Simulated wood duck maternity analysis results from COLONY and CERVUS

Modern genetic parentage methods reveal that alternative reproductive
strategies are common in both males and females. Under ideal conditions,
genetic methods accurately connect the parents to offspring produced by
extra-pair matings or conspecific brood parasitism. However, some breeding
systems and sampling scenarios present significant complications for
accurate parentage assignment. We used simulated genetic pedigrees to
assess the reliability of parentage assignment for a series of challenging
sampling regimes that reflect realistic conditions for many
brood-parasitic birds: absence of genetic samples from sires, absence of
samples from brood parasites, and female kin-structured populations. Using
18 microsatellite markers and empirical allele frequencies from two
populations of a conspecific brood parasite, the wood duck (Aix sponsa),
we simulated brood parasitism and determined maternity using two widely
used programs, CERVUS and COLONY. Errors in assignment were generally
modest for most sampling scenarios but differed by program: CERVUS
suffered from false assignment of parasitic offspring, whereas COLONY
sometimes failed to assign offspring to their known mothers. Reducing the
number of markers (9 loci rather than 18) caused the assignment error to
slightly worsen with COLONY but balloon with CERVUS. One potential error
with important biological implications was rare in all cases—few nesting
females were incorrectly excluded as the mother of their own offspring, an
error that could falsely indicate brood parasitism. We consider the
implications of our findings for both a retrospective assessment of
previous studies as well as suggestions for the best practices for future
studies.