Tuesday, November 8, 2016 Thesis Advisor Dr. Kostas Konstantinidis School of Civil and Environmental Engineering Committee Members Dr. Mark Borodovsky (Biomedical Engineering) Dr. I. King Jordan (Biological Sciences) Dr. Joel Kostka (Biological Sciences) Dr. James R. Tiedje (Michigan State University) ABSTRACT Metagenomics offers unique opportunities to close the gaps that exist between theory and empirical testing in microbial ecology given the windows it can provide into the large fraction of yet-unculturable organisms and species-level resolution. However, realizing the full potential of metagenomics requires the advancement of computational and statistical techniques for data analyses. In the first part of this dissertation I will present a general overview of the current state of quantitative methods for comparative metagenomics and introduce novel algorithms developed for these purposes. Particular emphasis will be placed on the issue of metagenomic coverage, i.e., what fraction of the microbial community was characterized by the sequencing effort, discussing its practical impact on metagenomic analyses and study design, and proposing a novel method for its accurate estimation, Nonpareil (Rodriguez-R and Konstantinidis, Bioinformatics 2014; -, ISME J 2014). Further, the Nonpareil approach can be leveraged for the estimation of extant diversity using a novel metric of sequence diversity, Nd, independent of databases, and with little impact of sequencing effort. In the second part of this dissertation, I will apply the computational and statistical toolbox described above to the characterization of community assembly processes in natural environments. First, I will describe the post-disturbance successional patterns following the deposition of large amounts of hydrocarbons in the shoreline of Florida following the 2010 Macondo Oil Spill in the Gulf of Mexico. We demonstrated that this secondary succession had only minor lasting effects in community composition post-recovery, and was mainly driven by hydrocarbon degradation and nutrient scavenging metabolic potentials in concert with the degree of specialization of community members (Rodriguez-R et al., ISME 2015). I will discuss these results as evidence in support of the specialization-disturbance ecologic hypothesis. Finally, I'll describe variations within the meta-community of a freshwater interconnected system without recent major documented disturbances composed of five lakes and two estuarine locations along the Chattahoochee River (Southeast USA) monitored for a span of six years. Our results revealed strong seasonal patterns together with significant distance decay and minor landscape effects, indicating that both historic and contemporary factors shape the community assembly within this system in similar proportions. I will discuss these results in the framework of classic (macro-organismal) biogeographic theory, and illustrate in which ways they contradict the contemporary view of the now classic Baas-Becking dictum:"everything is everywhere, and the environment selects."
