• Rodney Gordon Eastwood Rodney Gordon Eastwood

The butterfly family Lycaenidae (including the Riodinidae) contains an estimated 30% of all butterfly species and exhibits a diverse array of life history strategies. The early stages of most lycaenids associate with ants to varying degrees, ranging from casual facultative coexistence through to obligate association where the long-term survival of the butterfly is dependent on the presence of its attendant ants. Attendant ants guard the butterflies against predators and parasites during their vulnerable period of larval growth and pupation. The caterpillars, in return, reward the ants by providing attractive secretions from specialized glands in their cuticle. The prevalence of caterpillar-ant associations in the species rich Lycaenidae is in contrast with other Lepidoptera, where ant association appears only as isolated cases in otherwise non ant-associated lineages. This has led to the proposal that ant association may have influenced lycaenid diversification or even enhanced the rates of speciation in the group. In contrast, facultative ant-associated butterflies exhibit high levels of host plant integrity, so it is reasonable to assume that host plants may have played a significant role in their diversification. Since the influence of ants (or plants) on diversification is independent of geographic speciation modes such as vicariance or peripheral isolates, there is an underlying inference of sympatric speciation. Certain prerequisites thought to be important for sympatric speciation, such as mating on the host plant (or in the presence of the appropriate ant) as well as ant dependent oviposition preferences are characteristic of many obligate myrmecophiles. Not surprisingly, it has been suggested that evidence for sympatric speciation is more likely to be found in the Insecta since this additional mode of diversification could account for the large numbers of insect species. This thesis tested the diversification processes in obligate and facultative ant associated lycaenids using comparative methodologies in hierarchical molecular phylogenetic analyses. First, several hypotheses relating to the influence of ants on diversification in obligately ant associated lycaenid butterflies were tested in a phylogeographic analysis of the Australian endemic Jalmenus evagoras. The phylogeographic analysis revealed that regional isolation of butterfly subpopulations coincident with locally adapted ant taxa could generate a phylogenetic pattern in which related lycaenids would be seen to associate with related or ecologically similar ants. Likewise, ecological shifts in habitat preferences by lycaenids could lead to co-diversification with habitat specialist ants, even though in both cases, the ants may play only an incidental role in the diversification process. A comparative methodology was then applied in a molecular phylogenetic analysis of the genus Jalmenus to test for a signal of diversification consistent with shifts in ant partners, and to infer the processes by which ants could influence speciation. Several other specific hypotheses relating to monophyly and taxonomy were also examined. Comparative analysis of the Jalmenus phylogeny found that attendant ant shifts coincided with high levels of sympatry among sister species. This pattern could be explained by sympatric speciation; however, data suggested it was more likely that ant shifts occurred during butterfly population expansions as a result of vegetation and climate changes in the Pleistocene. Fragmentation of populations associating with novel ants could promote rapid ecological and behavioural changes and this could result in reproductive isolation of conspecifics when in secondary contact. Diversification would then continue in sympatry. In contrast, secondary contact of populations associating with the same ant species would result in homogenisation of the two lycaenid lineages or the extinction of one. A phylogeographic analysis of the facultative myrmecophiles, Theclinesthes albocincta/T. hesperia, was then undertaken to infer the evolutionary processes (such as the effects of host plant shifts) that could result in extant demographics. Species-specific questions of taxonomy, relative population ages and dispersal routes in arid Australia were also addressed. Results from the analysis suggested the two taxa were conspecific and had diversified in the late Pleistocene as a consequence of isolation in refugia in and around the arid areas of mainland Australia. However, as was the case in the J. evagoras population analysis in which attendant ant shifts were not detected, host plant shifts were not detected in the population analysis of T. albocincta/hesperia. Host plant or attendant ant shifts manifest more frequently at the species level, thus it was necessary to test the influence of host plant shifts at this higher level. The comparative methodology was then applied to a molecular phylogenetic analysis of the facultative ant-associated section Theclinesthes (comprising Theclinesthes, Sahulana and Neolucia) to test for modes of diversification consistent with host plant shifts. The relative importance of other influences on diversification was also assessed. Akin to the Jalmenus analysis, the prediction that sister species ranges should broadly overlap when a shift in host plants had taken place was upheld in the comparative analysis. Species in the genera Jalmenus and Theclinesthes were found to have diversified in the Pleistocene so were subject to the same climatic oscillations that influenced patterns of vegetation expansion and contraction across much of Australia. Thus, the similarity and predictability of relationships in the comparative analyses based on biological data suggested that host plant shifts have influenced diversification in facultative myrmecophiles by inhibiting gene flow in secondary contact in similar fashion to that of attendant ant shifts identified in the Jalmenus phylogeny. Interpretation of data in these analyses suggested that allopatric diversification was the most common mode of speciation. Isolation was inferred to be the result of fragmentation following long distance dispersal across wide expanses of marginal habitat, or vicariance following the closing of biogeographical barriers. However, attendant-ant and host-plant shifts clearly played an important role in the diversification process, and in the maintenance of species integrity among lycaenid butterflies. Furthermore, exceptions to the predicted patterns of range overlap and ecological shifts provided clues to additional modes of diversification including shifts in habitat preferences and an unusual temporal shift following changes in specific host plant phenology resulting in allochronic diversification. Inferring modes of diversification using comparative methods based on range overlap and biological traits in a phylogenetic context is not new; however, the interpretation presented in this thesis is in contrast with contemporary methods. It is clear that the patterns of species range overlap and the ecological preferences of sister taxa are intimately related among lycaenid species that diversified during the Pleistocene. As a result, different influences on diversification can be highlighted in phylogenies when applying existing comparative methodologies but without necessarily drawing the same conclusions about modes of diversification. A more inclusive explanation for patterns of range overlap among sister taxa is detailed, a consequence of which is a method for estimating rates of extinction in a phylogeny where comprehensive distributional, biological and taxonomic data are available. These patterns and predictions may be applicable to a range of taxa, especially those that have diversified in the Pleistocene. Plans for future studies are outlined.