Trophotypes of the human gut microbiome: discrete energetic states under a macroecological framework

Abstract

A central prediction of complex-systems ecology is that strongly interacting communities settle into a limited number of recurrent configurations -- attractors of their internal energy-flow dynamics -- rather than spreading continuously across the space of possible compositions. This has been confirmed in terrestrial mammals, in birds and mammals combined at global scale, and in marine communities, using guild richness as a state descriptor that integrates long-term energetic capacity. Here I extend the framework to the human gut microbiome. Using 8,960 faecal samples from the American Gut Project described by the richness of twelve metabolic guilds, I apply Average Membership Degree analysis (AMD) and identify four discrete Trophotypes separated by sparsely populated regions of the functional space. Principal component analysis, applied independently to the same matrix, identified the same three guilds --- primary generalist degraders, butyrate producers, and acetate producers --- as the main axes of variation, together accounting for 78% of total variance; a diagnostic random forest recovered the same partition structure. The four Trophotypes occupy the four quadrants of the energetic plane defined by input through primary degradation and retention through butyrate production, in a topology consistent with multiple stable configurations sustained by stoichiometric constraints. Host-level metadata predict membership weakly across three independent algorithms (Cohens {kappa} between 0.09 and 0.13). The human gut microbiome organises into a small set of recurrent functional states aligned with broad energetic axes, consistent with the multistability expected in systems governed by nonlinear network interactions. ImportanceThe human gut microbiome does not exist in a single state: different individuals harbour communities with qualitatively different functional organisations. Understanding why requires knowing how many distinct configurations exist and what governs them. This study applies a geometric framework previously used to characterise trophic structures of bird and mammal communities at global scale to a large cohort of human faecal microbiomes. Four discrete functional configurations emerge, defined by variation in the capacity to input energy through primary polysaccharide degradation and to retain it through butyrate and acetate production -- the two macroscopic axes that current theory identifies as the organising dimensions of the colonic fermentative network. Self-reported host metadata predict configuration membership only weakly, indicating either that the relevant control variables are not captured by survey data or that the configurations represent genuinely multistable attractors of the system.