Bacterial and Eukaryotic Small-Subunit Amplicon Data Do Not Provide a Quantitative Picture of Microbial Communities, but They Are Reliable in the Context of Ecological Interpretations

Abstract

High-throughput sequencing (HTS) of gene amplicons is a preferred method of assessing microbial community composition, because it rapidly provides information from a large number of samples at high taxonomic resolution and low costs. However, mock community studies show that HTS data poorly reflect the actual relative abundances of individual phylotypes, casting doubt on the reliability of subsequent statistical analysis and data interpretation. We investigated how accurately HTS data reflect the variability of bacterial and eukaryotic community composition and their relationship with environmental factors in natural samples. For this, we compared results of HTS from three independent aquatic time series (n ϭ 883) with those from an established, quantitative microscopic method (catalyzed reporter deposition-fluorescence in situ hybridization [CARD-FISH]). Relative abundances obtained by CARD-FISH and HTS disagreed for most bacterial and eukaryotic phylotypes. Nevertheless, the two methods identified the same environmental drivers to shape bacterial and eukaryotic communities. Our results show that amplicon data do provide reliable information for their ecological interpretations. Yet, when studying specific phylogenetic groups, it is advisable to combine HTS with quantification using microscopy and/or the addition of internal standards. IMPORTANCE High-throughput sequencing (HTS) of amplified fragments of rRNA genes provides unprecedented insight into the diversity of prokaryotic and eukaryotic microorganisms. Unfortunately, HTS data are prone to quantitative biases, which may lead to an erroneous picture of microbial community composition and thwart efforts to advance its understanding. These concerns motivated us to investigate how accurately HTS data characterize the variability of microbial communities, the relative abundances of specific phylotypes, and their relationships with environmental factors in comparison to an established microscopy-based method. We compared results obtained by HTS and catalyzed reporter deposition-fluorescence in situ hybridization (CARD-FISH) from three independent aquatic time series for both prokaryotic and eukaryotic microorganisms (almost 900 data points, the largest obtained with both methods so far). HTS and CARD-FISH data disagree with regard to relative abundances of bacterial and eukaryotic phylotypes but identify similar environmental drivers shaping bacterial and eukaryotic communities.

Quantifying the information content of lake microbiomes using a machine learning-based framework

“…Applying our framework to a microbial ecology dataset collected in a large-scale survey of European lakes, the results suggest that lower levels of microbial taxonomy share a higher level of information with lake parameters (figures 1 and 2B). This supports the hypothesis that OTUs from the same species can react to environmental change differently and might, therefore, inhabit different ecological niches [44,47,[84][85][86]. Furthermore, our results show that, for most physico-chemical parameters, higher levels of taxonomy do not contain information not already present on the OTU level (see figure 3B), which is in contrast to the findings of others [87].…”

Microdiversity and phylogeographic diversification of bacterioplankton in pelagic freshwater systems revealed through long-read amplicon sequencing

“…S4). While our approach can efficiently resolve microdiversity within individual lineages, it is not designed to reveal overall microbial community composition, for which conventional short-read amplicon sequencing or a direct quantification via fluorescence in situ hybridization is still a reasonable approach [80].…”

dadasnake, a Snakemake implementation of DADA2 to process amplicon sequencing data for microbial ecology

“…A second limitation is that relative abundances of ASVs are not reflective of the actual abundance of the sequenced taxa, which varied for the prokaryotic mock community and were equal in the fungal mock community (Figure 3 c&d). There are numerous reasons for misrepresentation of abundances by PCR-based analyses [8,42]. Of note, the variation in the relative abundance estimates is observed to be highest at low sequencing depths (Figure 3 c&d).…”

“…Computational methods have been refined in the recent years, especially with the shift to exact sequencing variants and better use of sequence quality data [2,3]. While amplicon sequencing can have severe limitations, such as limited and uneven taxonomic resolution [4,5], over-and underestimation of diversity [6,7], lack of quantitative value [8,9] and missing functional information, amplicon sequencing is still considered the method of choice to gain an overview of microbial diversity in a large number of samples [10,11]. Consequently, the sizes of typical amplicon sequencing datasets have grown.…”