Discussion
According to the ADH, complex life cycles are pervasive in nature
because they facilitate the independent evolution of traits expressed at
different life stages, thereby removing constraints that may otherwise
slow or prevent adaptation to novel environments (Moran, 1994). Evidence
for a key prediction of the ADH—that traits are decoupled across
metamorphosis—has been mixed ((Aguirre et al., 2014; Collet &
Fellous, 2019; Fellous & Lazzaro, 2011; Freda, Alex, Morgan, &
Ragland, 2017; Saenko et al., 2012; Sherratt et al., 2017) and
references therein). Here we argue that if metamorphosis is an
adaptation for optimizing genetic correlations across life stages,
genetic decoupling may evolve only for traits that experience
antagonistic selection across ontogeny. In other words, the ADH does not
necessarily predict that all traits will be decoupled, but rather that
trait decoupling should be somewhat predictable given sufficient
knowledge of stage-specific selection pressures. Therefore knowledge of
how selection pressures may differ on traits across development is
essential to predicting and interpreting patterns of decoupling for
individual gene-expression traits with respect to testing the ADH.
To evaluate the predictability of trait decoupling in light of the ADH,
we took advantage of a well-characterized hypermetamorphic insect, the
redheaded pine sawfly (Figure 1) to generate a priori predictions
regarding the decoupling of gene-expression traits. Consistent with our
predictions, we found that: 1) ecological differences between life
stages enabled us to accurately predict how transcriptome-wide levels of
gene-expression decoupling vary across development (Figure 2). 2) Levels
of decoupling vary among functionally different genes in the
transcriptome, with ecologically relevant genes being some of the most
highly and variably decoupled genes between life stages (Supplemental
Tables 3-5 and Figure 3). 3) Decoupling of gene-expression traits tended
to be more pronounced between developmental stages than between sexes
(Figures 2 and 3). Here, we compare our results to other studies of
trait decoupling under ontogenetically and sexually antagonistic
selection and, in light of this body of work and limitations of our own
study, make suggestions for future research on the ADH.