a–d, Summary of parameters describing line and cellular properties and the relationship among these parameters in ancestral (a) and derived (b) CE and ancestral (c) and derived (d) CP populations. Traits in the evolved populations (b, d) are depicted relative to their respective ancestral states (a, c): significant increase (large red circle), significant decrease (small blue circle), and no significant change (grey circle) using a generalized linear model (error structure: binomial; link function: logit) for line fitness and WS→SM with post hoc contrasts; and analysis of variance (ANOVA) for number of cells, number of WS, number of SM, and SM growth rate with post hoc contrasts. WS→SM, proportion of lines producing SM during phase I. Arrows indicate significant regressions and lines indicate significant correlations between traits, dashed lines indicate trends (0.05 < P < 0.09). The colour represents the direction of the relationship: red, positive; blue, negative. The significance level is P < 0.05 using Pearson and Spearman rank correlations, and regressions (line level fitness: generalized linear models; cell-level fitness and number of SM per μl (if present): general linear models). Individual cell properties displayed in a and c are identical for the ancestral state in both CE and CP regimes, but measures of line fitness are regime-specific, and transform the associations between parameters. Parameters that relate positively to line fitness in the CE regime negatively affect line fitness in the CP regime, and vice versa (a versus c). For example, in the CE regime, the number of SM cells and the rate at which WS cells give rise to SM cells positively regresses on line fitness (red arrows, a), whereas only the number of WS cells shows a positive regression with line fitness in the CP regime (red arrows, c). The relationships between parameters in the ancestral populations predict their evolutionary trajectory in each regime. After 10 generations of line selection the relationships between cell- and line-level parameters significantly altered in both CE and CP regimes (b and d). Line fitness improved in both regimes, thereby imposing selection on parameters that were linked to line fitness in their respective baselines. In the CE regime enhanced line fitness is explained by a significant increase in the capacity to transition from WS to SM and is not explained by enhanced performance of single cells: the fitness of single cells either remained unaltered or declined. Increased line fitness can be seen as a product of selection at the higher (group) level. In marked contrast is the CP regime where improved line fitness is readily explained by changes in traits that improve the competitive ability of individual cells. Enhanced line fitness in the CP regime can be interpreted as a by-product of selection at the lower (cell) level.