Ecotype Simulation 2: An improved algorithm for efficiently demarcating microbial species from large sequence datasets

Background: Microbial systematists have used molecular cutoffs to classify the vast diversity present within a natural microbial community without invoking ecological theory. The use of ecological theory is needed to identify whether or not demarcated groups are the ecologically distinct, fundamental units (ecotypes), necessary for understanding the system. Ecotype Simulation, a Monte-Carlo approach to modeling the evolutionary dynamics of a microbial population based on the Stable Ecotype Model of microbial speciation, has proven useful for finding these fundamental units. For instance, predicted ecotypes of Synechococcus forming microbial mats in Yellowstone National Park hot springs, which were previously considered to be a single species based on phenotype, have been shown to be ecologically distinct, with specialization to different temperature and light levels. Unfortunately, development of high-throughput DNA sequencing methods has outpaced the ability of the program to analyze all of the sequence data produced. Results: We developed an improved version of the program called Ecotype Simulation 2, which can rapidly analyze alignments of very large sequence datasets. For instance, while the older version takes days to analyze 200 sequences, the new version can analyze 1.92x105 sequences in about six hours. The faster simulation identified similar ecotypes as found with the slower version, but from larger amounts of sequence data. Conclusions: Based on ecological theory, Ecotype Simulation 2 provides a much-needed approach that will help guide microbial ecologists and systematists to the natural, fundamental units of bacterial diversity.