Harmful algal blooms are commonly thought to be dominated by a single genus, but they are not homogenous communities. Current approaches, both molecular and culture-based, often overlook fine-scale variations in community composition that can influence bloom dynamics. We combined homology-based searches (BLASTX) and phylogenetics to distinguish and quantify Microcystis host and phage members across a summer season during a 2014 Microcystis- dominated bloom that occurred in Lake Tai (Taihu), China. We found 47 different genotypes of the Microcystis-specific DNA-dependent RNA polymerase (rpoB), which included several morphospecies. Microcystis flos-aquae and Microcystis wesenbergii accounted for ~86% of total Microcystis transcripts, while the more commonly studied Microcystis aeruginosa only accounted for ~7%. Microcystis genotypes were classified into three temporal groups according to their expression patterns across the course of the bloom: early, constant and late. All Microcystis morphospecies were present in each group, indicating that expression patterns were likely dictated by competition driven by environmental factors, not phylogeny. We identified three primary Microcystis-infecting phages based on the viral terminase, including a novel Siphoviridae phage that may be capable of lysogeny. Within our dataset, Myoviridae phages consistent with those infecting Microcystis in a lytic manner were positively correlated to the early host genotypes, while the Siphoviridae phages were positively correlated to the late host genotypes, when the Myoviridae phages express putative genetic markers for lysogeny. The expression of genes in the microcystin-encoding mcy cassette was estimated using mcyA, which revealed 24 Microcystis-specific genotypes that were negatively correlated to the early host genotypes. Of all environmental factors measured, pH best described the temporal shift in the Microcystis community genotypic composition, promoting hypotheses regarding carbon concentration mechanisms and oxidative stress. Our work expounds on the complexity of HAB events, using a well-studied dataset to highlight the need for increased resolution of community dynamics.