Quantifying human-environment interactions through Bayesian modeling of species-resolved microbial transfer signatures: an exploratory proof-of-concept study

Background Microbial trace evidence offers potential for forensic reconstruction of human-environment interactions, but current methods lack standardized quantitative frameworks. While 2bRAD-M (type IIB restriction site-associated DNA markers for microbiomes) sequencing provides species-level resolution from low-biomass samples, its integration with robust statistical models for forensic applications remains unexplored. Methods We developed an integrated framework combining 2bRAD-M sequencing with a Bayesian hierarchical model to quantify microbial transfer patterns. The model incorporates geospatial parameters, substrate-specific persistence kinetics, and temporal decay functions. We generated 2bRAD-M data from host-associated (skin, saliva; n = 12) and environmental samples (personal devices, high-touch surfaces; n = 14), integrated with public 16S rRNA data (Qiita studies; n = 2,263 samples) for model training. Key findings The Bayesian model demonstrated preliminary accuracy in attributing microbial traces to their likely source environment categories (within ~100 meters in preliminary tests) and provided initial estimates for deposition time. Personal devices were found to retain taxa associated with host such as Staphylococcus hominis for extended periods (exceeding 72 h in our observations), suggesting persistent microbial transfer. Conclusion This proof-of-concept study suggests that integrating 2bRAD-M sequencing with Bayesian modeling could provide a framework for quantitative reconstruction of microbial transfer histories. The approach indicates potential for forensic applications but is not yet validated for casework. Extensive validation with larger, independent datasets is imperative to assess its reliability and admissibility standards.