DNA was extracted from different depth soils (0-5, 45-55 and 90-100 cm below surface) sampled at Lower East Fork Poplar Creek floodplain (LEFPCF), Oak Ridge (TN, USA). The presence of merA genes, encoding the mercuric reductase, the key enzyme in detoxification of mercury in bacteria, was examined by PCR targeting Actinobacteria, Firmicutes or beta/gamma-Proteobacteria. beta/gamma-Proteobacteria merA genes were successfully amplified from all soils, whereas Actinobacteria were amplified only from surface soil. merA clone libraries were constructed and sequenced. beta/gamma-Proteobacteria sequences revealed high diversity in all soils, but limited vertical similarity. Less than 20% of the operational taxonomic units (OTU) (DNA sequences > or = 95% identical) were shared between the different soils. Only one of the 62 OTU was > or = 95% identical to a GenBank sequence, highlighting that cultivated bacteria are not representative of what is found in nature. Fewer merA sequences were obtained from the Actinobacteria, but these were also diverse, and all were different from GenBank sequences. A single clone was most closely related to merA of alpha-Proteobacteria. An alignment of putative merA genes of genome sequenced mainly marine alpha-Proteobacteria was used for design of merA primers. PCR amplification of soil alpha-Proteobacteria isolates and sequencing revealed that they were very different from the genome-sequenced bacteria (only 62%-66% identical at the amino-acid level), although internally similar. In light of the high functional diversity of mercury resistance genes and the limited vertical distribution of shared OTU, we discuss the role of horizontal gene transfer as a mechanism of bacterial adaptation to mercury.