Comparison of fits to digestion time course with different fitting assumptions. (A and B) Best fit of random mixing model (left) and segregated model (right) with all parameters (τ_β3a, τ_β2, n, and sf) unconstrained. Table on the lower right of each panel displays the nominal and effective mole fraction of β2. (AB) The column shows, for each time point in A and B, the SSQ for the random mixing model (in red) and the segregated model (in black), with on bottom the summer SSQ for each digestion time course and then the total. (C and D) Best fit for each model with the additional constraint that n, the parameter for average number of IDs (n) per channel in the population, was set to 4. (CD) Comparison of SSQ at different time points across all conditions of injection ratios. (E and F) As in the main text, best fits for each model with only sf being a free parameter. (EF) SSQ comparisons at each time point and for each injection ratio. We note that, for the random mixing model, the change in SSQ going from 4 free parameters (A:0.0929) to 1 free parameter (E:0.0993) is minor compared with the difference between the random mixing model with 1 free parameter (E:0.0993) and the segregated model with 4 free parameters (B:0.1743). Irrespective of the fitting constraints, the random mixing model provides a better fit than the segregated model with identical numbers of free parameters. SSQ, sum of squares.