Evidentiary soil in an investigation can link an individual with the scene of a crime since the diversity and geospatial distribution of soils can make it highly probative. Recently, advanced techniques have been developed that allow a deeper investigation into bacterial communities and produce considerably more data than previous methods. This research used next-generation sequencing and statistical analyses to identify factors influencing soil bacterial communities and assess the feasibility for their use forensically. Soil samples were collected from a variety of habitats over different distances, depths, and times, DNAs were extracted, the 16S rRNA gene amplified, and DNAs sequenced on a Roche 454 platform. Five statistical procedures--nonmetric multidimensional scaling, hierarchical cluster analysis, integral library shuffle, unique fraction method, and k-Nearest Neighbor--were used to compare differences or changes in bacterial communities. Multiple similar and diverse habitats were differentiated with both multivariate statistics and pairwise comparisons. Additionally, changes in communities were indicated over time, horizontal space, and depth. Multivariate statistics generally suggested similar relationships though not always consistent with pairwise comparisons, which showed analogous results though the unique fraction method always found fewer differences. k-Nearest Neighbor could be forensically useful based on the correct classification accuracy of `unknown' samples from a non-ideal training set. This research elucidates the potential of next-generation sequencing for soil investigation, how samples should be collected, and what statistics would be useful to analyze the data.
Read
