Tuesday, 17 December 2013

Evidence for a persistent microbial seed bank throughout the global ocean

This paper attempts to answer the question, “Do bacterial taxa demonstrate clear endemism?”, are they restricted to specific geographical ranges to the extent that macroorganisms are, or can one sample site contain the total phylogenetic diversity of the world’s oceans.
The microbial community within an environment at any given time is determined by current environmental and biological factors, so it changes across space and time as niches open and close. Different species bloom and then decrease in certain conditions, but they must all be present at low levels for these blooms to occur, in what is known as the “microbial seed bank”.
For this study, across a 6 year period, around 10 million 16S rRNA sequences were recorded at a deep water sampling site in the West English Channel, known as the L4-DeepSeq dataset. Their first finding was that all operational taxanomic units (OTUs) identified throughout the 6 years were present at any time point within the 6 years. All taxa were present at all times, but their abundance varied dramatically. These results contrast the widely accepted model that it is presence/absence of taxa that controls community structure, suggesting instead that all taxa persist and it is their abundances that control the community composition.
From this they suggest that sufficient sequencing would allow the testing of the null hypothesis “all bacteria are found in any particular environment because of an immense and persistent microbial seed bank”, this would be falsified by evidence of clear endemism, the lack of certain taxa in some environments.
In order to begin investigating this, they compared the L4-DeepSeq dataset with the global International Census of Marine Microbes (ICoMM), which contains 356 datasets of bacterial 16S rRNA sequences ranging from pelagic and sediment samples to mangrove and sponge environments.
They found that the L4-DeepSeq dataset overlapped significantly with ICoMM datasets, containing 31.7-66.2% of taxa in any given ICoMM biome. They also discovered the relationship that the greater the sequencing depth in L4 samples, the greater their overlap with global ICoMM data. Sequencing depth here does not mean depth in the ocean but the number of sequences within a sample. By extrapolating the relationship, they suggested that 100% overlap between L4-DeepSeq data and ICoMM global data would be achieved when a 16S rRNA sequencing depth of 1.93x1011 reads was reached. Achieving this would require over 200L of seawater to be filtered, rather than the 2L per sample used in this study, in order to collect enough cells to allow a sequencing depth that great. It would also cost an estimated $1.34million in sequencing costs. Regardless of the feasibility of such a study, it seems incredibly implausible that 100% overlap would ever truly be found, I think it far more likely that the relationship between sampling depth and phylgenetic overlap would at some point ‘level off’, where increased sequencing effort would cease to produce additional operational taxomic units. Though it seems the opinion of the authors that every taxa will be able to be found in every environment, if the sequencing depth is great enough.
The main value of this paper is not so much that it proves the existence of microbial endemism, as such a thing seems obvious given the existence of highly specialised extremophiles. It is valuable because it shows that an arbitrary sample of a marine microbial community overlaps on average 44% with other samples from varying marine environments around the world, with the suggestion that increased detection resolutions may show even greater overlap. So rather than 100% of taxa being present in all environments, I think they provide evidence for a global seedbank for a large number of cosmopolitan taxa.
One weakness of the study in my opinion is that comparing one sample point in Britain against many sample points from around the world, and generating the overlap percentages is not the best method for investigating global taxa overlap of communities. I think that comparing each sample point in the ICoMM to every other point in the ICoMM would give a higher average overlap, rather than using L4-DeepSeq as the central reference point. The average overlap between all samples is likely greater than the average overlap between L4-DeepSeq with each other sample.
The use of extreme habitats such as hydrothermal vents, cold seeps and estuaries likely pulled the average down due to their large number of specialised and endemic taxa. Excluding them from the dataset may produce stronger evidence for a global microbial seedbank for the general ocean.

Gibbons, S. M., Caporaso, J. G., Pirrung, M., Field, D., Knight, R., & Gilbert, J. A. (2013). Evidence for a persistent microbial seed bank throughout the global ocean. Proceedings of the National Academy of Sciences, 110(12), 4651-4655.

1 comment:

  1. Hi Dave,

    I think the authors' assumption is an interesting point of view for the global distribution of microbial taxa, however, I agree with you that is rather difficult to draw conclusions when they only compare this one site from the British Channel with some other random environments. I think more sampling effort from more sites and much more analyses would be needed to confirm this trend. Moreover I agree with you that 100% would be very unlikely even if the sampling effort would be tremendous; even though the marine environment is a dynamic habitat and mixing occurs constantly, I find it impossible to have the same taxonomic group in every environment. Although many microbes can be tolerant to many environmental conditions, growth and survival of microbes is often limited to some extent. For me it is hard to imagine that all taxa are present in every environment (e.g. poles vs. tropics?), yet I find that the idea of a persistant microbial seed bank is an interesting thought and it would be worth to establish more comparisons with (as the authors suggested) greater sampling effort!
    When the authors compared their samples from the British Channel to the ones from the other habitats - do you know over how many years the databanks they used from the other samples have been built? Even though it is rather unlikely, but if the databanks they used were sampled for a longer or shorter period of time, it is important to consider natural events that might have occurred during this time and could have affected the distribution of some groups. I'm not sure how much that would matter for this research though.