Introduction
Phenotypic plasticity allows organisms to thrive in environments that are variable in space and time (Stearns, 1989). A key source of environmental variability is the seasonal cycle, which brings changes in temperature, weather, food availability and predation rates across the year. While often dramatic, seasonal changes can be predicted and anticipated through environmental signals, such as changes in the length of day versus night (Bradshaw & Holzapfel, 2007). Therefore, an organism may achieve high fitness at different times of year through seasonal plasticity: the controlled expression of seasonally appropriate body forms, physiologies, or behaviors (Shapiro, 1976; Moran, 1992; Varpe, 2017).
Plastic responses sometimes include suites of traits working together to form complex alternative strategies, as is seen with predator defense morphs in juvenile frogs (McCollum & Van Buskirk, 1996), paedomorphic versus metamorphic development in salamanders (Semlitsch et al., 1990), migratory polyphenism in locusts (Pener & Simpson, 2009), and indeed with environmentally controlled sex differentiation in many organisms (Ah-King & Nylin, 2010). Such examples suggest that in the presence of an adaptive plastic response on one trait axis, selection can be expected to favor co-adaptive fine-tuning on additional trait axes.
In the case of seasonal plasticity, many insects and other arthropods have the facultative ability to either enter diapause (a hormonally controlled resting state) at a given life stage, hence postponing reproduction until the following year, or develop directly to adulthood and attempt to reproduce (Tauber et al., 1986). Besides the various physiological changes inherent to diapause itself (increased stress tolerance; etc.), a decision to diapause or not also determines the amount of time stress placed on an individual relative to the end of the favorable season, which in turn predicts changes in the optimal values of core life history traits such as growth rate and body size (Abrams et al., 1996). In other words, an individual destined for diapause can afford to develop slowly, while averting diapause and attempting to fit an additional reproductive cycle into the same year may necessitate faster development, implying rapid growth and/or a smaller adult size. Diapausing and nondiapausing individuals will also experience different temporal environments as adults, which enables divergent selective pressures on adult size and morphology (Van Dyck & Wiklund, 2002). Correlations between diapause decision and life history traits, especially development rate and body size, have been predicted through optimality modelling (Kivelä et al., 2013) as well as empirically observed (Nylin, 1992; Blanckenhorn & Fairbairn, 1995; Aalberg Haugen et al., 2012; Friberg et al., 2012; Esperk et al., 2013; Aalberg Haugen & Gotthard, 2015). However, little is still known about the ontogeny and mechanistic details of these co-adapted diapause/nondiapause phenotypes.
For an insect using seasonal cues to determine whether or not to enter diapause, cues perceived later in life (or during a longer period of time) should be more likely to accurately predict the future selective environment, which suggests that it may be favorable to postpone the diapause decision to gather more information. However, a potential tradeoff lies in the fact that the later a decision is made, the smaller the scope for differentially expressing any co-adapted traits that diverge downstream of the diapause switch (Friberg et al., 2011). The severity of this tradeoff may depend, in turn, on the extent to which the set of traits constituting the diapause/nondiapause phenotypes are regulated independently of one another. One extreme possibility is that, once induced by seasonal cues, individuals are irreversibly channeled into one of two distinct developmental programs, each with a corresponding set of trait values (Nijhout, 2003). Alternatively, traits may be induced independently of one another, but by the same environmental cues as the diapause switch (Mather, 1955), or pathway choice may be reversible by cues experienced later in development. While these latter scenarios would allow for more time to develop optimal trait values, they may also increase the risk of producing intermediate phenotypes of low fitness, through developmental instability or conflicting cues (Moran, 1992; DeWitt et al., 1998). Finally, there is the possibility that the expression of one trait indirectly affects the expression of another. In particular, plastic regulation of development rate early in the insect’s life may determine which regulatory stimuli it becomes exposed to later on, hence phenotypes are molded into two overall responses under natural conditions (slow-growing diapausers versus fast-growing non-diapausers). This scenario would correspond to the “cascade”-style developmental switch described by West-Eberhard (2003).
In a study of three butterfly species, each from a lineage that has seemingly separately evolved diapause in the pupal stage, Friberget al. (2011) showed that the photoperiodic switch controlling whether or not to enter diapause is “locked in” relatively late in larval development. In other words, the daylength experienced last determined diapause decision, to a large extent overriding earlier experiences. However, for all three species, regulation was asymmetric: a decision to enter diapause could be reversed later in life than could a decision not to enter diapause, likely reflecting the relative amounts of time required to adequately prepare for each respective pathway. Further investigations in one of the species, Pieris napi , revealed that the diapause and non-diapause pathways differed in growth rate after, but not before, the final decision had been made, indicating that to a large extent pathway trait differences accumulate downstream of the diapause switch (Friberg et al., 2012). Here, we build on these previous results by examining the ontogeny of the diapause/nondiapause polyphenism in another of the three butterfly species, Pararge aegeria . By manipulating photoperiod regimes at different points during the larval period, we investigate how early the two pathways diverge in terms of development rate and body size. Furthermore, we test to what extent growth decisions are reversible by changes in photoperiodic information during development, and whether the strategic asymmetry seen in the regulation of diapause decision is also reflected in the regulation of these co-adapted life history traits.