Weitz Group @ Georgia Tech Theoretical Ecology and Quantitative Biology



New paper on robust estimation of microbial diversity in the Top Ten at ISME J's most downloaded

Posted by jsweitz

The inspiration for our new manuscript in ISME J lies, in part, in a recent claim within Mora et al's interesting study of diversity on Earth in which they stated that that there are at least 10,100 number of prokaryotic species. As Jonathan Eisen noted on his blog, this statement may be true, but it's not particularly helpful. The first point of our work is to demonstrate rigorously why such statements are both true and unhelpful. The reason is that a lower estimate of species richness may have no correspondence to actual values since the distribution of individuals observed in a sample is insensitive to whether or not there are many rare species in the community.

Moreover, we then show that one cannot even compare such statements about species richness. For example, imagine two studies of microbial species richness led by Alice and Bob - in which estimated species richness is inferred from a relatively small sample of the community. Alice estimates that there are least 5K microbial species in environment A and Bob estimates that there are at least 10K microbial species in environment B. As we have shown, the true value of species may be much bigger in A than in B! Hence the rank-order of lower estimates of species richness need not correspond to the true order of species richness.

Finally, we then show that Shannon/Simpson diversity of those environments can be estimated from the sample and that the estimates are robust, i.e., can be compared. We did this by constructing lower AND upper bounds for these diversity metrics (and for all Hill diversities) and showing the range between the bounds for these two metrics is small. The reason why the range is small is that Shannon/Simpson don't depend sensitively on the abundance of very rare species, due to the weighting of more abundant species in their notions of diversity.

The paper is one of two papers stemming from Weitz group work and collaborations that are listed in the Top Ten of all downloaded papers at ISME J as of late February, 2013.