Genetically distinct breeding populations
Given that signatures of population structure can be skewed by closely
related individuals, we used NGSrelate version 2 to identify and remove
individuals with with up to second-degree relationships (kinship
> 0.0884). We implemented principal components analysis
(PCA) and estimated individual admixture proportions using Pcangsd ,
which estimates individual allele frequencies to minimize bias from low
and variable sequencing depth. We determined the number of genetically
distinct breeding populations of American Redstarts by identifying
congruent geographic signatures of clustering in the PCA with groupings
of individuals based on admixture proportions. Posterior probabilities
of group membership from the admixture proportions were visualized on a
base map from Natural Earth (naturalearthdata.com) with each group
specified by a different colour, and clipped to the breeding range of
the American Redstart (Strimas-Mackey et al., 2021). Colour transparency
was scaled such that the highest posterior probability of group
membership is opaque while the smallest posterior probability is
transparent. Visualization was performed in R (R Core Team, 2021).
Genetic differentiation among the breeding populations was calculated by
creating site allele frequency files for each breeding population and
calculating FST in ANGSD ().