Offspring age (mean 14.9 days±8.8 s.d., minimum 5, maximum 42) was positively correlated with non-specific ROS level (r=0.50, p=0.035, n=16), but not with corresponding estimates of superoxide (r=−0.21, p>0.41). Non-specific ROS levels differed between adult females and juveniles (1.10±0.16 s.e. versus −0.48±0.11 s.e., in adults versus juveniles, respectively; t-test; t=8.16, p<0.00001, d.f.=32). The corresponding t-tests for superoxide with and without CCCP treatment showed corresponding differences (mean superoxide for hatchlings versus mothers, 11.6±0.64 s.e., n=47 and 14.5±1.15 s.e., respectively; t=2.31, d.f.=30.8 (Satterthwaites' approximation), p=0.024; mean superoxide+CCCP, 25.2±0.86 s.e., and 29.3±1.59, respectively; t=2.48, d.f.=30.8, p=0.016).
The heritability of non-specific ROS did not differ significantly from zero (F1,15=0.03, p=0.88). The parent–offspring regression of basal superoxide level was statistically significant (F1,14=8.3, p=0.012, R2=0.37, estimated heritability=0.45±0.16 s.e.), and even more pronounced subsequent to mitochondrial uncoupling by CCCP (F1,14=13.1, p=0.003, R2=0.48; estimated heritability=0.54±0.15). We then subtracted basal superoxide level from induced level to specifically isolate the effects of mitochondrial uncoupling for analysis. This revealed an even higher heritability (; F1,14=33.6, p<0.0001, R2=0.71; estimated heritability=0.82±0.14 s.e.).
Figure 1 The increase in superoxide level (SO) above basal level subsequent to mitochondrial uncoupling is strongly heritable, suggesting that parent–offspring similarity in the baseline level of SO is caused by maternal (mitochondrial genetic) effects (more ...)
Our full-sib analysis showed that, when we removed the variance from offspring age and date of measurement on ROS levels, all three of our ROS estimates showed significant family effects (non-specific ROS: h2=0.49, LR=5.1, d.f.=1, p<0.025; superoxide: h2=0.55, LR=6.3, d.f.=1, p<0.025; superoxide at CCCP treatment: h2=0.73, LR=9.7, d.f.=1, p<0.005).
The results of our allometric engineering experiment showed no effect of yolk manipulation on the levels of offspring superoxide (F1,15=0.01, p=0.91), whereas the effect of maternal identity was significant (likelihood ratio test: LR=5.5, p=0.019). The corresponding treatment effect on non-specific ROS showed a similar pattern (treatment F1,15=1.81, p=0.20; maternal identity LR=3.9, p=0.048). Thus, our manipulation of yolk level shows no effect on the levels of offspring ROS and, hence, variation in maternal yolk investment cannot explain any covariation between maternal and offspring ROS.