Novel use of statistics to determine precise relationship assignment to
estimate breeding output of a threatened amphibian with tadpole
genotypes
Abstract
Relatedness (rxy) measures are useful in molecular ecology studies as
they can provide a means to answer biological hypotheses where pedigree
information is valuable. Our understanding of the reproductive ecology
of the threatened amphibian Litoria aurea is not complete where applying
rxy measurements may provide further elucidation. Here we use SNPs to
identify which rxy estimators (or combination of) most precisely assign
relationships in L. aurea tadpoles to determine how many breeding pairs
contribute to producing propagules in explosive breeding events. We
aimed to (1) use simulated L. aurea genotype data to determine the
precision of six rxy estimators, (2) compare the precision of
relationship assignment thresholds between rxy estimators computed by
discriminant function analysis (DFA) and test if using multiple
estimators improved precision, and (3) Apply the best performing DFA
model to assign relationships in wild tadpoles to quantify how many
mating pairs reproduced in explosive breeding events. We hypothesised
that each tadpole cohort produced during explosive breeding events would
be conceived by >|2| mating pairs. The
triadic maximum likelihood estimator had the highest Pearson’s R2 (0.92)
and the lowest amount of misclassifications in the DFA (16.00%). A
multi-variate DFA that included three rxy estimators further reduced the
misclassifications down to 11.88%. There was evidence that more than
one mating pair contributed to each explosive breeding event (n = 11).
We show that the multi-variate DFA enabled precise relationship
classification of free-living tadpoles which improved knowledge on
amphibian reproductive ecology. We recommend this method for
relationship assignments in SNP genotyped datasets.