×Iberian and North African Podarcis wall lizards constitute a cryptic species complex for which different morphological studies have produced largely conflicting results regarding the number, taxonomic rank and distribution of evolutionary entities. Recently, the assessment of mitochondrial DNA (mtDNA) diversity demonstrated the existence of highly differentiated and geographically consistent clusters, some of which corresponded neatly to partitions previously suggested by morphological analyses. As a result of this concordance, systematic reevaluations were carried out, and at present five different species are recognised: P. bocagei, P. carbonelli, P. vaucheri, P. hispanica and P. atrata, although the latter has been shown to be closely related to one of the various phylogroups that constitute the polytypic (and, from a mitochondrial DNA perspective, paraphyletic) P. hispanica. Given this framework, the first goal of this thesis was to characterize in more detail the dynamics of genetic differentiation among forms of this species complex. Previous descriptions of genetic variability within the complex relied on a single marker – mitochondrial DNA –, and biogeographical inferences were heavily dependent upon poorly supported estimates of relationships. On one hand, we aimed at obtaining a more robust mtDNA-based phylogenetic tree in order to validate previous hypotheses regarding the biogeographical processes governing variability within the clade. On the other hand, we sought to complement these data with information from the nuclear genome, in order to independently assess the distinctiveness of the various clades previously described. Additionally, by examining patterns of nuclear-mitochondrial concordance and analysing in detail the dynamics of a contact zone between two species, we were expecting to evaluate levels of reproductive isolation among forms of the species complex. The reevaluation of mitochondrial phylogenetic relationships was accomplished by selecting a few individuals from each lineage and extending the amount of included sequence data. In addition to the two mtDNA gene regions already published, we sequenced three other fragments, yielding a total of 2425bp. A robust phylogenetic tree was recovered, with a significant proportion of nodes receiving bootstrap support close to 100%. This rigorous assessment of mtDNA relationships between lineages confirmed some of the results obtained in previous studies, such as the paraphyly of P. hispanica. However, other aspects of the phylogeny differ, offering alternative scenarios for the timing and polarity of the colonization of North Africa by wall lizards. In particular, we suggest that the opening of the Strait of Gibraltar might have caused vicariant isolation of Iberian from North African forms, and that a subsequent transmarine colonization could have occurred from North Africa to Iberia. An additional important contribution of this study was the detection of a previously undescribed lineage inhabiting southern Spain, which further exemplifies the evolutionary complexity within this group. Two major lines of research were pursued regarding the description of nuclear gene variation: allozyme electrophoresis and sequencing of nuclear introns. The study of allozyme variation in 30 populations representing all known mtDNA lineages documented a high degree of genetic divergence between most of them, with groups within P. hispanica generally presenting similar levels of differentiation to those observed between fully recognised species. Although this constitutes evidence corroborating the validity of mtDNA lineages as evolutionary distinct units, the application of recent model-based clustering approaches revealed some important discordant patterns with respect to the mtDNA, which may be indicative of the occurrence of gene flow between forms. In this context, two hybrid individuals between the broadly sympatric P. bocagei and P. hispanica type 1 were detected. Summary viii Conversely, the study of two nuclear gene genealogies (β-fibint7 and 6-Pgdint7), demonstrated that, in general, species and mtDNA lineages are strikingly nonmonophyletic, which was unexpected given the high levels of differentiation detected by the use of other markers. An important issue that required clarification was, then, if this pattern emerged as the result of incomplete lineage sorting of ancestral polymorphism (motivated by the four-times higher effective population size of nuclear genes when compared to the mtDNA) or if it resulted from high levels of gene flow. In order to disentangle between these two non- mutually exclusive hypotheses, we estimated levels of gene flow between all pairs of lineages using recent methods of analyses based on coalescent theory that allow the study of the dynamics of admixture and divergence in the same framework. In contrast to the information obtained using classic, F-statistic-based estimators of gene flow, this approach revealed that only a few pairs of lineages have exchanged genes since their divergence, and even fewer show important levels of gene flow, suggesting that in general the various lineages are overall differentiated. Therefore, although gene flow has certainly played an important role in the evolution of the species complex (and as a result some mtDNA lineages may have even lost their nuclear identity), the polyphyletic pattern observed in nuclear genealogies is especially influenced by the persistence of ancestral polymorphism. These studies addressed the subject of gene flow among forms superficially, mostly based on the patterns of mitochondrial/nuclear discordance. In order to characterize in more detail the dynamics of gene flow, a battery of nuclear and mitochondrial markers was studied along a transect crossing the contact zone between P. bocagei and P. carbonelli. This information was complemented with analyses of morphology and fertility. Applying model-based individual multilocus genotype clustering approaches, we document abundance of individuals showing signs of admixture in the locality where the two species meet. However, strong Hardy-Weinberg and linkage disequilibria were observed. This clear bimodality suggests the existence of strong barriers to gene flow, the nature of which still remain obscure. For example, our data suggests neither obvious reductions in hybrid fertility nor the verification of Haldane’s rule. However, divergent natural selection acting on morphology is suggested by the fact that genetically admixed individuals are clearly assigned to one species or the other based on morphological characters. Bimodality within a hybrid zone is usually suggestive of a nearly complete process of speciation; while these results cannot be promptly generalized to other contact zones, they are in accordance with previous suggestions that, although Podarcis lineages are not fully reproductive isolated, levels of gene flow among them do not seem to challenge the ongoing process of differentiation. A second major goal of this thesis was to describe the phylogeographic structure within selected forms of the species complex using multiple molecular markers. In particular, we attempted to evaluate the response of the species to the Pleistocene climatic oscillations, comparing the patterns detected with phylogeographic scenarios described for other species occupying the same regions, and interpreting contrasting signatures left by the Ice Ages on distinct species of Podarcis. The study of mtDNA variation in P. bocagei, P. carbonelli and P. vaucheri, three species with parapatric distributions that replace each other along a latitudinal gradient, allowed us to test two predictions relative to the differential influence of Quaternary climatic oscillations on distinct latitudes: i) northerly distributed species are expected to bear lower levels of genetic subdivision and diversity than species distributed in the south because they were able to survive in fewer, smaller and less stable patches of favourable habitat during glaciations; ii) species distributed in the south should retain signs of long-term effective population size stability, while northern species, because they were more Summary ix confined during glacial stages, should have experienced a rapid demographic growth concomitant with a post-glacial colonization of newly available habitats. Our results show that P. bocagei presents remarkably low levels of diversity and subdivision, a shallow coalescent history (the estimated coalescence time was ~100,000-~70,000 years) and a strong signature of demographic growth. P. vaucheri, on the other hand, presents large levels of genetic diversity even at small geographic scales, strong geographic subdivision, no evidence of demographic growth and an ancient coalescence time, probably dating back to the initial stages of the Pleistocene. The intermediately distributed P. carbonelli presents average values of all studied variables. Taken together, these results entirely fit to our main predictions and demonstrate that different effects of the Ice Ages can be detected at small geographic scales. The analyses of nuclear markers (allozymes and a set of microsatellite loci specifically developed during this study) in populations of P. bocagei and P. carbonelli confirmed, on one hand, the low levels of population subdivision observed in both species from the study of mitochondrial DNA. Moreover, levels of differentiation are higher among populations of P. carbonelli than among populations of P. bocagei, which is also in accordance with higher levels of persistence and a consequent longer time for differentiation in the former inferred from mtDNA. In P. bocagei, we report a progressive loss of genetic diversity in a northwards fashion, consistent with a recent geographic expansion from a reduced source as suspected from mtDNA-based evidence of a rapid demographic growth. Moreover, based on a spatial interpolation of genetic data and on a Bayesian model-based multilocus genotype clustering algorithm, we were able to identify probable expansion routes and to pinpoint with some degree of certainty the area that has probably functioned as a glacial refugium for the species during the last glaciations. Applying the same methodology in P. carbonelli we document some degree of association between genetic clusters and geography, but with only partial correspondence to inferences based on mtDNA; to explain these discordant patterns, we hypothesise that recent fragmentations in the species’ distribution might have led to a loss of the genetic signatures that are typically found in co-distributed species, and that the patterns that are presently observed are more a byproduct of recent genetic drift than of Pleistocene climatic oscillations. Globally, these results constitute a major improvement regarding previous knowledge on evolutionary relationships, stages of differentiation, levels of gene flow and patterns of intraspecific subdivision in Iberian and North African Podarcis. At a more general level, the large-scale study of nuclear gene variation on these lizards, documenting polyphyly and gene flow among mtDNA lineages, illustrates that these correspond to very incipient species with permeable boundaries, highlighting the need for the assessment of multiple genetic markers prior to any taxonomic reevaluations. Moreover, the results have a double importance for the understanding of the Iberian and North African Quaternary biogeography: on one hand, by helping to confirm the transversality of patterns that appear in response to climatic fluctuations, such as glacial fragmentation and post-glacial expansion; on the other hand, by documenting how species-specific such patterns may be even among closely-related species.
