For each of the selection regimes, we have tested the ecological stability of crossfeeding in a random set of 10 runs, for populations taken from 3 time points. The total number of runs, including the ones that lack crossfeeding, is shown in the second column. The first time point is chosen early in the simulations, after the innovation event. The other two time points are indicative of the long term evolutionary dynamics. Each entry gives the fraction of runs with crossfeeding that show ecologically stability. A run is labeled as ecologically stable if crossfeeding is maintained in short ecological simulations (25 000 time steps, no mutations).