Abstract Developmental plasticity enables organisms to cope with environmental heterogeneity, such as seasonal variation in climatic conditions, and is thought to affect a species’ capability to adapt to environments with novel seasonal and ecological dynamics. We studied developmental plasticity of the widespread tropical butterfly, Bicyclus safitza, which reaches the southern edge of its distribution in the temperate zone of South Africa. In wet–dry seasonal environments in tropical Africa, adults of Bicyclus butterflies are present all year round and exhibit discrete seasonal forms in alternating generations. We demonstrate that a population that colonized a more temperate climate region has adopted a different strategy to cope with the local environment. No active adults were encountered during the temperate winter. The flight season coincided with a period when evaporation stress was lowest and temperatures were higher in the South African population. Butterflies collected from the field did not express seasonal polyphenism or show full expression of the tropical wet season form phenotype. Reaction norm experiments comparing stocks from South Africa and Uganda indicated that local adaptation of this tropical butterfly to a more temperate climate involved changes in the degree of developmental plasticity, such that a more robust development in response to thermal variation was observed for a broad suite of morphological and life-history traits. Our findings have implications for understanding the mechanisms that facilitate expansion into a novel ecological niche in seasonally variable climatic conditions. INTRODUCTION Natural environments are heterogeneous, and many insect species have evolved strategies to cope with spatial and temporal environmental variation. Typically, these strategies involve dispersal from unfavourable to favourable conditions, either in space by means of migration, or in time by inducing a state of arrested development or diapause (Bohonak & Jenkins, 2003). Compelling examples of both phenomena include the annual migrations of monarch butterflies (Danaus plexippus) and the winter diapause in many other temperate butterflies, such as the speckled wood butterfly (Pararge aegeria), which is induced when changes in day length indicate that winter is coming (Aalberg Haugen & Gotthard, 2015). An alternative strategy to cope with environmental variability is seasonal polyphenism, which is an extreme example of developmental plasticity whereby different forms of a species are produced at different times of the year. Seasonal forms have adaptive traits specific to the environment in which they occur (Shapiro, 1976). In contrast to spatially or temporally dispersing insects in the tropics, seasonally polyphenic species remain active throughout the year and, thus, demonstrate less dramatic seasonal peaks in activity and abundance. Bicyclus butterflies have become a hallmark example of adaptive developmental plasticity and seasonal polyphenism because they have proved a tractable system in which to study the environmental regulation of development in both natural (Brakefield & Reitsma, 1991; Windig et al., 1994) and laboratory populations (de Jong et al., 2010; Oostra et al., 2014a; van Bergen et al., 2017). The genus consists of > 100 extant species that inhabit a wide range of tropical habitats in sub-Saharan Africa (Aduse-Poku et al., 2015). Recent work has suggested that a single Bicyclus species began to colonize more open and seasonal habitats during the Miocene epoch, when much of the transcontinental forests began to open up (Aduse-Poku et al., 2015). Such environments, with highly distinct wet and dry seasons, are characterized by predictable patterns of variation in temperature, rainfall and humidity that are closely associated with changes in vegetation cover and host plant availability (e.g. Brakefield & Larsen, 1984). The rains of the wet season result in a luxuriant growth of herbs, including species of grass, which the larvae of most Bicyclus species use as host plants. Adult butterflies in the middle of the warm, wet season are highly active and reproduce quickly (Brakefield & Reitsma, 1991). Larvae of this generation develop in increasingly arid and cooler conditions towards the dry season when the ground vegetation, including larval host plants, dies back to become a layer of brown leaf litter. The next generation of adult butterflies emerges around the transition between the wet and dry seasons, and then survives the unfavourable conditions as active, but reproductively dormant, adults before reproducing at the beginning of the next wet season (Brakefield & Reitsma, 1991; Windig et al., 1994; van Bergen et al., 2016). In the laboratory, development of phenotypes similar to the seasonal forms found in nature can be induced by manipulating the temperature during a sensitive phase of pre-adult development (Kooi & Brakefield, 1999). Larvae reared at lower temperatures, which typically represent the environmental conditions of the tropical dry season, develop into relatively large individuals, which allocate resources towards a more durable body and demonstrate cryptic patterning of the ventral wings that are exposed when at rest. In contrast, wet season form individuals, which are induced by high developmental temperatures (e.g. Windig et al., 1994; Roskam & Brakefield, 1996), have a series of conspicuous marginal eyespots on their ventral wing surfaces (Brakefield & Reitsma, 1991) and demonstrate an increased investment in reproduction (Oostra et al., 2011; van Bergen et al., 2017). The distinct wing patterns serve an important fitness function in terms of coping with changing predatory threats between seasonal environments (Lyytinen et al., 2004; Prudic et al., 2015). The common bush brown, Bicyclus safitza (Westwood, 1850), is one of the most widely distributed species of Bicyclus butterflies (e.g. Larsen, 2005). Whereas all other species of Bicyclus are restricted to tropical climate zones, the distribution of B. safitza extends far into more temperate climate zones in southern Africa. One of the most fundamental differences between temperate and tropical ecosystems is the ambient temperature during winter, which can even drop below freezing in temperate regions, and populations of tropical butterflies that successfully colonized more temperate climate regions are predicted to have evolved a suite of adaptations to cope with local environmental conditions. For example, recent field studies have revealed that populations of B. safitza in these temperate regions have a strong preference for shaded forests habitats (Nokelainen et al., 2016), which may buffer seasonal fluctuations in temperature and humidity, whereas in tropical biomes B. safitza is mainly found in semi-open woodland and forest edge habitats (Brakefield & Reitsma, 1991; Windig et al., 1994). Here, we explore whether a population of B. safitza that occurs in Eastern Cape of South Africa has become locally adapted to temperate climatic conditions. To gain a more complete understanding of a species’ potential to colonize novel ecological conditions, we address this question using an integrative and comparative approach. First, we study the phenology and the expression of seasonal polyphenism of the temperate population by conducting a longitudinal survey in the Eastern Cape, South Africa, and compare these with a well-studied tropical population of B. safitza (Brakefield & Reitsma, 1991; Windig et al., 1994). Second, using the ratios of stable isotopes of oxygen obtained from the exoskeleton of field-trapped individuals as well as measurements of local water evaporation during the larval stage, we explore whether evaporation stress might constrain the butterfly niche. We also aim to find explanations for the strong population-specific preference for more shaded habitats in temperate climates (Nokelainen et al., 2016). Finally, we conduct a comparative reaction norm study, using populations from temperate and tropical climate regions, to investigate the extent of local adaptation in response to different thermal environments. Using data from these two populations, we aim to provide insights into the mechanisms that facilitate the colonization of novel ecological niches in seasonally variable climatic conditions. MATERIAL AND METHODS Field sites and butterfly monitoring We monitored butterflies at the southern-most edge of the species’ range in the Eastern Cape province of South Africa. The Eastern Cape spans a multitude of climatic regions, including cold and temperate interior parts as well as temperate and sub-tropical coastal regions (Mucina & Rutherford, 2006). Our study sites were chiefly in a temperate zone that is characterized by open, semi-arid grasslands, whereas afromontane forests and coastal thickets provide more humid, shaded habitats. The field sites in the vicinity of Grahamstown and more humid coastal and riverine environments were chosen using satellite images (Google Earth; Google Inc., Mountain View, CA, USA) to detect suitable habitats, after which the areas were visited to confirm the presence of B. safitza. Three field sites were used in this study: Bathurst (33°30′S, 26°46′E), Kapriver (33°21′S, 26°52′E) and Kasouga (33°39′S, 26°44′E). The Bathurst site represents a riparian bush habitat, with more open areas along the edges. The Kapriver site is an open, grassy hilltop, which transitions into riparian forest in a lower lying river gorge. The Kasouga site is characterized by coastal thickets and bordered by pastureland to the north and by the shores of the Indian Ocean to the south. To investigate the seasonal phenology of B. safitza in temperate regions, we monitored field populations by conducting monthly trapping sessions between November 2014 and October 2015. Nine traps (Megaview, DC0017, Pop-up Butterfly Bait Trap, cone type) were placed at each of the three field sites and equally distributed among three habitat types (open grasslands, forest fringes and under shaded canopy) within each site (for further details and habitat preference comparisons, see Nokelainen et al., 2016). The traps were baited with fermented banana once a month and emptied on the next day. Wild-caught individuals were stored in entomological envelopes until further processing. Data on the phenology and wing pattern plasticity of tropical populations have been published by Brakefield and Reitsma (1991) and Windig et al. (1994), and here we use the same methodologies to quantify habitat use and variation in butterfly occurrence and wing patterning. Evaporation stress measures Evaporation stress from wild-caught butterflies was first studied using the ratio of stable isotopes of oxygen (δ18O) present in the exoskeleton, which have been shown to reflect the mean atmospheric conditions surrounding the insect before moulting (Ellwood et al., 2011). Briefly, the more common, lighter 16O isotope evaporates more readily than the 18O isotope, which leads to enrichment of 18O in sample tissues and thus, more positive δ18O values. To quantify the δ18O values of the specimens collected at the field sites, two legs were placed into silver capsules, sealed and loaded into an auto-sampler. The tissue within the capsule was pyrolysed at 1200 °C using a Thermo Finnigan TC/EA attached to a Thermo Delta V mass spectrometer via a ConFlo 3. Reference standards from International Atomic Energy Agency (IAEA) in Vienna were run at intervals throughout the sequence, and these values were used to calibrate to the international standards of 18O/16O [δ18O Vienna Standard Mean Ocean Water (V-SMOW)]. Analyses were conducted at the Godwin Laboratory for Palaeoclimate Research, Department of Earth Sciences at University of Cambridge, UK. We measured larval rates of water loss with a Li-Cor 6400 photosynthesis system (Li-Cor Biosciences, Lincoln, NE, USA). Three separate runs were conducted using a total of nine similar-sized third instar larvae of B. safitza (x̅ = 22.5 mg, SD = 3.6 mg). Larvae originated from an F1 laboratory stock, initially collected from the Kasouga field site, and were used only once. For the measurements, three larvae were placed together (for better measurement accuracy) in a small mesh cage that was inserted into the leaf chamber of the photosynthesis system in order to obtain rates of water loss. The device measures the exchange of CO2 and H2O between the organism and the atmosphere, controlling the ambient CO2 concentration, temperature and relative humidity, and hence, the vapour pressure deficit (VPD; the difference between the amount of moisture in the air and how much moisture the air can hold when it is saturated). The conditions in which an organism maintains its water balance during temperature changes are more clearly shown by noting the VPD than the relative humidity (Anderson, 1936). Based on this, we measured larval evaporation rates in response to five different VPD conditions [VPD (in kilopascals) = 1.5, 2.0, 2.5, 3.0 and 3.5] at a constant ambient temperature of 25 °C. Evaporation measures were started at the lowest VPD and successively increased. Once a target VPD was attained, five water loss measures (10 s in duration) were recorded and averaged. Evaporation values were used to calculate average water loss of larvae as a percentage of larval body mass per hour (% H2O g−1 h−1) and fitted along the linear regression (i.e. sum of squares) line against VPD. Then, measurements of relative humidity and temperature, and thus VPD, were obtained from each of the three habitats at all three sites using data loggers (Maxim DS1922T iButton Temperature Logger, San Jose, CA, USA) to describe climatic conditions in the wild. Climate data recorded at 3 h intervals over the course of 6 months (N = 8878) at each of the study sites were used to calculate the natural range of VPDs (Table 1) and to predict the larval water loss based on the relationship established in the laboratory (see Supporting Information Table S1). Table 1. Variation in climatic conditions during a field survey of Bicyclus safitza in Eastern Cape, South Africa Abiotic factor Minimum Mean Maximum SD Temperature Open 11.05 21.50 43.54 4.80 Edge 8.05 21.40 52.08 5.21 Closed 9.10 20.49 39.60 3.90 Relative humidity Open 12.59 77.80 100 16.75 Edge 12.57 79.54 100 16.08 Closed 17.25 84.16 100 13.45 Vapour pressure deficit Open 0 0.71 7.41 0.82 Edge 0 0.67 12.01 0.86 Closed 0 0.45 6.00 0.51 Abiotic factor Minimum Mean Maximum SD Temperature Open 11.05 21.50 43.54 4.80 Edge 8.05 21.40 52.08 5.21 Closed 9.10 20.49 39.60 3.90 Relative humidity Open 12.59 77.80 100 16.75 Edge 12.57 79.54 100 16.08 Closed 17.25 84.16 100 13.45 Vapour pressure deficit Open 0 0.71 7.41 0.82 Edge 0 0.67 12.01 0.86 Closed 0 0.45 6.00 0.51 The air temperature (in degrees Celsius), relative humidity (expressed as a percentage) and vapour pressure deficit (in kilopascals) range values and their standard deviations are shown for three habitat types along the coastal range. Vapour pressure deficit is the difference between the amount of moisture in the air and how much moisture the air can hold when it is saturated, and is important in understanding the conditions in which an organism maintains its water balance during temperature changes. View Large Temperature reaction norm experiment To study the geographical variation in the degree of developmental plasticity, we conducted a reaction norm experiment using two populations of B. safitza and four constant thermal regimes. The laboratory populations of B. safitza were established in 2013 from eggs collected at a single location in the Semuliki National Park in Uganda (0°50′N, 30°9′E) and the Kasouga field site in South Africa (33°39′S, 26°44′E). The eggs from at least ten females contributed to each stock population. Thus, the Ugandan colony was derived from a population in the tropics, whereas the South African colony originated from a temperate population at the poleward margin of the species range. After about four generations of laboratory rearing, eggs were collected from both laboratory stocks and larvae randomly divided over four climate-controlled chambers (21, 23, 25 and 27 °C) within 1 day after hatching. In these chambers [Sanyo/Panasonic MLR-350H; 70% relative humidity (RH), 12 h–12 h light–dark cycle], larvae were reared in sleeve-like gauze cages on young wheat (Triticum aestivum) plants; a host plant that is frequently used to rear newly established laboratory populations of Bicyclus butterflies (Oostra et al., 2014a; van Bergen, 2017). Pre-pupae were collected daily, and 1 day after pupation they were weighed to the nearest 0.1 mg (Fisherbrand PS-60) and individually placed in transparent pots until they eclosed. On the first day after eclosion, the adults were sacrificed by freezing and carefully dissected. In addition to wing pattern measurements (see next subsection), we recorded the larval and pupal development times, calculated the growth rate, and measured adult dry mass, relative fat content and abdomen ratio. Data from the Ugandan population were included in a comparative study on developmental plasticity in mycalesine butterflies; for details on methodology, see van Bergen et al. (2017). Wing pattern measurements The ventral surface of one hind- and one forewing of each individual were photographed using a Leica DFC495 digital camera with a Leica M125 stereomicroscope. The images were analysed with the image processing package Fiji (Schindelin et al., 2012). We followed the Comstock–Needham system to refer to wing veins and cells (see Miller, 1970). On the ventral hindwing, the area of the yellow outer ring, the black inner disc and the white focus of the eyespot in cell Cu1 were measured. The relative distance of the proximal edge of the median band along the second wing vein was taken as a measure of the width of the band. The measurements on the ventral forewing included the yellow, black and white areas of the eyespot in cell M1 and the area of the black inner disc of the larger eyespot in cell Cu1. For all wings, an area enclosed by three clear landmarks was used as a proxy of wing size (Fig. 1). The same protocol was used for both experimental and field-caught individuals. Of the latter, we could only analyse those butterflies with a good wing-wear condition and which were collected during the first 6 months. Figure 1. View largeDownload slide Wing pattern elements measured in all specimens of Bicyclus safitza. The images represent the typical phenotype of individuals reared at 27 (rows 1 and 3) and 21 °C (rows 2 and 4). The first two columns represent females (left) and males (right) of the population from the Kasouga field site in South Africa. The last two columns represent individuals of the population from Semuliki National Park in Uganda. For each individual, we obtained 11 wing measurements corresponding to three categories of traits: ventral eyespots (1–3), ventral bands (4 and 5) and wing areas (6 and 7). Different letter codes were used to refer to the corresponding yellow rings (y), black discs (b) and white pupils (w). References to wing veins and cells follow the Comstock–Needham system (Miller, 1970). The icons were provided by Manuel Marques-Pita and adjusted from Mateus et al. (2014). Figure 1. View largeDownload slide Wing pattern elements measured in all specimens of Bicyclus safitza. The images represent the typical phenotype of individuals reared at 27 (rows 1 and 3) and 21 °C (rows 2 and 4). The first two columns represent females (left) and males (right) of the population from the Kasouga field site in South Africa. The last two columns represent individuals of the population from Semuliki National Park in Uganda. For each individual, we obtained 11 wing measurements corresponding to three categories of traits: ventral eyespots (1–3), ventral bands (4 and 5) and wing areas (6 and 7). Different letter codes were used to refer to the corresponding yellow rings (y), black discs (b) and white pupils (w). References to wing veins and cells follow the Comstock–Needham system (Miller, 1970). The icons were provided by Manuel Marques-Pita and adjusted from Mateus et al. (2014). Statistical analyses Before analyses, all wing pattern elements were corrected for wing size, and the nine ventral wing pattern measurements (traits 1–5; see Fig. 1) were reduced using a principal component analysis, pooling all available data. The first principal component (PC1) explained 60% of the total variation and was strongly associated with the effect of the developmental temperature and month of capture. The second principal component (PC2) explained 15% of the variation and was correlated with sex rather than seasonality. All development times from the temperature reaction norm experiment were log10-transformed to improve normality and homoscedasticity. Statistical analyses were performed with the R Statistical Package v 3.1.2 (R Core Team 2015) and IBM SPSS Statistics (v22). To investigate seasonal phenology of the butterflies, we tested the monthly occurrence of butterflies (i.e. frequency of captured butterflies) across months (N = 13) using a general linear model (with Poisson distribution), with the number of butterflies as a dependent variable and month as an explanatory variable. For the field-collected data, general linear models were used to analyse the influence of temperature (monthly mean temperature, as continuous), sex (male/female) and sampling site (Bathurst, Kapriver, Kasouga) on wing pattern morphology (PC1) and butterfly dry mass. Full models were fitted including temperature, sex, sampling site and their interactions, before successive removal of non-significant terms [based on the lowest Akaike information criterion (AIC) value]. To investigate evaporation stress, we analysed data from adult butterflies and data for larval water loss separately. For stable oxygen isotope data collected from adults, we used a general linear model, where the δ18O values were used as the dependent variable and monthly mean temperature, sex and sampling site and their interactions as explanatory variables. To investigate larval water loss, we used a linear mixed effects model for the effect of monthly mean temperature and habitat on predicted larval water loss (i.e. proportional loss of body mass per hour). In this analysis, the repeated measures from data loggers were nested within sites and treated as a random factor. For the reaction norm experiment, three-way ANOVAs were used to analyse the effect of developmental temperature, sex and population on each phenotypic trait of interest. Again, full models were fitted initially, before successive removal of non-significant terms. The degree of plasticity was estimated by calculating the effect size (Hedges’ g), using the means and standard deviations of the data from 21 and 27 °C, for each sex and population separately. RESULTS Seasonal phenology and plasticity in the wild Butterfly monitoring revealed that the number of butterflies varied significantly across months (N = 13, Z = −14.65, SE = 0.014, P < 0.001), with a clear absence of adult activity between May and August (Fig. 2). Out of 490 B. safitza recorded, males were overrepresented, with approximately 3:1 ratio in comparison to females. Figure 2. View largeDownload slide Seasonal phenology of Bicyclus safitza. A, monthly numbers of butterflies captured during the longitudinal survey in the Eastern Cape, South Africa (2014–2015) are given in log2 scale, and the red dots represent monthly mean temperatures for a period of 30 years (1980–2009). In the temperate climate zone, a flight period (October–April) was followed by a period in which no active adults were encountered (May–August). B, in tropical climate zones, here represented by data from Zomba in Malawi (1988–1989), adults of B. safitza are actively present throughout the year. Data presented in B were derived from Windig et al. (1994). Figure 2. View largeDownload slide Seasonal phenology of Bicyclus safitza. A, monthly numbers of butterflies captured during the longitudinal survey in the Eastern Cape, South Africa (2014–2015) are given in log2 scale, and the red dots represent monthly mean temperatures for a period of 30 years (1980–2009). In the temperate climate zone, a flight period (October–April) was followed by a period in which no active adults were encountered (May–August). B, in tropical climate zones, here represented by data from Zomba in Malawi (1988–1989), adults of B. safitza are actively present throughout the year. Data presented in B were derived from Windig et al. (1994). Variation in ventral wing pattern morphology (N = 178) of field-collected specimens of South African B. safitza was best explained by the monthly average temperature, while there was also an interaction with site; the Kapriver population showed least variation in the wing patterning (Table 2A). Adult dry mass (N = 230) showed interactions between temperature, sex and site (Table 2B); apart from one site, females were heavier than males, and female dry mass varied significantly depending on the month of capture, whereas it remained similar in males throughout the survey. Table 2. Testing wing pattern (A) and weight (B) responsiveness to seasonal temperature on different sites in Bicyclus safitza Source Estimate SE Test value P-value (A) Wing pattern (principal component 1)† Temperature 0.36 0.05 6.51 < 0.001 Site [Bat] 7.55 6.65 1.13 0.257 Site [Kap] 5.94 2.23 2.66 0.008 Temperature *Site [Bat] −0.40 0.34 −1.17 0.241 Temperature *Site [Kap] −0.29 0.11 −2.56 0.011 (B) Dry mass† Temperature 17.64 3.69 4.77 < 0.001 Sex 65.99 10.28 6.41 < 0.001 Site [Bat] 758.58 457.14 1.69 0.098 Site [Kap] 391.25 136.42 2.86 0.004 Temperature *Site [Bat] −40.15 23.74 −1.69 0.092 Temperature *Site [Kap] −20.64 6.88 −3.00 0.003 Sex* Site [Bat] −158.82 65.02 −2.44 < 0.015 Sex* Site [Kap] 17.70 19.04 0.93 0.353 Source Estimate SE Test value P-value (A) Wing pattern (principal component 1)† Temperature 0.36 0.05 6.51 < 0.001 Site [Bat] 7.55 6.65 1.13 0.257 Site [Kap] 5.94 2.23 2.66 0.008 Temperature *Site [Bat] −0.40 0.34 −1.17 0.241 Temperature *Site [Kap] −0.29 0.11 −2.56 0.011 (B) Dry mass† Temperature 17.64 3.69 4.77 < 0.001 Sex 65.99 10.28 6.41 < 0.001 Site [Bat] 758.58 457.14 1.69 0.098 Site [Kap] 391.25 136.42 2.86 0.004 Temperature *Site [Bat] −40.15 23.74 −1.69 0.092 Temperature *Site [Kap] −20.64 6.88 −3.00 0.003 Sex* Site [Bat] −158.82 65.02 −2.44 < 0.015 Sex* Site [Kap] 17.70 19.04 0.93 0.353 The factors refer to monthly mean temperature (as a continuous variable), site (Kasouga, Bathurst [Bat] or Kapriver [Kap]) and their interaction (*). The reported models are the best models according to the smallest Akaike information criterion (AIC) value. †Intercept includes factor levels: sex [male] and site [Kasouga]. Significant P-values are denoted in bold. View Large Evaporation stress Stable isotopes of oxygen (δ18O) indicated that month and site both influenced evaporation stress as measured from adult butterflies (N = 232; Table 3A). There was also an interaction between the month and site, and the main flight season coincided with a period when δ18O values were lowest (minimum = 20.03), in contrast to the end of the season with highest δ18O values (maximum = 32.84). This effect, however, was not as strong in the Kapriver population. The lowest δ18O values were recorded at Kasouga (x̅ = 24.84, N = 160, SD = 2.57), followed by Bathurst (x̅ = 25.24, N = 7, SD = 2.28), whereas butterflies from Kapriver mirrored a higher and more variable evaporation stress (x̅ = 25.61, N = 76, SD = 2.63). Table 3. Evaporation stress measures: testing evaporation stress as butterfly oxygen isotope values (A) and larval water loss (B) in Bicyclus safitza Source Estimate SE Test value P-value (A) Butterfly water loss† Temperature −3.51 0.34 −10.09 < 0.001 Site [Bat] −40.23 19.64 −2.04 0.041 Site [Kap] −53.00 8.54 −6.20 < 0.001 Temperature*Site [Bat] 2.03 0.94 2.14 0.032 Temperature*Site [Kap] 2.66 0.41 6.34 < 0.001 (B) Larval water loss‡ Habitat [edge] −3.12 × 10−2 8.01 × 10−2 −0.39 0.721 Habitat [open] −2.58 × 10−1 8.01 × 10−2 −3.22 0.045 Temperature −1.41 × 10−2 5.76 × 10−2 −2.45 0.014 Source Estimate SE Test value P-value (A) Butterfly water loss† Temperature −3.51 0.34 −10.09 < 0.001 Site [Bat] −40.23 19.64 −2.04 0.041 Site [Kap] −53.00 8.54 −6.20 < 0.001 Temperature*Site [Bat] 2.03 0.94 2.14 0.032 Temperature*Site [Kap] 2.66 0.41 6.34 < 0.001 (B) Larval water loss‡ Habitat [edge] −3.12 × 10−2 8.01 × 10−2 −0.39 0.721 Habitat [open] −2.58 × 10−1 8.01 × 10−2 −3.22 0.045 Temperature −1.41 × 10−2 5.76 × 10−2 −2.45 0.014 In A, the factors refer to monthly mean temperature (as a continuous variable, temperature), site (Kasouga, Bathurst [Bat] or Kapriver [Kap]) and their interaction (*), whereas in B, the factors are habitat (open, edge or shade), and the random structure accounts for repeated measures from data loggers at sites (i.e. logger nested within site). The reported models are the best models according to smallest Akaike information criterion value. †Intercept includes factor levels: sex [male] and site [Kasouga]. ‡Intercept includes factor levels: habitat [shade] and a nested random structure data, logger within site. Significant P-values are denoted in bold. View Large To explore further how evaporation stress may constrain the butterfly niche, we used the laboratory-established physiological relationship to predict larval water loss with respect to abiotic conditions through the monitoring period. The main effects of month and habitat influenced the predicted larval water loss (N = 8878 climatic measurements; Table 3B). Fluctuations in both temperature and humidity throughout the survey period were less dramatic under the shaded forest canopy than in forest fringes or grasslands (Table 1). Plastic responses to developmental temperature In total, 728 individuals were reared in the temperature reaction norm experiment. We found significant interactions between population and temperature for most life-history traits and wing pattern elements, indicating that populations respond differently to developmental temperature (see Supporting Information Table S2 for all minimum adequate models). For all traits, except for the pupal development time in females, the degree of plasticity was larger in the Ugandan population (Fig. 3), and the phenotypic differences between populations were wider at higher temperatures. Not all traits were equally plastic in their response to developmental temperature. Relative to other wing pattern elements, the plastic response of the large eyespot on the forewing (trait 2) and the width of the ventral bands (traits 4 and 5) was less pronounced in both populations (Fig. 3). Figure 3. View largeDownload slide The degree of plasticity for a suite of phenotypic traits in populations of Bicyclus safitza from temperate South Africa (blue) and tropical Uganda (red). Filled symbols represent the degree of plasticity, calculated as the effects size (Hedges’ g) between 21 and 27 °C in males, while females are represented by lighter shaded symbols. The black circles in the shaded area at the bottom of the figure represent the effect size in Hedges’ g. For details of the icons and codes for the wing pattern elements, see Figure 1. Figure 3. View largeDownload slide The degree of plasticity for a suite of phenotypic traits in populations of Bicyclus safitza from temperate South Africa (blue) and tropical Uganda (red). Filled symbols represent the degree of plasticity, calculated as the effects size (Hedges’ g) between 21 and 27 °C in males, while females are represented by lighter shaded symbols. The black circles in the shaded area at the bottom of the figure represent the effect size in Hedges’ g. For details of the icons and codes for the wing pattern elements, see Figure 1. The relationships between developmental temperature and phenotypic variation in the South African population were linear or continuous for most traits. In contrast, the Ugandan population showed more discontinuous responses to temperature (Fig. 4; Supporting Information Fig. S1). For example, the conspicuousness of the wing pattern elements (PC1) increased dramatically between 23 and 25 °C in the Ugandan population and, as a consequence, all individuals reared at the extremes of the temperature gradient showed a close phenotypic resemblance. The difference in the shape of the reaction norm was even more pronounced when using developmental time as a proxy for environmental variation (Fig. 4). Figure 4. View largeDownload slide Effects of developmental temperature (A, B) and development time (C, D) on the first principal component (PC1) of nine ventral wing pattern elements (see Fig. 1) of Bicyclus safitza butterfly. A and C represent the data from the South African population, whereas B and D represent the Ugandan population. Coloured dots represent the values for individuals reared at 21 (purple), 23 (blue), 25 (pink) and 27 °C (red). In C and D, non-linear sigmoidal curves were fitted and explained > 49 and 80% of the variation in the population from South Africa and Uganda, respectively. Figure 4. View largeDownload slide Effects of developmental temperature (A, B) and development time (C, D) on the first principal component (PC1) of nine ventral wing pattern elements (see Fig. 1) of Bicyclus safitza butterfly. A and C represent the data from the South African population, whereas B and D represent the Ugandan population. Coloured dots represent the values for individuals reared at 21 (purple), 23 (blue), 25 (pink) and 27 °C (red). In C and D, non-linear sigmoidal curves were fitted and explained > 49 and 80% of the variation in the population from South Africa and Uganda, respectively. DISCUSSION We show that a natural population of B. safitza in temperate South Africa displays a different seasonal phenology and reduced developmental plasticity in comparison to a population from a tropical region in Uganda. In the tropics, populations of B. safitza are typically found in wet–dry seasonal habitats, and active butterflies are present throughout the year. They exhibit discrete seasonal forms, one with conspicuous and one with cryptic wing patterns (Windig, 1991; Brakefield & Reitsma, 1991; van Bergen et al., 2016). In contrast, our data show that field-collected specimens in South Africa displayed continuous morphological variation throughout the year, with no individuals showing the large eyespots typical of the wet season form in the tropical regions. Moreover, a flight season, in which adult butterflies were active, was followed by a period when no active adults were encountered. Populations of B. safitza that inhabit tropical environments typically survive the unfavourable conditions of the dry season as semi-active adults that will feed opportunistically on fruit, and freely and continuously enter fruit-baited traps (Brakefield & Reitsma, 1991; Windig et al., 1994). Our results reveal that a population that colonized a more temperate climate region in southern Africa has adopted a different strategy to cope with local environmental conditions. In South Africa, two large activity peaks in November and February, and a smaller increase in butterfly numbers in April, were followed by a quiescent period, in which no butterflies were caught until late September. These data suggest that this tropical butterfly species takes on a different ‘overwintering’ strategy in temperate climates, possibly with butterflies aestivating as fully dormant individuals in shelters or surviving in an arrested stage of pre-adult development. The adjusted phenology, as well as the preference for more shaded habitats at the range margin of the species’ distribution, may be associated with different evaporation constraints. Stable isotopes of oxygen, which in butterflies reflect evaporation rates during the late larval development (van Bergen et al., 2016), indicated that individuals caught at the beginning of the flight season experienced high evaporation rates during development. In contrast, we observed less evaporation stress in the middle of the summer months, and larval evaporation rates rose again closer towards the end of the flight season. The quiescent period of adults could thus reflect coping with evaporation stress as well as targeting the more favourable environmental conditions for reproduction (Brakefield & Larsen, 1984; Brakefield & Reitsma, 1991). Moreover, larval evaporation rates were predicted to be significantly higher in forest fringes and open grassland compared with more shaded habitats. In the shaded forests, the temperature was mild, humidity high and vapour pressure deficit low, which provides a buffer against the weather extremes (Addo-Bediako, Chown & Gaston, 2001; Chown, Sørensen & Terblanche, 2011). Thus, it is possible that the maintenance of the body water balance, together with potential specialization to shade-adapted host plants (Braschler & Hill, 2007; Nokelainen et al., 2016), constrains populations of B. safitza to microclimates provided by shaded habitats in the temperate zone. Our results show that the successful colonization of this tropical butterfly species to more temperate climatic conditions in southern Africa involved changes in the degree of developmental plasticity. Field surveys confirmed the absence of polyphenism in the wild and, when compared in the laboratory with a population from the tropics, the South African population showed more robust development in response to thermal variation for a broad suite of morphological and life-history traits (see Supporting Information Table S1 and Fig. S1). In addition to geographical variation in the degree of plasticity (i.e. the steepness of the reaction norm), we also observed clear differences in the shapes of reaction norms between the two populations (Fig. 3). The expression of ventral wing pattern elements responded in a discontinuous manner to the temperature gradient, which is typical of polyphenism, whereas the relationship between developmental temperature and phenotypic variation in the South African population was more-or-less linear for these traits. The evolution of developmental plasticity in this group of butterflies has been studied by conducting artificial selection experiments using Bicyclus anynana, a species closely related to B. safitza and a model system in the field of eco-evo-devo (Brakefield, Beldade & Zwaan, 2009). Rapid responses to selection were observed with respect to the height of reaction norms (intercept), while developmental plasticity for eyespot size was retained in the selection lines (Brakefield et al., 1996). In contrast, attempts to change the slope (steeper or shallower) or the shape of the reaction norms were largely unsuccessful (Wijngaarden, Koch & Brakefield, 2002), which suggested that the slope of reaction norms is unlikely to evolve as readily as the intercept. Subsequent studies using two different tropical populations of B. anynana from different latitudes revealed parallel reaction norms for a suite of traits, and no obvious genotype-by-environment interactions (de Jong et al., 2010), confirming the results obtained in the laboratory. Moreover, recent work confirms that intrapopulation genetic variation for plasticity is highly depleted in B. anynana (Oostra et al., 2017), which might hinder expansions of this species into environments with different seasonal and ecological dynamics. The striking differences in plasticity among populations described in the present study indicate that natural populations of B. safitza contain sufficient genetic variation in the response to thermal variation for developmental plasticity to evolve. In insects, including B. anynana (Koch, Brakefield & Kesbeke, 1996; Mateus et al., 2014; Monteiro et al., 2015), developmental plasticity is often mediated by endocrine signalling (Nijhout, 1999; Zera, Harshman & Williams, 2007). The evolution of environmentally sensitive traits, as shown here for B. safitza, is likely to have involved evolutionary changes in the levels and timing of systemic hormone titres in response to external cues or in the degree and timing of the sensitivity of hormonal receptors in the developing target tissues (Oostra et al., 2014b). Finally, phenotypic plasticity not only enables organisms to cope with environmental heterogeneity, such as seasonal variation in climatic conditions, but it may also enable dispersal into regions with climates to which organisms are not adapted at source (West-Eberhard, 2003; Wund et al., 2008; Gibert, 2017). Upon exposure to novel environmental variation, plasticity provides an immediate shift in phenotypic variation, leading to increased population persistence and providing time for adaptive evolution to take place. Based on the data from the two populations studied here, we postulate that developmental plasticity of the ancestral population of B. safitza might have facilitated the process of local adaptation to temperate climates. Interestingly, individuals from the extant populations mated and produced viable offspring in the laboratory (E. van Bergen, personal observations). However, hybrid females demonstrated signs of reduced fertility, which is in line with Haldane’s rule: the preferential hybrid sterility of the heterogametic sex. This may imply that the South African population of B. safitza is becoming genetically isolated and may be on its way to evolving full reproductive isolation. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Figure S1. Reaction norm representations of the effect of population, developmental temperature and sex on a suite of phenotypic traits in populations of Bicyclus safitza from Uganda and South Africa, related to Figure 3 and Supporting Information Table S1. Table S1. Extrapolated larval water loss of Bicyclus safitza in Eastern Cape, South Africa. Table S2. Minimum adequate models of the effect of population, developmental temperature and sex on a suite of phenotypic traits in populations of Bicyclus safitza from Uganda and South Africa, related to Figure 3 Supporting Information Figure S1. ACKNOWLEDGEMENTS We thank two anonymous reviewers for their helpful comments. We are grateful to Andrew Balmer, Lea van de Graaf, Dave Osbaldeston and James Rolfe for practical assistance and technical support. All authors conceived and designed the experiments. O.N. and E.v.B. performed the experiments, analysed the data and wrote the manuscript. All authors read and approved the final version of the manuscript for publication. This work was supported by the European Research Council grant EMARES (250325) and the John Templeton Foundation grant #60501 to P.M.B. During the final stages of the project, O.N. was funded through an Emil Aaltonen Foundation’s young researcher grant and Academy of Finland Postdoctoral Fellow Grant (21000038821) and E.v.B. was supported by funding from the Portuguese Foundation for Science and Technology (FCT; ref: PTDC/BIA-EVF/0017/2014). We have no conflict of interest to declare. REFERENCES Aalberg Haugen IM, Gotthard K. 2015. Diapause induction and relaxed selection on alternative developmental pathways in a butterfly. Journal of Animal Ecology 84: 464– 472. Google Scholar CrossRef Search ADS PubMed Addo-Bediako A, Chown SL, Gaston KJ. 2001. Revisiting water loss in insects: a large scale view. Journal of Insect Physiology 47: 1377– 1388. Google Scholar CrossRef Search ADS PubMed Aduse-Poku K, Brattström O, Kodandaramaiah U, Lees DC, Brakefield PM, Wahlberg N. 2015. Systematics and historical biogeography of the old world butterfly subtribe Mycalesina (Lepidoptera: Nymphalidae: Satyrinae). BMC Evolutionary Biology 15: 167. Google Scholar CrossRef Search ADS PubMed Anderson DB. 1936. Relative humidity or vapor pressure deficit. Ecology 17: 277– 282. Google Scholar CrossRef Search ADS van Bergen E, Barlow HS, Brattström O, Griffiths H, Kodandaramaiah U, Osborne CP, Brakefield PM. 2016. The stable isotope ecology of mycalesine butterflies: implications for plant–insect co-evolution. Functional Ecology 30: 1936– 1946. Google Scholar CrossRef Search ADS van Bergen E, Osbaldeston D, Kodandaramaiah U, Brattström O, Aduse-Poku K, Brakefield PM. 2017. Conserved patterns of integrated developmental plasticity in a group of polyphenic tropical butterflies. BMC Evolutionary Biology 17: 59. Google Scholar CrossRef Search ADS PubMed Bohonak AJ, Jenkins DG. 2003. Ecological and evolutionary significance of dispersal by freshwater invertebrates. Ecology Letters 6: 783– 796. Google Scholar CrossRef Search ADS Brakefield PM, Beldade P, Zwaan BJ. 2009. The African butterfly Bicyclus anynana: a model for evolutionary genetics and evolutionary developmental biology. Cold Spring Harbor Protocols 4: 1– 10. Brakefield PM, Gates J, Keys D, Kesbeke F, Wijngaarden PJ, Monteiro A, French V, Carroll SB. 1996. Development, plasticity and evolution of butterfly eyespot patterns. Nature 384: 236– 242. Google Scholar CrossRef Search ADS PubMed Brakefield PM, Larsen TB. 1984. The evolutionary significance of dry and wet season forms in some tropical butterflies. Biological Journal of the Linnean Society 22: 1– 12. Google Scholar CrossRef Search ADS Brakefield PM, Reitsma N. 1991. Phenotypic plasticity, seasonal climate and the population biology of Bicyclus butterflies (Satyridae) in Malawi. Ecological Entomology 16: 291– 303. Google Scholar CrossRef Search ADS Braschler B, Hill JK. 2007. Role of larval host plants in the climate-driven range expansion of the butterfly Polygonia c‐album. Journal of Animal Ecology 76: 415– 423. Google Scholar CrossRef Search ADS PubMed Chown SL, Sørensen JG, Terblanche JS. 2011. Water loss in insects: an environmental change perspective. Journal of Insect Physiology 57: 1070– 1084. Google Scholar CrossRef Search ADS PubMed Ellwood MDF, Northfield RGW, Mejia-Chang M, Griffiths H. 2011. On the vapour trail of an atmospheric imprint in insects. Biology letters 7: 601– 604. Google Scholar CrossRef Search ADS PubMed Gibert JM. 2017. The flexible stem hypothesis: evidence from genetic data. Development Genes and Evolution 227: 297– 307. Google Scholar CrossRef Search ADS PubMed de Jong MA, Kesbeke FMNH, Brakefield PM, Zwaan BJ. 2010. Geographic variation in thermal plasticity of life history and wing pattern in Bicyclus anynana. Climate Research 43: 91– 102. Google Scholar CrossRef Search ADS Koch PB, Brakefield PM, Kesbeke F. 1996. Ecdysteroids control eyespot size and wing color pattern in the polyphenic butterfly Bicyclus anynana (Lepidoptera: Satyridae). Journal of Insect Physiology 42: 223– 230. Google Scholar CrossRef Search ADS Kooi RE, Brakefield PM. 1999. The critical period for wing pattern induction in the polyphenic tropical butterfly Bicyclus anynana (Satyrinae). Journal of Insect Physiology 43: 201– 212. Google Scholar CrossRef Search ADS Larsen TB 2005. Butterflies of West Africa . Brill, Leiden, Netherlands: Apollo Books. Lyytinen A, Brakefield PM, Lindström L, Mappes J. 2004. Does predation maintain eyespot plasticity in Bicyclus anynana? Proceedings of the Royal Society of London B: Biological Sciences 271: 279– 83. Google Scholar CrossRef Search ADS Mateus AR, Marques-Pita M, Oostra V, Lafuente E, Brakefield PM, Zwaan BJ, Beldade P. 2014. Adaptive developmental plasticity: compartmentalized responses to environmental cues and to corresponding internal signals provide phenotypic flexibility. BMC Biology 12: 97. Google Scholar CrossRef Search ADS PubMed Miller L. 1970. Nomenclature of wing veins and cells. Journal of Research on the Lepidoptera 8: 37–48. Monteiro A, Tong X, Bear A, Liew SF, Bhardwaj S, Wasik BR, Dinwiddie A, Bastianelli C, Cheong WF, Wenk MR, Cao H, Prudic KL. 2015. Differential expression of ecdysone receptor leads to variation in phenotypic plasticity across serial homologs. PLoS Genetics 11: e1005529. Google Scholar CrossRef Search ADS PubMed Mucina L, Rutherford MC. 2006. The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19: 1– 30. Nijhout HF 1999. Control mechanisms of polyphenic development in insects. BioScience 49: 181– 192. Google Scholar CrossRef Search ADS Nokelainen O, Ripley BS, van Bergen E, Osborne CP, Brakefield PM. 2016. Preference for C4 shade grasses increases hatchling performance in the butterfly, Bicyclus safitza. Ecology and Evolution 6: 5246– 5255. Google Scholar CrossRef Search ADS PubMed Oostra V, Brakefield PM, Hiltemann Y, Zwaan BJ, Brattström O. 2014a. On the fate of seasonally plastic traits in a rainforest butterfly under relaxed selection. Ecology and Evolution 4: 2654– 2667. Google Scholar CrossRef Search ADS Oostra V, de Jong MA, Invergo BM, Kesbeke F, Wende F, Brakefield PM, Zwaan BJ. 2011. Translating environmental gradients into discontinuous reaction norms via hormone signalling in a polyphenic butterfly. Proceedings of the Royal Society B: Biological Sciences 278: 789– 797. Google Scholar CrossRef Search ADS Oostra V, Mateus AR, van der Burg KRL, Piessens T, van Eijk M, Brakefield PM, Beldade P, Zwaan BJ. 2014b. Ecdysteroid hormones link the juvenile environment to alternative adult life histories in a seasonal insect. The American Naturalist 184: E79– E92. Google Scholar CrossRef Search ADS Oostra V, Saastamoinen M, Zwaan BJ, Wheat CW. 2017. Extensive phenotypic plasticity in a seasonal butterfly limits potential for evolutionary responses to environmental change. bioRxiv . doi: 10.1101/126177. Prudic KL, Stoehr AM, Wasik BR, Monteiro A. 2015. Eyespots deflect predator attack increasing fitness and promoting the evolution of phenotypic plasticity. Proceedings of the Royal Society B: Biological Sciences 282: doi: 10.1098/rspb.2014.1531. Roskam JC, Brakefield PM. 1996. A comparison of temperature-induced polyphenism in African Bicyclus butterflies from a seasonal savannah-rainforest ecotone. Evolution 50: 2360– 2372. Google Scholar PubMed Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, Saalfeld S, Schmid B, Tinevez J-Y, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona A. 2012. Fiji: an open-source platform for biological-image analysis. Nature Methods 9: 676– 682. Google Scholar CrossRef Search ADS PubMed Shapiro AM. 1976. Seasonal polyphenism. In: Hecht MK, Steere WC, Wallace B, eds. Evolutionary Biology , 9th edn. New York, NY: Plenum Press; Springer US, 259– 333. Google Scholar CrossRef Search ADS West-Eberhard MJ. 2003. Developmental plasticity and evolution . New York, NY: Oxford University Press. Wijngaarden PJ, Koch PB, Brakefield PM. 2002. Artificial selection on the shape of reaction norms for eyespot size in the butterfly Bicyclus anynana: direct and correlated responses. Journal of Evolutionary Biology 15: 290– 300. Google Scholar CrossRef Search ADS Windig JJ 1991. Quantification of Lepidoptera wing patterns using an image analyzer. Journal of Research on the Lepidoptera 30: 82– 94. Windig JJ, Brakefield PM, Reitsma N, Wilson JGM. 1994. Seasonal polyphenism in the wild: survey of wing patterns in five species of Bicyclus butterflies in Malawi. Ecological Entomology 19: 285– 298. Google Scholar CrossRef Search ADS Wund MA, Baker JA, Clancy B, Golub JL, Foster SA. 2008. A test of the “flexible stem” model of evolution: ancestral plasticity, genetic accommodation, and morphological divergence in the threespine stickleback radiation. The American Naturalist 172: 449– 462. Google Scholar CrossRef Search ADS PubMed Zera AJ, Harshman LG, Williams TD. 2007. Evolutionary endocrinology: the developing synthesis between endocrinology and evolutionary genetics. Annual Review of Ecology, Evolution, and Systematics 38: 793– 817. Google Scholar CrossRef Search ADS © 2017 The Linnean Society of London, Biological Journal of the Linnean Society
Biological Journal of the Linnean Society – Oxford University Press
Published: Feb 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera