Harmful blooms of the haptophyte Phaeocystis globosa can substantially reshape surrounding microbial communities, with potential consequences for bloom dynamics and biogeochemical cycling. However, distributional breadth patterns across archaeal, bacterial, viral and eukaryotic groups, as well as their cross-domain co-occurrence patterns, remain poorly characterized. In this study, surface waters were collected from a P. globosa bloom-affected site (Tiaoshun Bridge, TS) and a non-bloom reference site (Jinshawan, JSW) in Zhanjiang Bay, China, and analyzed via shotgun metagenomic sequencing, sample-associated distributional breadth estimation, and FDR-corrected Spearman correlation network analysis. Chlorophyll a was markedly higher at TS than at JSW (49.53 vs. 9.93 μg·L⁻¹), consistent with greater P. globosa colony abundance (36 vs. 8 colonies·L⁻¹), confirming strong spatial heterogeneity in bloom intensity. The eukaryotic community at TS was dominated by Haptophyta and Phaeocystaceae, whereas Ascomycota, Bacillariophyta, Bacteroidota, and Nucleocytoviricota were enriched at JSW. Generalist taxa predominated among archaea and bacteria, whereas microbial eukaryotes exhibited stronger distributional specialization. The highly modular network was bacteria-dominated (87.3% of all nodes). Notably, under the specified thresholds, P. globosa showed only two significant positive associations, exclusively with two Pseudoalteromonas taxa. These findings indicate that bloom-associated microbial networks are structured primarily by bacterial covariation rather than by direct algal-centered linkages, and identify Pseudoalteromonas as candidate bloom-associated bacteria for future hypothesis testing.