Bacterial diversity assessment in Antarctic terrestrial and aquatic microbial mats: a comparison between bidirectional pyrosequencing and cultivation

Abstract

The application of high-throughput sequencing of the 16S rRNA gene has increased the size of microbial diversity datasets by several orders of magnitude, providing improved access to the rare biosphere compared with cultivation-based approaches and more established cultivation-independent techniques. By contrast, cultivation-based approaches allow the retrieval of both common and uncommon bacteria that can grow in the conditions used and provide access to strains for biotechnological applications. We performed bidirectional pyrosequencing of the bacterial 16S rRNA gene diversity in two terrestrial and seven aquatic Antarctic microbial mat samples previously studied by heterotrophic cultivation. While, not unexpectedly, 77.5% of genera recovered by pyrosequencing were not among the isolates, 25.6% of the genera picked up by cultivation were not detected by pyrosequencing. To allow comparison between both techniques, we focused on the five phyla (Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes and Deinococcus-Thermus) recovered by heterotrophic cultivation. Four of these phyla were among the most abundantly recovered by pyrosequencing. Strikingly, there was relatively little overlap between cultivation and the forward and reverse pyrosequencing-based datasets at the genus (17.1-22.2%) and OTU (3.5-3.6%) level (defined on a 97% similarity cut-off level). Comparison of the V1-V2 and V3-V2 datasets of the 16S rRNA gene revealed remarkable differences in number of OTUs and genera recovered. The forward dataset missed 33% of the genera from the reverse dataset despite comprising 50% more OTUs, while the reverse dataset did not contain 40% of the genera of the forward dataset. Similar observations were evident when comparing the forward and reverse cultivation datasets. Our results indicate that the region under consideration can have a large impact on perceived diversity, and should be considered when comparing different datasets. Finally, a high number of OTUs could not be classified using the RDP reference database, suggesting the presence of a large amount of novel diversity.

Figure 1. Overview of the distribution of the phyla per sample for the forward sequencing dataset.

Circle area is a log₂ transformation of the number of sequences ([log₂(N)*5/PI], with N the number of sequences in that phylum). Color intensity reflects the number of OTUs per phylum (total OTUs/total sequences), with a darker hue indicating a higher relative richness. The first two columns show the total number of sequences and diversity of each phylum for pyrosequencing and cultivation separately. The phyla are ordered according to decreasing total number of sequences. The yellow to red scale shows pyrosequencing data, the blue-purple scale the cultivation data.

Figure 2. Overview of the distribution of the phyla per sample for the reverse sequencing dataset.

Circle area is a log₂ transformation of the number of sequences ([log₂(N)*5/PI], with N the number of sequences in that phylum). Color intensity is an approximation for the number of OTUs per sequence (total OTUs/total sequences). The first two columns show the total number of sequences and diversity of each phylum for pyrosequencing and cultivation separately. The order of the phyla is as in Figure 1 and additional phyla were added at the bottom. The yellow to red scale shows pyrosequencing data, the blue-purple scale the cultivation data.

Figure 3. Bar chart illustrating the number of OTUs picked up from one or more samples for the forward dataset.

The number of OTUs is log₂ transformed. Blue bars, total sequences (pyrosequences plus cultivated sequences); red bars, pyrosequences only; green bars, cultivation sequences only.

Figure 4. Rank-abundance plot showing the distribution of genera in a sample, illustrating the difference between techniques.

Sequence numbers are plotted on a log scale. Blue bars are pyrosequencing based, red bars are cultivation based.