We address the controversy over the processes causing divergence during speciation. Host races of the fruit fly Tephritis conura attack the thistles Cirsium oleraceum and Cirsium heterophyllum. By studying the genetic divergence of T. conura in areas where host plants are sympatric, parapatric and allopatric, we assessed the contribution of geography in driving host-race divergence. We also evaluated the relative importance of genetic drift and selection in the diversification process, by analysis of the geographic distribution of genetic variation. Host races were significantly diverged at five out of 13 polymorphic allozyme loci. Variance at two loci, Hex and Pep D, was almost exclusively attributable to host-plant affiliation in all geographic settings. However, Hex was significantly more differentiated between host races in sympatry/parapatry than in allopatry. This result might be explained by selection against hybridisation or against incorrect host choice in contact areas. Linkage disequilibrium tests suggest the latter: gene flow in contact areas may occur from males of the host-race C. heterophyllum to females of the host-race C. oleraceum, whereas incorrect oviposition events were never observed. The distinctive patterns of genetic differentiation at the two highly differentiated loci implicate the action of selection (acting directly or on linked loci) rather than genetic drift. Despite their restricted interactions in sympatry, we conclude that host races are stable and that the major diversification process took place before species arrived in today's geographical settings.

Social cooperative spiders from diverse taxonomic families share life-history and demographic traits, including highly inbred colony structure. The combination of traits suggests constrained pathways for social evolution in spiders. The genus Stegodyphus has three independently evolved social species, which can be used as replicate samples to analyse population constraints in evolutionary time. We tested colony structure and population history of the social S. mimosarum from South and East Africa using mitochondrial DNA variation, and we compared the results to published data for the independently evolved social congener S. dumicola. S. mimosarum had many and diverse haplotypes (5-7% sequence divergence for ND1) but colonies were monomorphic and genealogically similar haplotypes occurred in abutting regions. These findings are nearly identical to results for S. dumicola and imply similar colony-level processes over evolutionary time in independently evolved social species. These population dynamics are discussed with respect to the apparent lack of cladogenesis in social spiders.

The fruit fly Tephritis bardanae infests flower heads of two burdock hosts, Arctium tomentosum and A. minus. Observations suggest host‐associated mating and behavioural differences at oviposition indicating host‐race status. Previously, flies from each host plant were found to differ slightly in allozyme allele frequencies, but these differences could as well be explained by geographical separation of host plants. In the present study, we explicitly test whether genetic and morphological variance among T. bardanae are explained best by host‐plant association or by geographical location, and if this pattern is stable over a 10‐year period. Populations of A. tomentosum flies differed significantly from those of A. minus flies in (i) allozyme allele frequencies at the loci Pep‐A and Pgd, (ii) mtDNA haplotype frequencies and (iii) wing size. In contrast, geographical location had no significant influence on the variance estimates. While it remains uncertain whether morphometric differentiation reflects genotypic variability or phenotypic plasticity, allozyme and mtDNA differentiation is genetically determined. This provides strong evidence for host‐race formation in T. bardanae. However, the levels of differentiation are relatively low indicating that the system is in an early stage of divergence. This might be due to a lack of time (i.e. the host shift occurred recently) or due to relatively high gene flow preventing much differentiation at loci not experiencing selection.

Colonial social spiders experience extreme inbreeding and highly restricted gene flow between colonies; processes that question the genetic cohesion of geographically separated populations and which could imply multiple origins from predecessors with limited gene flow. We analysed species cohesion and the potential for long‐distance dispersal in the social spider Stegodyphus dumicola by studying colony structure in eastern South Africa and the cohesion between this population and Namibian populations previously published. Data from both areas were (re)analysed for historic demographic parameters. Eastern South African S. dumicola were closely related to an east Namibian lineage, showing cohesion of S. dumicola relative to its sister species. Colony structure was similar in both areas with mostly monomorphic colonies, but haplotype diversity was much reduced in eastern South Africa. Here, the population structure indicated recent population expansion. By contrast, Namibia constitutes an old population, possibly the geographic origin of the species. Both the comparison of the eastern South African and Namibian lineages and the distribution within eastern South Africa show the potential for long‐distance dispersal in few generations via colony propagation.

The population genetic structure of the butterfly Melitaea didyma was studied along the northern distribution range border in Central Germany by means of allozyme electrophoresis. Individuals were sampled from a total of 21 habitat patches from four regions, and two provinces. Sampling was designed to estimate local vs. regional differentiation. High levels of variability were found, He= 0.14‐0.21. The mean expected sample heterozygosity from one region, Mosel, was significantly lower than from the Hammelburg region, He= 0.17 and 0.19, respectively. Two hierarchical levels of genetic differentiation were found. Within regions individuals sampled from different patches behaved as belonging to one population with high levels of gene flow (Hammelburg FST= 0.015, Mosel FST= 0.044), though local isolation barriers did create a substructuring of these populations. The inbreeding coefficients, FIS, were constant over all sample levels, suggesting a similar distribution of habitat patches within regions. Between regions gene flow was limited. An isolation by distance analysis indicated that the hierarchical structure, at the provincial level, may be breaking down due to isolation of regional populations. A more general observation was that the sampling design may greatly have influenced the estimation of genetic differentiation. Depending on which samples were included, overall FST estimates ranged from 0.059‐0.090.