Statistical analyses
All analyses were conducted using R programming environment (R Core Team
2017). The effect of egg type (mimetic vs. non=mimetic) on
corticosterone levels was analyzed using a linear model. Corticosterone
levels were natural log-transformed to conform to the expectations of
normal distribution (log-transformed corticosterone: Shapiro-Wilk
W=0.966, p = 0.238). In addition to the egg type, we included the date
of the manipulation as a fixed factor in the analyses, because maternal
hormone levels often show strong seasonality (Tyrrell & Cree 1998;
Jawor et al. 2006; Hauber et al. 2020a) and in this study
corticosterone showed a significant decrease over the breeding season
(see below). We then used a linear model with date as a covariate to
test if corticosterone levels differed between females who rejected or
accepted a non-mimetic egg at the two-hour mark.
For RNAseq analysis, all paired-end reads were mapped to the Swainson’s
thrush genome (Accession GCF_009819885.1) as it was the closest
available full genome assembly at the time of this analysis. Paired-end
mapping was completed using rsem (Li & Dewey 2011) and bowtie 1.0.0
(Langmead et al. 2009) with default parameters. To exclude
unexpressed genes and genes not present in robins, only genes with at
least 10 identified read counts in over one-third of samples were
included for analysis. Differential expression was analyzed using DESeq
(Anders & Huber 2010) for differences between treatments and
differences between individuals who rejected and those that did not
reject the mimic egg. For all analyses, the threshold for significance
was set at a false-discovery-rate of 5%.
We calculated instantaneous heart rate each minute, for the first 10
minutes, after the bird arrived at the nest (first audible detection).
The average instantaneous heart rates over this 10-minute period were
then log-transformed to satisfy assumptions of normality (log-minute
averages: Shapiro-Wilk W = 0.992 p-value = 0.058). We then used linear
mixed models using R package nlme (Pinheiro et al. 2015) to
analyze the effect of model egg color on average heartrate, adding time
since arrival, treatment date, and treatment order as covariates, and
bird ID as a random factor. Time since arrival was included because
robin heart rate was typically high immediately after the arrival at the
nest (presumably due to the metabolic demands of flight) following which
the heart rate rapidly decreased. To test if individuals differed in the
rate of which their heart rate declined following the arrival we asked
if addition or random slope terms to the model resulted in a better
model fit. However, random slope models had a higher AIC, therefore the
final models did not include the random slope term. Only one female
rejected the non-mimetic beige egg within the 10 min heart rate
recording window and we, therefore, did not include the rejection as a
covariate in the model. However, we ran a separate model asking if
females that rejected the non-mimetic beige egg within two hours had a
higher heart rate in response to the beige eggs compared to females that
did not reject these eggs within this time period.