I'd say there are 2 fundamentally different questions here:

1. Do rare sequences = rare microbes?

2. Are the truly rare microbes always rare?

As for 1, there are a few answers here relating to this. The bogusphere has already been mentioned, and it is indeed part of any sequencing project and a large part of our databases as well. Specifically chimeras are all to common in database, and when I was still going manually through sequences I was able to find two different chimeric sequences in my own collection that had a 97-98% match to chimers in the database and yet where clearly chimeric. The problem of PCR biases is also recognized although all to often it is ignored (see for example the "next base effect" on PCR results). So a rare sequence in your collection may not be all that rare in the environment. Finally, rare sequences may be of bacteria that have drifted from other environments, or even from leftover DNA (very stable stuff, DNA) and so represent a bug that is not in effect active in the studied environment.

The other question has more to do with time. Assuming your rare bug is indeed rare, it may very well be that under more favorable conditions it will become very abundant, likely at the expense of some other bug that has now becomes rare. This relates more to the original hypothesis of the rare biosphere, and in dynamic environments such as wastewater treatment systems it is fairly easy to observe such transitions.

So I think the best answer to your question would be to stop sampling single time points and start looking at time series. A rare sequence that keeps coming up is at least likely to be real, while one that suddenly becomes un-rare is probably of ecological significance as well.

That is an evolving question that is slowly being answered by the thousands of researchers.

Those rare species are probably have weird metabolic pathways, utilize even weirder substrates, possibly co-metabolizing xenobiotics and other contaminants.

If you have the time to spend on doing the rare species then it will probably help future research but otherwise just save the DNA fragmnets somewhere and focus on understanding the most active organisms. On that note, trying to get a transcriptomic data will really push our understanding of each ecological niche.

Rare microbes may have specific and unique metabolism which could be usefull for ecosystems facing stresses (pollution, temperature variations...).

But the question of their utility is still not clear.

See for archeal studies : https://www.researchgate.net/profile/Didier_Debroas/publications/

Really interesting question about the "Rare biosphere" concept. Do organisms that appear as singleton or doubleton sequences in massive sequencing approaches to the microbial diversity of the weirdest sample you can think of functionally matter? I really do not know and will keep an eye in this post see if anyone shares a bit of light on it.

Something wonders me somehow and perhaps may lead us to an answer to this question. I'm in massive sequencing of microbial (archaea and bacteria) assemblages through 16S rRNA and functional genes. Many past papers using these techniques (454, illumina,...) focused a significant part of the discussion on the rare biosphere (sometimes more than 40% of the obtained sequences, occurred less than twice), now we tend to eliminate rare sequences (up to 10 sequences per OTU) on community analysis. Why?

As I said, a difficult to answer but really interesting question. The unexplored microbial diversity is vast and I agree with Pierre, a lot of unique metabolic activities are yet to be discovered. But, the other side of the coin is the relevance of these bacteria in ecology. To my opinion there’s a huge divergence in the scales we measure microbial diversity and environmental variables that forces us to consider the abundant microbes in ecology studies. We collect few grams of soil and get thousands of sequences, but, at the same time, we make a single measurement for most ecologically relevant variables (pH, conductivity, O2, nutrient composition, temperature,…). Do all the “species” we detect at low abundances in the soil sample thrive at the estimated pH? Are they relevant in the decrease of the measured ammonium (talking about nitrification as an example)? I don’t think so. I hope the rare microbes will push us to invest more time and technology in implementing micro-scale measurements of environmental variables in our studies.

Whether "rare species" are really rare in the environment is an open question, because the limitation of current techniques (especially what primers to use ) might bias toward the common species, and thus miss a lot of uncommon species which we assume as rare species. I think it is urgent to redesign new primer sets from the large pool of common and uncommon sequence data so that we will have a better idea whether the rare species are really rare.

If you are using next gen sequencing techniques, I'd be more worried them "rare biosphere" things being more of a "rare bogusphere". You have to apply very stringent filtering to your data before you can be sure that the "unprofilable microbes" are real and not a product of PCR-/sequencing- errors. The way bases are called in next gen sequencing, especially the new technology, means that we are not fully equipped with ways of eliminating all possible forms of error. I would suggest reading "Wrinkles in the rare biosphere" by Hugenholtz and co. before you decide to describe new rare organisms.

Here is another interesting paper that goes in the same direction Özgül was heading.

The authors show that dormany of organisms is an important aspect that has to be considered in the analysis of communities and function. Just because there is a lot of some organisms does not necessarily mean that they are active.

http://www.pnas.org/content/early/2010/03/08/0912765107

Is the contribution of microbes to ecological functions measured only on a metabolic level? Here we must consider also other mechanisms, through which the structure of microbial community affects its functioning. Metabolic contribution of rare microbes to community functioning might be minimal, but one must have in mind, that rare microbes comprise most of the diversity in a community and as such provide a heterogeneous gene pool which is available to other bacteria through horizontal gene transfer. E.g. a presence of a single antibiotic resistant bacteria can result in a wide spread resistance across different species in the community, exposed to sub-inhibitory levels of antibiotics. In this case a rare trait quickly becomes the prevalent under certain conditions and directs the dynamics of bacterial community. Therefore rare microbes provide a library of genes which might come in handy for the whole community.

I agree in that many of the "rare microbes" are just rare sequences, not identical to real organisms. http://csu-cvmbs.colostate.edu/Documents/Microbe-2007Dec.pdf

Sequence-based species identification is not straight-forward, and DNA detection does not distinguish between accidentally present mo and those that are part of an active community. We get a lot of information on mo ecology from DNA analyses, but not the taxa Identity. GeneBank is full of misidentifications.

I'd say there are 2 fundamentally different questions here:

1. Do rare sequences = rare microbes?

2. Are the truly rare microbes always rare?

As for 1, there are a few answers here relating to this. The bogusphere has already been mentioned, and it is indeed part of any sequencing project and a large part of our databases as well. Specifically chimeras are all to common in database, and when I was still going manually through sequences I was able to find two different chimeric sequences in my own collection that had a 97-98% match to chimers in the database and yet where clearly chimeric. The problem of PCR biases is also recognized although all to often it is ignored (see for example the "next base effect" on PCR results). So a rare sequence in your collection may not be all that rare in the environment. Finally, rare sequences may be of bacteria that have drifted from other environments, or even from leftover DNA (very stable stuff, DNA) and so represent a bug that is not in effect active in the studied environment.

The other question has more to do with time. Assuming your rare bug is indeed rare, it may very well be that under more favorable conditions it will become very abundant, likely at the expense of some other bug that has now becomes rare. This relates more to the original hypothesis of the rare biosphere, and in dynamic environments such as wastewater treatment systems it is fairly easy to observe such transitions.

So I think the best answer to your question would be to stop sampling single time points and start looking at time series. A rare sequence that keeps coming up is at least likely to be real, while one that suddenly becomes un-rare is probably of ecological significance as well.

Several responses noted that "rare" microbes can provide unique metabolic capabilities and may become common as the environment changes. There is a real-world example of that: the response of microbes during the Deepwater Horizon oil spill in the Gulf of Mexico - see: Hazen TC, Dubinsky EA, DeSantis TZ, Andersen GL, Piceno YM, et al. (2010) Deep-sea oil plume enriches indigenous oil-degrading bacteria. Science 330: 204–208. doi: 10.1126/science.1195979.

Presumably, the ecosystem will return to "normal" after this even when hydrocarbons are degraded. Thus, the "rare" biosphere has the potential to contribute to ecosystem resistance, resilience, and recovery - important ecosystem services.

Perhaps we should distinguish microbes that are rare at a site from those that occur at few sites. Microbes with few, perhaps dormant individuals at a site can proliferate if appropriate substrates or symbionts become available. They then become less 'rare' in this sense, as in the oil example cited by Parke Rublee. Many microbes are so good at persisting, dispersing and proliferating that most microbial processes can be simulated based on substrate-availability and other aspects of the environment rather than needing to invoke microbial population dynamics. However, microbes that are occur at few sites may be very important for certain processes - for instance, it is sometimes necessary to inoculate with an appropriate Rhizobium strain when growing soya at a new site. Most ecosystem 'functions' are dependent on the presence of particular species rather than on species-richness per se. So yes, rare microbes can be important.

I agree with Daniel, when he mentions the need of new primers. Indeed I believe that we are missing some species when we confine the search with primers directed to most common species. On the other hand, the importance of differentiating real new sequences from sequencing errors has also to be taken in account. In my opinion, the duplicates, triplicates or quadruplicates of experiments may be an asset in that quality control of the sequencing technique, in one hand, and, on the other, to discriminate if the singletons do appear in the repeted experiments. If so, I think they should not be excluded and are, indeed, important for the ecossystem they are inserted in. We all know the intrincate relations and importance of all organisms in a ecossystem.