Bias in bacterial diversity as a result of Nycodenz extraction from bulk soil

Nycodenz density centrifugation (NDC) is an isolation method that allows extraction of both culturable and unculturable bacterial cells from soil, to be used in further downstream analysis; however, to date there has been a lack of information concerning the efficiency of this method. The aim of this study was therefore to investigate the overall efficiency of NDC extractions from soil and to identify sampling bias, if any. Bacterial cells were extracted from three soil plots from the Danish CRUCIAL field trial using an already established NDC protocol. To evaluate all aspects of the NDC procedure, DNA was extracted directly from soil, from NDC-extracted cells, and from the soil pellets left after NDC. Bacterial diversity was assessed by PCR amplification of the V4-V6 regions of the 16S rRNA from the extracted DNA followed by sample-tagged amplicon-pyrosequencing using the 454 Genome Sequencer FLX system. Sequences were processed and analyzed using the Ribosomal Database Project's (RDP) Pyrosequencing Pipeline tools. In this study, we show that extraction of bacteria from soil using NDC can result in significant biases in the form of either over- or underrepresentation of specific bacterial phyla commonly found in soil. Furthermore, rarefaction analysis, analysis of similarity, multidimensional scaling plots and analysis of variance showed that the diversity in the NDC-extracted sample was reduced significantly compared to both the original soil sample and the remaining NDC-pellet. To further study the soil diversity a mathematical model was employed to estimate how many sequences would be required in order to find 95% of all operational taxonomic units (OTUs) in the soil. The model estimated that the soil contains approximately 29,400 OTUs and that just 351,500 sequences are needed to cover 95% of the bacterial biodiversity, the equivalent of one full standard GS FLX run. (C) 2011 Elsevier Ltd. All rights reserved.