Assessment of microbial diversity bias associated with soil heterogeneity and sequencing resolution in pyrosequencing analyses.

It is important to estimate the true microbial diversities accurately for a comparative microbial diversity analysis among various ecological settings in ecological models. Despite drastically increasing amounts of 16S rRNA gene targeting pyrosequencing data, sampling and data interpretation for comparative analysis have not yet been standardized. For more accurate bacterial diversity analyses, the influences of soil heterogeneity and sequence resolution on bacterial diversity estimates were investigated using pyrosequencing data of oak and pine forest soils with focus on the bacterial 16SrRNA gene. Soil bacterial community sets were phylogenetically clustered into two separate groups by forest type. Rarefaction curves showed that bacterial communities sequenced from the DNA mixtures and the DNAs of the soil mixtures had midsize richness compared with other samples. Richness and diversity estimates were highly variable depending on the sequence read numbers. Bacterial richness estimates (ACE, Chao 1 and Jack) of the forest soils had positive linear relationships with the sequence read number. Bacterial diversity estimates (NPShannon, Shannon and the inverse Simpson) of the forest soils were also positively correlated with the sequence read number. One-way ANOVA shows that sequence resolution significantly affected the α-diversity indices (P<0.05), but the soil heterogeneity did not (P>0.05). For an unbiased evaluation, richness and diversity estimates should be calculated and compared from subsets of the same size.