Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Fitness Costs of Thermal Reaction Norms for Wing Melanisation in the Large White Butterfly (Pieris brassicae)

  • Audrey Chaput-Bardy ,

    Affiliations Muséum National d’Histoire Naturelle, UMR 7205 Institut Systématique Evolution Biodiversité, Paris, France, INRA, Equipe Ecotoxicologie et Qualité des Milieux Aquatiques, UMR 985 Ecologie et Santé des Ecosystèmes, INRA-Agrocampus, Rennes, France

  • Simon Ducatez,

    Affiliations Muséum National d’Histoire Naturelle, UMR 7205 Institut Systématique Evolution Biodiversité, Paris, France, Department of Biology, McGill University, Montreal, Quebec, Canada

  • Delphine Legrand,

    Affiliations Muséum National d’Histoire Naturelle, UMR 7205 Institut Systématique Evolution Biodiversité, Paris, France, Station d’Ecologie Expérimentale du CNRS à Moulis, CNRS USR 2936, Moulis, France

  • Michel Baguette

    Affiliations Muséum National d’Histoire Naturelle, UMR 7205 Institut Systématique Evolution Biodiversité, Paris, France, Station d’Ecologie Expérimentale du CNRS à Moulis, CNRS USR 2936, Moulis, France

Fitness Costs of Thermal Reaction Norms for Wing Melanisation in the Large White Butterfly (Pieris brassicae)

  • Audrey Chaput-Bardy, 
  • Simon Ducatez, 
  • Delphine Legrand, 
  • Michel Baguette


The large white butterfly, Pieris brassicae, shows a seasonal polyphenism of wing melanisation, spring individuals being darker than summer individuals. This phenotypic plasticity is supposed to be an adaptive response for thermoregulation in natural populations. However, the variation in individuals’ response, the cause of this variation (genetic, non genetic but inheritable or environmental) and its relationship with fitness remain poorly known. We tested the relationships between thermal reaction norm of wing melanisation and adult lifespan as well as female fecundity. Butterflies were reared in cold (18°C), moderate (22°C), and hot (26°C) temperatures over three generations to investigate variation in adult pigmentation and the effects of maternal thermal environment on offspring reaction norms. We found a low heritability in wing melanisation (h2 = 0.18). Rearing families had contrasted thermal reaction norms. Adult lifespan of males and females from highly plastic families was shorter in individuals exposed to hot developmental temperature. Also, females from plastic families exhibited lower fecundity. We did not find any effect of maternal or grand-maternal developmental temperature on fitness. This study provides new evidence on the influence of phenotypic plasticity on life history-traits’ evolution, a crucial issue in the context of global change.


Knowing the mechanisms allowing organisms to cope with rapid variations of their environmental conditions is a research priority in the current era of anthropogenic global changes and species extinction [1], [2]. Adaptation is the only alternative to extinction for those organisms that are unable to move fast enough to track the shift of their climatic envelopes [3]. Adaptation occurs through two main molecular mechanisms, polymorphism of gene expression (phenotypic plasticity) and polymorphism of gene sequences (allelic fixation) [4]. Phenotypic plasticity consists in changes in phenotypic expression of a genotype in response to environmental factors [5], [6] and has been shown to have significant evolutionary consequences [7], [8]. Plasticity is adaptive if the phenotypes produced in two different environments result in higher average fitness across both environments than either fixed phenotype would [9], [10]. Adaptive phenotypic plasticity thus allows organisms to maintain fitness in rapidly changing environments [11][15].

In spite of this benefit, phenotypic plasticity is not ubiquitous at all [16]. Plasticity of organisms is limited in the range of environments they can respond to [17]. There are indeed some constraints such as limits or costs of plasticity that are not straightforward to show [18]. Plasticity costs are defined as any fitness reduction incurred by a plastic individual compared with a non-plastic individual that expresses the same trait value [19]. Studies attempting to quantify these costs have found either no or limited support to fitness decrease in plastic organisms [16], [18]. After a thorough review of studies that measure the costs of plasticity, Auld et al. [16] concluded that a potentially common correlation between environment-specific trait values and the magnitude of trait plasticities (i.e. multi-collinearity) could result in imprecise and/or biased estimates of costs. However, plasticity costs are expected to be common because developmental responses to environmental changes show evidence of imperfect adaptation [12][18]. Phenotypic plasticity creates resource allocation trade-offs during development, i.e. increased investment into one trait may decrease investments in others [17]. Given the occurrence of phenotypic plasticity and the impact that a cost of plasticity could have on phenotypic evolution, identifying costs of plasticity is thus an important issue to evolutionary ecologists [18], [20].

Phenotypic plasticity is commonly estimated as the response of an organism to an environmental gradient. Importantly, this response can be affected by environmental conditions experienced by the parents of this organism [12], [21]. Integrating sources of non genetic inheritability such as the parental environmental effects on phenotypic variation [22], [23] might thus bring information of major interest to better understand the evolution of phenotypic plasticity. Especially, we could expect non genetic transgenerational effects to affect the way organisms cope with phenotypic plasticity costs [24].

Here we tackle these issues by investigating the cost of plasticity in wing melanisation in response to temperature variation. We use the large white butterfly, Pieris brassicae L., as model species. This butterfly shows seasonal polyphenism, the spring morph being usually darker than the summer one because of larger melanised areas on the fore and the hind wings [25]. Studies in other Pierid butterflies have reported that wing melanisation have a high to moderate heritability [26], [27]. This variation in wing melanisation heritability may be due to genotype-environment interactions and then we can expect that those differences in mean wing melanisation can be adaptive. The synthesis of melanin, which is a complex nitrogen-rich polymer with a heavy molecular weight, involves costs in butterflies, especially since the allocation of nitrogen-rich pigment material is constrained within the closed metabolic system of a developing pupa, as shown both by various indirect and direct evidences [28][32]. Melanin and components of the melanin synthesis pathway ensure a wide range of functions in insects, from thermoregulation to immune defence and tegument coloration but also in traits as diverse as wound healing, cuticle sclerotisation (hardening) and egg tanning [29], [33][35]. Given the complexity of the melanin synthesis pathway and the various roles played by melanin in insect homeostasis, we expect significant trade-offs between the melanin production per se, and fitness-related traits [29].

In Pieris brassicae, the experimental exposition of full-sibling caterpillars to contrasted temperatures during their development is supposed to induce the expression of the spring or the summer morphs [25]. Such plasticity of wing melanisation in response to photoperiod and/or temperature is supposed to be adaptive in Pierid butterflies with respect to thermoregulation [26], [29], [36]. Increased melanisation may increase the direct absorption of solar radiation, thus favoring a faster heating rate or giving the ability to reach higher body temperatures. The large white butterfly typically basks by opening its wings at angles of 5–75° to the incident sunlight. This behavior allows dorsal wings to act as mirrors that reflect the solar radiation onto the thorax and/or the more melanised part of the wings [35], [37], [38]. When butterflies use reflectance basking, melanisation of the central and distal parts of the dorsal forewing is thus expected to influence body temperature [35], [37], [38]. Decreased melanisation of spots and distal parts of forewings is expected to favor the efficiency of reflectance basking [35], [37], [38]. In a related species to P. brassicae, Pieris rapae, Stoehr and Goux found a plastic response of the distal part of dorsal forewings to temperature variation [36]. In addition, dorsal forewing patterns are expected to play an important role in sexual selection and mate choice in Pieris butterflies. Especially, UV reflectance is known to affect mate detection and choice in P. rapae [39]. As UV reflectance is strongly affected by both wing melanin patterns and external light conditions [40], we may also expect variation in melanisation across time and space according to the expected light conditions [40]. Overall, dorsal wing melanisation in P. brassicae may thus be affected by various, potentially opposed, selection pressures. Consequently, plasticity in wing melanisation may depend on combined effects of temperature and light, and can be affected by selection on thermoregulation efficiency and sexual selection.

In this study, we address the following questions:

  1. What is the relative importance of the additive genetic variation as opposed to the environmental variation in the wing melanisation? We used ‘animal model’ methods to estimate the heritability of wing melanisation and to partition the relative roles of these two sources of variation.
  2. Does the level of plasticity in wing melanisation vary among full-sib families? Phenotypic plasticity was quantified from variance between environments and from the slope of the reaction norm, which is the phenotypic trait value expressed as a function of environmental conditions [13].
  3. Is there a fitness cost associated with the plasticity in wing melanisation (after taking into account direct costs associated with mean melanisation), and is it affected by non-genetic parental effects? We assumed plasticity costs would appear when the strength of the reaction norm negatively affects fitness independently from the character values expressed within single environments [19].

Materials and Methods

Ethics Statement

No permits were necessary for the sampling of individuals and the experimental work on the large white butterfly, and this project did not involve endangered or protected species.

Rearing Experiment

Experiments were performed at the Muséum National d’Histoire Naturelle in Brunoy, France, using butterflies from a recently established breeding. The stock originated from eggs collected in Brunoy (Essonne, France, 48°42′00″N, 2°30′00″E), Moulis (Ariège, France, 42°57′40″N, 01°05′27″E) and Mesnil-Eglise (Belgium, 50°10′00″N, 45°58′00″E) during the summer 2007.

Twenty mating pairs were isolated from the tenth generation of captive breeding, 1507 eggs were collected from these mating pairs and served as basis for the first generation of our experiment (F1). The rearing experiment consisted of three temperature treatments over three generations to estimate melanin heritability and to test the existence of transgenerational effects of parental and grand-parental developmental temperature on offspring plasticity. Full-sibling eggs were randomly divided over three climate rooms set to constant temperatures of 18°C (cold treatment), 22°C (moderate treatment), and 26°C (hot treatment). Optimum temperature for development seems to be between 20°C and 26°C [41]. Moreover, a pilot experiment with F9 individuals showed that larvae reared above 26°C and below 18°C experienced high mortality (>90%). Larvae were separated in familial groups of thirty individuals. They were reared on cabbage, Brassica oleracea provided ad libitum, in 15*9*9 cm boxes under a constant light cycle (Light : Dark 14∶ 10 h) that induces direct development without pupal diapause. Temperature, light brightness and hygrometry were controlled daily within each climate rooms (POL-EKO ST500), using the same thermometer, hygrometer and light meter, and the only significant difference we observed between the rooms was the temperature (which is the manipulated variable here). Rearing boxes with caterpillars were moved from their position in a climate room every day to avoid a position effect in the room. After emergence, imagoes were placed in 60*60*60 breeding cages, at 25°C ±1°C with unrelated individuals, where their survival was recorded daily. Mating pairs were isolated in 20*20*20 cm laying cages to produce the next generation (F2). Every day, the eggs laid by each female on cabbage leaves were counted before splitting them over the three temperature treatments. Eggs from the same clutches were full siblings as females could mate only once. The number of offspring at each developmental stage was also recorded. We used the same protocol to produce the next generation (F3). The developmental temperature of F1 and F2 generations was manipulated in order to test the importance of maternal and grand-maternal effects on F3 generation, while the developmental temperature of F3 generation was manipulated to test potential costs of phenotypic plasticity.

Picture Acquisition and Processing

We developed a system to take standardised digital photography of living imagoes after their emergence. Each individual was photographed on the day of its emergence in order to limit wing attrition effect on colouration and area of wing. Butterflies were anaesthetized with nitrogen monoxide using an Inject+Matic Sleeper TAS and held between two transparent plastic pieces. The butterfly was then placed into a light tent on a standard white background. The tent diffused homogeneous and constant light from two lamps on both sides. Grey reflectance standards were included in pictures at the start of a photography session [42]. A decimeter arranged under the butterfly allowed us to control the scale of each photograph. The digital camera was placed on a tripod and in a dark room at a distance of 30 cm from the butterfly. We used a Nikon D300 digital camera equipped with a 105 mm macro lens. The shutter lag time was 10 seconds to impede vibrations. We manually set the white balance according to the light intensity and we maintained a constant integration time and a constant lens aperture [42].

The left and the right forewings were extracted from the digital picture (4288×2488 pixels) with the Gimp shareware (, and wing variables were measured with the ImageJ shareware ( The total areas of the right and left forewing of each individual were measured. Melanised areas had a grey reflectance score above the threshold of 120 (where 0 = white, 255 = black). This threshold reliably separated melanised from unmelanised scales on the wings according to the grey reflectance standards [36]. Based on a subsample of 30 individuals, wing measures were highly repeatable (99% of the variation in repeated measurements, including repeated photographing, is due to between individual variations).

Inheritance of Wing Melanisation

A total of 829 butterflies (403 males and 426 females) on 3 000 harvested eggs over three generations were photographed. Adult size was estimated by averaging the area of left and right forewings (i.e. ‘size’ variable). Total black area was also averaged over both forewings for each side (i.e. ventral and dorsal, see Figure 1). As ventral melanised area was correlated with dorsal melanised area (Pearson’s correlation test, r2 = 0.87, t = 51.21, df = 827, p<0.0001), we used dorsal melanised area for subsequent analyses.

Figure 1. Dorsal wings of female (a) and male (b) Pieris brassicae.

The colouration of ventral wings is the same for both sexes; a female is shown here (c). Total black areas of dorsal and ventral forewings were measured.

We estimated wing melanisation heritability from the pedigree data (829 butterflies over 3 generations) fitting animal models with the R package MCMCglmm [43]. The animal model is a mixed-effect model that allows assessing the genetic and non genetic components of phenotypic variation. Individuals’ pedigree is included as an explanatory random variable to estimate genetic variance. This approach allowed to partition the additive genetic variance (animal effect), the variance due to maternal effects, and the residual environmental variance of melanisation. The forewing melanised area constituted the response variable. Forewing area and sex were treated as fixed effects and animal effect (pedigree), rearing temperature, mother temperature and grand-mother temperature were treated as random effects. The MCMC chains were run for 1 300 000 iterations with a burn-in period of 300 000 to ensure satisfactory convergence [44]. Parameters, estimated standard errors and confidence intervals were performed by sampling 1 000 times the posterior parameter distribution. The deviance information criterion (DIC) was used for model selection. The difference between DIC-values of two competing models should not exceed 10 [45]. Adding a random effect ‘rearing box’ nested in the animal effect did not improve the models, and we thus excluded this effect from the models.

Wing Melanisation Plasticity

For analyses of wing melanisation plasticity we used the offspring of 9 females (9 families, 585 individuals) from the F3 generation. Thereby we measured wing melanisation plasticity at the family level. As estimators of phenotypic plasticity, we considered: (i) the variance of melanin proportion (i.e. the ratio between dorsal melanised area and total wing area) between temperatures, and (ii) the absolute slope of the linear regression between the melanin proportion and temperature in each family. We used the absolute slope of the reaction norm, as we were interested in the magnitude of the variation, and the direction of the variation was indicated by a supplementary binary variable (i.e. the ‘slope sign’).

We tested the effect of wing melanisation plasticity for dorsal forewings on adult lifespan, using linear mixed effect models (LME) with sex, melanin proportion, rearing temperature, rearing cage, rearing temperature of the mother and the grand-mother, mating status and first order interactions as supplementary fixed effects, and family as a random effect. The mating status, hereafter called ‘mate’ indicates whether an individual mated or not. This variable is known to negatively affect adult lifespan due to an energetic cost [46], [47]. The melanin proportion was added to models in order to verify that lifespan was influenced by melanin plasticity and not by individual melanisation only. We performed a model selection approach using the Akaike Information Criterion (AIC, [48]) and a model averaging procedure. Akaike weights (Wi) that represent the relative probability for a model i to be the best among considered models, were calculated for the subset of models having (AICbest −AICi) ≤2. In a model averaging procedure, averaged parameters and their corresponding unconditional standard errors were calculated from the smallest subset of AIC-ranked models for which ∑Wi was ≥0.95. The relative importance of each variable within the averaged model was estimated by adding Wi-values of those models within the 95% confidence set containing that variable [48][50].

We tested the effect of wing melanisation plasticity on female fecundity (30 females from F3 generation) using generalized linear mixed models (GLMM) with the number of hatching eggs as response variable, and wing melanisation plasticity, melanin proportion, rearing temperature, rearing temperatures of the mother and the grand-mother, female lifespan and their interactions as fixed effects, and family as a random effect. The number of hatching eggs was analyzed using a quasi-Poisson error and a logarithmic function link (a Poisson GLM correcting for overdispersion [51]). As in quasi-Poisson models, the AIC is not defined, we performed analysis of deviance to compare two nested models (full and nested models [52]). The best model was obtained by stepwise deletion of non significant terms (p>0.05). No transformations of response variables were needed to meet the assumptions of normality and homoscedasticity. Statistical analyses were performed using R, version 2.11.1 (The R Foundation for Statistical Computing, Vienna).


Heritability of Wing Melanisation

The most parsimonious model included forewing area and sex as fixed effects and animal effect and rearing temperature as random effects (Table 1). The animal model provided evidence for equivalent effects of additive genetic variation (CVa = 40.03; 95% Highest Posterior Density Interval (HDPI) = 20.00–62.26), developmental temperature variation (CV = 35.19; 95% HDPI = 10.13–527.08) and environmental (residual) variation (CVe = 48.91; 95% HDPI = 36.54–60.98) on wing melanisation, which resulted in a rather low heritability estimate for melanin (h2 = 0.18; HDPI = 0.01–0.41). It is noteworthy that maternal and grand-maternal temperatures were not included in the best model.

Table 1. Model selection for ‘animal models’ of melanin variation to estimate forewing melanisation heritability.

Thermal Reaction Norm

The 9 families showed contrasted reaction norms (i.e. the relationship between melanin and temperature, Figure 2). Mean reaction norms exhibited positive, flat and negative slopes (Figure 2). Testing the effect of interaction between temperature and family on wing melanisation indicated that variations between environments were significantly different across families (Likelihood Ratio Chi-Square = 1190, DF = 8, p<0.0001). Wing melanisation increased significantly with temperature in 4 families (0.006< slope of the reaction norm <0.021), decreased significantly in 4 families (−0.007< slope of the reaction norm<−0.027), and did not significantly vary in one family (slope = −0.002, p = 0.136, Table 2).

Figure 2. Thermal reaction norms (mean ± standard error) for melanin proportion of the dorsal forewing in 9 full-sib families of Pieris brassicae.

Table 2. Effect of rearing temperature on melanin proportion in 9 butterfly families (GLM results).

Effect of Plasticity on Adult Lifespan

We did not find a parental or grand-parental effect of rearing temperature on adult lifespan, as mother and grand-mother rearing temperatures were not included as random factors in the best models. Wing melanisation plasticity (the slope of the reaction norm and the variance between environments), melanin proportion, rearing temperature, sex, mating status and interactions accounted for adult lifespan (Table 3). Only the magnitude of the thermal reaction norm (the absolute slope of the reaction norm) accounted for wing melanisation plasticity, as the direction of the reaction norm (the slope sign) did not affect lifespan.

Table 3. Effects of forewing melanisation plasticity (‘plasticity’), melanin proportion (‘melanin’), sex, mating and rearing temperature on lifespan in Pieris brassicae: results of model averaging on linear mixed models with family as a random effect.

In order to better understand the effects of interactions between melanin plasticity, sex and rearing temperature on lifespan (Table 3), we made post-hoc analyses by sex within each temperature. These post-hoc analyses were linear models with lifespan as response variable and plasticity (i.e. variance between environments) as fixed effect (Table 4 and Figure 3). Lifespan was differently affected by temperature according to the sex. The lifespan of males tended to be longer in high wing melanisation plasticity families at 18°C and shorter at 26°C (Table 4 and Figure 3). The lifespan of females was also shorter at 26°C when plasticity increased (Table 4 and Figure 3).

Figure 3. Effect of the interaction between wing melanisation plasticity (i.e. variance between environments) and rearing temperature on mean adult lifespan (± standard error) in females (a) and males (b).

The continuous black line and closed circles correspond to 26°C, the continuous grey line and grey circles to 22°C and the dotted black line and open circles to 18°C.

Table 4. Effect of forewing melanisation plasticity (i.e. within family variance in forewings melanisation between temperatures) on lifespan in females and males at a rearing temperature of 18, 22 and 26°C.

Effect of Plasticity on Females’ Fecundity

Females’ fecundity was also negatively influenced by wing melanisation plasticity although this effect was weaker considering variance of melanin between temperatures instead of the slope of the thermal reaction norm (Table 5). There was no developmental and parental effect of temperature on fecundity. Females from plastic families produced fewer eggs. Females that lived longer produced more eggs than those with a shorter lifespan (Table 5).

Table 5. Effects of forewing melanisation plasticity and female lifespan on fecundity.


Roff [53] compared the heritability of suites of life history, behavioural, physiological and morphological traits in a large panel of organisms and showed that life-history traits are always less heritable than other traits. We report here a rather low heritability of wing melanisation in P. brassicae (h2 = 0.18), which is in the range of heritabilities of life-history traits rather than of morphological or physiological traits. This result is in contradiction with studies on other Pierids that showed a high to moderate heritability of melanin [26], [27]. We suggest that this low heritability of melanin deposition on the wings reflects more developmental plasticity associated with the multiple functions of melanin in the homeostasis of insects [29], [33][35] rather than the transmission of pure morphological or physiological attributes. The effect of environment on wing melanisation revealed here by the animal model is also in line with adaptive plasticity reported in Pierid butterflies with respect to thermoregulation [26], [29], [36]. This adaptive plasticity does not seem to be influenced by the developmental temperature experienced by the previous generations. This result confirms findings of other cross-generation studies in which effects of parental temperature on offspring life-history traits were weak or absent [54], [55].

Wing melanisation is classically supposed to decrease with increasing temperature [56], [57]. Here we found that some butterfly families were darker at warmer temperatures. Besides, the reaction norms of these 9 families were markedly different. Our results seem to support a quantitative genetic variation in thermal reaction norms for wing melanisation, rather than a simple, general mechanistic relationship linking temperature to melanin deposition. Families of a same population can diverge not only in the average amount of plasticity expressed but also in their patterns of inter-individual variation in plasticity. The maintenance of this variation may be explained by the existence of different life history and ecological strategies within populations, or by differences in individuals’ exposure to selective pressures in the wild. Such a reversal in the temperature-phenotype rule, together with between-family variation in thermal reaction norms were also observed for the body size in natural populations of Pieris rapae [58]. Adaptive plasticity consists in producing an optimum phenotype corresponding to the prevalent local environmental conditions [59]. We would expect individuals that experienced particular environmental conditions during their larval development to be better adapted to similar conditions during their adult stage. However, we showed here that adult lifespan was not higher when larvae developed at 26°C, which is close to adult rearing temperature (25°C ±1°C). Similarly to ours, studies on thermal reaction norms based on the breeding of larvae at contrasted temperatures [29], [36], [58] investigated the adaptability of adult phenotypes at only one temperature regime. Alongside with the recommendation of Gilchrist and Huey [55], we suggest that it would be highly informative to study the relationship between life-history traits and temperature on adults manipulating adult breeding temperatures according to the ranges of developmental temperatures.

We have highlighted the coexistence of high and low wing melanisation plasticity families in our experimental population (Figure 2). Among males, high plasticity seemed to confer a lifespan benefit at low temperature but not at moderate temperature, and became costly at high temperature. Among females, high wing melanisation plasticity incurred a lifespan cost at high temperature and a fecundity cost regardless the developmental temperature. Overall, in our experimental setting, wing melanisation plasticity was costly for females, whereas males seemed to benefit from plasticity at low temperature. The evolutionary dynamics of this potential trade-off certainly deserves further research; we suggest again that this exciting issue could be fully addressed only by comparing fitness of adults bred under the same ranges of temperatures than those experienced by larvae during their growth.

Here, adult butterflies experienced a homogeneous thermal environment across 13 generations in rearing conditions. Such stable conditions probably maintained constant the costs of wing melanisation plasticity. Differences in fitness between high and low wing melanisation plasticity families tend to confirm that homogeneous environmental conditions could select against plasticity or favour canalisation [18]. Plastic individuals must indeed invest resources in maintaining the molecular/physiological ‘machinery’ needed to detect, monitor and respond to various environmental conditions [16] and therefore may be counter-selected under stable conditions.

Temperature is one of the environmental cues that influences the expression of many phenotypic traits [13]. We reported here on fitness costs of thermal reaction norms for wing melanisation in a butterfly, which potentially entail selection against the more plastic individuals. Making quantitative predictions on the extent to which the homogenisation of the environmental conditions [60] associated with the current era of global changes will affect phenotypic plasticity is thus an important challenge.


We thank Marielle Peroz and Sandrine Chertouk for rearing maintenance. We also acknowledge Charlotte Joreau, Lucas Auer and Christine Li who helped us in pictures’ acquisition and Gabriel Nève (University of Provence), Vincent Debat (MNHN) and Marc Théry (CNRS) for their advice in picture analysis. Pierre-Yves Henry (MNHN), Céline Teplitsky (MNHN) and Thomas Tully provided us with statistical advice. Finally, we thank Lhorine François (University of Bordeaux) for her English corrections.

Author Contributions

Conceived and designed the experiments: ACB MB. Performed the experiments: ACB SD. Analyzed the data: ACB SD. Contributed reagents/materials/analysis tools: ACB SD MB. Wrote the paper: ACB SD DL MB.


  1. 1. Hoffmann A, Sgro C (2011) Climate change and evolutionary adaptation. Nature 470: 497–485.
  2. 2. Bell G, Gonzalez A (2011) Adaptation and evolutionary rescue in metapopulations experiencing environmental deterioration. Science 332: 1327–1330.
  3. 3. Berg MP, Kiers ET, Driessen G, Vand Der Heijden M, Kooi BW, et al. (2010) Adapt or disperse: understanding species persistence in a changing world. Glob Chang Biol 16: 587–598.
  4. 4. Pigliucci M, Murranand CJ, Schlichting CD (2006) Phenotypic plasticity and evolution by genetic assimilation. J Exp Biol 209: 2362–2367.
  5. 5. Schlichting CD (1986) The evolution of phenotypic plasticity in plants. Annu Rev Ecol Syst 17: 667–693.
  6. 6. Bradshaw AD (1965) Evolutionary significance of phenotypic plasticity in plants. Adv Genet 13: 115–155.
  7. 7. Schlichting CD (2004) The role of phenotypic plasticity in diversification. In: DeWitt TJ, Schneider SM, editors. Phenotypic plasticity: Functional and Conceptual Approaches. Oxford: Oxford University Press. 191–200.
  8. 8. Murren CJ, Denning W, Pigliucci M (2005) Relationships between vegetative and life history traits and fitness in a novel field environment: impacts of herbivores. Evol Ecol 19: 583–601.
  9. 9. van Kleunen M, Fisher M (2005) Constraints on the evolution of adaptive phenotypic plasticity in plants. New Phytol 166: 49–60.
  10. 10. Davidson AM, Jennions M, Nicotra AB (2011) Do invasive species show higher phenotypic plasticity than native species and, if so, is it adaptive? A meta-analysis. Ecol Lett 14: 419–431.
  11. 11. Schlichting CD, Pigliucci M (1998) Phenotypic evolution: a reaction norm perspective. Sunderland, MA: Sinauer. 387 p.
  12. 12. West-Eberhard MJ (2003) Developmental plasticity and evolution. New york: Oxford University Press. 794 p.
  13. 13. Via S, Gomulkiewicz R, de Jong G, Schneider SM, Schlichting CD, et al. (1995) Adaptive phenotypic plasticity: consensus and controversy. Trends Ecol Evol 10: 212–217.
  14. 14. Moran NA (1992) The evolutionary maintenance of alternative phenotypes. Am Nat 139: 971–989.
  15. 15. Pigliucci M (2001) Phenotypic plasticity: beyond nature and nurture. Baltimore, Maryland: John Hopkins University Press. 344 p.
  16. 16. Auld JR, Agrawal AA, Relyea RA (2010) Re-evaluating the costs and limits of adaptive phenotypic plasticity. Proc R Soc Biol Sci Ser B 277: 503–511.
  17. 17. Moczek AP (2010) Phenotypic plasticity and diversity in insects. Philos. Trans R Soc B Biol Sci 365: 593–603.
  18. 18. Van Buskirk J, Steiner UK (2009) The fitness costs of developmental canalization and plasticity. J Evol Biol 22: 852–860.
  19. 19. DeWitt TJ, Sih A, Wilson DS (1998) Costs and limits of phenotypic plasticity. Trends Ecol Evol 13: 77–81.
  20. 20. Pigliucci M (2005) Evolution of phenotypic plasticity: where are we going now? Trends Ecol Evol 20: 481–486.
  21. 21. Mousseau TA, Fox CW (1998) The adaptive significance of maternal effects. Trends Ecol Evol 13: 403–407.
  22. 22. Bonduriansky R, Day T (2009) Non genetic inheritance and its evolutionary implications. Annu Rev Ecol Evol Syst 40: 103–125.
  23. 23. Danchin E, Wagner RH (2010) Inclusive Heritability: combining genetic and nongenetic information to study animal behavior and culture. Oikos 119: 210–218.
  24. 24. Ducatez S, Baguette M, Stevens VM, Legrand D, Fréville H (2012) Complex interactions between paternal and maternal effects: parental experience and age at reproduction affect fecundity and offspring performance in a butterfly. Evolution 66: 3558–3569.
  25. 25. Feltwell J (1982) Large white butterfly, the biology, biochemistry and physiology of Pieris brassicae (Linnaeus). The Hague, the Netherlands: Dr. W. Junk Publisher. 535 p.
  26. 26. Kingsolver JG, Wiernasz DC (1991) Seasonal polyphenism in wing-melanin patterns and thermoregulatory adaptation in Pieris butterflies. Am Nat 137: 816–830.
  27. 27. Ellers J, Boggs CL (2002) The evolution of wing color in Colias butterflies: heritability, sex linkage, and population divergence. Evolution 56: 836–840.
  28. 28. Stoehr AM (2006) Costly melanin ornaments: the importance of taxon? Funct Ecol 20: 276–281.
  29. 29. Stoehr AM (2010) Responses of disparate phenotypically-plastic, melanin-based traits to common cues: limits to the benefits of adaptive plasticity. Evol Ecol 24: 287–298.
  30. 30. Talloen W, Van Dyck H, Lens L (2004) The cost of melanization: butterfly wing coloration under environmental stress. Evolution 58: 360–366.
  31. 31. Lee KP, Simpson SJ, Wilson K (2008) Dietary protein-quality influences melanization and immune function in an insect. Funct Ecol 22.
  32. 32. Ma W, Chen L, Wang M, Li X (2008) Trade-offs between melanisation and life-history traits in Helicoperva armigera. Ecol Entomol 33: 37–44.
  33. 33. Nappi AJ, Christensen BM (2005) Melanogenesis and associated cytotoxic reactions: applications to insect innate immunity. Insect Biochem Mol Biol 35: 443–459.
  34. 34. Sugumaran M (2002) Comparative biochemistry of eumelanogenesis and the protective roles of phenoloxidase and melanin in insects. Pigm Cell Res 15: 2–9.
  35. 35. Kingsolver JG (1985) Thermoregulatory significance of wing melanization in Pieris butterflies (Lepidoptera: Pieridae): physics, posture, and pattern. Oecologia 66: 546–553.
  36. 36. Stoehr AM, Goux H (2008) Seasonal phenotypic plasticity of wing melanisation in the cabbage white butterfly, Pieris rapae L. (Lepidoptera: Pieridae). Ecol Entomol 33: 137–143.
  37. 37. Kingsolver JG (1985) Thermal ecology of Pieris butterflies: A new mechanism of behavioral thermoregulation. Oecologia 66: 540–545.
  38. 38. Kingsolver JG (1987) Evolution and coadaptation of thermoregulatory behavior and wing pigmentation pattern in pierid butterflies. Evolution 41: 472–490.
  39. 39. Obara Y, Majerus MEN (2000) Initial mate recognition in the British cabbage butterfly, Pieris rapae rapae. Zoolog Sci 17: 725–730.
  40. 40. Obara Y, Koshitaka H, Kentaro A (2008) Better mate in the shade: enhancement of male mating behaviour in the cabbage butterfly, Pieris rapae crucivora, in a UV-rich environment. J Exp Biol 211: 3698–3702.
  41. 41. Maercks H (1934) Untersuchungen zur ökilogie des kohlweisslings (Pieris brassicae L.), die temperaturreaktionen und das feuchtigkeitoptimum. Z Morph Ökol Tiere 28: 692–721.
  42. 42. Stevens M, Parraga CA, Cuthill IC, Partridge JC, Troscianko TS (2007) Using digital photography to study animal coloration. Biol J Linn Soc 90: 211–237.
  43. 43. Hadfield JD (2010) MCMC methods for multiple-response generalized linear mixed models: MCMCglmm R package. J Stat Softw 33: 1–22.
  44. 44. Wilson AJ, Réale D, Cements MN, Morrissey MM, Postma E, et al. (2010) An ecologist’s guide to the animal model. J Anim Ecol 79: 13–26.
  45. 45. Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc Ser B 64: 583–639.
  46. 46. Fowler K, Partridge L (1989) A cost of mating in female fruitflies. Nature 338: 760–761.
  47. 47. Partridge L, Harvey PH (1988) The Ecological context of life history evolution Science. 241: 1449–1455.
  48. 48. Burnham KP, Anderson DR (2002) Model selection and multimodel inference: A practical information-theoretic approach. New York: 2nd ed. Springer-Verlag. 488 p.
  49. 49. Burnham KP, Anderson DR (2004) Multimodel inference: Understanding AIC and BIC in model selection. Sociol Methods Res 33: 261–304.
  50. 50. Grueber CE, Nakagawa S, Laws RJ, Jamieson IG (2011) Multimodel inference in ecology and evolution: challenges and solutions. J Evol Biol 24: 699–711.
  51. 51. Richards SA (2008) Dealing with overdispersed count data in applied ecology. J Appl Ecol 45: 218–227.
  52. 52. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009) Mixed effects models and extensions in Ecology with R. New York: Springer. 574 p.
  53. 53. Roff DA (2002) Life History Evolution. Sunderland, MA: Sinauer Associates. 465 p.
  54. 54. Crill WD, Huey RB, Gilchrist GW (1996) Within- and between-generation effects of temperature on the morphology and physiology of Drosophila melanogaster. Evolution 50: 1205–1218.
  55. 55. Gilchrist GW, Huey RB (2001) Parental and developmental temperature effects on the thermal dependence of fitness in Drosophila melanogaster. Evolution 55: 209–214.
  56. 56. Kingsolver JG (1983) Thermoregulation and flight in Colias butterflies: elevational patterns and mechanistic limitations. Ecology 64: 534–545.
  57. 57. Van Dyck H, Matthysen E (1998) Thermoregulatory differences between phenotypes in the speckled wood butterfly: hot perchers and cold patrollers? Oecologia 114: 326–334.
  58. 58. Kingsolver JG, Massie KR, Ragland GJ, Smith MH (2007) Rapid population divergence in thermal reaction norms for an invading species: breaking the temperature–size rule. J Evol Biol 20: 892–900.
  59. 59. Moczek AP (2011) Evolutionary biology: the origins of novelty. Nature 473: 34–35.
  60. 60. Western D (2001) Human-modified ecosystems and future evolution. Proc Natl Acad Sci U S A 98: 5458–5465.