澱んだ水の中のささやき : 生態系の健全性を支えるバクテリアプランクトンの相互作用ネットワーク

Accurate monitoring and prediction of aquatic ecosystems at multiple scales is crucial to maintaining the biosphere, especially given the role of rivers, estuaries, and oceans in species habitat provision and biogeochemical cycling in the land-atmosphere continuum. Communities of microscopic biota - the microbiome - perform many of these ecosystems regulating functions, emitting informative signals of environmental pressure via patterns in their collective structure and function. Still, the collective dynamics of the microbiome remains poorly understood. What might the characteristics of community structure and function be under optimal conditions? How does the environment destabilize community structure or function? Does an optimally structured microbiome imply a resilient microbiome? Are the many signals reducible to a habitat-specific portfolio that characterizes ecosystem health? In this dissertation I advance our understanding and inform ecohealth assessment and engineering by extracting phylogenetic, structural, and functional patterns from the collective dynamics of the microbiome, using a combination of information theory, network theory, and statistical physics, thereby simplifying ecosystem complexity. I also present the utility of envisioning the environment similarly: as an assembly of interacting elements whose destabilizing impact emerges from patterns within the network. In Chapter 2, I demonstrate how it is possible to extract many signals from patterns in community phylogenetic relatedness, population abundance distributions, and species interactions within the Bacteriome network inferred from information fluxes, all informative of ecosystem state. Among these are the new info-theoretic Kleiber’s Law between bacterioplankton phylum co-predictability (directed interactions) and population and community abundance uncertainty, with an average exponent φ ∼ 2/3 in striking accordance with theoretical expectations; as well as the Phylogenetic Separation Rate (ρP S = 0.01) showing that communities accrue new functional groups much slower than new species (ρSR = 0.4). In Chapter 3, I present the novel Eco-Evo Mandala, a multiscale map of the Bacteriome considering habitat-defined distributions, species interactions, and phylogeny, which signals community - and likely ecological - departures from relative theoretical optimality. The Mandala confirms that these departures are habitat-specific, mostly considering the structural and functional traits related to bacterioplankton abundance and interaction distributions (reflected by ε and λ as power law and exponential distribution parameters, respectively), which are not linearly associated with each other. In Chapter 4, I describe how the Envirome - the collective assemblage of environmental drivers defined by environmental interactions - can pinpoint factors responsible for community disorganization, an idea which has been applied elsewhere but not ecological research. Disorganization within the Envirome meant higher environmental impacts causing larger disorganization within the Bacteriome toward random interaction topologies. I end by envisioning a possible future for ecological research and practice that incorporates the perspective underlying this work. This work emphasizes how functional diversity regulating community optimality is different from taxonomic richness, and it must be combined with probabilistic bacterio-environmental co-variations to create a fulsome picture of ecosystem state, particularly because full biological knowledge is presently too cumbersome to obtain. Consequently, eco-engineering policies and technologies should focus less on moderating the whispers from individual populations or environmental drivers, and more on orchestrating all components into a symphony of order and health