Nematodes play an important role in ecosystem processes, yet the relevance of nematode species diversity to ecosystem ecology is still an enigma.  Because nematode identification of all individuals at the species level using standard techniques is difficult, laborious, and extremely time consuming, the characterization of nematode communities continues to be resolved at higher than the species level leaving ecological analysis partially ambiguous or superficial.  Novel cloning-independent pyrosequencing may offer a potentially rapid tool to inventory nematode fauna at previously unparalleled levels of resolution at faster speeds and lower cost. The main objective of our study was to assess the suitability of massively parallel sequencing using GS FLX technology for nematode species identification from metagenomic samples. Two ~ 400 bp fragments of rRNA loci flanked by “universal” primer pairs (NF1 - 1573R 18S and D3A - D3B 28S) were used as barcoding regions. We hand-picked 44 known nematode species in known frequencies to set up 4 artificial metagenomic samples. Two metagenomic samples consisted of DNA amplified via multiplex PCR reactions (all nematode species together: 18S mPCR  and 28S mPCR). Two additional samples came from pooling PCR products from single nematode species PCR reactions (18S sPCR and 28S sPCR).  All metagenomic samples were A-Amplicon sequenced on GS FLX and run separately on an 8-chambered plate.
Results/Conclusions The total number of reads ranged from 4159 to 14771 per sample. Out of all reads, ~81% comprised reads of >199 bp (minimum length for species ID) and within those reads ~88% were identified as matching our referenced 44 species. While ~52% of reads gave 100% identity match, about 30% of reads varied by 1-2 bp, and the rest of the reads varied by >3 bp. Although neither barcode recovered all nematode species, 18S resulted in higher species recovery (~90%) than 28S barcode (~80%) and the use of both barcodes improved the detection level of nematode species. Of the >199 bp reads that were unmatched to our nematode database (~12%), the majority (~9%) were identified as fungal and ~3% were identified as nematode contaminants from previous sequencing projects in our lab. The frequency distribution of reads did not mirror the frequency distribution of nematode species.
Overall, results strongly support the suitability of massively parallel sequencing technology for identification of all nematode individuals from environmental samples. At this point, however, the use of the distribution reads for inferring the relative abundances of species within a nematode community is premature.