The main broad conclusion of this work (“relic DNA contributes minimally to the characterization of microbial community structure”) is not well-supported by the data presented.
First, the efficiency of relic DNA removal with the DNAse-based approach used in this manuscript does not appear to have been tested empirically. That is, there were no experiments reported to conclusively validate whether the DNase treatment effectively removed extracellular DNA or DNA from dead cells. The lack of such experimental controls makes it difficult to interpret their results, as there is extensive literature showing that DNA bound to soil/sediment components (minerals or organic matter) can be highly resistant to DNAse digestion (e.g. Romanowski, et al., 1991, Paget, et al.,1992, Cai, et al., 2006, Khanna and Stotzky, 1992, Lorenz and Wackernagel 1987, Nielsen et al., 2006). If DNAse treatment is not effectively removing relic DNA pools, it would reduce the apparent amounts of relic DNA and reduce the apparent influence of relic DNA on estimates of community structure.
Second, the authors analyzed 6 unique samples from each ecosystem type (e.g. 6 unique soil samples and 6 unique lake water samples, etc.). They then assessed whether there were consistent increases or decreases in richness across all 6 samples from a given ecosystem. The problem with this approach is that this effectively obscures any effects that removal of relic DNA might have on estimated richness within individual samples. This is important to account for as we know from our work (Carini et al. 2016. Nature Micro.) that in 20% of our soil samples, removal of relic DNA had no significant effect on richness estimates. Thus, by running the analyses across all 6 samples combined, instead of quantifying relic DNA effects on a per-sample basis, it likely obscures any effects of relic DNA. This is apparent from their Figure 3, where there appears to be a subset of samples where removal of relic DNA reduced richness as the mean richness ratio is >1 for three of the sample types. Thus, instead of concluding that “relic DNA contributes minimally to estimates of microbial diversity”, it would be more accurate if the authors had concluded that the effects of relic DNA are variable and relic DNA does not always introduce biases.
Just to be clear: we do not advocate that relic DNA is always important to consider when conducting DNA-based analyses of environmental samples. As we detailed previously (Carini et al. 2016), relic DNA is unlikely to obscure the ability to say that two distinct communities are indeed distinct with respect to both richness and community composition. Likewise, relic DNA effects are not going to be equally important across all samples from a given ecosystem type. Depending on the temporal variability, community turnover rates and the residence time of relic DNA, failure to remove relic DNA could have no effects or it could introduce significant biases. We still maintain that there are common situations where failure to remove relic DNA could obscure patterns in community structure (e.g. trying to detect temporal changes in communities or the short-term effects of environmental perturbations on microbial communities). While removal of relic DNA can complicate laboratory analyses and the presence of relic DNA can alter how we interpret our estimates of microbial community structure, the data presented here do not provide enough evidence to justify ignoring the potential importance of relic DNA.