Monday, March 25 2019
3:00pm
Room 1232, U.A. Whitaker Building, 313 Ferst Dr NW, Atlanta, GA 30332
Add To My Calendar
Guilds Identification in a Temperate Freshwater Chronoseries using Dynamic Modeling and Network Mining of 460 Metagenome-assembled Genomes

Theme: Aquatic Microbial Systems
Miguel Rodriguez
Konstantinidis Lab

ABSTRACT
Guilds are a foundational concept in community ecology, complementing phylogeny, food webs, and niches on the understanding of community assembly, biogeography, succession, and metabolic theory, among others. However, guild definition and identification in microbial ecology has been traditionally relegated to industrially productive taxa and coarsely-defined forms of metabolism, notably primary producers and specific roles in biogeochemical cycles. Here we aim to systematically identify and evaluate the consistency of guilds in a microbial freshwater meta-community.

First, we present a chronoseries (69 metagenomes) from seven locations along the Chattahoochee River basin (Southeastern USA) and a novel iteratively subtractive binning methodology. Next, we use smoothed abundance profiles to infer directed interactions from Lotka-Volterra models, differentiating predation from mutualism, commensalism, amensalism, and competition. Finally, we apply hierarchical link clustering to the resulting network and select modules, equated to guilds, maximizing partition density.

Using this framework and the resulting high-quality genomes from 463 distinct species capturing ~50% of sampled communities, we were able to recover de novo guilds representing photoautotrophs, plant-degraders, and nitrogen fixers, as well as less expected groups such as phage-associated species. Using label permutation, we demonstrate that detected guild functional specialization is significantly higher than expected by chance (P < 0.002).
 
In this work, we present a novel methodology for microbial guild identification without metabolic assumptions, and showcase selected guilds and their overall conceptual consistency. We also introduce methodological innovations on metagenome binning, dynamic modeling and network mining, and provide a genome collection representing an unprecedentedly large fraction for a freshwater meta-community.