Effects of insects and soils on the assembly of universal microbial decomposers and prediction of postmortem interval

Establishing a timeline of events associated with death is fundamental to medicolegal death investigations. The postmortem interval (PMI) is a component of the investigative timeline that can be used to assess the veracity of statements, identify the deceased, and facilitate closure for the bereaved. However, estimating PMI is difficult and can require the use of biological-based evidence such as microorganisms and insects. Over the last decade, the proposal investigators have demonstrated the power of microbiome-based estimates of PMI for carrion and human cadavers. Recent research by the proposal investigators generated a machine learning model utilizing microbiome data associated with 36 human cadavers from three forensic facilities that predicts PMI within approximately +/- 3 days over the first 21 days postmortem. This research also revealed many of the microbial taxa that are highly predictive of PMI are likely brought to cadavers by insects. Our preliminary data from indoor-decomposed cadavers with delayed insect exposure resulted in decreased machine learning model performance for estimating PMI. This suggests insect-seeded microbial taxa are important features in the model. Furthermore, the source of several other highly predictive microbial taxa in the PMI models is unclear, but may be rare members of the soil microbiome. Therefore, we aim to address knowledge gaps about how microbial decomposer communities depend on insect and soil sources (goal 1), and how the absence of these potential sources affects PMI accuracy (goal 2). The proposed work includes a replicated, block-controlled experiment utilizing swine in an outdoor, terrestrial environment. The first goal includes conducting a field experiment in which swine carrion are exposed to treatments of i) endogenous insects and soils ii) endogenous soil with macro insects excluded iii) endogenous insects with soils excluded iv) soil and macro insects excluded, as well as determining sources of decomposer microbes by collecting host skin and fecal swabs, insects, and soils at the start of experiment. For the second goal, the team will develop PMI models for each treatment to understand how carrion exposure to different sources affects the model performance and important features, as well as comparing machine-learning and artificial intelligence algorithms. The purpose of the proposed research is to create a microbial-based model to predict PMI across multiple environmental contexts, and thus increase knowledge about physical evidence. The products will include publicly available microbiome data, modeling code, and open-access peer-reviewed publications. We will further disseminate results and research products via presentations to stakeholders. CA/NCF