Complete metamorphosis is often treated either as a developmental reset that disconnects juvenile and adult phenotypes or as a stage boundary across which early-life effects simply “carry over.” Both interpretations are too coarse. Evidence from holometabolous insects shows that larval temperature, nutrition, crowding, predator exposure, microbial conditions, toxicants, and social information can influence later phenotypes, yet the affected adult traits, directions, and conditions vary markedly. This perspective argues that metamorphosis should instead be viewed as a selective filter acting on biological states with different capacities to survive, be reconstructed, be compensated, or become conditionally expressed. Persistence is most plausible when a relevant substrate, state, or downstream consequence bridges metamorphic remodeling; disappearance may reflect tissue replacement, state reinitialization, or selective loss of affected individuals; compensation requires a later adjustment rather than mere absence of an adult difference; and phenotypic reversal should be reserved for sign changes in the same homologous response dimension. A further class comprises latent effects that become detectable only under adult challenge. The proposed selective-filter model therefore separates observed adult outcomes from their possible mechanisms and from fitness or adaptive interpretation. It yields comparative predictions based on tissue fate, exposure timing, sex, taxon, developmental environment, and adult context while emphasizing that most existing studies measure cohort-level phenotypes rather than continuous individual trajectories. The framework is intended to organize, not replace, mechanistic tests.
