The Secret Hidden in Dust: Uncovering the Potential to Use Metabarcoding of Dust for Provenance Determination, and the Capacity to Utilise Existing Soil Reference Databases

The ubiquitous nature of dust, along with localised chemical and biological signatures, makes it an ideal medium for provenance determination in a forensic context. Metabarcoding of dust can yield biological communities unique to the site of interest, similarly, geochemical and mineralogical analyses can uncover elements and minerals within dust than can be matched to a geographic location. Combining these analyses presents multiple lines of evidence as to the origin of collected dust samples. In this work, we investigated whether the time an item spent at a site collecting dust influenced the ability to assign provenance. We then integrated dust metabarcoding of bacterial and fungal communities into a framework amenable to forensic casework, (i.e., using calibrated log-likelihood ratios to predict the origin of dust samples) and assessed whether current soil metabarcoding databases could be utilised to predict dust origin. Furthermore, we tested whether both metabarcoding and geochemical/mineralogical analyses could be conducted on a single sample for situations where sampling is limited. We found both analyses could generate results capable of separating sites from a single swabbed sample and that the duration of time to accumulate dust did not impact site separation. We did find significant variation within sites at different sampling time periods, showing that bacterial and fungal community profiles vary over time and space – but not to the extent that they are non-discriminatory. We successfully modelled soil and dust samples for both bacterial and fungal diversity, developing calibrated log-likelihood ratio plots and used these to predict provenance for dust samples. We found that the temporal variation in community composition influenced our ability to predict dust provenance and recommend reference samples be collected as close to the sampling time as possible. Thus, our framework showed soil metabarcoding databases are capable of being used to predict dust provenance but the temporal variation in metabarcoded communities will need to be addressed to improve provenance estimates.