Variation in the chloroplast genome of 44 accessions representing 14 Eucalyptus L'Hér. species from the series Viminales (sensu Pryor and Johnson 1971) was investigated. Southern analysis of the chloroplast genomes restricted with 12 enzymes revealed 20 restriction-site polymorphisms of which 7 were autapomorphic to individual trees. The 13 informative restriction-site polymorphisms were distributed between individuals of different species, but none was species-specific. Fourteen chloroplast haplotypes were identified for south-eastern Australian individuals. In endemic Tasmanian species, five haplotypes were identified. Chloroplast haplotypes appear to have a mosaic distribution in south-eastern Australia, more closely associated with geographical regions than with morphological species boundaries. The biogeographic distribution of chloroplast haplotypes may be explained by a combination of interspecific hybridisation and introgression, and convergent evolution. The lack of species-specificity of cpDNA variation indicates that, although cpDNA is not appropriate for species-level phylogeny analysis in the series Viminales, it may provide useful information in studies of biogeography and gene flow in Eucalyptus.

Patterns of variation in the Eucalyptus globulus Labill. complex are reassessed by combining capsule measurements from an earlier study with recent collections, mainly of subspecies globulus. Four groups of populations are apparent and can be ascribed to the four subspecies maidenii, pseudoglobulus, bicostata and globulus. Intergrade populations between the latter three subspecies are widespread and mainly occur in the Otway Ranges and west Gippsland. There is a continuum in capsule morphology between the three-fruited subspecies, pseudoglobulus and bicostata. Subspecies globulus intergrades with these three-fruited intermediates. Three-fruited intergrade populations occuning north and south of the range of core pseudoglobulus can be differentiated and probably represent intergrades between pseudoglob~rlus and bicostata and between pseudoglobulus and globul~is respectively. Reports of bicostata in the Furneaux Group and southern Victoria are thus probably erroneous and result from convergence in capsule morphology. The previously described taxon E. stjohnii (R. T. Bak.) R. T. Bak. is part of the continuum between subspecies pseudoglobulus and bicostata, but closer to pseudoglobulus. Populations phenotypically intermediate between and significantly different from globulus and the three-fruited intergrades are highly variable and occur in western Tasmania, on the northern end of Flinders Island, in the Otway Ranges and in west Gippsland. An isolated population on Rodondo Island is highly variable and has closest affinities to pseudoglobulus despite being within the geographical range of core globulus. The population from King Island is intermediate between the Otway phenotype and core globulus. The climatic regimes of the subspecies are markedly different and most three-fruited and globul~cs intergrade populations have closer climatic affinities to pseudoglobulus and globulus respectively. Hypotheses relating to the origin of the pattern of variation in E. globulus are discussed.

E. risdonii is a rare Tasmanian endemic which occurs as a series of small disjunct populations within a more-or-less continuous population of a closely related species, E. amygdalina. In localized areas (e.g. Risdon, Tasmania), patches of high phenotypic diversity are encountered, with individuals encompassing the complete phenotypic range between these two species. Progeny trials indicate a large heritable component to this variation. Open-pollinated progenies from intermediate mothers exhibit greater variability than those from either pure species, which strongly suggests that these intermediate phenotypes are a result of hybridization. Progenies from pure species mothers near a hybrid swarm are more variable than those from pure stands. There is a greater proportion of seedlings which match the artificially produced F1 in open-pollinated progenies from E. amygdalina than from E. risdonii mothers. In addition, progenies from intermediate mothers show a bias toward E. risdonii types, which implies that they are predominantly outcrossing to E. risdonii. This evidence suggests an asymmetrical flow of genes by pollen migration from E. risdonii into the hybrid swarm and surrounding E. amygdalina. Little difference in seedling vigour or mortality occurred between seedlings from the various parental phenotypes, although differences in seed output per capsule and germination were apparent. The specific identity of E. risdonii and E. amygdalina is usually maintained in parapatry by a range of mechanisms including their specific ecological preference, reduced hybrid fitness and differences in flowering phenology. The reasons for the major zones of hybridization occurring at boundaries on ridge tops as opposed to those on the dry slopes are discussed.

Regeneration of a hybrid zone between E. amygdalina and E. risdonii and pure species stands following wildfire is reported, as well as the reproductive and vegetative fitness of parental and hybrid phenotypes. E. risdonii phenotypes dominated the seed rain and seedling cohort and there was clearly a marked fitness differential between E. amygdalina and E. risdonii at their boundary. When the F1 type hybrid is in competition with both parental types it is generally reproductively the least fit, although frequently vegetatively vigorous. Reduced fitness appears to extend to advanced generations as hybrid phenotypes tending.toward either species are, on average, less fit than the corresponding parental type. The pattern of phenotypic fitness suggests that the species' boundary is in disequilibrium and it is argued that E. risdonii is invading the range of E. amygdalina by both pollen and seed migration. There is an asymmetric distribution of F1 type hybrids across the boundary and the hybrid swarm examined is being invaded by E. risdonii genes. It is suggested that hybridization may be associated with natural disequilibrium and, where seed migration is limited, boundary movements may be preceded by a wave of hybridization due partly to pollen swamping of the least fit species. Hybrid swarms may develop but, at the boundary of large stands, are probably transitory. There is a marked inertia in the population response to the prevailing selective regime due to the extremely slow population turnover and limited dispersal potential. This is discussed in the broader context of non-equilibrium models where it is argued that dispersal may be the factor limiting population response to perturbation of a shallow environmental gradient. This is due to large geographical shifts in the position of the null point and would be accentuated in a patchy environment where migration as a front is prevented.

Fine-scale genetic structure in Eucalyptus globulus ssp. globulus native forest was detected using 69 randomly amplified polymorphic DNA (RAPD) markers. The association between genetic similarity and geographic distance was studied among 51 trees from the Tinderbox locality in Tasmania (distance ranging from 2 m to 4 km apart) and compared to 18 trees from localities up to 100 km away. Twenty pedigreed F1s were used as controls to scale the RAPD similarity among individuals to pedigree similarity. The association between genetic similarity and geographic distance was weak, yet at Tinderbox, highly related trees were shown to occur within 25 m of one another. There is an abrupt drop in average similarity after about 25 m, with no significant change with distances up to 14 km. Nevertheless, Tinderbox trees outside the 25 m genetic patches are still more similar to each other than they are to trees from the Mayfield Bay locality 100 km away. These results suggest that E. globulus native forests have a family group structure, superimposed on a noisy, background level of lower relatedness which extends over a wider geographical range. This study is unique in demonstrating the congruence between fine-scale genetic structure as revealed by molecular data and previous quantitative genetic data.