Bacteria form heterogeneous communities with specific spatial structures. The heterogeneity of bacteria can be genetic or phenotypic and even within one strain of bacteria there can be differences in the metabolism and behavior between the cells. The most common bulk method, the amplification and sequencing of 1–3 hypervariable regions of the 16S rRNA gene, reveals the assemblage of the bacterial community, from which the functional potential of the community can be inferred. In this thesis, I used 16S rRNA gene sequencing to examine how two commonly used agrochemicals, glyphosate-based herbicide (GBH) and phosphate, affect the endophytic bacterial communities of various potato, faba bean and oat tissues in the early vegetative and the late flowering growth stages (Chapter I). Changes in the community diversity and/or composition were seen in nearly all the plant tissues, in at least one of the growth stages, more so in response to phosphate than GBH. I also used the 16S rRNA gene sequencing approach to study how the gut microbiota of two odonate species, Lestes sponsa and Sympetrum vulgatum, varied in response to environmental factors, species, sex, sampling site and season, and diet (Chapter II). The diversity of the gut microbiota was higher in S. vulgatum than in L. sponsa, potentially in response to a more diverse diet. The diet affected the microbiota assemblages more than the environmental factors, but the effect was not strong, potentially due to large variation between individuals. As the 16S rRNA gene sequencing approach and other bulk methods cannot uncover details about the heterogeneity and spatial structuring of the microbiome, novel single-cell methods are required. Chapter III presents Prider, a computational tool we developed for high-throughput primer and probe designing. Chapter IV describes a single-cell sequencing based method we developed that combines porous polyacrylamide beads and split-pool barcoding. The method does not require specialised equipment, such as microfluidic systems, and once fully optimised, can be used for the analysis of both the microbial genetic heterogeneity as well as microbial interactions in communities, expanding the kind of information that can be gained from microbiome research, including studies such as those described in Chapter I and Chapter II.