Parentage analyses
COLONY was run with 23 loci in the same configuration as for marker set
calibration, to assign candidate mother-offspring relationships and to
infer the genotype of unknown parents. For genotyped mothers (n=107), an
offspring assignment probability threshold of 0.9 was used based on the
results of the above calibration. For genotypes that were reconstructed,
COLONY only provides a probability per locus. A probability threshold of
0.9 is usually a good choice for conservative results, however, in our
case, this threshold would discard 93% of the genotypes. Thus,
probability thresholds from 0.4 to 0.9 were examined to determine the
best trade-off between conservative and robust results (i.e. number of
samples retained) (Supplementary Data 2). Finally, only reconstructed
loci with a probability superior to 0.8 were retained for the analyses,
while the others were considered as missing data. Like genotyped
individuals, individuals with missing data at more than 3 loci were
discarded.
Finally, the dataset was checked for inconsistencies (for example, when
more than one mother was found for a single nest) and the least likely
relationship (i.e. with the lowest probability) was removed. Ultimately,
a parent-offspring relationship was only retained when both parents
(assigned or reconstructed) passed the different thresholds: 0.9 for
assigned mothers, 0.8 for reconstructed loci, and less than 3 missing
loci for a given genotype.
From parentage assignments, OSR, female and male reproduction
frequencies were determined. Additionally, within-season parameters such
as multipaternity, female nesting interval, and the number of partners
for males, were reported. The effect of multipaternity on fitness
parameters (clutch size, number of hatchlings, number of dead embryos)
was investigated by performing a two-tailed Student’s t-test on the
average.