Overwintering success in adults of the Japanese common grass yellow Eurema mandarina
Abstract
The evolution of reproductive diapause is controversial in males as compared to females that must overwinter to leave offspring, because late‐autumn males can obtain offspring by pre‐overwintering copulation. The Japanese common grass yellow Eurema mandarina is suitable to examine the evolution of male reproductive diapause, because direct comparisons are possible between males that do and do not exhibit reproductive diapause. Approximately one‐half of males are insensitive to diapause‐inducing conditions, and emerge as non‐diapause summer‐form. Most autumn‐form females mate with summer‐form males in late autumn. Females that have overwintered re‐mate with autumn‐form males before the onset of oviposition. Because last‐male‐precedence is general in sperm competition in Lepidoptera, it is unclear why half of males emerge as summer‐form in late autumn. A potential adaptive benefit for emerging as summer‐form is increased sperm overwintering success, if autumn‐form females have a higher overwintering success than autumn‐form males. In the present study, overwintering success was estimated for both sexes of autumn‐form adults by rearing under seminatural conditions and a mark–release–recapture technique. Both approaches estimated an overwintering success of approximately 5% for both sexes. The absence of difference in overwintering success between the sexes suggests that pre‐overwintering copulation does not increase sperm overwintering success. However, a considerably low overwintering success may explain, at least partly, the presence of summer‐form males in late autumn. The degree of overwintering success might be more important than the sexual differences of overwintering success in the evolution of male reproductive diapause.
Introduction
Reproductive diapause is the hormonally induced arrested development of reproductive organs, and has evolved as a seasonal adaptation in overwintering adult insects (Nylin 2013). Overwintering insects are subjected to unfavorable conditions such as low temperatures and snow, and limited food and water availability (Calvert et al. 1983; Danks 1987; Brower et al. 2004), therefore, successful overwintering requires nutrients such as lipid reserves (Tuskes & Brower 1978; Pullin 1987; Alonso‐Mejia et al. 1997). Reproductive diapause is important for females, because they must endure such conditions and invest in their own survival during winter to lay eggs the following spring. However, it is not always important for male insects, because two alternative strategies are possible (Pener 1992; Kubrak et al. 2016). One is a combination of reproductive diapause and mating after overwintering, similar to the strategy in females. The other is production of offspring without a reproductive diapause by pre‐overwintering copulation. In Lepidoptera, males of the pierid butterfly Gonepteryx rhamni (L., 1758) show the former strategy (Wiklund et al. 1996), whereas most males of the lycaenid butterfly Curetis acuta Moore, 1901 show the latter strategy (Shirouzu 2006). In the monarch butterfly Danaus plexippus (L., 1758) (Leong et al. 2012) and the Japanese common grass yellow Eurema mandarina (de l’Orza, 1869) (Kato 1986), mating occurs in both autumn and spring. Factors for the evolution of male reproductive diapause are controversial.
Eurema mandarina is a suitable species for investigating the evolution of male reproductive diapause, because direct comparisons are possible between males emerging in the same season with and without reproductive diapause. This species shows multivoltinism and seasonal polyphenism, consisting of summer and autumn forms (Fukuda et al. 1982). Eurema mandarina adults emerge as the summer form from June to September. A short photoperiod and low temperature during the larval stage induce the development of autumn‐form adults with reproductive diapause (Yata 1974). Some males, however, are insensitive to short photoperiods and low temperatures, and emerge as the summer‐form even under these conditions (Kato & Sano 1987). Thus, a considerable number of summer‐form males emerge with autumn‐form adults from October to November in central Japan (Kato 1986), even though they are from the same cohort. In contrast to autumn‐form adults, summer‐form males cannot overwinter (Kato 2005).
Summer‐form males and autumn‐form females show high mating activity in late autumn (Kato 1989), although ovary development is clearly arrested until the following spring in autumn‐form females (Konagaya & Watanabe 2015). Thus, most autumn‐form females accept courtships of summer‐form males before winter (Kato 1986). In contrast, most autumn‐form males show low mating activity in late autumn (Kato 2005), and their mating activity increases in the following spring (Kato 1989). Konagaya and Watanabe (2015) reported that most autumn‐form females re‐mate with autumn‐form males before the onset of oviposition in spring. Therefore, sperm competition must occur between the summer‐ and autumn‐form males. Because higher second‐male sperm precedence is generally shown in Lepidoptera (Simmons 2001), autumn‐form males may have an advantage over summer‐form males. Deterioration of sperm derived from summer‐form males is also possible during winter, but half of males emerge as the summer‐form under autumn‐like conditions in the laboratory (Kato & Sano 1987), suggesting that summer‐form males benefit from avoiding overwintering.
Emerging as a summer‐form male in autumn seems advantageous because it improves the “overwintering success of sperm” (Watanabe 2016). If males suffer higher mortality than females during winter, overwintering success of sperm is higher in females than in males. Under these conditions, natural selection may favor males that copulate before overwintering, resulting in an increase of summer‐form males. Although information on overwintering success in E. mandarina is deficient, Kato (1989) reported that the sex ratio of autumn‐form adults did not differ before and after winter. However, the sex ratio in individuals collected from the field is too sensitive to catchability of both sexes, depending on the activity of each sex and field conditions (e.g. weather and time of day). Therefore, experiments under seminatural conditions, such as using cages outdoors, are required to examine sexual differences in overwintering success in this species. In addition, the mark–release–recapture technique is valuable to estimate the degree of overwintering success in wild E. mandarina populations.
Materials and Methods
Overwintering success under seminatural conditions
In central Japan, autumn‐form adults usually enter diapause in late November (Kato 1986). To obtain autumn‐form adults immediately before overwintering, we patrolled the margins of rice fields that were surrounded by deciduous and coniferous forests in the afternoon of 16 November 2015 in Iwakura, Kyoto, Japan (35.09°N, 135.79°E). Seventeen males and 15 females of E. mandarina that were flying alone or foraging on nectar plants were caught with a net, individually placed into glassine envelopes, and brought to the laboratory at room temperature.
On 17 November, their forewing length was measured as an index of body size using digital Vernier calipers (0.1 mm accuracy), and their body mass was measured using an electric balance (0.01 mg accuracy; AG285; Mettler Toledo, Greifensee, Switzerland). The overwintering experiment was started on 17 November based on Wiklund et al. (2003). Butterflies were individually placed in plastic cups (101 mm diameter and 44 mm depth) with net caps. The cups were placed in plastic insect cages (210 × 320 × 180 mm3, with a netlike lid) with shallow water for keeping high humidity. To avoid wind and direct sunlight, the cages were covered with newspaper and placed under the eaves of a building on Yoshida Campus of Kyoto University (35.03°N, 135.78°E). The butterflies were not allowed to feed during the experiment.
The butterflies were observed every Monday, and dead individuals were removed. The water was replenished as necessary to maintain high humidity. The experiment finished on the 16th week (7 March 2016), when the maximum temperature exceeded 20°C, because autumn‐form adults increase their flight activity at this temperature.
For butterflies that died before 7 March, the factors that affected their survival were examined using generalized linear mixed models, with sex, forewing length and body mass included as factors. Body mass was expected to be correlated with forewing length, and the models did not include both body mass and forewing length to avoid multicollinearity. The most appropriate model was selected based on the small‐sample‐size‐corrected version of the Akaike Information Criterion (AICc), and forewing length and body mass were compared by Student’s t‐test. The analyses were carried out using R version 3.0.2 (R Development Core Team 2013).
Mark–release–recapture experiment
A field survey for the mark–release–recapture experiment was undertaken at the foot of Mt. Hokyo (461 m a.s.l.), Tsukuba, Japan (36.15°N, 140.11°E; Fig. 1). The survey area mainly consisted of deciduous forest, rice fields, and abandoned fields that were dominated by the Japanese silver grass Miscanthus sinensis Andersson and the goldenrod Solidago canadensis L. Mountains and a few residential areas with a two‐lane road surrounded the survey area. The survey area contained many sunny slopes, where autumn‐form adults of E. mandarina overwinter (Itoh 2014). Several small ponds for irrigation were scattered over the survey area, and the host plant Lespedeza juncea (L.f.) Pers. grew along the margins of the forests and rice fields. Nectar plants for adults, Bidens frondosa L. and Viola sp., appeared in autumn and spring, respectively. We defined seven survey sites within the survey area to apply the mark–release–recapture method within 30 min, due to the area’s topographical and vegetational features. Survey sites A–E were near rice paddy fields or grasslands, and sites F and G were in a deciduous forest. Adult butterflies cannot enter most deciduous forests, because of dense bamboo grasses on the forest floor. However, mountain trails supplied a suitable space for flying at sites F and G.
The field survey was carried out from October to November 2013 and from March to May 2014. Each sampling effort was performed from 10:00 to 15:00 JST, on sunny days with gentle or moderate winds. One to four persons surveyed each site for at least 15 min in the order A, B, G, C, D, E and F. When we encountered butterflies, we captured them gently using a net and put them in a cylindrical net cage. After sampling at each site, we recorded their forewing length, sex and seasonal form. The wing wear of each individual was used to classify the butterflies into five age classes (see Watanabe & Ando 1993): FF, intact wings; F, fresh wings with fine tears but less lustrous scales; B, some wing tears and scale loss; BB, notched tears and many scales lost; and BBB, broken or extensive tears and many scales lost. After being marked with a black felt‐tip pen, the butterflies were anesthetized with carbon dioxide to avoid “capture–release trauma”, such as escape reactions or other abnormal dispersal behaviors, and released (Watt et al. 1977).
Tukey’s test was used to compare forewing length among summer‐form males, autumn‐form males, and autumn‐form females before and after overwintering. The statistical analyses were performed using R version 3.0.2 (R Development Core Team 2013).
The popan procedure in the program mark (version 8.x) was used to generate linear models with constraints under the Jolly–Seber method, which assumes an open population (Cooch & White 2017). Popan estimated primarily the daily apparent survival rate (ϕ i), catchability (p i) and probability of entrance (Pent i). The probability of entrance is an index of recruitment. Then, daily population size (N i) was also estimated. Because summer‐form and autumn‐form adults have clearly different demographies, the analysis was conducted separately for each form.
The apparent survival rate, catchability and probability of entrance may vary with time. For apparent survival rates and probabilities of entrance, a constant model (.), a full‐time dependent model (t), a linear trend model (T) and a quadratic trend model (TT) can be used. These parameters may also differ among autumn, winter and spring (season), or between winter and the other seasons (winter). Effects of seasons or winter on these parameters could coincide with the linear or quadratic trend. In contrast, we assumed that catchabilities are always full‐time dependent, because the detailed environment of the study area and the sampling effort may vary with time.
These parameters may also differ between the sexes (sex), therefore, models including sex and interactions between sex and time components were also run for the apparent survival rate and probability of entrance of autumn‐form adults. Catchability and population size were assumed to be sex‐dependent, based on the general pattern observed in butterflies (e.g. Watt et al. 1977). The most appropriate model was selected based on AICc values.
Results
Overwintering success under seminatural conditions
During the experiment, the maximum temperature reached 20°C on 20 November, 14 February, and 5 and 6 March. The survival rate in the 12th week of the experiment (just before the week including 14 February) was 11.8% (95% confidence interval (CI), 1.46–36.44%) and 20.0% (95% CI, 4.33–48.09%) for males and females, respectively (Fig. 2). At the end of the experiment, the survival rate was 5.58% (95% CI, 0.15–28.69%) and 6.67% (95%, CI 0.12–31.95%) for males and females, respectively. The survival rates were not significantly different between the sexes, neither in the second week of February (P = 0.645, by two‐tailed Fisher’s exact test), nor at the end of the experiment (P = 1, by two‐tailed Fisher’s exact test).
The optimal model revealed that winter survival depended only on body mass (Table 1). In this model, the mean body mass regression coefficient was 0.017 ± 0.007 (± SE), indicating that heavy butterflies had longer survival times than light butterflies. The mean male body mass was 51.16 ± 13.76 mg (± SD), and the mean female body mass was 57.42 ± 15.01 mg (± SD); the difference was not statistically significant (t30 = 1.231, P = 0.228).
| Model | AICc | ΔAICc | d.f. | Deviance |
|---|---|---|---|---|
| Body mass | 49.58 | 0 | 27 | 42.66 |
| Sex + body mass | 51.69 | 2.11 | 26 | 42.10 |
| Null model | 53.27 | 3.69 | 28 | 48.83 |
| Sex * body mass | 54.25 | 4.67 | 25 | 41.75 |
| Forewing length | 55.14 | 5.56 | 27 | 48.21 |
| Sex | 55.62 | 6.04 | 27 | 47.71 |
| Sex + forewing length | 57.59 | 8.01 | 26 | 47.99 |
| Sex * forewing length | 60.46 | 10.88 | 25 | 47.96 |
Forewing length may also affect survival, but models including forewing length as an independent variable had higher AICc values than those including body mass, and the null model (Table 1). Therefore, survival during winter depended on body mass rather than forewing length. The mean male forewing length was 22.50 ± 1.65 mm (± SD), and the mean female forewing length was 23.21 ± 1.48 mm (± SD); the difference was not statistically significant (t30 = 1.255, P = 0.219).
Recapture rate, forewing length and wing wear
We captured 53 summer‐form males, 111 autumn‐form males and 96 autumn‐form females in the autumn. After the winter, 80 autumn‐form males and 83 autumn‐form females were captured. Four autumn‐form males captured in the autumn were recaptured in the following spring. Summer‐form males, autumn‐form males and autumn‐form females were handled 88, 266 and 219 times, respectively. Of the individuals released, 23 summer‐form males (43.4%), 48 autumn‐form males (26.8%) and 34 autumn‐form females (19.5%) were recaptured at least once.
The forewings of summer‐form males were significantly shorter than those of autumn‐form males and females in the autumn (Table 2), indicating that autumn‐form adults were larger than summer‐form males. For autumn‐form adults, no significant difference was found between sexes in any season. Forewing length did not significantly differ before and after overwintering in either sex of autumn‐form adults, suggesting that overwintering success was not dependent on forewing length.
| Survey period | Summer‐form males | Autumn‐form males | Autumn‐form females |
|---|---|---|---|
| Autumn | 21.3 ± 0.19 (53)a | 22.1 ± 0.12 (109)b | 22.1 ± 0.14 (95)b |
| Spring | – | 22.2 ± 0.12 (75)b | 22.2 ± 0.12 (81)b |
- Different superscript letters indicate significant differences (P < 0.01, Tukey test). –, Not applicable.
A few summer‐form males were age FF, but over half of them were age B, BB or BBB (Fig. 3a). In contrast, almost all of the autumn‐form adults collected in the autumn were age FF or F (Fig. 3b,c), and over half of adults had fresh wings even in mid‐April. After late April, most adults were classified as age BB or BBB without any remarkable differences between the sexes, suggesting that autumn‐form adults fly with damaged wings exclusively in spring.
Abundance, apparent survival rate, probability of entrance and catchability
The optimal model revealed that there were no sexual differences in survival rate or the probability of entrance for autumn‐form adults (Table 3), although these parameters differed between winter and the other seasons. Catchability was full‐time dependent, indicating that the parameter changed from day to day. The parameter was also sex‐dependent, but without an interaction with time. Differences in AICc values between the most appropriate model and the other models were >2, suggesting considerable support for real differences (Cooch & White 2017). The total model set is shown in Table S1.
| Model | AICc | ΔAICc | AICc weights | Model likelihood | No. of parameters | Deviance |
|---|---|---|---|---|---|---|
| ϕ(winter) p(sex + t) Pent(winter) N(sex) | 1205.609 | 0.0000 | 0.31363 | 1.0000 | 35 | −1308.38 |
| ϕ(winter) p(sex + t) Pent(T + winter) N(sex) | 1207.961 | 2.3519 | 0.09676 | 0.3085 | 36 | −1322.32 |
| ϕ(sex*winter) p(sex + t) Pent(T) N(sex) | 1208.911 | 3.3015 | 0.06019 | 0.1919 | 37 | −1323.73 |
- ϕ, apparent survival rate; N, population size; p, catchability; Pent, probability of entrance. Terms in parentheses represent the model’s structure: sex, parameter differed between sexes; t, parameter changed from day to day; T, linear trend; winter, parameter differed between winter and other seasons.
For summer‐form males, the optimal model had a linear trend survival rate, a full‐time‐dependent catchability, and a constant probability of entrance (Table 4). The top three models had the same structure for catchability and probability of entrance. Although different structures for survival rates among the top three models were reported, differences in AICc values between the most appropriate model and the other models were <2. The total model set is shown in Table S2.
| Model | AICc | ΔAICc | AICc weights | Model likelihood | No. parameters | Deviance |
|---|---|---|---|---|---|---|
| ϕ(T) p(t) Pent(.) N(.) | 226.3241 | 0.0000 | 0.30331 | 1.0000 | 12 | −112.270 |
| ϕ(.) p(t) Pent(.) N(.) | 226.4470 | 0.1229 | 0.28523 | 0.9404 | 11 | −109.461 |
| ϕ(TT) p(t) Pent(.) N(.) | 227.2252 | 0.9011 | 0.19329 | 0.6373 | 12 | −111.369 |
- ϕ, apparent survival rate; N, population size; p, catchability; Pent, probability of entrance. Terms in parentheses represent the model’s structure: full point (.), parameter was constant throughout the experiment; t, parameter changed from day to day; T, linear trend; TT, quadratic trend.
In the autumn, the number of summer‐form males and autumn‐form adults of both sexes gradually decreased with time (Fig. 4). The maximum population size was 259.003 individuals (95% CI, 180.730–371.175) and 391.396 individuals (95% CI, 254.482–601.970) for autumn‐form males and females, respectively, and that of summer‐form males was 52.722 individuals (95% CI, 30.693–90.561) on the same day; autumn‐form comprised 83.1% of males. The sex ratio of summer‐ and autumn‐form males to autumn‐form females was estimated as 0.796.
The population in the spring was clearly smaller than that in the autumn for both sexes of the autumn‐form (Fig. 4). The population size during the first survey in the spring was 5.2% of that during the last survey in the autumn, for both sexes. Although the number of autumn‐form males and females increased with time, the maximum population size in the spring was only 37.7% (males) and 38.2% (females) of each population size immediately before overwintering. No sexual difference was detected in the decrease pattern of the population size.
The daily apparent survival rate during the winter was estimated as 0.979 (95% CI, 0.970–0.986) for both sexes of the autumn‐form in the most appropriate model (Table S3), which was higher than the rate of 0.923 (95% CI, 0.896–0.944) that was estimated for the autumn and spring. Because the interval between the last survey in the autumn and the first survey in the spring was 141 days, the apparent survival rate during the winter was calculated as 5.0% (95% CI, 1.3–13.7%) for both sexes of the autumn‐form.
According to the optimal model, the probability of entrance in the autumn and spring was estimated as 0.017 (95% CI 0.013–0.022) for both sexes of the autumn‐form (Table S4). Summer‐form males had a relatively high probability of entrance (0.054; 95% CI, 0.019–0.144) in the autumn. However, the probability of entrance during winter approached zero (95% CI, 0.000–0.000) for both sexes of autumn‐form adults, suggesting higher residency during the winter.
Summer‐form males had higher catchability than autumn‐form adults in the autumn (Table S5), and autumn‐form males always had higher catchability than autumn‐form females. The catchability of both sexes of the autumn‐form tended to be higher in spring than in autumn, indicating that changes in behavior had occurred between before and after overwintering.
Discussion
The objective of the present study was to ascertain the adaptive significance of the production of summer‐form males in late autumn in Eurema mandarina. We hypothesized that summer‐form males increase the overwintering success of their sperm by engaging in pre‐overwintering copulations with autumn‐form females. If autumn‐form females show higher overwintering success than autumn‐form males, the hypothesis can explain why summer‐form males emerge in late autumn. However, both the results of the experiment under seminatural conditions and those in the mark–release–recapture experiment showed that overwintering success was similar between autumn‐form males and females. Therefore, another hypothesis is required to explain the emergence of summer‐form males in autumn. A possible hypothesis is the lack or relaxation of last‐male precedence in the sperm competition the following spring. Although paternity of eggs laid by autumn‐form females remain unclear, eupyrene sperm derived from summer‐form males showed its motility in the spermatheca of autumn‐form females after overwintering (T. Konagaya, unpubl. data., 2013). No sexual differences in overwintering success have been reported in laboratory experiments conducted in other adult overwintering butterflies, such as Polygonia c‐album (L., 1758) and Inachis io (L., 1758), and adults of these species generally mate after overwintering (Wiklund et al. 2003).
In the present study, we examined the overwintering success of autumn‐form adults in E. mandarina not only in an experiment under seminatural conditions in 2015–2016, but also in a mark–release–recapture experiment in 2013–2014. The apparent survival rate as estimated by the Jolly–Seber method is a function of the true survival rate and the residence rate. Although autumn‐form adults sometimes fly on warm days during winter (Itoh 2014), emigration during winter is expected to be low because of the low temperature. Our analysis also revealed a low probability of entrance during winter. Therefore, the apparent survival rate during winter is an appropriate indicator of the true survival rate. We found that the overwintering success estimated for the wild population (1.3–13.7%) was similar to that found under seminatural conditions (males, 0.15–28.69%; females, 0.12–31.95%).
The field census by Kato (1986) also suggested a low overwintering success in autumn‐form adults of E. mandarina. Eurema mandarina produces three direct developing generations and an overwintering generation throughout a year in central Japan. All males and females emerge as summer form in the former three generations. Without any mark–release–recapture survey, Kato (1986) reported a strong increase in the abundance with generations from spring to autumn, suggesting a low mortality rate during these seasons. Failure of overwintering may be a main mortality factor in E. mandarina throughout a year.
Autumn‐form adults must endure unfavorable conditions during winter, and have evolved more suitable traits for survival in such conditions than have summer‐form adults (Brakefield & Zwaan 2011). It is important for butterfly activities to maintain the wings in good condition without any visible damage. Kato (1986) reported that wing wear in E. mandarina summer‐form adults increases linearly with age. We found that most autumn‐form adults had undamaged wings until early April, despite their emergence before winter, in contrast to summer‐form males. Because autumn‐form adults show higher foraging activity than summer‐form males in autumn (Kato 1989), the undamaged wings of overwintered adults may be due to the greater durability of their wings. Alternatively, reproductive activities such as searching females or oviposition might induce wing wearing. There was no clear sexual difference in the wing‐wearing process in autumn‐form adults throughout their lifespan.
Body size affects overwintering success in insects, and we used forewing length and body mass as indices of body size in the present study. Although autumn‐form adults had longer forewings than summer‐form males, forewing length did not affect their overwintering success under seminatural conditions. In addition, the mark–release–recapture experiment showed a slight or no effect of forewing length in the overwintering success of autumn‐form adults. However, survival during the winter did depend on body mass under seminatural conditions. This agrees with the results in Pullin (1987) that heavy I. io and Aglais urticae (L., 1758) butterflies have high lipid content and high overwintering success. Considering the fact that half of the lipid reserve of adult D. plexippus is consumed during winter (Tuskes & Brower 1978; Alonso‐Mejia et al. 1997), lipid content may also be important for overwintering for autumn‐form E. mandarina adults.
For overwintering butterflies, investing in survival by increasing their lipid content may negatively affect their investment in reproduction. Karlsson et al. (2008) reported that autumn‐form females have a lower fecundity than summer‐form females in P. c‐album. Although the cost of overwintering has not been examined in E. mandarina, summer‐form males should invest most of their resources in reproduction, because they avoid overwintering. This is supported by the fact that summer‐form males produce slightly longer eupyrene sperm than autumn‐form males (Konagaya & Watanabe 2015). The avoidance of overwintering may improve ejaculate quality, and result in the relaxation of second‐male precedence in the sperm competition in the following spring.
The overwintering success of E. mandarina recorded in the present study (approximately 5%) was considerably lower than the 90–95% overwintering success reported in P. c‐album and I. io in Sweden (Wiklund et al. 2003), even though the same methods were used to estimate overwintering success under seminatural conditions. In A. urticae, Pullin (1987) reported an overwintering success in the laboratory of higher than 80% when the butterflies were fed on a sufficient amount of honey solution before overwintering. On the other hand, D. plexippus is another butterfly species that has been reported to have a low overwintering success (Calvert et al. 1983; Culotta 1992). Brower et al. (2004) reported that 75% of an overwintering D. plexippus colony was killed by a storm. Danaus plexippus and E. mandarina have similar overwintering success rates, and their mating systems are also similar: in D. plexippus, some males in the overwintering generation copulate with females before winter (Leong et al. 2012). Curetis acuta also shows pre‐overwintering copulation and relatively low winter survival (<20%) (Umedzu 2016). Therefore, the considerably low overwintering success may relate to the presence of non‐diapause males (summer‐form males) in late autumn.
Female receptivity for mating before winter may relate to the evolution of summer‐form males in autumn (Pener 1992), especially when overwintering success is low. Severe winter requires many resources reserved for overwintering adults and reduces chances to encounter mates in the following spring due to the reduction of population density. Therefore, females in such species might evolve to increase receptivity for mating in autumn to obtain sperm or nutrients (Kato 1986), because a butterfly male transfers a spermatophore containing spermatozoa and nutrition as a nuptial gift during copulation (Boggs & Gilbert 1979). Conversely, if winter mortality is low, females are expected to have low receptivity to males before winter, because sperm storage during winter often bring costs for insect females (Roth & Reinhardt 2003). The degree of overwintering success may be more important than the sexual differences of overwintering success in the evolution of male reproductive diapause.
Acknowledgments
We thank N. Sasaki, N. Tokuda, Y. Ichikawa and G. Takahashi for their assistance in the field survey. The study was supported in part by JSPS KAKENHI (Grant No. 15J00738).
