In the past decade, the characterization of microbial community in soil of contaminated sites was primarily done by high-throughput short-read amplicon sequencing. However, the short-read approach often limits the microbial composition analysis at the species level due to the high similarity of 16S rRNA and ITS genes amplicon sequences. Here, we simultaneously performed full-length (PacBio platform) and short-read (Illumina platform) amplicon sequencing to clarify the adaptation mechanisms of different microbial taxa to soil pollution from a high-resolution perspective. We found that (1) full-length 16S rRNA gene sequencing from PacBio platform gave better resolution for bacterial identification at all levels (especially at the level of genus and species), while there was no significant difference between the two platforms for fungal identification in some samples. (2) abundant taxa dominated the microbial communities, and abundant fungal species such as Mortierella alpine, Fusarium solani, Mrakia frigida, and Chaetomium homopilatum served as the keystone species. (3) heavy metal and soil texture affected microbial community structure significantly, and abundant taxa preferred deterministic processes, whereas rare taxa randomly formed due to weak selection. Importantly, our study for the first time characterized soil microbiota in contaminated sites with a superior resolution at the species level, emphasizing that abundant taxa, especially abundant fungi, played the keystone role in co-occurrence networks. Overall, these findings expand current understanding of the ecological mechanisms and microbial interactions in contaminated site ecosystems and demonstrate that full-length sequencing has the potential to provide more details of microbial community.