Temporal dynamics and stability of the human microbiome for forensic individual identification

The human microbiome is ubiquitous across nearly all body sites and exhibits biological characteristics that enable stable detection under diverse environmental conditions. Microbial DNA can be analyzed even when human DNA is present in trace amounts or is severely degraded, supporting its potential as a complementary approach to conventional DNA-based identification. However, systematic validation of whether the microbiome simultaneously maintains individual specificity and long-term stability remains limited. In this study, we conducted a longitudinal analysis of the microbiome from the scalp, cheek, hand, and saliva of five healthy Korean participants over a period of up to three years. Intra-individual temporal stability and inter-individual variability were evaluated across multiple temporal scales, ranging from daily to annual intervals. Microbial community dynamics were assessed using relative abundance analysis, beta diversity metrics, and the theta-YC similarity index. Individual identification performance was evaluated for each sampling source using an XGBoost-based machine learning approach. Skin sites and saliva represented distinct ecological niches, with intra-individual similarity exceeding inter-individual similarity across sampling sites and temporal intervals. Although transient community shifts were observed in frequently exposed sites such as the scalp and hand under the four-season climate of the Republic of Korea, individual-specific microbial signatures were maintained over time. The XGBoost-based identification models achieved high accuracy, particularly for saliva (93.3%) and cheek (92.9%) samples. These findings support the potential of the skin and saliva microbiome as complementary tools for individual identification, particularly in forensic contexts where conventional human DNA analysis is limited.