Table 1. Statistical analyses of fly size variation at the start vs the end of positive size selection.

Effect	Population mean sizes			Upper quartile sizes
	F	DF	p	F	DF	p
10 kPa vs 21 kPa: Generations 1 vs 11, during truncation selection for large size
	Females			Females
aPO2	69.09	2, 15	<0.0001	89.75	2, 15	<0.0001
Generation	95.98	2, 15	<0.0001	77.98	2, 15	<0.0001
aPO2×Generation	23.28	2, 15	<0.0001	24.07	2, 15	<0.0001
	Males			Males
aPO2	45.32	2, 15	<0.0001	95.52	2, 15	<0.0001
Generation	39.52	2, 15	<0.0001	157.58	2, 15	<0.0001
aPO2×Generation	9.18	2, 15	<0.0025	14.18	2, 15	<0.0004
21 kPa vs 40 kPa: Generations 1 vs 11, during truncation selection for large size
	Females			Females
aPO2	0.05	2, 15	0.9531	4.36	2, 15	<0.0322
Generation	52.14	2, 15	<0.0001	36.20	2, 15	<0.0001
aPO2×Generation	3.04	2, 15	0.0781	1.52	2, 15	0.2500
	Males			Males
aPO2	0.921	2, 15	0.4197	0.71	2, 15	0.5084
Generation	73.46	2, 15	<0.0001	62.90	2, 15	<0.0001
aPO2×Generation	7.23	2, 15	<0.0063	3.33	2, 15	0.0636

Repeated measures ANOVA statistics for the first and last generations that experienced directional selection for larger size, comparing hypoxic-reared (10 kPa, top) or hyperoxic-reared flies (40 kPa, bottom) to the control or normoxic-reared flies (21 kPa). Significant p values are boldfaced. In all cases, hypoxic-reared flies were significantly smaller than normoxic-reared flies, and responded differently than normoxic-reared flies. 10 kPa flies had a lesser increase in mass with size selection, indicated by significant aPO₂×Generation terms. (F = F-ratio; DF = degrees of freedom).