Comparative study on diatom morphology and molecular identification in drowning cases.

In the field of criminal investigations, in the event that a body is found in water, the ability to differentiate whether the cause of death was drowning or the body was murdered then dumped into water elsewhere is difficult but important for case detection. Detecting diatoms in human organs can be used to effectively identify if the cause of death was drowning. At present, diatom detection methods are roughly divided into morphological and molecular detection methods, but both methods have different limitations. In this study, a total of 79 samples from 23 victims in 19 known drowning deaths were collected. The diatom morphological identification combined with DNA metabarcoding technology was used to compare the reliability of the diatom detection method. Microscopic observations revealed that the positive detection rate of diatoms was 52.6 %, 26.3 % and 58.8 % respectively in the kidney, liver and lung samples. DNA metabarcoding analysis found that the positive detection rate of diatoms was 31.6 %, 31.6 % and 35.3 % respectively in kidney, liver and lung samples. When compared with barcode BacirbcL, barcode 18S605 detected more diatoms, while diatoms in BacirbcL were more consistent with environmental samples. The comparative analysis found that microscopic observations were not highly correlated with the identification results of DNA barcoding technology. There were no obvious differences in the effect of internal organs on diatom enrichment, and different organs should be tested at the same time. At present, the DNA barcode reference sequence is gravely insufficient and has many errors, which leads to restrictions in the application of this technology, resulting in many OTU not being accurately identified. This explains why the success rate of molecular identification is not higher than that of microscopic identification. Construction of a reliable diatom DNA barcode reference sequence database is an urgent task for drowning forensics.