Primer biases and implications
Most known primers used in amplifying microbial eukaryotic communities are those reported by Stoeck et al. (2010) and Bradley et al. (2016) but also suffer in their biases, which has implications in inferring ecological insights on microbial communities. We then generated new primer sets (Wolf 938 and Wolf 964) to complement the limitations of the existing ones and evaluated their efficiency in silico using known reference sequences, in vitro amplification using mock communities, and their applicability in situ using environmental samples.
Consistently, the different primer sets we tested (Wolf 938, Stoeck, Wolf 964) were not able to amplify or provide high resolving power in all target taxonomic groups, although all sets amplified some if not all abundant taxa we expected based on microscopy (Lalande et al., 2016; Nöthig et al., 2015). Relative to the three most dominant and significant phytoplankton groups in the Arctic, the Stoeck and Wolf938 primers seemed to be good in detecting all three groups, but differences were found at the sequence level. For example, while primers Stoeck and Wolf938 were able to show ecotype-level variations in chlorophytes, they did not exhibit sensitivity to the haptophytes. In contrast, Wolf964 demonstrated the alternating abundances between Emiliania andPhaeocystis but did not perform well in detecting the chlorophytes. These results suggest that although we might come close to generating a broader taxonomy-binding primer, a ‘universal one’ is still not available. Primer usage then will ultimately depend on the questions being asked and the groups being targeted, or a combination of primer sets in case of exploratory work. This is especially true for the marine environment, which possesses one of the highest microbial eukaryotic diversities (see de Vargas et al., 2015).
We further observed disparity between in silico and in situ tests. For example, Wolf938 did not perform well in silicoespecially in detecting diatoms but was able to amplify sequences and even revealed patterns between Chaetoceros phylotypes from environmental samples. This indicates that what works in silicomight not necessarily readily work in environmental samples and vice-versa. These could be due to efficiency of nucleic acid extraction, priming specificity, PCR biases, and sequencing-related issues, PCR inhibition due to the presence of humic and fulvic acids, preservation of the samples (Lever et al., 2015; Metfies et al., 2017), and competition reaction between known and unknown taxa. It is important to note however that limitations of currently used primers do not render previous and ongoing efforts in investigating eukaryotic microbes wrong or inaccurate. On the contrary, knowing such limitations provide new insights, perspectives, and contexts in analyzing and interpreting HTS-generated data. It is certain however that in some cases, these biases in primers could limit our understanding of natural ecosystems or even provide wrongful conclusions in interpreting data by missing out certain taxa if not considered during analysis.