Speaker
Description
The past decades of experimental evolution in controlled laboratory conditions have established that microbial evolutionary dynamics are largely complicated by clonal interference and hitchhiking, making the process of adaptation hard to model and predict. Harnessing evolutionary knowledge for applied purposes is further complicated by how the various relevant microbial populations in health and industry differ greatly in their evolutionary parameters (e.g. population size, mutation rate, and spatial structure), thus calling for empirical observation and monitoring of the specific biological systems of interest. Bioethanol production in Brazil relies on the fermentation of sugarcane products by the yeast Saccharomyces cerevisiae, a process carried out in large open fermentors that hold population sizes as large as 10^17 individuals. Starter strains are first grown from small aliquots by propagation companies, and are then used to inoculate the process at the start of the season, which runs continuously for eight months every year. Productivity varies through the season, and so does fermentation-relevant metrics (e.g. flocculation, foaming, amount of bacteria). While previous studies using low-resolution karyotyping have demonstrated the occasional substitution of starter strains by wild strains, it is unclear how prevalent that is, or how it depends on the specific starter strains used. We also do not know the relevance of selection on novel mutations in the timescale of an industrial season. Here, we have collected samples through two seasons across two bioethanol production sites, and used whole-population and single-clone whole-genome sequencing to infer a comprehensive picture of the evolutionary dynamics of lineages during the course of these four independent time-series. We find that as the industrial season progresses, starter and invading strains compete and change the population composition in a way that varies year to year, and site to site, with some starter strains being more often substituted than others. We also find little connection between the observed lineage dynamics and the changes in productivity throughout the season. Despite the clear signature of strong selection from the strain-level dynamics, we observe no signs of selection for novel mutations, indicating that clonal interference is a major barrier for new segregating mutations in these very large populations. Our study sheds light on the specific evolutionary dynamics relevant to the bioethanol production process, as well as illustrates the potential of evolutionary monitoring of industrial populations for guiding production choices.
