Understanding how organisms adapt to complex environments is a central goal of evolutionary biology and ecology. This issue is of special interest in the current era of rapidly changing climatic conditions. Here, we investigate clinal variation and plastic responses in life history, morphology, and physiology in the butterfly Pieris napi along a pan-European gradient by exposing butterflies raised in captivity to different temperatures. We found clinal variation in body size, growth rates and concomitant development time, wing aspect ratio, wing melanisation, and heat tolerance. Individuals from warmer environments were more heat-tolerant, had less melanised wings and a shorter development but still they were larger than individuals from cooler environments. These findings suggest selection for rapid growth in the warmth and for wing melanisation in the cold, and thus fine-tuned genetic adaptation to local climates. Irrespective of the origin of butterflies, the effects of higher developmental temperature were largely as expected, speeding up development, reducing body size, potential metabolic activity, and wing melanisation, while increasing heat tolerance. At least in part, these patterns likely reflect adaptive phenotypic plasticity. In summary, our study revealed pronounced plastic and genetic responses, which may indicate high adaptive capacities in our study organism. Whether this may help such species though to deal with current climate change needs further investigation, as clinal patterns have typically evolved over long periods.

For this study, we collected freshly-eclosed, spring generation females from a total of nine populations along a latitudinal gradient from northern Italy to Sweden (Fig. 1). We sampled three replicate populations each in northern Italy (I: Torino 45.11°N / 7.48°E, Pavia 45.21°N / 9.27°E, Mantova 45.21°N / 10.75°E), northern Germany (G: Wahrenholz 52.64°N / 10.61°E, Rathenow 52.65°N / 12.44°E, Strausberg 52.60°N / 13.86°E), and Sweden (S: Örebro 59.29°N / 15.01°E, Eskilstuna 59.36°N / 16.54°E, Stockholm 58.95°N / 17.58°E). Mean annual (Italy: 13ºC, Germany, 9ºC, Sweden: 6ºC) and the mean temperature during the vegetation period (May to September; Italy: 21ºC, Germany: 16ºC, Sweden: 14ºC) followed a temperature gradient across latitudes while precipitation is higher in Italy than in Germany and Sweden (Supplementary Table S1). The minimal straight distance between two populations was 73 km, the total latitudinal gradient spanned ca. 1660 km. We collected a total of 74 females from Italy, 94 from Germany, and 76 from Sweden between 19^th of April and 14^th of June 2016. All females were afterwards transferred to Greifswald University for egg-laying.

Experimental design

Field-caught females were kept individually in translucent 1 litre plastic pots covered with gauze, which were placed into a climate chamber set at a constant temperature of 25°C, 65 % relative humidity, and a photoperiod of L18:D6. Females where fed ad libitum with water, a 10% sugar solution, and additionally flowers (e.g. Sambucus nigra, Taraxacum spec., Senecio spec.). For oviposition, they were provided daily with a fresh cutting of A. petiolata. The resulting eggs were collected daily, counted, and kept separated by oviposition day and female under the above conditions until hatching. Eggs were collected until the death of the females. After hatching, all larvae were transferred individually to translucent plastic pots (250 mL) lined with moist tissue and cuttings of A. petiolata ad libitum. Host-plants were replaced on a daily basis. On day 3 after hatching, larvae were randomly divided among two rearing temperatures (18 and 25°C), using a split-brood design. These temperatures were chosen to mimic summer conditions (July) in Germany / Sweden and Italy (Supplementary Table S1). Larval rearing took place in climate cabinets (Sanyo MLR-351H; Bad Nenndorf, Germany) at 60% relative humidity and L18:D6. For each individual, we recorded larval development time, pupal development time, and pupal mass (one day after pupation, KERN ABJ-120-4M; 0.1 mg accuracy). Larval growth rate was calculated as the natural logarithm of pupal mass / larval time. One day after adult eclosion, butterflies were subjected to a heat knock-down assay. Therefore, they were individually placed into translucent plastic pots (100 ml) in a randomized block design, and exposed to a constant temperature of 43°C (climate cabinet Sanyo MIR-553; Bad Nenndorf, Germany). The time until physical knock down, characterized by an inability to move in a coordinated manner, was recorded. Butterflies were afterwards frozen at -80°C for later analyses.

Frozen butterflies were later thawed and their adult fresh mass was measured with a digital scale (KERN ABJ-120-4M). Then, the head, wings and legs were removed, and the thorax and abdomen were separated before being weighed. The thorax-abdomen ratio was calculated as the thorax mass divided by the abdomen mass. Butterfly wings were used to measure wing morphology and melanisation. Therefore, we took a photograph of one dorsal fore- and one ventral hindwing per individual under standardized illumination with a PC microscope camera (Veho MS-004 Discovery Deluxe USB Microscope). The ventral hindwing was used as its melanisation is known to influence heat gain during lateral basking (Heinrich, 1996), whilst dorsal forewing melanisation influences the heat gain during dorsal basking (Kingsolver, 1987). To score wing area, we used the “lasso” function in Adobe Photoshop CS6. Wing melanisation was defined as the percentage of black wing area (Supplementary Figure S2). We used a threshold approach with a fixed value of 128 (on a scale of 0 to 255), turning each pixel on the butterfly wing into either black or white (Adobe Photoshop CS6). We additionally calculated fore- to hindwing ratio (forewing area divided by hindwing area), wing loading (total body mass divided by forewing area), and wing aspect ratio to examine wing shape (4 x forewing length² divided by forewing area; Berwaerts et al., 2002).

Physiological parameters

We measured the following parameters related to oxidative stress: (1) potential metabolic activity (PMA), (2) two markers of antioxidant defences: total hemolymph antioxidant capacity (OXY) and glutathione (a non-enzymatic antioxidant; GSH), and (3) three markers of oxidative damage: hydroperoxide concentration (a Reactive Oxygen Metabolite; ROM), malondialdehyde concentration (a marker of lipid peroxidation; MDA), and DNA damage (8OHdG). To keep the size of the experiments manageable, PMA, OXY and ROM were only measured in males. Frozen abdomen were cut into two similar-sized parts and weighed (± 0.1 mg). The first half was used to measure PMA, while the second half was used to measure OXY and ROM. For measuring GSH, MDA, and 8OHdG, we used thoraces and abdomen from other males and females. For details on measuring physiological parameters, see Supplementary Material S3.