Extended Data Figure 2: Spatial and geometric relationships for modelled aggregate geometries (well mixed to segregated) as a function of relative diffusivity (the ratio of growth rates to growth yields and diffusivity; see Supplementary Information) within the intermediate exchange model.

Slow diffusion is on the left (equivalent to roughly half the relative diffusivity of hydrogen compared to measured growth rates in our system) and fast on the right (equivalent to 10³ times faster relative diffusivity than hydrogen compared with measured growth rates; see Supplementary Information). a, Total aggregate activity normalized to the group maximum as a function of the J spatial metric showing a strong dependency on geometry favouring well mixed (low J value) geometries under slow relative diffusion (left) and almost no relationship with J in fast-diffusion models (right). The average activity, normalized across all of the regimes rather than within a single regime, also changes dramatically from 0.002 to 0.99 as the relative diffusivity is increased. b, Total normalized archaeal population activity plotted against the total bacterial population activity within the same modelled aggregate. The total number of in silico consortia for rows a and b is 23. c, The normalized (z-score) activity for archaea (red) and bacteria (green) plotted against the distance to the nearest three partners. d, The z-score activity for archaea (green) and bacteria (red) plotted against the distance to environment-aggregate interface (that is, aggregate surface). In plots c and d the r-squared values for each correlation are given at the top of each plot in colours that correspond to the two cell types. The number of modelled in silico bacterial and archaeal cells from c and d plotted in the columns from left to right are: 1,138 bacterial and 1,162 archaeal cells; 1,163 bacterial and 1,137 archaeal cells; and 1,153 bacterial and 1,147 archaeal cells. As diffusion is increased in these models from left to right, the organisms within consortia become less dependent on each other and instead become less syntrophically coupled, relying on environmental exchange. This leads to the highest average activity rates per consortia (compare the top panel a to b).