Logo Logo
Switch Language to German
Morinière, Jerome; de Araujo, Bruno Cancian; Lam, Athena Wai; Hausmann, Axel; Balke, Michael; Schmidt, Stefan; Hendrich, Lars; Doczkal, Dieter; Fartmann, Berthold; Arvidsson, Samuel; Haszprunar, Gerhard (2016): Species Identification in Malaise Trap Samples by DNA Barcoding Based on NGS Technologies and a Scoring Matrix.
In: PLOS ONE 11(5), e0155497


The German Barcoding initiatives BFB and GBOL have generated a reference library of more than 16,000 metazoan species, which is now ready for applications concerning next generation molecular biodiversity assessments. To streamline the barcoding process, we have developed a meta-barcoding pipeline: We pre-sorted a single malaise trap sample (obtained during one week in August 2014, southern Germany) into 12 arthropod orders and extracted DNA from pooled individuals of each order separately, in order to facilitate DNA extraction and avoid time consuming single specimen selection. Aliquots of each ordinal-level DNA extract were combined to roughly simulate a DNA extract from a non-sorted malaise sample. Each DNA extract was amplified using four primer sets targeting the CO1-5' fragment. The resulting PCR products (150-400bp) were sequenced separately on an Illumina Mi-SEQ platform, resulting in 1.5 million sequences and 5,500 clusters (coverage >10;CD-HIT-EST, 98%). Using a total of 120,000 DNA barcodes of identified, Central European Hymenoptera, Coleoptera, Diptera, and Lepidoptera downloaded from BOLD we established a reference sequence database for a local CUSTOM BLAST. This allowed us to identify 529 Barcode Index Numbers (BINs) from our sequence clusters derived from pooled Malaise trap samples. We introduce a scoring matrix based on the sequence match percentages of each amplicon in order to gain plausibility for each detected BIN, leading to 390 high score BINs in the sorted samples;whereas 268 of these high score BINs (69%) could be identified in the combined sample. The results indicate that a time consuming pre-sorting process will yield approximately 30% more high score BINs compared to the nonsorted sample in our case. These promising results indicate that a fast, efficient and reliable analysis of next generation data from malaise trap samples can be achieved using this pipeline.