Nearly 30 MYA, a subfamily of termites, Macrotermitinae, started an obligate mutualistic relationship with the basidiomycete fungus Termitomyces. Fungus-growing termites maintain the ectosymbiont in optimal growth conditions and Termitomyces provides the food source and an external plant decomposition factory for the termites. This association ensured such a successful history since the origin of fungiculture that none of the two symbionts has abandoned the system. The adoption of the fungus by the eusocial termites has triggered changes in the colony, and here I focus on the responsive changes in the bacterial and fungal communities. To do this, we used pyrosequencing, fluorescent in situ hybridisation, light and confocal microscopy, enzymatic assays, chemical extractions, in vitro assays, and feeding experiments in this thesis work to elucidate these predicted changes in fungus-growing termite microbial communities. This thesis shows that fungusgrowing termite gut microbiotas are compositionally consistent over time. Fungus comb structures, where Termitomyces is cultivated in the nests, harbour bacterial communities that change over time. These fungus comb microbiotas are largely termite species-specific due to major contributions from gut deposits during comb formation, but the environment affects which gut bacteria dominate comb communities at a given point in time. We also find that gut bacterial communities are structured according to termite caste and species rather than colonies; thus, reflecting different caste roles and diets. Royal pair (queens and kings) microbiotas are, unlike the rich and diverse sterile caste microbiotas, extremely skewed and dominated by a few bacterial taxa, reflecting the specialised dietary intake and unique, reproduction-centred lifestyle of the queen and king. We therefore propose that division of labour extends beyond the termites to their gut bacterial communities. We also show that Termitomyces-fed cockroaches undergo compositional changes in their gut microbiotas. A step-wise gradient pattern in gut bacterial community that correlated with an increase in the proportion of fungal material provided to the cockroaches. However, gut microbiotas remained distinct from those of termites after Termitomyces-feeding, indicating that a fungal diet can play a role in structuring gut community composition, but at the same time exemplifies how original community compositions, and possibly gut microenvironment constrain the magnitude of change. This thesis also characterises the fungus comb fungal communities (mycobiotas) in fungusgrowing termites, and shows that non-Termitomyces fungi were essentially absent in combs, and that Termitomyces fungal crops are maintained in monocultures as heterokaryons with two or three abundant ITS variants in a single fungal strain. Additional Termitomyces variants were also detected with low frequency. We also tested for the presence of antifungal metabolites in Termitomyces or comb bacteria and found that both Termitomyces and chemical extracts of fungus combs inhibit a number of antagonistic fungi, but not the same ones. This indicates that there is likely a bacterial contribution to antifungal defences in fungus-growing termite nests. We finish the main part of the thesis by arguing that an integrated view of homogenous comparative genomic analyses and insectassociated microbiotas is required to improve our understanding of social insect disease-defences and how eusociality influenced insect immunity.