Paternal age effects on offspring fitness in four populations
Paternal age effect in a sibling comparison design, adjusting for average paternal age, maternal age, number of siblings, number of older siblings, being last born, birth cohort, and parental loss.
We find that the children of older fathers consistently have fewer childrens in all four populations, when compared to their siblings.
Much reporting on increasing parental ages tends to focus on ages at first birth (dashed lines) and hence seems at times unduly alarmist. Parental ages at all births (solid lines) show an increase since 1970 too, but are still lower than in the four Swedish regions of our historical Swedish data (1737-1880). The biggest difference is in fact not the slight delay of age at first birth in modern Sweden, but the much earlier reproductive stoppage (ages at last birth, dotted lines).
Hence, contrary to many people’s intutitions, in modern Sweden (1947-1959, gray) the average child was born to a younger parent than in our pre-industrial populations (1737-1880, green). This is mostly because people stopped having children earlier and to a lesser degree because people started having children ca. 1 year earlier. In contemporary Sweden (2010), the average parental age for both mothers and fathers is still lower than in our historical Swedish data.
In the below plot, we have compared all four populations ages at first and last births. The data for modern Sweden focuses on children born 1947 to 1959. Especially, mothers’ ages at last births stand out: In all historical datasets, it was still common for mothers to reproduce beyond the age of 40, reproductive stoppage for both men and women occurs much earlier in modern Sweden. Further descriptive comparisons on reproductive timing and other parameters.
Comparison of descriptives across populations.
Comparison of the main models across populations.
Comparison of the selective episode models across populations.
Comparison of the results of robustness checks in different populations.
How do the obtained effect sizes on number of children compare to estimates of genome-wide deleterious mutation rate times the mean selection coefficient against a deleterious heterozygous mutation (Hayward et al., 2015) and how to the effect sizes on infant survival compare to calculations of the mutation-attributable effect size computed in Gratten et al. (2016).