Abstract
Microbial communities play a vital role in cadaver decomposition and serve as reliable tools for postmortem interval (PMI) estimation. However, current research focuses primarily on bacterial/fungal communities, though viruses-as Earth's most abundant biological entities-play key roles in biogeochemical cycles by regulating bacterial communities via lysis-lysogeny switching. Viral succession patterns during decomposition remain insufficiently characterized, and their PMI biomarker potential is unexamined. We present metagenomic analysis of viral succession during 35-day decomposition of buried rat cadavers, revealing stage-specific dynamics: early dominance of Peduoviridae (0-3 days), mid-stage proliferation of Herelleviridae (7-21 days), and late-stage resurgence of Peduoviridae (28-35 days). Viral α-diversity exhibits a fluctuating downward trend. β-Diversity analysis (PCoA, ANOSIM, PERMANOVA; P < 0.001) confirmed PMI as a major structural driver (27% variance explained). Nine viral families exhibited significant PMI correlations (P < 0.05): Zierdtviridae, Casjensviridae, Schitoviridae, and Ackermannviridae showed strong negative correlations (r = -0.82 to -0.78), while Straboviridae correlated positively (r = 0.59). Using integrated viral abundance data, our Extremely Randomized Trees model achieved exceptional PMI prediction accuracy (test set: R² = 0.96, MAE = 2.54 days). Spearman correlations between dominant bacterial phyla (Bacteroidota, Bacillota, etc.) and viral families, combined with Procrustes analysis (M² =0.3385, P = 0.001) and Mantel tests (r = 0.8059, P = 0.001), confirmed strong virus-bacteria community consistency. This indicates viruses may regulate decomposition by targeting bacteria for lysis, releasing nutrients (e.g., organic nitrogen/phosphorus) to drive bacterial succession. Our study establishes a novel virus-based PMI prediction tool and discusses ecological drivers of decomposition. IMPORTANCE: We present a viral succession-based framework for estimating PMI in buried remains. Our study identifies stage-specific viral biomarkers and identified nine viral families significantly correlated with PMI. By combining metagenomics and machine learning, we developed an Extremely Randomized Trees (ERT) model that achieved a low prediction error (test set: R² = 0.96, MAE = 2.54 days). Furthermore, our findings demonstrate that viral and bacterial communities exhibit significant consistency and correlation during cadaver decomposition. This study not only provides a novel tool for the accurate estimation of forensic PMI but also advances our insight into viral regulation of bacteria and their interactions during cadaver decomposition.