The continental island system of the Aegean region comprises a very interesting research area for some fundamental questions in evolutionary biology, since its paleogeological history is characterized by repeatedly fragmentation and merging of landmasses caused by changing sea levels during the Plio/Pleistocene climatic oscillations. The Aegean Nigella arvensis complex comprises six ecologically nonspecialized species (12 taxa) of the phrygana vegetation and is distributed across various islands of the Aegean Sea (e.g., Crete, Rhodos, Kiklades) and the surrounding Greek/Turkish mainland. Most members of the complex are outcrossing and interfertile (N. arvensis alliance) and are allopatrically distributed. Only two taxa are reproductively isolated, selfing species, one (N. doerfleri) is partly sympatric with an outcrosser (N. degenii) in the Kiklades. Following ITS analyses (Bittkau & Comes, 2008; Comes et. al., 2008) the complex is a monophyletic group, that likely originated between the Messinian and the Mid-Pleistocene (6.2-1.3 mya), and radiated in the Late Pleistocene (0.78-0.16 mya), mainly triggered by allopatric speciation. CpDNA analyses of the N. arvensis alliance (Bittkau & Comes, 2005) confirmed well-known phytogeographical patterns of the Aegean region (e.g. “Rechingers line”, a floristic boundary between Europe and Asia) and were broadly consistent with the present taxonomical circumscription of (sub)species, except several instances of haplotype sharing across taxonomical boundaries. The present AFLP study confirmed the genetical distinctiveness of most taxa of the N. arvensis complex based on genetic distance analyses among populations, with the exception of N. arvensis ssp. brevifolia, which was found to comprise two different lineages from Crete and Rhodos. Although basal genetic relationships among species were not resolved, further evidence for the existence of a hierarchical phylogenetic s tructure was provided by considering successively increasing numbers of genetic clusters (K) inferred from a Bayesian structure analysis. Interpreted as a spatial-temporal process, the resulting pattern suggested that the diversification of the complex might have proceeded from the Central (Kiklades) to the Eastern Aegean region and further to the Greek/Turkish mainland and/or its offshore islands. Due to their genetic distinctiveness in all analyses, the present data also argue for an independent origin of the selfers (N. doerfleri, N. stricta) from an unknown (but probably 'degenii-like') outcrossing ancestor. In light of the present AFLP-data incomplete lineage sorting instead of recent gene flow was regarded as the best 105 explanation for most instances of cpDNA haplotype sharing across (sub)species boundaries as previously reported by Bittkau & Comes (2005). However, one instance of putative homoploid hybrid origin was detected involving ssp. arvensis (Makedonia), which revealed admixture between its parapatric conspecifics ssp. aristata and ssp. glauca. Furthermore, the occurence of an Anatolian cpDNA haplotype in N. carpatha from Karpathos (Bittkau & Comes, 2005) was best explained by introgression/chloroplast capture following a long-distance dispersal event from Anatolia (or Samos) to Karpathos. Overall, the phylogeographical analyses supported well established phytogeographic patterns (e.g., “Rechinger's line”, “Kikladian window”). However, in contrast to the phylogeographic break across the Strait of Marmaris as registered by cpDNA (Bittkau & Comes, 2005) the AFLPs detected affinities of both N. carpatha from Kassos/Karpathos and N. arvensis ssp. brevifolia from Rhodos to ssp. glauca from (mainly) the Turkish mainland. The isolating effect of islands on population genetic structure was reflected by consistently lower gen etic diversity and higher population differentiation in the island taxa compared to those from the mainland. As expected by theory (Charlesworth & Pannell, 2001), both selfing species (N. doerfleri, N. stricta) revealed considerably higher total diversity in relation to mean within-population diversity values than the outcrossing taxa. Moreover, comparing levels of among-population differentiation in two outcrossing mainland/island taxa (N. arvensis/N. degenii) and the selfing island taxon N. doerfleri still revelaed stronger population differentiation in the selfer (sc = 0.24) than in the outcrossers (sc = 0.12/0.10) after accounting for a potential bias introduced by geographic (mainland/island) substructure. Together, these latter results show that the breeding system has an overriding effect on population substructure and diversity, and regardless of whether Nigella populations occur on the mainland or in the islands.