Ancient insect vision tuned for flight amongst rocks and plants underpins natural flower colour diversity - rock, mineral, stick, bark, leaf, bird- and insect-flower petal reflectance spectra

Understanding the origins of flower colour signalling to pollinators is fundamental to evolutionary biology and ecology. Flower colour evolves under pressure from visual systems of pollinators, like birds and insects, to establish global signatures among flowers with similar pollinators. However, an understanding of the ancient origins of this relationship remains elusive. Here, we employ computer simulations to generate artificial flower backgrounds assembled from real material sample spectra of rocks, leaves, and dead plant materials, against which to test flowers’ visibility to birds and bees. Our results indicate how flower colours differ from their backgrounds in strength, and the distributions of salient reflectance features when perceived by these key pollinators, to reveal the possible origins of their colours. Since Hymenopteran visual perception evolved before flowers, the terrestrial chromatic context for its evolution to facilitate flight and orientation consisted of rocks, leaves, sticks, and bark. Flowers exploited these pre-evolved visual capacities of their visitors, and in response evolved chromatic features to signal to bees, and differently to birds, against a backdrop of other natural materials. Consequently, it appears that today’s flower colours may be an evolutionary response to the vision of diurnal pollinators navigating their world millennia prior to the first flowers.

Methods

Material samples, data collection
Background materials data (non-floral surfaces): rocks, minerals, sand, shells, dry leaves, dry bark, wood, dry seeds, and green leaves.

Background surface data samples were collected across a 2400 km range from the tropical north (16.9°, 145.7°) to the temperate southern tip (39.1°, 146.2°) of mainland Australia.
N = 507 total natural background surface data samples were collected: n = 65 green leaves; n = 96 dry leaves, bark, wood etc. samples; n = 346 rocks/mineral samples.

Flower data (petal surfaces): insect- and bird-pollinated flowers.

Flower data has previously been described and analysed 9,29 and is made available here.

Spectral reflectance measurement
Reflectance curves were measured for background material and floral samples from 300-700 nm (see below – Marker point calculations) using a spectrophotometer with quartz optics and a PX-2 pulsed xenon light source (USB2000+, Ocean Optics Inc., Dunedin, FL, USA) attached to a computer running SPECTRA SUITE software. Reflectance profiles were measured relative to a Lambertian PTF WS-1 reflectance standard (Ocean Optics, USA). Multiple readings, usually three, were taken from within a square region of each surface (see associated article, Fig. 1B for examples). Where sample surfaces were conspicuously heterogeneous to human visual systems several sections were sampled individually and used in this study. In the case of fresh green leaves specifically, one measurement was taken near the base, one in the middle, and one near the tip of each. These three spectra were then used to calculate an average reflectance spectrum for the leaf.

To assist us in making accurate spectrophotometer readings, we built a black, curved-wall sample enclosure and covered this with a black cardboard lid. The lid was perforated with a tiny hole through which a fibre-optic light was channelled which prevented ambient light from hitting the sample during measurement. Room lighting was turned off. The optical fibre was held in an aluminium block that also helped eliminate stray illumination. The fibre was held approximately 6 mm above each sample, illuminating a circular patch 3-4 mm in diameter. Sample data points recorded between 300 and 700 nm in wavelength that represent reflectance spectra for three background surfaces are illustrated (see associated article, Fig. 1B).

Marker point calculations
Marker points are locations on surface spectral reflectance curves centrally located within sudden changes in spectral reflectance (see associated article, Fig. 1B) 14, SuppRef-1. The severity of the jump is measured as a threshold over which it occurs from its base reflectance to its peak.  For this study, marker points were identified as the midpoint of any change in reflectance of at least a pre-specified percentage threshold value, occurring within a wavelength range of < 50 nm. We calculated thresholds for each background sample at 5% reflectance jump and for each flower sample at 20% reflectance jump, in order to compare the properties of different materials. The 20% threshold for flower spectra was taken from the literature as being suited to such evolutionarily enhanced signals16. The 5% threshold for natural backgrounds that do not constitute evolutionarily enhanced reflectance curves was determined empirically to provide a similar mean number of marker points per sample as the 20% threshold had done for the flowers (see associated article, Supp Table 1).

Marker point calculations were performed using the Open Source Spectral-MP softwareSuppRef-1 with parameters: Threshold 5% (backgrounds), 20% (flowers); Range 50 nm; Smoothing window 21 points; Lookahead 5 points; Interval 300-700 nm. The study interval of 300-700 nm was selected to ensure we analysed regions in which the spectrophotometer readings were reliable, noting that the extremities of these regions are outside the range of sensitivity of avian and hymenopteran pollinator vision (see associated article, Fig. 1B).

References
(Citation numbering refers to original article bibliographic data, reproduced here without change for clarity)

9. Shrestha, M., Dyer, A. G., Boyd-Gerny, S., Wong, B. B. M. & Burd, M. Shades of red: bird pollinated flowers target the specific colour discrimination abilities of avian vision. New Phytologist 198, 301–310 (2013).

14. Chittka, L. & Menzel, R. The evolutionary adaptation of flower colors and the insect pollinators' color vision systems. Journal of Comparative Physiology A 171, 171-181 (1992).

16. Dyer, A. G. et al. Parallel evolution of angiosperm colour signals: common evolutionary pressures linked to hymenopteran vision. Proc. Royal Soc. London B 279, 3605-3615 (2012).

29. Burd, M., Stayton, C. T., Shrestha, M. & Dyer, A. G. Distinctive convergence in Australian floral colours seen through the eyes of Australian birds. Proc. R. Soc. B 281 (2014).

SuppRef-1. Dorin, A., Shrestha, M., Herrmann, M., Burd, M. & Dyer, A. G. Automated calculation of spectral-reflectance marker-points to enable analysis of plant colour-signalling to pollinators. MethodsX 7, 1-9, https://doi.org/10.1016/j.mex.2020.100827 (2020).