2.1. Photoperiodism and Temperature

The longstanding assumption that temperature was the abiotic factor responsible for the phenotypic switch in A. levana was elegantly refuted by Müller (1955). Before presenting his findings, he remarked on the matter of temperature: “In the field, it cannot be the cause, as it is on average about the same during the crucial developmental phases of the two generations” ^[18]. In his experiment, he assigned offspring from both generations into four treatment groups. He then reared one group from each parental generation under two separate light regimes but otherwise identical conditions, in particular, at the same temperature. The two groups exposed to more than 16 h of light per day developed exclusively into subitaneous pupae and thus into the prorsa form, whereas the two groups reared under short-day conditions (8 h of light per day) invariably developed into diapause pupae and thus into the levana form, regardless of the parental generation ^[18].

Because the potentially modifying influence of temperature was still unclear at this point, Müller subsequently investigated the effects of the distinct light regimes at two different temperatures: 20 and 30 °C ^[24]. At 20 °C, all larvae developed into the levana form if exposed to fixed light regimes of 4–15 h per day, whereas there was an inverse relationship between light duration and the proportion of prorsa individuals in the same photoperiodic range when larvae were reared at 30 °C. Specifically, longer photoperiods led to a steady decline in the proportion of prorsa individuals. When the day length was 6 h, the proportions of prorsa and levana adults were approximately equal, but when the day length was 12 h the ratio was 3% prorsa to 97% levana. However, at both temperatures, a switch occurred between day lengths of 15.5 and 16.5 h. More than 16.5 h resulted in the complete inhibition of levana development, yielding 100% prorsa adults. When the photoperiod falls below 15–16 h (daylight lasts for 15.5 h between the middle of May and late July, in this study at app. 51°8′ N 11°1′ E), the temperature is, therefore, used as an additional cue to determine whether direct development or diapause is preferred.

Later work showed that the critical photoperiod is longer at lower temperatures, with a temperature regime of 15 °C shifting the photoperiod needed for direct development towards longer days ^[25]. Exposure to 16 h of daylight at this temperature still committed little more than half of all larvae to direct development. These findings indicate that longer photoperiods are required to induce direct development at lower temperatures, whereas shorter photoperiods are sufficient at higher temperatures, although the latter only applies to photoperiods of less than 12 h. In an ecological context, this means that warmer temperatures tip the risk–benefit ratio in favor of direct development (betting on continued beneficial conditions), whereas cooler temperatures have the opposite effect (betting on an overwintering strategy). The increase in critical day length is more likely to prevent direct development and consequently the formation of a potential third butterfly generation. In Central Europe, this reflects conditions in the wild, where prorsa larvae develop from mid-May to mid-July at day lengths of at least 16.5 h and mean temperatures of 15–18 °C ^[18][24][25] (Figure 1). In contrast, levana larvae develop in August and September, when the mean temperature is initially ~19 °C but quickly declines to ~11 °C by the beginning of October ^[18]. Day length during the same period declines from 15.5 to 11.5 h (Figure 1). This suggests that the photoperiod takes precedence as the key climate predictor with the ultimate decision-making role, but it can be modified and fine-tuned by prevailing temperatures.

The development of subitaneous or diapause pupae depends on the day length in the mid (but not early or late) larval stages, with a critical photoperiod of ~15.5 h ^[24][26]. These findings have been modified by a more recent study ^[27], although direct comparisons are not possible because the data provided in the original studies are not precise. Larvae were reared at 23 °C under short-day conditions (12 h photoperiod) or long-day conditions (20 h photoperiod), and subsets were transferred between conditions in both directions in each of the five instars. Transfer from long-day to short-day conditions during the first three larval stages invariably led to diapause development, whereas the transferred fourth-instar larvae yielded ~40% prorsa adults and transferred final-instar larvae yielded 100% prorsa adults ^[27]. A fixed number of long days (18 h photoperiod) is necessary for direct development, representing up to half of the entire larval development period (~23 days) at 20 °C ^[25]. This indicates that there is a point of commitment during the fourth-instar stage beyond which diapause development is no longer possible, an advantageous strategy given the additional preparations needed to survive winter, such as general physiological changes, the formation of denser tissues, and the thickening of the cuticle ^[27]. On the other hand, it also implies that natural selection favors a decision made late in larval development, when larvae have the most current information about their position in the season. Friberg and colleagues also showed that transfer from short-day to long-day conditions during the first four larval stages led reproducibly to 100% prorsa adults, and even when switched during the final larval stage, there was still a 50% likelihood of prorsa development. In nature, longer day lengths correspond to spring and early summer (Figure 1). In years with early-season high temperatures, imagoes may emerge ahead of time, as reported in 1990 in the south-west of Germany ^[28]. After mating and oviposition, larval development may, therefore, start when day lengths are below the critical photoperiod for prorsa development of >15.5 h. This threshold is likely to be even higher, given that the mean early-season temperatures are still comparatively low even in unusually warm years and low temperatures require longer day lengths in order to achieve direct development (contrast with the modification of critical day length by low temperatures, as discussed above). In such cases, it would, therefore, be beneficial if larvae were able to identify and respond to a “switch from short days to long days” and accordingly favor the subitaneous pathway over diapause throughout larval development.

Temperature can also modify photoperiod effects at later stages of development. As with many nymphalids and other butterflies, the influence of higher or lower temperatures during early pupal development can lead to a brightening or darkening of wing color patterns in both Araschnia generations. However, a complete change from levana to prorsa or vice versa is not possible ^[29]. For subsequent development, only the temperature is relevant because both pupae and imagoes are profoundly insensitive to day length. During diapause, pupae must undergo a cool period (0–10 °C) lasting at least 3 months before eclosion can be induced by spring temperatures of 12–24 °C ^[26].

2.2. Food Quality

Not only day length and temperature change seasonally but also availability, composition, and quality of food sources. These may, therefore, have an impact on the phenotype in their own right. In both generations, A. levana larvae are strictly monophagous and feed exclusively on leaves of the stinging nettle Urtica dioica. However, the nutritional quality of this plant deteriorates as the nettle matures and thus also varies seasonally ^[30][31][32]. Seasonal variations in adult food sources are also conceivable, suggesting that food quality may influence the fitness of A. levana, which may contribute to polyphenism. Access to carbohydrates, nitrogen, and amino acids across juvenile and adult stages is known to influence the adult size, longevity, and fecundity in many lepidopteran species ^[30][33]. Adult female A. levana f. prorsa reared on a low-quality larval diet preferred a high-quality nectar mimic containing both essential and non-essential amino acids ^[33]. The authors proposed that such a preference implies that adult resources are more important when larval reserves are poor and that butterflies may compensate for adverse larval conditions by selective adult feeding. They further demonstrated a negative relationship between emergence mass and amino acid preference regardless of the larval diet, with amino acid preference diminishing as female mass increased. Interestingly, male larvae did not display any food preference regardless of the larval feeding state or emergence mass, suggesting this trait is linked to female fecundity or sex determination.

A follow-up study tested whether the use of nectar amino acids by adult female A. levana f. prorsa increased fecundity ^[30]. The authors evaluated the effect of low-quality and high-quality larval diets (based on the U. dioica leaf nitrogen content) combined with adult high-quality or low-quality nectar mimics (with or without amino acids). They compared the effects of these four diets on multiple fecundity parameters, including the number of eggs laid, egg mass, longevity, and hatching rate. Female emergence mass (reflecting larval food quality), adult nectar diet, and the amount of nectar consumed all had significant effects on the number of eggs laid. Individuals produced fewer eggs only if they were reared on a low-quality larval diet and, as adults, nectar lacking amino acids. The egg number was on par in the three other groups, including individuals reared on a low-quality larval diet but switched to the amino acid-rich high-quality nectar as adults. However, the egg mass, carbon to nitrogen (C/N) ratio, and hatching rate stayed the same even under adverse conditions, indicating that the fitness cost is purely quantitative. Furthermore, there was no significant effect on longevity in any treatment group, and the C/N ratio of abdomens from female specimens did not vary. The authors reported a positive correlation between butterfly emergence mass and the total number of eggs laid. Emergence mass depends on larval nutrition, and poor-quality foliage is typical for wild A. levana larvae feeding later in the year. This supports the presence of polyphenism in addition to compensatory adult feeding.

Differences have been reported in the body composition of pupae and imagoes depending on sex and/or light regime applied during the rearing of larvae ^[32]. Groups were reared with a 16 h photoperiod representing the spring and early summer conditions of the subitaneous prorsa generation and a 12 h photoperiod representing the late summer and early autumn conditions of the diapausing levana generation ^[32]. The water content of male and female pupae was lower in levana than prorsa individuals, and the male levana pupae also featured lower concentrations of lipids, but there were no significant differences in sugar or protein content and dry weight. Interestingly, the differences were not apparent in the adults. Indeed, the water content was higher in levana than prorsa imagoes at eclosion, and there was no seasonal difference in lipid content. These findings were put forward as evidence for phenotypic traits associated with pupal diapause and overwintering ^[32]. For example, diapausing levana pupae may be adapted specifically to prevent further dehydration, allowing them to enter pupation with a lower water content, protecting them against freezing. The authors did not report an increase in sugar content, but suggested that more subtle changes in sugars related to freeze tolerance in levana pupae may have been overlooked. This is because their methods were insensitive to variations in sugars associated with regular metabolism (such as glucose) and sugars that act as cryoprotectants (such as trehalose) but also polyols such as glycerol with a similar role.

The levana adults weighed less than their prorsa counterparts, which may reflect the metabolic cost of diapause and overwintering ^[32]. Notably, the lower body mass primarily affected the adult head, thorax, and wings. Because levana adults also emerged with lower protein concentrations, the flight musculature (and, by extension, flight capacity) is likely to be affected, consistent with findings in field-caught butterflies ^[32][34][35]. Both morphs in the study were fed on the same larval diet of freshly-picked U. dioica collected between July and August, representing the mediocre to poor food quality typically encountered by levana larvae. The study design is, therefore, likely to have masked any polyphenic effects related to differences in food quality.

Recently, a study by Esperk and Tammaru (2021) reported comparisons of various parameters of larval growth schedules in a 2 × 2 × 2 crossed design with photoperiod, temperature, and host plant quality as the varied factors ^[36]. Specifically (among other findings), they showed that levana larvae spent more time in both final and penultimate larval instars. Lower (but not higher) temperatures, also promoted lower levana pupal masses. In contrast, prorsa larvae displayed higher growth rates—a pattern that was consistent across different rearing conditions, sexes, and larval instars. These authors concluded that their findings demonstrated that the between-generation differences in development have a significant element of anticipatory plasticity and thus should be considered adaptive.

The latter study elegantly demonstrated how the approaches of the other studies described above could be combined to disentangle further the responses to food regimes as well as polyphenic adaptations to predictable seasonal variations in nutrition.