Table 1. Summary of four-part modeling for each of the five
univariate response variables considered in the core analyses
implemented in this research. Each variables were examined in four
different ways using linear mixed models: across sampling dates (i.e.,
including sampling date as a predictor) with larvae and pupae pooled
together as “immatures” (a); late stage only (where pupae uniquely
occurred) removing “Sampling date” and including “Immature stage” as
a predictor (c). Models (b) and (d) use as their response variables the
studentized residuals from (a) and (c) for each respective model to
examine the explanatory power of additional biotic or abiotic “latent”
variables (measured at the log scale) not explained by base models a and
c.