Friday, August 04, 2006

3,290, 8,699, or more than 20,000 species of bacteria per liter of seawater?

Deep-sea News summarized the findings of a recently published survey of bacterial diversity around hydrothermal vents (Sogin et al. 2006):
  • 20,000…Number of species of bacteria found per 1L of seawater in a new study
  • 2,000…Number of species the researchers expected to find per liter
  • 550-4400…Depths in meters at which samples were taken
  • 5,000,000 to 10,000,000…The number of species likely to be contained in the Earth's oceans
  • 90…Percentage of biomass that bacteria comprises in the ocean
  • 1,200…Number of ocean samples the researchers want to extend the analysis to
The results are indeed astounding, partly because until recently there were only about 5,000 identified species of bacteria. Thanks to recent advances in technology, though, we now know that the 5,000 species number is far too low.

In the past, identifying a species of bacteria required culturing it in the lab and then analyzing its morphological (and molecular) characteristics. The problem with this technique, however, is that many bacteria don't grow under common lab conditions, and thus the vast majority of bacterial species went unidentified simply because we could not culture them.

However, that's all changing now. The current rage in microbiological taxonomy is to forget about trying to rear bacteria, and instead just take samples from the field, lyse the cells, PCR-amplify portions of the bacterial DNA (typically rRNA encoding sequences), and then sequence the DNA and compare it to known sequences. If researchers find a novel sequence, they know they've got a new species. Using these types of techniques, researchers have now estimated that we've identified (and cultured) less than 1% of all bacterial species.

One problem with most current molecular taxonomy techniques is that rare species' sequences are often lost (or not sequenced) during analysis. Sogin et al. (2006) used a new molecular technique that enabled them to sequence vastly more PCR amplification products (they ended up sequencing more than 115,000 PCR amplification products from only 14 samples), and thus they could identify even extremely rare bacteria that prior surveys would have overlooked. Here's their primary data table:

Table 2: Summary table from Sogin et al. (2006); each sample is 1L of water taken from a specifc point and depth. OTUs are "operational taxonomic units."
Sample ID Total reads Trimmed tags Unique tags OTUs
53R 6,505 5,000 2,656 1,184
55R 18,439 13,902 7,187 2,555
112R 12,916 9,282 5,752 2,135
115R 14,731 11,005 5,777 2,049
137 18,137 13,907 6,752 2,480
138 18,451 14,374 7,168 2,550
FS312 6,605 4,835 2,769 1,362
FS396 22,994 17,666 8,699 3,290

The introduction of the paper reports that operational taxonomic units (OTUs) "correspond to species or kinds of organisms." Thus, based on this table it looks like the maximum number of species of bacteria Sogin et al. found in any given sample was 3,290. That's far fewer than the 20,000 species per liter of water that's being reported elsewhere (e.g., Deep-sea News, BBC, this press release).

Now, I'm no molecular biologist, but it appears to me that while there were indeed more than 20,000 "total reads" in one sample, this number is just the total number of PCR ampilfication products that were obtained from the sample. Many of these products appear to have been duplicates of each other, as the very same sample reported only 8,699 "unique tags." This implies that while Sogin et al. may have gotten 20,000 PCR amplification products from one liter of water, they actually observed far fewer species.

Thus, I will close with a question for my molecular-savvy readers: am I misunderstanding this, or are the reports in the press entirely inaccurate?

[Update 8:30 pm: The mistake is mine; Dr. Sogin has replied to the author of Deep-sea News via e-mail, and has confirmed that the 20,000 species-per-liter figure is accurate. While the authors didn't actually obtain sequences from 20,000 species in any given liter of water, they were able to estimate that there were 20,000 species-per-liter present.]

Sogin ML, HG Morrison, JA Huber, DM Welch, SM Huse, PR Neal, JM Arrieta, and GJ Herndl. 2006. Microbial diversity in the deep sea and the underexplored "rare biosphere" PNAS 103:32 12115–12120. (abstract, full-text PDF)

No comments: