Fig. 1.

Experimental design. Each combination of treatment (‘ + ’ = selection for large size, ‘0’ = unselected control, ‘ − ’ = selection for small size) and environment (‘Still’ and ‘Mixed’) was replicated with 21 populations (‘tubes’, a). For statistical analyses, five colonies from each tube were isolated (b) and propagated for three generations (c) in fresh, still medium to minimize environmental effects. Trait values were calculated as within-tube means from measurements on five colonies from each tube (d). Differences among colonies within tubes (b, c, d) could include environmental and maternal effects; comparisons of means among tubes (a) are independent comparisons representing genetic differences.