Fly pollination drives convergence of flower coloration

Abstract

Plant-pollinator interactions provide a natural experiment in signal evolution. Flowers are known to have evolved colour signals that maximise their ease of detection by the visual systems of important pollinators such as bees. Whilst most angiosperms are bee pollinated, our understanding on how the second largest group of pollinating insects, flies, may influence flower colour evolution is limited to the use of categorical models of colour discrimination that do not reflect the small colour differences commonly observed between and within flower species. Here we show by comparing flower signals that occur in different environments including total absence of bees, a mixture of bee and fly pollination within one plant family (Orchidaceae) from a single community, and typical flowers from a broad taxonomic sampling of the same geographic region, that perceptually different colours, empirically measured, do evolve in response to different types of insect pollinators. We show evidence of both convergence among fly-pollinated floral colours but also of divergence and displacement of colour signals in the absence of bee pollinators. Our findings give an insight into how both ecological and agricultural systems may be affected by changes in pollinator distributions around the world.

Sigmoidal-shaped Colour Discrimination Functions of the Humming Bird Hawkmoth (Macroglosum Stellatarum)

“…These sigmoidal shaped functions have now been demonstrated to reliably predict colour choices in hawkmoths (Fig. 2), bees (Garcia et al 2017) and a variety of other animal pollinators (Garcia et al 2021(Garcia et al , 2022, and physiological evidence suggests that the way neurons respond to perceptually (dis)similar colour stimuli (Komatsu and Ideura 1993; Dyer and Chittka 2004c) is a likely explanation for such observations. Initial estimates of colour discrimination thresholds for M. stellatarum (Telles et al, 2016) were based on the receptor-noise threshold model (Vorobyev and Osorio 1998) which assumes that colour discrimination is limited by purely physiological aspects of vision namely, the xed signal to noise ratio imposed by the density and number of photoreceptor available in the visual system of the observer (Vorobyev et al 2001).…”

“…We tted two sigmoidal functions (Equations 2 and 3) of different complexity to each dataset, as the shape of these functions has been shown to successfully describe colour discrimination by hymenopterans and dipteran pollinators (Garcia et al 2017(Garcia et al , 2018(Garcia et al , 2022.…”

“…In Equations 2 and 3, the variable π denotes the probability of accurate discrimination of the colour difference (ΔC). We expressed ΔC as the Euclidean distance of two loci in the Maxwell triangle, although other metrics can also be implemented (Garcia et al 2017(Garcia et al , 2018(Garcia et al , 2022, and the constants de ne the unique shape of the function.…”

“…Finally, we modelled chromatic adaptation by M. stellatarum using a standard von Kries type scaling function (Garcia et al 2017). Colour difference between reference and each tested stimuli pair is here expressed as the Euclidean distance between their respective loci in the Maxwell colour space, although the formulation of colour discrimination function as a continuous function can accommodates other colour distance metrics; see for example (Garcia et al 2017(Garcia et al , 2018(Garcia et al , 2022. All colour modelling was done using custom written scripts for Matlab release 2017b.…”

The genetic mechanisms underlying the convergent evolution of pollination syndromes in the Neotropical radiation of Costus L.

“…Convergent shifts in floral features resulting in observed shifts in pollinators are prevalent in the literature, including changes in corolla shape in Iochrominae (Solanaceae; Smith and Kriebel, 2018 ) and changes in stamen and stigma shape in Salvia L. (Lamiaceae; Kriebel et al, 2020 , 2021 ). While many studies have investigated discrete shifts in flower color in relation to pollinators ( Smith et al, 2008 ; McCarthy et al, 2015 ; Roguz et al, 2020 ; Garcia et al, 2021 ), evidence of significant pollinator-mediated selection on continuously varying flower color is lacking ( Trunschke et al, 2021 ). Fewer studies have looked at the molecular evolution of genes responsible for pigment production, including anthocyanins ( Lu and Rausher, 2003 ; Ho and Smith, 2016 ; Wheeler et al, 2021 ) or carotenoids ( Livingstone and Anderson, 2009 ; Clotault et al, 2012 ; Cao et al, 2021 ).…”