Fierer Lab

Exploring the structure and function of microbial communities

Pet Peeves

By: Noah Fierer

Papers using 16S rRNA gene sequencing to characterize bacterial communities are now a dime a dozen. Every day a new ‘microbiome’ is described using these methods and microbiologists like to view themselves as Indiana Jones hacking through the jungle of microbial diversity wielding sequencers for machetes.

Many of these papers are quite good and there is undoubtedly a lot of fascinating microbial natural history out there waiting to be discovered. But there are aspects of this literature that can be annoying.

At risk of being viewed as an old curmudgeon, here is my running list of pet peeves:

  1. Don’t brag about how many sequences you have. Really – no one cares. Just because you may have used a super-duper, next-millenium sequencing approach that gave you 5 terabases of 16S rRNA sequence data from a single sample, I’m not impressed. Today’s ‘ultra high-throughput’ sequencing platform is tomorrow’s mimeograph machine. Besides, there is a good chance you would have observed the same general patterns with only 100 sequences per sample.


  1. Remember that there were papers written before 2005 and many of them are quite relevant. Culture-dependent papers are still valuable – in fact, they are the best source of information on the actual physiology and ecology of those taxa you find in your sample. Although many taxa are clearly difficult to cultivate, there are thousands of described strains and many of these strains will be closely related to those taxa that are abundant in your samples. If you dig deep into the literature, you will probably find that a lot more is known about the microbes found in your samples than you might think. Anyone can report patterns in data – it takes a microbiologist armed with many decades of research to explain why those patterns might exist.


  1. Don’t bother reporting a Chao 1 estimate of microbial diversity. It is almost always just your number of observed taxa multiplied by two. More generally, there are important caveats and issues associated with trying to estimate the total number of microbial ‘species’ found in a given sample so this is often best avoided altogether (a topic for another blog post).


  1. Ordination plots are not statistical analyses. Showing me clouds of points does not tell me if samples have distinct microbial communities. Moreover, ordination plots are not always the most appropriate way to visualize the relevant patterns. Yes – they are easy to make and they can be helpful for exploring a dataset, but they do not always need to go into the resulting paper or presentation. Ordination plots are like Ambien for audiences.


  1. Steer clear of the term ‘beta diversity’ – it is confusing to many readers and it is a term that has numerous definitions making it confusing even to other ecologists. In most cases, when microbial ecologists use the term ‘beta diversity’ we really mean community composition (or differences in community composition) – so just say that.


  1. Watch out for contaminants – particularly if you are working with low biomass samples. Include (and sequence) negative controls at all steps in the process – this has always been a good idea – anyone who is familiar with PCR using universal or nearly-universal bacterial primers knows that it is very difficult to avoid contamination with bacterial DNA. I do not want to read another paper describing a soil bacterial community that is dominated by Staphylococcus, Streptococcus, Propionibacterium, and other skin bacteria.


  1. At some point in your manuscript, mention what bacteria you actually have in your samples and why they might be there. I review a surprising number of papers with few, if any, details on what bacteria were found in the samples. This is akin to describing plant communities and never mentioning what plants were actually found in the surveyed plots.


  1. Remember that many of the 16S rRNA genes in a given sample may come from dead (or at least dormant) organisms. DNA is remarkably persistent in the environment and on surfaces, so just because you amplified DNA from an organism does not mean that the organism was actually living in your sample. My guess is that this is particularly true for many studies focusing on the microbes of the built environment (e.g. this study).


I admit that I have been guilty in the past of some of these transgressions, but don’t hold that against me….