Metabolic network reconstruction based on envirome data. The selected EMs presented in Table 2 and 3 and in Additional File 1 were used to reconstruct reduced metabolic networks for each phase by excluding non-participating reactions. (a) Consistency analysis of metabolic flux estimations. The consistency index h Wang and Stephanopoulos , and the χ2value (with degrees of freedom equal to the number of redundant measurements) are shown for original and simplified networks. Simplified networks have more degrees of freedom since excluding reactions is equivalent to adding null flux constraints. As shown, all phases have a h value lower than the corresponding χ2value, meaning that the assumed biochemistry is consistent with the measured extracellular rates. (b) Intracellular flux estimations for the first metabolic phase of fed-batch 1. Excluded reactions are indicated with an arrow, showing null or negligible fluxes in the original network. See Figure 2b for an overview of all reactions and Table 6 for details. (c) Histogram showing the frequency of reaction exclusion in all metabolic phases of the three fed-batch cultures. These reactions correspond to the catabolism of some amino acids. The ones with low exclusion frequency should be controlled at stoichiometric levels in the medium to avoid ammonia accumulation. For the reactions excluded during each metabolic phase, see Table 6.