Comparing the Influence of Assembly Processes Governing Bacterial Community Succession Based on DNA and RNA Data

Abstract

Quantifying which assembly processes structure microbiomes can assist prediction, manipulation, and engineering of community outcomes. However, the relative importance of these processes might depend on whether DNA or RNA are used, as they differ in stability. We hypothesized that. RNA-inferred community responses to (a)biotic fluctuations are faster than those inferred by DNA; the relative influence of variable selection is stronger in RNA-inferred communities (environmental factors are spatiotemporally heterogeneous), whereas homogeneous selection largely influences DNA-inferred communities (environmental filters are constant). To test these hypotheses, we characterized soil bacterial communities by sequencing both 16S rRNA amplicons from the extracted DNA and RNA transcripts across distinct stages of soil primary succession and quantified the relative influence of each assembly process using ecological null model analysis. Our results revealed that variations in α-diversity and temporal turnover were higher in RNA- than in DNA-inferred communities across successional stages, albeit there was a similar community composition; in line with our hypotheses, the assembly of RNA-inferred community was more closely associated with environmental variability (variable selection) than using the standard DNA-based approach, which was largely influenced by homogeneous selection. This study illustrates the need for benchmarking approaches to properly elucidate how community assembly processes structure microbial communities.

Keywords: 16S rRNA gene; community turnover; dispersal; drift; ecological modeling; selection.

Figure 1

Workflow applied in this study. Procedures for analyzing RNA samples are shown in red, procedures for analyzing DNA samples are shown in blue, and common procedures are shown in black.

Figure 2

β-diversity patterns of bacterial communities displaying differences according to soil successional stages and sequencing approach (i.e., DNA- or RNA-based). Weighted (a,c,e) and unweighted (b,d,f) UniFrac distances were used to calculate β-diversity, which was visualized using principal coordinate analysis (PCoA). PCoA results are illustrated in different ordinate axes. Colors represent successional stages (i.e., 0, 10, 40, 70 and 110 years of succession), and different shapes represent RNA- and DNA-based approaches. The percentages in the axes show the variation of species composition explained by each ordinate.

Figure 3

Boxplots displaying the temporal variation of RNA- and DNA-inferred communities along the soil successional stages. The temporal variation was based on weighted UniFrac distances of communities across distinct sampling time points. *** indicates p < 0.001 in Wilcoxon signed-rank test.

Figure 4

Line plots displaying the dynamic changes in relative abundances of bacterial phyla (mean relative abundance > 0.1%) based on DNA- and RNA-inferred communities. The x-axis displays successional stages in years (i.e., 0, 10, 40, 70 and 110 years of succession), and the y-axis displays the phyla relative abundance (in % of the total abundance). Points indicate average, and error bars represent standard errors from the averages.

Figure 5

Pie charts and stacked-bar plots displaying the relative influences of assembly processes governing community turnover. (a)The pie charts show the relative influence of distinct assembly processes determining the spatiotemporal variation of RNA- and DNA-inferred communities. (b) Stacked-bar plots display the relative influences of distinct assembly processes structuring the temporal variation of bacterial communities in each successional stage based on both RNA- and DNA-based approaches. * indicates the influence of homogenizing dispersal and undominated processes for the turnover of RNA-inferred communities to be 0.62% and 0.06%, respectively. ** indicates the influence of homogenizing dispersal and undominated processes for the turnover of DNA-inferred community to be 0.68% and 0.45%, respectively.