Statistical analysis
All statistical analyses were performed using R statistical software v.
3.6.0 (R Core Team 2014). We used Generalized Linear Mixed Models (GLMM)
to analyze host development time, host survival (HS), and parasitism
rate (PR) as response variables. Replicates were initiated over the
course of five blocks of time; thus, we applied a mixed model approach,
in which block was used as a random intercept. For host survival and
parasitism rates, we included an observation level random effect (OLRE)
to meet overdispersion and heteroscedasticity model assumptions. In all
further analyses, the alternative host species treatment of D.
birchii and D. sulfurigaster were combined to form its own
unique species identification (D. birchii-D. sulfurigaster ).
We modeled host development times using data from control vials (i.e.,
no parasitoids) only, with fixed factors of temperature, competition,
and host species combination, and all potential interactions between the
three factors. Log transformed development time data improved model fit
and helped meet assumptions of normality. We modeled host survival and
parasitism rates as a function of the cofactors using a Binomial GLMM
with a logit link function. All statistical models included a three-way
interaction between temperature, resource competition, and phenological
delay. We then compared models that incorporated interactions with host
or parasitoid species and included all nested two-way interactions. We
selected the model which minimized AICc using the bblme package
in R
(Bolker
& Bolker 2019). We used the DHARMa(Hartig
2019) package to statistically test whether any assumptions of
normality, non-constant error variance, and overdispersion were
violated. DHARMa simulates scaled (quantile) residuals for mixed
models and provides built-in tests to inspect model assumptions.
Post-hoc multiple comparisons were performed using the emmeanspackage
(Lenth
2019) and P-values were adjusted using the Tukey method when necessary.
All figures were generated using the ggplot2 R package
(Wickham
2011).