights, and, despite the very long lengths of the Markov chains already sampled, no single run can be yet considered to present satisfactory effective sample size (ESS) values, stationarity or convergence. Overall, across runs, sequence model parameters such as base frequencies, substitution rates and heterogeneity parameters (G and I), as well as clock rates, do apparently achieve convergence and high ESS values, while gene-tree likelihoods, gene-tree heights, as well as overall likelihood and posterior almost never do. Likewise, convergence at models for tree priors and population sizes was never achieved despite the large number of (large) runs performed. Figure 3.6. Posterior density of species-trees (cloudogram) from *BEAST analyses for the Iberian and North African Podarcis species for all loci, for a chain length/tree prior of 341M/Yule. Each thin line corresponds to a sampled tree, so darker areas correspond to higher density of trees in agreement. Blue sets of trees represent those with the same topology as the most popular tree, the next most popular set appearing in red, and the third most popular green. Remaining trees are all dark green. Uncertainty in node heights is shown by smears around the mean node height. Maximum clade credibility tree and posterior probabilities of support above 50 are show in blue. P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. virescens P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. carbonelli P. liolepis 2 P. liolepis 1 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. liolepis 3 P. atrata 2 P. atrata 1 P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. hispanica “Batna” P. hispanica “Azazga” P. vaucheri “Morocco/Algeria 1” P. muralis 1 P. muralis 2 P. hispanica “Galera” P. taurica P. tiliguerta P. sicula P. erhardii PVS PVMA2 PH2 PH1B PH1A PB PC PL2 PL1 PHSS1 PHSS2 PL3 PA1 PA2 PHAM1 PHAM2 PHGAL PHJS PHTA PHBAT PHAZA PVMA1 PM1 PM2 PT PE PS PTA 0,94 1 0,84 0,89 0,8 1 0,88 0,86 0,93 0,63 0,76 0,59 0,55 0,99 0,73 0,76 0,94 0,91 1 1 0,73 P. hispanica North Africa Southeastern Iberia Eastern Iberia Western and Central Iberia P. vaucheri 1 44 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 4.3 Incomplete lineage sorting versus gene flow between Iberian and North African Podarcis Incomplete lineage sorting arises from the retention of ancestral polymorphism through speciation events. One of the aspects the data now collected further highlights is that Podarcis species have diverged very rapidly, with lineages not having time to achieve reciprocal monophyly at most loci. Incomplete lineage sorting seems thus to be the most widely accepted hypothesis for the abundant haplotype sharing, as also inferred from previous work (Pinho et al., 2008). Nevertheless, gene flow was also inferred. Results from STRUCTURE obtained in this thesis showed cases of gene flow among forms/species, some of which had never been reported before (as well as gene flow between probably intraspecific forms). Twenty-two individuals were identified as having a possibly admixed genotype, and a few more were assigned to a different group than suggested by its mitotype Some of these cases were already expected because they involve species that are found in sympatry: for example, gene flow was detected between P. bocagei and P. guadarramae lusitanica, in similarity to various previous reports (Pinho et al., 2007a, 2008; Pinto, 2013). There was also one individual carrying the P. hispanica sensu stricto mtDNA lineage that was assigned to P. hispanica “Galera”, a situation that had also been reported before (Pinho et al., 2008; Renoult et al., 2009). Other cases were totally new, such as the finding of an individual admixed between P. vaucheri “Spain” and P. carbonelli in the sympatric zone of Doñana. Individuals presenting signs of admixture between P. vaucheri “Spain” and P. virescens were also found, although far from the putative contact zone between these species. Individuals showing signs of admixture were also detected between the geographically close P. hispanica “Galera” and P. hispanica “Albacete/Murcia”. Probably one of the most striking results was the high number of admixed individuals between P. liolepis, P. hispanica sensu stricto and P. atrata. These individuals are hard to explain taking into account hybridization, not only because of its triple nature but also because P. atrata is an island form. Other cases of gene flow were identified between clusters found within mtDNA lineages. All these cases most likely reflect the overall lack of differentiation between these clusters and their probably intraspecific nature. Additionally, there is one case where it was observed one individual of P. carbonelli mtDNA lineage with signs of admixture with the North African P. hispanica group. This result can only be an artifact because is impossible that gene flow occurs between these forms. As the species-tree inference methods here used all assume absence of gene flow between “species” and because the approach used of removing admixed individuals as detected by FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 45 STRUCTURE may fail to detect (older) historical gene flow, we further applied the coalescent model of divergence with gene flow, IMa2 (Hey and Nielsen, 2007), to these groups of Podarcis in a pairwise manner, in a similar way as Pinho et al. (2008). Surprisingly, gene flow was now detected among most of the species. Yet, these results have to be considered with caution as our inference scheme, performing pairwise tests between all units (even those which are clearly not sister taxa), is a clear violation of the IM model assumptions (Strasburg and Rieseberg, 2009). In many of the pairwise comparisons, the possibility of unaccounted gene flow from external taxa may strongly limit the validity of our inferences. Moreover, a thorough evaluation of levels of historical gene flow between Iberian and North African Podarcis was outside the scope of this thesis. Thus, we provisionally decided not to take these results into account until we can more accurately test for the presence of gene flow without the possibility of too many false positive estimates. Nevertheless, we cannot disregard the possibility of unaccounted gene flow in “species-tree” inference, which effect can be both false sister relationships, and the inference of divergence times biased towards more recent times (Leaché et al., 2014). 4.4 Phylogenetic relationships of Podarcis “species” This work is the first attempt so far at inferring Iberian and North African Podarcis phylogeny from a multilocus nuclear DNA sequence data. For this, a number of different approaches were used, representative of the main categories of these methods; a distance method (NJst), a maximum “pseudolikelihood” method (MP-EST), and a probabilistic method (Guenomu) (super-tree methods), and finally a fully probabilistic bayesian co-estimation method (*BEAST ). Starting by the “outlier”, the topology inferred by Guenomu is the less resolved one and does not seem to be biologically realistic in many aspects. In fact, the strict consensus of the recovered species-trees distribution was a full polytomy. Despite the fact that some of the inferred groups and/or sister relationships (extended consensus) do make sense and agree with other methods (eg. P. liolepis and P. atrata; P. bocagei and P. guadarramae lusitanica; the African P. hispanica forms, P. hispanica “Albacete-Murcia” and P. hispanica “Galera”; P. muralis), no relationship is actually supported, showing that there is almost no common clade in the 1000 species-trees used to calculate the consensus. Further, and contrary to all other methods, the ingroup is not recovered as monophyletic. Possible reasons for the clear lower performance of this method are explored below (section 4.7) 46 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Regarding inferences from other methods, overall, the results from NJst, MP-EST, and *BEAST show many similarities between them and also with the mtDNA phylogeny (Kaliontzopoulou et al., 2011), as well as some particular differences. For a start, all estimates agree in a highly supported “ingroup”, to the exclusion of P. muralis, whose two “groups” also define a monophyletic clade. Then, basal relationships within the ingroup are mostly unresolved, and thus different (and never supported) across methods, but some groups of closely related “species” were commonly recovered, with high support, and are described below. One group that is always recovered with high support by NJst and by MP-EST and which *BEAST runs tend to recover also, is the Eastern Iberian clade of P. hispanica sensu stricto, P. liolepis and P. atrata. Within this group, the two lineages of P. hispanica sensu stricto and of P. atrata are, respectively, recovered with support as monophyletic, but remaining relationships are recovered differently and unsupported across methods. The interesting exception is MP-EST (consensus), which recovers very well supported relationships within this group, with a non-monophyletic P. liolepis, and P. liolepis 2 lineage as basal to the whole group. Nevertheless, and especially given that this MP-EST consensus does not represent real “bootstrap” values, this results must be taken with caution and remain as an hypothesis to be further tested. The remaining species from the Iberian Peninsula (Western-central Iberian group) were sometimes recovered as a clade (although never well supported) – NJst – or at least part of its “species”- MP-EST, *BEAST – generally with unsupported basal relationships. Again, within the group some sister-relationships do are well supported across methods, as the sister relationship between P. bocagei and P. guadarramae lusitanica. With MP-EST, P. carbonelli is also recovered with high support as sister taxa of P. guadarramae guadarramae, something which is not recovered with any other method. The other highly consistent result across all methods was the inference of the clade containing all P. hispanica mtDNA lineages from North Africa. In the relationships within this group there were some differences across methods, yet, the lineage from Jebel Sirwah was always recovered as the most basal within this clade. The most evident difference between the present estimates of relationships and the ones using mtDNA is a swapping in sister taxa relationships among P.liolepis, P.hispanica “Galera”, P.hispanica “Albacete/Murcia” and P.hispanica sensu stricto. Indeed, current nuclear data places P. liolepis as closely related (perhaps conspecific or sister) to P. hispanica sensu stricto and P. hispanica “Albacete/Murcia” as sister to P. hispanica “Galera”. However, mtDNA recovers as sister taxa the pairs P. hispanica sensu stricto/P. hispanica “Albacete/Murcia” and P.liolepis/P. hispanica “Galera”, the two pairs very distantly related. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 47 It is not the first time that this close relationship between P.liolepis and P.hispanica sensu stricto is suggested. It was also inferred with allozymes (Pinho et al., 2007a), low divergence time estimated from nuclear introns (Pinho et al., 2008) and from the lack of morphological and genetic differentiation across the mtDNA lineage’s contact zone (Renoult et al., 2009). Using allozymes, these two mtDNA lineages could not be well distinguished, in similarity to the present analyses using STRUCTURE, which despite revealing some differentiation recovers fuzzy boundaries between these taxa, with prevalent cytonuclear discordance and admixture, It thus seems possible that they are in fact conspecifics. A few hypotheses can explain these discordances: 1) it could imply ancestral (or not so) gene flow between divergent taxa with a probable dilution of the nuclear genome but not the mitochondrial one (with the mtDNA genealogy representing the original relationships between taxa) or 2) the capture of a foreign mtDNA lineage, corresponding to a now extinct nuclear “unit” (without any corresponding obvious signature in the nuclear genome). This last scenario was also one of the hypotheses proposed by Renoult et al., (2009), which suggests an ancient mitochondrial introgression originating from an evolutionary unit presently absent from the study area. Remarkably, the close relationship inferred between P. hispanica “Albacete/Murcia” and P. hispanica “Galera” implies that a similar phenomenon of an ancient mtDNA capture happened in parallel in this species pair. Another possibility is, then, that it was the same mtDNA lineage, the ancestral to one of the pairs (e.g. P. hispanica sensu stricto and P. hispanica Albacete/Murcia) that was captured by both species of the other pair (e.g. by P. liolepis and P. hispanica “Galera”), and that it has diverged in the two species (generating the two divergent mtDNA lineages) since then. If this was the case, it is difficult to explain such double introgression without invoking an adaptive nature. Curiously, the idea above can be seen as the recycling of one proposal by Renoult et al., (2009), which our data actually dismiss. Indeed, these authors also suggested a double introgression, but much more recent, in both cases involving the “modern” P. hispanica sensu stricto: one into P. liolepis (as our data also may suggest) and another into P. hispanica “Galera”, which our data do not support. Although we do find instances of cytonuclear discordance involving the P. hispanica sensu stricto mtDNA lineage and P. hispanica “Galera”, the discordance is sporadic, not general, as the two forms can be clearly distinguished even when occupying nearby localities. Another difference between nDNA and mtDNA estimates of relationships, but this time involving relationships within a clade and not between major clades, is that P. guadarramae (previously P. hispanica 1A and 1B) is probably paraphyletic since P. bocagei appears always highly supported as the sister taxa of P. g. lusitanica. 48 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Inconsistencies between our nuclear trees were also observed. Podarcis vaucheri “Spain” (that correspond to mtDNA lineages P. vaucheri “Southern-central Spain” and P. vaucheri “Southern Spain”) is recovered by NJst (both consensus plus “full” dataset) as closely related to P. vaucheri “Morocco/Algeria” 1 and 2 and sister of the all other lineages from North Africa (although not strongly supported) but in the *BEAST results this taxa appears most often as closely related to species from Western and central Iberia. With MP-EST (consensus), one of the African P. vaucheri lineages is inferred as closely related to the North African P. hispanica forms, while the other two lineages forms a separate group (without a supported relationship to any other group). At *BEAST runs, some of the topologies recovered related P. vaucheri (African) lineages with the North African forms of P. hispanica, but also P. vaucheri from Spain was often related to the West-central Iberian clade. Overall, this seems to support a relationship between P. vaucheri and other North African forms, with the placement of P. vaucheri from Spain close to the Western Iberian group possibly caused by some degree of unaccounted-for gene flow (possibly with P. carbonelli or P. virescens). As a last mention to aspects of the presented phylogenies, at some *BEAST runs P. vaucheri “Morocco/Algeria 1” and P. atrata 2 were sometimes clearly distant from all the others species, as also P. liolepis 3 tended to be inferred as related to quite different taxa across runs. These “species” include only one (P. vaucheri “Morocco/Algeria 1” and P. liolepis 3) or two individuals (P. atrata 2) and we hypothesize that is the cause of their behavior and of their position being more difficult to estimate with this co-estimation method. Due to the fact that none of the *BEAST runs here presented achieved satisfactory converge, we do not have much confidence in these topologies overall, although we believe the results support some of the inferences made from other methods, especially regarding consistently recovered relationships. Moreover, sampling from the prior only yielded very different results from those obtained with data, indicating also that there is some information in the trees so far been obtained. 4.5 Biogeographic implications Another interesting aspect of the evolutionary history of the Iberian and North African Podarcis is the biogeography of the group around the Strait of Gibraltar. For some species, the opening of the Strait was the probable cause for separation between Iberian and North African species (Maia-Carvalho et al., 2014), but in the case of Podarcis the successive mtDNA phylogenies suggest otherwise. Indeed, the Strait has not worked as a complete barrier to migration, and the distribution of genetic variation requires two independent events, either two FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 49 transmarine colonizations or one vicariant event followed by the crossing of the Strait (Harris et al., 2002; Pinho et al., 2006; Kaliontzopoulou et al., 2011). Either we accept or not that P. vaucheri and the North African P. hispanica forms are related, the results are concordant with the two-event scenario suggested by the mtDNA, since there is one Iberian form (P. vaucheri “Spain”) grouped within a North African clade. This observation thus implies, again, at least one transmarine colonization. However, the directionality of the colonization cannot be inferred from the present estimates of relationships. 4.6 Taxonomic implications A taxonomic reevaluation of the clade is clearly beyond the scope of this thesis. However, it is possible to make some reflections on this subject. In general our results support the distinctiveness of currently accepted species. Particularly, P. bocagei, P. carbonelli and P. virescens are clearly diagnosable genetically and morphologically. P. vaucheri is also clearly distinct from other species, although it is difficult to evaluate the taxonomic status of forms inhabiting South Iberia and North Africa. P. guadarramae, on the other hand, is recovered as paraphyletic, since P. guadarramae lusitanica (former P. hispanica type 1A) is placed as sister to P. bocagei in our analyses. If true, this suggests that the taxonomy of these forms, which was recently redefined (Geniez et al., 2014), will require a new reevaluation. However, the sister taxa relationship between P. bocagei and P. guadarramae lusitanica could eventually be biased by high levels of gene flow between these sympatric species. With respect to P. liolepis, the genetic proximity and in some cases paraphyly with respect to P. atrata may indicate that these two forms are closely related or even conspecific, as suggested previously by Harris and Sá-Sousa, (2002) and Renoult et al., (2010). However, the distinctiveness of P. atrata (and of different island groups within it) raises the possibility that it may well deserve a different taxonomic status. P. liolepis seems to be also closely related to populations exhibiting the P. hispanica sensu stricto mtDNA lineage. This is true for populations from the Northern area of the distribution of this lineage, in the provinces of Cuenca and Valencia, as suggested by Renoult et al., (2009) but also for the South of its distribution, in the provinces of Jaén and Granada. Given the low level of differentiation and the prevalence of admixed individuals between clusters ascribed to P liolepis, P. atrata and P. hispanica sensu stricto, it is probably more conservative to assume that these clusters all correspond to phylogroups within P. liolepis. Further studies of the contact zones between 50 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards clusters will be highly valuable in order to understand whether there are reproductive barriers among them. The recovered phylogenetic relationships also highlight that the forms that still remain under the designation of P. hispanica are unrelated. It includes at least two groups, one from Southern Iberian Peninsula and another from North Africa. Within each group, it is difficult to assess whether or not the taxa deserve the species status. On one hand, all the mitochondrial DNA lineages correspond to genetically diagnosable groups; on the other hand, both the pair “Albacete/Murcia” and “Galera” and the set “Tunisia and Algeria”, “Jbel Siroua”, “Batna” and “Azazga” are obviously closely related. It is this difficult to evaluate, with the data in hands, their taxonomic status. Again this would require an extensive study of contact zones, which might be possible in the first case given their contiguous distribution but likely impossible in the second, given the fragmented distribution patterns of these lineages. 4.7 Comparison between methods: factors affecting species-tree inference The choice of the method can often have a large impact in the analyses, as different methods have different accuracies in different kinds of datasets. It is thus of most importance to be aware of the expected error rates and specific biases of the methods we use in the specific characteristics of the datasets we have on hands. NJst, MP-EST and Guenomu are simple and extremely fast methods compared with full probabilistic approaches such as *BEAST and seem to provide good approximations for large amounts of data, often outperforming the other methods of the same class. Constructing a species phylogeny with NJst and MP-EST methods take only a few minutes (given, of course, gene-trees are already obtained from a separate method at the users choice). Yet, from a practical point of view, both often require the manipulation of large amounts of data, and sometimes the performance of not-so-trivial operations such as rooting a big amount of genetrees, often not with the same taxa (MP-EST), therefore requiring some scripting abilities, and thus, the time-investment. Yet, they are very fast, and perfectly usable at scales of hundreds of loci; contrary to *BEAST, for which datasets of few tens of loci and “species” already reveal problematic. Guenomu takes slightly more time (especially if we account for the Bayesian gene-tree search), but yet runs in a few hours, which is remarkable compared with *BEAST that may need months to achieve convergence. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 51 The authors of these methods have tested them extensively on simulated data sets, which typically include few species and few individuals per species (especially true in the case of NJst and MP-EST; (Liu et al., 2010; Liu and Yu, 2011). Further research with multilocus data has been done to compare these methods in the presence of various levels of incomplete lineage sorting, showing some problems that could influence their accuracy (Yang and Warnow, 2011; Bayzid and Warnow, 2013; Mirarab et al., 2014). These studies concluded that in general the results can be consistently improved as the number of genes increases, but also that the methods are highly sensitive to violations of their assumptions caused for example by error in gene-tree inference and introgressive hybridization (Leaché et al., 2014). Yet, simulations also show that for trees involving very short branch lengths (or at least a proportion of branches which is very short) and/ or for larger population sizes (which increase the average proportion of genetree/species-tree discordance), it may not be possible to infer the “true” species-tree (Leaché and Rannala, 2011; Mirarab et al., 2014). In the case of Podarcis we are dealing with closely related “species”, where genetic distances are very small and often zero. This is likely a reason for high levels of gene-tree error in our estimates and consequently high species-tree estimation error. This was perhaps one of the reasons of Guenomu to have failed in our dataset, for example, given that this program performs better at simulations than other supertree methods, including STAR (Liu et al., 2009b), which is almost identical to NJst. Additional reasons for the lower performance of Guenomu may involve the lack of convergence of Bayesian gene-tree distributions. Although runs were quite long and convergence was carefully examined, decision about runs convergence and stabilization is often not straightforward, Finally, it may also be that the method itself (built for datasets involving both duplications and losses and ILS) does not perform well when only ILS is present in the dataset - being a “reconciliation” method (in the sense that it tries to optimize some cost function), dup-losses are almost always very informative. Yet, some of the simulations performed with the method (de Oliveira Martins et al., 2014), and across which this method still outperformed others, involved very low rates of dup-loss and very high rates of ILS. It is certainly interesting to further explore the performance of this method in other cases of ILS only and to investigate the reasons for its apparent very low performance in this dataset. *BEAST is expected to have better performance than supertree methods and also seems to produce more accurate trees than the other methods of the same class, like BUCKY and STEM (Kubatko et al., 2009). In our case, although *BEAST seems to be walking towards a similar solution to that obtained using NJst or MP-EST, and many common aspects can be found, it is clearly much slower, and so far, after multiple runs of several weeks, most parameters still did not stabilize or converge. This was expected since this approach is computationally intensive and the size of the dataset seems to be a crucial factor for full probabilistic methods. Dealing 52 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards with this problem and improving the scalability of *BEAST is the object of current research and some strategies are being developed (Zimmermann et al., 2014), which may be definitely worth to try. As this thesis clearly illustrates, the applicability of these methods in cases of very shallow divergence with extensive lineage sorting and haplotype sharing can be very challenging. Because these methods are relatively new and hence poorly tested on empirical data, it remains to be seen whether the difficulties inherent to the analysis of this dataset are shared by other case studies, suggesting that is difficult to apply these methods on real data, under complex scenarios like those represented here. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 53 Chapter 5 Conclusions and Future Perspectives Iberian and North African Podarcis wall lizards are an example of a complex group that clearly illustrates how fast diversification coupled with gene flow can cause complex patterns that make the evolutionary history extremely difficult to infer. The nuclear data here analysed increased our knowledge about the history of this group and allowed us to obtain another perspective into the levels of genetic polymorphism and distinctiveness of evolutionary units, as well as into their phylogenetic relationships. Inferences based on mtDNA or morphological characters were in some cases corroborated and in others rejected. In summary, our results suggest that: 1) Evolutionary lineages still share an important proportion of alleles and have not had time to achieve reciprocal monophyly at the majority of the genome. 2) Despite this lack of single-locus differentiation lineages can be easily diagnosed using a multilocus framework. 3) Gene flow among lineages is prevalent, particularly among sympatric forms (P. bocagei and P. guadarramae lusitanica, P. carbonelli and P. vaucheri, P. hispanica “Galera” and populations carrying the P. hispanica sensu stricto mitotype) but also between parapatric forms such as P. hispanica “Galera”/P. hispanica “Albacete/Murcia”, or P. vaucheri and P. virescens. A high number of admixed individuals were also observed for P. liolepis/P. hispanica sensu stricto/P. atrata, which may indicate conspecificity. 4) A well-supported Eastern Iberian clade, composed by P. hispanica sensu stricto, P. liolepis and P. atrata, is typically recovered by the different methods, in similarity to a Southeastern Iberia clade placing P. hispanica “Galera” and the phylogroups within P. hispanica “Albacete/Murcia” as sister taxa. This is a clear difference compared with mtDNA that recovers as sister taxa the pairs P. hispanica sensu stricto/P. hispanica “Albacete/Murcia” and P.liolepis/P. hispanica “Galera”, suggesting repeated, perhaps adaptive, phenomena of mtDNA capture and/or extensive nuclear swamping. 60 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Heled J, Bryant D, Drummond AJ (2013). Simulating gene trees under the multispecies coalescent and time-dependent migration. BMC Evol Biol 13: 44. Heled J, Drummond AJ (2010). Bayesian inference of species trees from multilocus data. Mol Biol Evol 27: 570–580. Hey J, Nielsen R (2007). Integration within the Felsenstein equation for improved Markov chain Monte Carlo methods in population genetics. Proc Natl Acad Sci USA 104: 2785– 90. Hoban S (2014). An overview of the utility of population simulation software in molecular ecology. Mol Ecol 23: 2383–401. Hoegg S, Vences M, Brinkmann H, Meyer A (2004). Phylogeny and comparative substitution rates of frogs inferred from sequences of three nuclear genes. Mol Biol Evol 21: 1188– 2000. Holland BR, Benthin S, Lockhart PJ, Moulton V, Huber KT (2008). Using supernetworks to distinguish hybridization from lineage-sorting. BMC Evol Biol 8: 202. Hudson RR, Turelli M (2003). Stochasticity overrules the "three-times rule": genetic drift, genetic draft, and coalescence times for nuclear loci versus mitochondrial DNA. Evolution 57: 182–190. Huson DH, Klöpper T, Lockhart PJ, Steel MA (2005). Reconstruction of Reticulate Networks from Gene Trees. Research in Comput Mol Biol. Springer Berlin Heidelb: 233–249. Jewett EM, Rosenberg NA (2012). iGLASS: an improvement to the GLASS method for estimating species trees from gene trees. J Comput Biol 19: 293–315. Joly S, Mclenachan PA, Lockhart PJ (2009). A Statistical Approach for Distinguishing Hybridization and Incomplete Lineage Sorting. Am Nat 174: E54–E70. Kaliontzopoulou A, Carretero MA, Llorente GA (2012). Morphology of the Podarcis wall lizards (Squamata: Lacertidae) from the Iberian Peninsula and North Africa: patterns of variation in a putative cryptic species complex. Zool J Linn Soc 164: 173–193. Kaliontzopoulou A, Pinho C, Harris DJ, Carretero MA (2011). When cryptic diversity blurs the picture: a cautionary tale from Iberian and North African Podarcis wall lizards. Biol J Linn Soc 103: 779–800. Katoh K, Standley DM (2013). MAFFT: Multiple Sequence Alignment Software Version 7': Improvements in Performance and Usability Article Fast Track. Mol Biol Evol 30: 772– 780. Kluge AG (1989). A Concern for Evidence and a Phylogenetic Hypothesis of Relationships Among Epicrates (Boidae, serpentes). Syst Zool 38: 7–25. Kubatko LS, Carstens BC, Knowles LL (2009). STEM: species tree estimation using maximum likelihood for gene trees under coalescence. Bioinformatics 25: 971–973. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 61 Kubatko LS, Gibbs HL, Bloomquist EW (2011). Inferring species-level phylogenies and taxonomic distinctiveness using multilocus data in sistrurus rattlesnakes. Syst Biol 60: 393–409. Larget BR, Kotha SK, Dewey CN, Ané C (2010). BUCKy: Gene tree / species tree reconciliation with Bayesian concordance analysis. Bioinformatics 26: 2910–2911. Leaché AD, Harris RB, Rannala B, Yang Z (2014). The influence of gene flow on species tree estimation: a simulation study. Syst Biol 63: 17–30. Leaché AD, Rannala B (2011). The accuracy of species tree estimation under simulation': A comparison of methods. Syst Biol: syq073. Li N, Stephens M (2003). Modeling linkage disequilibrium and identifying recombination hotspots using single-nucleotide polymorphism data. Genet Soc Am 165: 2213–2233. Librado P, Rozas J (2009). DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25: 1451–1452. Lima A, Pinho C, Larbes S, Carretero MA, Brito JC, Harris DJ (2009). Relationships of Podarcis wall lizards from Algeria based on mtDNA data. Amphibia-Reptilia 30: 483– 492. Liu L (2008). BEST: Bayesian estimation of species trees under the coalescent model. Bioinformatics 24: 2542–2543. Liu L, Yu L (2011). Estimating species trees from unrooted gene trees. Syst Biol 60: 661–667. Liu L, Yu L, Edwards S V (2010). A maximum pseudo-likelihood approach for estimating species trees under the coalescent model. BMC Evol Biol 10: 302. Liu L, Yu L, Kubatko L, Pearl DK, Edwards S V (2009). Coalescent methods for estimating phylogenetic trees. Mol Phylogenet Evol 53: 320–8. Liu L, Yu L, Pearl DK, Edwards S V (2009). Estimating species phylogenies using coalescence times among sequences. Syst Biol 58: 468–77. Löytynoja A, Goldman N (2008). Phylogeny-aware gap placement prevents errors in sequence alignment and evolutionary analysis. Science 320: 1632–5. Maddison WP (1997). Gene trees in species trees. Syst Biol 46: 523–536. Maia-Carvalho B, Gonçalves H, Ferrand N, Martínez-Solano I (2014). Multilocus assessment of phylogenetic relationships in Alytes (Anura, Alytidae). Mol Phylogenet Evol 79: 270–8. Mallet J (2005). Hybridization as an invasion of the genome. Trends Ecol Evol 20: 229–37. Mallo D, de Oliveira Martins L, Posada D (2014). Estimation of Species Trees. eLS Martin DP, Lemey P, Lott M, Moulton V, Posada D, Lefeuvre P (2010). RDP3: a flexible and fast computer program for analyzing recombination. Bioinformatics 26: 2462–3. McBreen K, Lockhart PJ (2006). Reconstructing reticulate evolutionary histories of plants. Trends Plant Sci 11: 398–404. 62 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Meade A (2011). BayesTrees v.1.3. Sch Biol Sci Univ Reading, Reading, United Kingdom. Mirarab S, Bayzid MS, Warnow T (2014). Evaluating summary methods for multilocus species tree estimation in the presence of incomplete lineage sorting. Syst Biol 0: 1–15. Mirarab S, Reaz R, Bayzid MS, Zimmermann T, Swenson MS, Warnow T (2014). ASTRAL: genome-scale coalescent-based species tree estimation. Bioinformatics 30: i541–i548. Müller K (2005). SeqState-primer design and sequence statistics for phylogenetic DNA data sets. Appl Bioinformatics 4: 65–69. Nylander JAA, Wilgenbusch JC, Warren DL, Swofford DL (2008). AWTY (are we there yet?): a system for graphical exploration of MCMC convergence in Bayesian phylogenetics. Bioinformatics 24: 581–3. O’Meara BC (2010). New heuristic methods for joint species delimitation and species tree inference. Syst Biol 59: 59–73. de Oliveira Martins L, Mallo D, Posada D (2014). A Bayesian supertree model for genomewide species tree reconstruction. Syst Biol: syu082. Pereira C, Couto A, Luís C, Costa D, Mourão S, Pinho C (2013). Twenty-one new sequence markers for population genetics, species delimitation and phylogenetics in wall lizards (Podarcis spp.). BMC Res Notes 6: 299. Pérez-Mellado V (1981). La lagartija de Bocage, Podarcis bocagei (Seoane, 1884): Primeros datos sobre su distribución, colorido y ecología. Amphibia–Reptilia 3: 253–268. Pérez-Mellado V (1981). Nuevos datos sobre la sistemática y distribución de Podarcis bocagei (Seoane, 1884) (Sauria, Lacertidae) en la Península Iberica. Amphibia-Reptilia 265: 259– 265. Pérez-Mellado V, Galindo MP (1986). Sistemática de Podarcis (Sauria, Lacertidae) ibéricas y norteafricanas mediante técnicas multidimensionales. In: Série Manuales Universitários, (Spain: Univ. Salamanca). Pinho C, Ferrand N, Harris DJ (2006). Reexamination of the Iberian and North African Podarcis (Squamata: Lacertidae) phylogeny based on increased mitochondrial DNA sequencing. Mol Phylogenet Evol 38: 266–73. Pinho C, Harris DJ, Ferrand N (2003). Genetic polymorphism of 11 allozyme loci in populations of wall lizards (Podarcis sp.) from the Iberian Peninsula and North Africa. Biochem Genet 41: 343–59. Pinho C, Harris DJ, Ferrand N (2007a). Comparing patterns of nuclear and mitochondrial divergence in a cryptic species complex': the case of Iberian and North African wall lizards (Podarcis, Lacertidae). Biol J Linn Soc 91: 121–133. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 63 Pinho C, Harris D, Ferrand N (2007b). Contrasting patterns of population subdivision and historical demography in three western Mediterranean lizard species inferred from mitochondrial DNA variation. Mol Ecol 16: 1991–1205. Pinho C, Harris DJ, Ferrand N (2008). Non-equilibrium estimates of gene flow inferred from nuclear genealogies suggest that Iberian and North African wall lizards (Podarcis spp.) are an assemblage of incipient species. BMC Evol Biol 8: 63. Pinho C, Kaliontzopoulou A, Carretero MA, Harris DJ, Ferrand N (2009). Genetic admixture between the Iberian endemic lizards Podarcis bocagei and Podarcis carbonelli: evidence for limited natural hybridization and a bimodal hybrid zone. J Zool Syst Evol Res 47: 368– 377. Pinho C, Rocha S, Carvalho BM, Lopes S, Mourão S, Vallinoto M, et al. (2010). New primers for the amplification and sequencing of nuclear loci in a taxonomically wide set of reptiles and amphibians. Conserv Genet Resour 2: 181–185. Pinto C (2013). Hybridization between Podarcis hispanica type 1A and Podarcis bocagei in areas of syntopy. University of Porto. Polzin T, Daneshmand SV (2003). On Steiner trees and minimum spanning trees in hypergraphs. Oper Res Lett 31: 12–20. Pritchard JK, Stephens M, Donnelly P (2000). Inference of population structure using multilocus genotype data. Genetics 155: 945–959. Purvis A, Gittleman JL, Cowlishaw G, Mace GM (2000). Predicting extinction risk in declining species. Proc Biol Sci 267: 1947–52. Rambaut A (2014). Figtree version 1.4.2. Available from http://beast.bio.ed.ac.uk/figtree. Rambaut A, Drummond A (2007). Tracer v1.4. Available from http://beast.bio.ed.ac.uk/tracer. Rannala B, Yang Z (2003). Bayes estimation of species divergence times and ancestral population sizes using DNA sequences from multiple loci. Genetics 164: 1645–1656. Renoult JP, Geniez P, Bacquet P, Guillaume CP, Crochet PA (2010). Systematics of the Podarcis hispanicus-complex (Sauria, Lacertidae) II: the valid name of the north-eastern Spanish form. Zootaxa 2500: 58-68. Renoult JP, Geniez P, Bacquet P, Benoit L, Crochet PA (2009). Morphology and nuclear markers reveal extensive mitochondrial introgressions in the Iberian Wall Lizard species complex. Mol Ecol 18: 4298–315. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, et al. (2012). MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol 61: 539–42. Rousset F (2008). genepop’007: a complete re-implementation of the genepop software for Windows and Linux. Mol Ecol Resour 8: 103–106. 64 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Sá-Sousa P (2001). Comparative chorology between Podarcis bocagei and P. carbonellae ( Sauria: Lacertidae ) in Portugal. Rev Española Herpetol 15: 85–97. Sá-Sousa P, Pérez-Mellado V, Martínez-Solano I (2009). Podarcis carbonelli. The IUCN Red List of Threatened Species. Version 20142 <www.iucnredlist.org> Downloaded 24 Sept 2014. Sá-Sousa P, Vicente L, Crespo E (2002). Morphological variability of Podarcis hispanica (Sauria: Lacertidae) in Portugal. Amphibia-Reptilia 23: 55–69. Sá‐Sousa P, Almeida AP, Rosa H, Vicente L, Crespo EG (2000). Genetic and morphological relationships of the Berlenga wall lizard (Podarcis bocagei berlengensis: Lacertidae). J Zool Syst Evol Res 38: 95–102. Salvi D, Harris DJ, Kaliontzopoulou A, Carretero M a, Pinho C (2013). Persistence across Pleistocene ice ages in Mediterranean and extra-Mediterranean refugia: phylogeographic insights from the common wall lizard. BMC Evol Biol 13: 147. Sawyer S (1989). Statistical tests for detecting gene conversion. Mol Biol 6: 526–538. Seo T-K (2008). Calculating bootstrap probabilities of phylogeny using multilocus sequence data. Mol Biol Evol 25: 960–71. Simmons MP, Ochoterena H (2000). Gaps as characters in sequence-based phylogenetic analyses. Syst Biol 49: 369–81. Simmons MP, Ochoterena H, Carr TG (2001). Incorporation, relative homoplasy, and effect of gap characters in sequence-based phylogenetic analyses. Syst Biol: 454–462. Smith JM, Smith NH (1998). Detecting recombination from gene trees. Mol Biol Evol 15: 590– 599. Srivastava DS, Cadotte MW, MacDonald a AM, Marushia RG, Mirotchnick N (2012). Phylogenetic diversity and the functioning of ecosystems. Ecol Lett 15: 637–48. Stamatakis A (2014). RAxML Version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30: 1312–1313. Stephens M, Smith NJ, Donnelly P (2001). A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68: 978–89. Strasburg J, Rieseberg L (2009). How robust are “Isolation with Migration” analyses to violations of the IM model? A simulation study. Mol Biol Evol 27: 297–310. Szöllősi GJ, Daubin V (2012). Modeling gene family evolution and reconciling phylogenetic discord. Evol Genomics Humana Press: 29–51. Uetz PP, Hosek J (2014). The Reptile Database. <www.reptile-database.org>. Watterson G (1975). On the number of segregating sites in genetical models without recombination. Theor Popul Biol 7: 276. Whelan NV (2011). Species tree inference in the age of genomics. Trends Evol Biol 3: e5. FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 65 Wiley EO (1981). Phylogenetics: the theory and practice of phylogenetic systematics. John Wiley Sons. Willis KJ, Birks HJB (2006). What is natural'? The need for a long-term perspective in biodiversity conservation. Science 314: 1261–1265. Wu Y (2012). Coalescent-based species tree from gene tree topologies under incomplete lineage sorting by maximum likelihood. Evolution 66: 763–775. Yang Z, Rannala B (2010). Bayesian species delimitation using multilocus sequence data. Proc Natl Acad Sci USA 107: 9264–9. Yang J, Warnow T (2011). Fast and accurate methods for phylogenomic analyses. BMC Bioinformatics 12: S4. Young ND, John H (2003). GapCoder automates the use of indel characters in phylogenetic analysis. BMC Bioinformatics 6: 1–6. Yu Y, Than C, Degnan JH, Nakhleh L (2011). Coalescent histories on phylogenetic networks and detection of hybridization despite incomplete lineage sorting. Syst Biol 60: 138–49. Zhang D-X, Hewitt GM (1996). Nuclear integrations: challenges for mitochondrial DNA markers. Tree 11: 247–251. Zhang D-X, Hewitt GM (2003). Nuclear DNA analyses in genetic studies of populations: practice, problems and prospects. Mol Ecol 12: 563–584. Zimmermann T, Mirarab S, Warnow T (2014). BBCA: Improving the scalability of *BEAST using random binning. BMC Genomics 15: S11. 66 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 67 Appendix Table A1. List of samples used in this study (mtDNA lineage, sample code and the GenBank accession numbers) is accessible in https://www.dropbox.com/s/wbijjldyarud69j/Appendix_GenBank.pdf?dl=0 68 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Table A2. Samples of Podarcis used in this study. mtDNA lineage, sample code, locality information, including province/district and country and latitude/longitude. mtDNA lineage Sample Code Locality Province/District Country Latitude Longitude P. atrata 4H Columbrete Grande Castellón Spain 39.898760 0.684864 10 Columbrete Grande Castellón Spain 39.898760 0.684864 F8 Foradada, Columbretes Castellón Spain 39.875071 0.670547 L3 Lobo, Columbretes Castellón Spain 39.875041 0.671748 M13 Mancolibre, Columbretes Castellón Spain 39.895511 0.690119 P. bocagei 3.120 Madalena Porto Portugal 41.103983 -8.661383 3 166 Montesinho Bragança Portugal 41.979267 -6.795317 Sar1 Sarria Lugo Spain 42.783333 -7.400000 3.221 Subportela Viana do Castelo Portugal 41.687433 -8.718117 3.123 Vila Pouca de Aguiar Vila Real Portugal 41.445833 -7.672183 3.56 Gião Porto Portugal 41.312950 -8.691633 3.341 Gerês Braga Portugal 41.718333 -8.166667 Tab1 Taboadela Ourense Spain 42.233333 -7.816667 3 281 Tanes Astúrias Spain 43.211167 -5.402533 DB4292 Torneros de la Valdería León Spain 42.225422 -6.239401 Mad11 Madalena Porto Portugal 41.103983 -8.661383 M6 Montesinho Bragança Portugal 41.979267 -6.795317 bta5 Tanes Astúrias Spain 43.211167 -5.402533 V12 Vairão Porto Portugal 41.330383 -8.672400 P. carbonelli 4.97 Cabo Raso Lisboa Portugal 38.700000 -9.466667 4 159 El Acebuche Huelva Spain 37.047740 -6.565696 MC16 Monte Clérigo Faro Portugal 37.339880 -8.853810 SPM9 S. Pedro de Moel Leiria Portugal 39.750000 -9.016667 N292 S.Pedro do Sul Viseu Portugal 40.750000 -8.066667 VR16 Villasrubias Salamanca Spain 40.316667 -6.616667 Av8 Aveiro Aveiro Portugal 40.631750 -8.746350 4.43 Berlengas Peniche Portugal 39.415211 -9.508529 ATL6 Carriço Leiria Portugal 39.966667 -8.800000 Esm18 Esmoriz Aveiro Portugal 40.61666 -8.750000 Alb8 La Alberca Salamanca Spain 40.466667 -6.083333 Albc7 La Alberca Salamanca Spain 40.466667 -6.083333 MC3 Monte Clérigo Faro Portugal 37.339880 -8.853810 PR3 Playa del Rompeculos Huelva Spain 37.106177 -6.756996 SPM2 S. Pedro de Moel Leiria Portugal 39.750000 -9.016667 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 69 P. erhardii DB3820 Menites Andros Greece 37.826416 24.900352 DB3819 Thirasia islet Santorini Greece 36.447830 25.344021 P. g. lusitanica 5 143 Alvão Vila Real Portugal 41.350000 -7.866667 5 259 Ledesma Salamanca Spain 41.091750 -5.997900 Anc5 Los Ancares León Spain 42.669633 -6.726967 5.23 Moledo Viana do Castelo Portugal 41.838567 -8.874067 5 121 Vale de Rossim, Serra da Estrela Guarda Portugal 40.383333 -7.516667 5 150 Vila de Rua Viseu Portugal 40.950000 -7.566667 Cel1 Celanova Ourense Spain 42.150000 -7.966667 5 180 Gerês Braga Portugal 41.718333 -8.166667 Pen21 Pendilhe Viseu Portugal 40.883333 -7.816667 5 225 Tudera Zamora Spain 41.416890 -6.210430 PG2 Atlantic Islands ONS Spain 42.386001 -8.930933 Anc2 Los Ancares León Spain 42.669633 -6.726967 Mon1 Montesinho Bragança Portugal 41.979267 -6.795317 Mon2 Montesinho Bragança Portugal 41.979267 -6.795317 Mon8 Montesinho Bragança Portugal 41.979267 -6.795317 Pen2 Pendilhe Viseu Portugal 40.883333 -7.816667 Pen8 Pendilhe Viseu Portugal 40.883333 -7.816667 FT12 Tua Bragança Portugal 41.218333 -7.368333 Rua1 Vila de Rua Viseu Portugal 40.950000 -7.566667 P. g. guadarramae 5 206 Alba de Tormes Salamanca Spain 40.825590 -5.515460 5 194 Ciudad Rodrigo Salamanca Spain 40.592950 -6.536333 GuaI1 Guadarrama Madrid Spain 40.683333 -4.083333 Oro1 Oropesa Toledo Spain 39.919900 -5.174650 TC1 Torrejón de la Calzada Madrid Spain 40.200000 -3.800000 6 291 Trujillo Cáceres Spain 39.460667 -5.881500 Are1 Arévalo Ávila Spain 41.062071 -4.720288 HLA2 La Alberca Salamanca Spain 40.466667 -6.083333 HLA5 La Alberca Salamanca Spain 40.466667 -6.083333 Vil6 Villacastín Segóvia Spain 40.783333 -4.416667 HAlb1 La alberca Salamanca Spain 40.466667 -6.083333 Trj1 Trujillo Cáceres Spain 39.460667 -5.881500 Vil3 Villacastin Segóvia Spain 40.783333 -4.416667 Vil8 Villacastin Segóvia Spain 40.783333 -4.416667 P. hispanica “Albacete/Murcia” ALB1=9.76 Cañada de Provencio Albacete Spain 38.518033 -2.353150 DB1841 El Pardal Albacete Spain 38.485309 -2.287438 DB1878 Montealegre del Castillo Albacete Spain 38.830755 -1.339061 76 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome PKM2int5 RAG1 RAG2 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 77 Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome Pod11 Pod13 RAG2 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 78 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome Pod14 Pod14b 1 chromosome 1 mutation 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 79 Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome Pod15 Pod15b 1 chromosome 1 mutation Pod16 1 chromosome 1 mutation 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 80 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome Pod17 Pod21 Pod20 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation Pod25 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 81 Figure A1. (Continuation) 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome Pod25 Pod33 Pod38 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 82 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure A1. (Continuation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome Pod43 Pod55 1 chromosome 1 mutation Pod69 1 chromosome 1 mutation 1 chromosome 1 mutation Pod72 1 chromosome 1 mutation 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 1 mutation 1 chromosome 83 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards 2.0 82 99 64 96 31 95 8 69 11 42 95 77 47 100 61 10 99 68 99 16 100 37 84 49 100 81 P. taurica P. tiliguerta P. sicula P. erhardii P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. hispanica “Batna” P. hispanica “Azazga” P. vaucheri “Spain” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. virescens P. carbonelli P. liolepis 2 P. liolepis 3 P. liolepis 1 P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. muralis 1 P. muralis 2 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. atrata 2 P. atrata 1 Eastern Iberia P. hispanica North Africa Southeastern Iberia Western and Central Iberia P. vaucheri Figure A2. Phylogeny of Iberian and North African Podarcis species as estimated with the NJst method using each gene ML tree topology from the “full” dataset; a) “best” NJ species-tree; b) NJst consensus consensus (branch lengths are transformed and do not reflect any amount of evolution). Trees were inferred unrooted and rooted for visualization in the branch leading to P. erhardii, P. taurica, P. sicula and P. tiliguerta. The numbers on the nodes indicate multilocus bootstrap support values for branches calculated following Seo, 2008. 2.0 82 99 64 96 31 95 8 69 11 42 95 77 47 100 61 10 99 68 99 16 100 37 84 49 100 81 P. taurica P. tiliguerta P. sicula P. erhardii P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. hispanica “Batna” P. hispanica “Azazga” P. vaucheri “Spain” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. virescens P. carbonelli P. liolepis 2 P. liolepis 3 P. liolepis 1 P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. muralis 1 P. muralis 2 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. atrata 2 P. atrata 1 Eastern Iberia P. hispanica North Africa Southeastern Iberia Western and Central Iberia P. vaucheri b) a) FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North Africa Podarcis Wall Lizards 84 Figure A3. Posterior density of consensus species-trees (cloudogram) from *BEAST analyses for the Iberian and North African Podarcis species for 21 loci, for a chain length/tree prior of a) 605M/Yule; b) 168M/coalescent. Consensus for every topology with branch length calculated as the average of the branch length for all trees with the same topology. P. carbonelli P. vaucheri “Spain” P. liolepis 2 P. liolepis 1 P. atrata 2 P. liolepis 3 P. atrata 1 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. virescens P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. hispanica “Jebel Sirwah” P. hispanica “Batna” P. hispanica “Tunisia/Algeria” P. hispanica “Azazga” P. taurica P. tiliguerta P. sicula P. erhardii P. muralis 1 P. muralis 2 P. hispanica North Africa Southeastern Iberia Eastern Iberia Western and Central Iberia P. vaucheri b) P. carbonelli P. vaucheri “Spain” P. liolepis 2 P. liolepis 1 P. atrata 2 P. liolepis 3 P. atrata 1 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. hispanica “Galera” P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. virescens P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. vaucheri “Morocco/Algeria 2” P. vaucheri “Morocco/Algeria 1” P. hispanica “Jebel Sirwah” P. hispanica “Batna” P. hispanica “Tunisia/Algeria” P. hispanica “Azazga” P. taurica P. tiliguerta P. sicula P. erhardii P. muralis 1 P. muralis 2 P. hispanica North Africa Southeastern Iberia Eastern Iberia Western and Central Iberia P. vaucheri a) 85 FCUP Multilocus Phylogenetics: Inferring the Species-Tree of the Iberian and North African Podarcis Wall Lizards Figure A4. Posterior density of consensus species-trees (cloudogram) from *BEAST analyses for the Iberian and North African Podarcis species for all loci, for a chain length/tree prior of 341M/Yule. Consensus for every topology with branch length calculated as the average of the branch length for all trees with the same topology. P. vaucheri “Spain” P. vaucheri “Morocco/Algeria 2” P. virescens P. guadarramae guadarramae P. guadarramae lusitanica P. bocagei P. carbonelli P. liolepis 2 P. liolepis 1 P. hispanica sensu stricto 2 P. hispanica sensu stricto 1 P. liolepis 3 P. atrata 2 P. atrata 1 P. hispanica “Albacete/Murcia 2” P. hispanica “Albacete/Murcia 1” P. hispanica “Jebel Sirwah” P. hispanica “Tunisia/Algeria” P. hispanica “Batna” P. hispanica “Azazga” P. vaucheri “Morocco/Algeria 1” P. muralis 1 P. muralis 2 P. hispanica “Galera” P. taurica P. tiliguerta P. sicula P. erhardii PVS PVMA2 PH2 PH1B PH1A PB PC PL2 PL1 PHSS1 PHSS2 PL3 PA1 PA2 PHAM1 PHAM2 PHGAL PHJS PHTA PHBAT PHAZA PVMA1 PM1 PM2 PT PE PS PTA Western and Central Iberia Eastern Iberia Southeastern Iberia P. hispanica North Africa P. vaucheri