B13H-04
Revealing and analyzing networks of environmental systems

Monday, 14 December 2015: 14:25
2004 (Moscone West)
Damien Eveillard, University of Nantes, Nantes, France
Abstract:
Understanding the interactions between microbial communities and their environment well enough to be able to predict diversity on the basis of physicochemical parameters is a fundamental pursuit of microbial ecology that still eludes us. However, modeling microbial communities is a complicated task, because (i) communities are complex, (ii) most are described qualitatively, and (iii) quantitative understanding of the way communities interacts with their surroundings remains incomplete. Within this seminar, we will illustrate two complementary approaches that aim to overcome these points in different manners. 

First, we will present a network analysis that focus on the biological carbon pump in the global ocean. The biological carbon pump is the process by which photosynthesis transforms CO2 to organic carbon sinking to the deep-ocean as particles where it is sequestered. While the intensity of the pump correlate to plankton community composition, the underlying ecosystem structure and interactions driving this process remain largely uncharacterized Here we use environmental and metagenomic data gathered during the Tara Oceans expedition to improve understanding of these drivers. We show that specific plankton communities correlate with carbon export and highlight unexpected and overlooked taxa such as Radiolaria, alveolate parasites and bacterial pathogens, as well as Synechococcus and their phages, as key players in the biological pump. Additionally, we show that the abundances of just a few bacterial and viral genes predict most of the global ocean carbon export’s variability. Together these findings help elucidate ecosystem drivers of the biological carbon pump and present a case study for scaling from genes-to-ecosystems.

Second, we will show preliminary results on a probabilistic modeling that predicts microbial community structure across observed physicochemical data, from a putative network and partial quantitative knowledge. This modeling shows that, despite distinct quantitative environmental perturbations, the constraints on community structure could remain stable.