Fig. 6: Cox proportional hazards regression models showing
predicted values of hazard ratios (i.e. mortality risk) with 95%
confidence intervals as a function of a) tarsus length, and b) telomere
length for each sex. Differences between sexes were not significant.
Larger individuals had lower long-term mortality, and model selection
indicated a tendency for both short and long telomeres to be associated
with lower mortality.
Table 5: Estimates with 95% confidence intervals (CI) and
standard errors (SE) from the highest ranked generalized linear models
(see Table S2.5) fitted using a Poisson distribution of variation in
lifetime reproductive success (LRS) corrected for lifespan for each
population (high /low ). Random intercepts for brood
identity and year were included as random factors in all models. Only
individuals that were not removed during the artificial selection and
that survived until breeding were included in these analyses.