Photography
We photographed 684 butterfly specimens from 343 species held in the Lepidoptera collections at the Natural History Museum in London (NHMUK). This includes species from all six families, covering approximately 70 % of all species found in Europe. The distribution of analysed species ranged from 34° to 70° N and 10° W to 44° E. The mean temperature over a species’ distribution varied from -2.7 to 18 °C, and the mean precipitation varied from 353 to 1544 mm per year. For each species, we selected and photographed two specimens whenever available and chose specimens with well-preserved wings and body. For the polymorphic species, we photographed each morph once. We focused on capturing inter- rather than intra-specific variation because we were primarily interested in evolutionary factors that have shaped broad-scale interspecific ecogeographic patterns. We note that most specimens were collected before 1980; however a previous study found no effect of specimen age on visible and near-infrared reflectance in butterflies (Munro et al. 2019) and any potential degradation will contribute variation within species but is unlikely to mask interspecific variation or affect biological conclusions at the scale of our analysis.
Photography was done in a dim room using two light bulbs simultaneously: an LED bulb (True-light LED 12W E27, Frankfurt, Germany CRI index 98, spectral power distribution for this bulb is provided in Fig. S1) and a 3,000K tungsten-halogen lamp (150W, Long Life Lamp Company, Harrow, UK). The tungsten-halogen lamp emitted all UV, visible, and NIR light. We set up these two bulbs approximately 60 cm above the photographic spot. A full-spectrum converted DSLR camera (Nikon D7000 converted by Lifepixel, Mukilteo, WA, USA; maximum) was set directly above the photographic area. We positioned the camera slightly below the bulbs to avoid light reaching directly to the camera lens (Jenoptik UV-VIS-IR 60 mm 1:4 APO Macro lens; transmission waveband is between 290 to 1500 nm). We used lens filters (Baader, Mammendorf, Germany) to capture ultraviolet (U-Venus filter), visible (UV/IR cut filter), and near-infrared wavelengths ranges (IR-Pass filter). These filters completely blocked wavelengths outside the transmission range and enabled us to photograph each specimen in three spectral ranges: ultraviolet (320 – 380 nm), visible (400 – 680 nm), and near-infrared (670 – 1050 nm; 1050 nm is the maximum sensitivity of the camera sensors provided by the manufacturer) ranges, respectively. Our analysed spectral range (320 – 1050 nm) captures approximately 80% of the energy in solar irradiation. We changed lens filters with a minimum disturbance to the camera body using a combination of magnetic lens adapters and filter holders (Manfrotto, Cassola, Italy).
To fix each specimen, we used a square paper box floored with styrofoam (17 x 10 x 5 cm). We placed each specimen at the centre with a 99% reflectance standard (WS-1-SL, Labsphere, NH, USA) on the upper-right corner of the box. Camera settings were constant (ISO 400, F8) except for the shutter speed which varied depending on the type of filters (1.3 s for ultraviolet, 1/500 s for visible, 1/800 s for near-infrared photos). This produced 4,104 images saved in raw format (see Fig. S2 for sample images).