Body size and diapause decision
Accumulating the materials to build an adult body takes time; an organism reaching adulthood within a shorter period of time must therefore either mature at a smaller size, compensate for the lost growing time by accumulating mass at a faster rate, or some combination thereof (Abrams et al., 1996; Davidowitz & Nijhout, 2004). P. aegeria skews strongly toward the latter option: the considerable variation in development rates between the treatments was matched to a large extent by variation in growth rates (defined as average weight gain per unit time) (Fig. S1), resulting in comparatively small variation in final size.
To the extent that body size differences between treatments emerged, they were mostly found in the Öland population. This is consistent both with earlier observations (Aalberg Haugen & Gotthard, 2015) and with adaptive expectations. The Stockholm population is univoltine, rarely expressing the nondiapause pathway in the field, hence the body size-diapause correlation is presumably not maintained by selection (Aalberg Haugen et al., 2012; Aalberg Haugen & Gotthard, 2015). While both of the other two populations are bivoltine, Öland experiences shorter summers, and only produces a partial second generation (Lindestad et al., 2019). Such seasonal time stress at voltinism boundaries is precisely the condition under which non-diapausing individuals are most expected to trade off fast development for smaller size (Mousseau & Roff, 1989).
The difference in body size between long-day and short-day Öland individuals was not detectable at any life stage earlier than pupae (Fig. S2), suggesting that it emerges at some point during the fourth instar. (An apparent difference between long-day and short-day individuals can be seen as early as the third instar in Fig. 4, but this was due to a coincidental hatching size difference between the two control groups; no overall photoperiod effect was seen when considering all six treatments groups.) These results are reminiscent of those obtained in scarce swallowtail butterflies, where the size polyphenism is reversed (non-diapause individuals are larger), but the size difference arises from higher growth rates only at the end of the last larval instar (Esperk et al., 2013). Intensive studies of the mothManduca sexta have revealed that body size is determined by the interplay of three parameters: basal growth rate, a critical weight, and the delay period from when a larva reaches the critical weight to when the resulting hormonal cascade arrests growth and triggers preparations for the molt to the pupal stage (Davidowitz & Nijhout, 2004; Nijhout et al., 2006; Callier & Nijhout, 2013). While each larval instar increases in size by the same multiple, so that molts between instars occur at predictable weights, the final instar “overshoots” the critical weight by continuing to grow during the delay period (Nijhout et al., 2006). IfP. aegeria functions along similar lines, it is possible that the plastic size difference between developmental pathways is achieved by modulating the length of the delay period; this would leave the size at each previous molt the same for both pathways, as was observed.
When comparing the results for body size to those obtained for larval development rate, an apparent paradox emerges. One may expect that, if the size difference is only established late in the larval period, it should be responsive to adjustment by changes in photoperiod earlier during life. However, this was not observed: final sizes tended to correspond to the initial photoperiod regime, and Öland individuals switching to diapause development in response to shortening days pupated at a smaller size than those reared in constant short days (Fig. 4). In other words, body size diverged between photoperiods later in life, but was nonetheless less adjustable to the environmental conditions experienced late in life, than development rate.
A possible interpretation is that the between-pathway size divergence that occurs in the fourth instar is driven by physiological mechanisms that are primed earlier during life, resulting in a delayed and inflexible effect of photoperiod on weight accumulation. Growth rates in general were seen to be highly flexible and responsive to photoperiod treatment (Fig. S1b), but if the diapause-pathway size difference utilizes a separate mechanism (as discussed above), this need not be a contradiction, and may explain how the development rate polyphenism can be shared across Scandinavian populations (Lindestad et al., 2019) while the body size polyphenism is not (Aalberg Haugen & Gotthard, 2015). Another possible explanation is that subjecting larvae to such unnaturally drastic shifts in daylength (and hence imposing rapid shifts in developmental strategy) resulted in physiological stress, which may have manifested as decreased final size, at least in individuals forced to change to the more resource-demanding diapause pathway. As growth rate was only coarsely measured here, with a single weighing per instar, a more detailed investigation of these mechanisms will need to await further study.