How biotic and abiotic effects colour flowers in a land Down Under

This article is a Commentary on Dalrymple et al. (2020), 228: 1972–1985.

higher than 2500 km away in the southern island of Tasmania. Similarly, inland species of the continent receive about a third of the precipitation compared to coastal species, but exhibit c. 4% more colour contrast.
How then do abiotic stressors impact flower coloration? Production of anthocyanins and flavonols can increase as a response to solar radiation (Falcone Ferreyra et al., 2012), which may lead to more saturated flowers. Pigments can have specific functions in protecting flowers from abiotic stressors. For example, increasing the amount of ultraviolet absorbing pigments can reduce the amount of (harmful) ultraviolet light that is reflected to the pollen (Koski & Ashman, 2015). Alternatively, though not mutually exclusive, is that changes in flower pigmentation are a side effect of the stress response on the plant. Warren & Mackenzie (2001) found that plants with higher anthocyanin levels throughout the entire plant have higher fitness under drought stress, whereas plants with low anthocyanin levels performed better in the well watered conditions. Nonetheless, the amount of pigment cannot fully explain differences in colour contrast as found by Dalrymple et al., as colour contrast is also mediated by the type of pigment and where it is located inside the flower (van der Kooi et al., 2019a). Currently however, we know little about how the synthesis of other floral pigments, such as carotenoids and betalaines, relate to abiotic stress.
Australia is well known for its temperature variation and extremes. An intuitive role of colour with regard to temperature is that flowers with darker hues will convert more radiation to heat than flowers with light hues, although this may be more clearly linked to solar radiance than ambient temperature. Further, (chemical) developmental processes of plants as well as pollinator foraging activities are temperature dependent (van der Kooi et al., 2019b). Perhaps surprising, therefore, is that in their large dataset Dalrymple et al. found no association between ambient temperature and colour contrast. The absence of an effect may be because colour contrast does not directly scale with pigmentation in a simple way. The optical properties of pigments that produce salient colours like yellow that contrast well with a (green) background may only require a bit of pigmentation, and increasing pigmentation to increase photon catch (and so passive heating of the flower) may actually decrease visual colour contrast to the background (van der Kooi et al., 2019a). It is further likely that plant physiological processes related to thermal tolerance will play a key role here also.
Many flowers display colour patterns, which can function as a guide for pollinators to find the pollen and nectar. Dalrymple et al. flower colour patterns converged with increasing plant species diversity. With increasing (pollinating) bird species richness, colour patterns also become more similar in hue; colour patches shift to orange-red wavelength ranges instead of the more contrasting orange-purple. There was no clear effect of insect community diversity on within-flower colour contrast.
At least three nonmutually exclusive factors can explain the observed convergence in flower colour patterns. First, floral signal convergence could be explained by increased sharing of pollinators in diverse communities (i.e. more generalized pollination), because plants with more generalized pollination systems are more successful in such environments than specialists. Convergence of floral (visual) signals will then be particularly useful when coflowering plants are neighbours andfrom a distanceconstitute a large floral display. Second, other pollinator-attracting signals, such as scent, size or shape, may become increasingly important in diverse communities. Third, pollinators may use fine colour differences more in highly diverse communities, meaning that (small) differences in colour are more important in species-rich than species-poor communities. Testing the relative importance of these different predictions requires detailed characterization of pollinators in the focal communities, which will be challenging owing to the vast geographic range. In this regard, within Australia, another recent community study revealed flower colour signals tend to converge towards the visually mediated colour preferences of key insect pollinators such as bees or flies (Shrestha et al., 2019).
Adding to the complexities of comparing biotic and abiotic influences at a continental scale was Dalrymple et al.'s observation of no clear effect of insect community diversity on within-flower colour contrast. Insect colour vision greatly varies between taxa, and insect foraging can further be guided by colour preferences and achromatic (green) contrast (Fig. 1, van der Kooi et al., 2019a). Together, this suggests that particularly for insect-pollinated flowers, understanding the evolution of flower colours requires incorporating pollinator visual systems. Dalrymple et al. did not include different visual systems in their analysis, which is legitimate given the breadth of their research and our current poor knowledge of colour vision in many (Australian) pollinators.
Another factor that determines the visual signal of flowers is the amount of the reflected light, which humans perceive as brightness. The amount of the reflected light depends on both pigments and flower structure. Light is reflected by structures inside the flower (e.g. cell walls, air gaps, vacuoles), and reflected light is modulated by wavelength-selective absorption by floral pigments (van der Kooi et al., 2019a). To humans, for example, white and yellow flowers appear brighter than red flowers. Dalrymple et al. found that the amount of light reflected by flowers increases with vegetation cover (quantified using leaf area index). In this regard the findings by Dalrymple et al. differ from those from Binkenstein & Schaefer (2015), who found no difference between colours of flowers between forests and open grasslands in Germany. The amount of the reflected lightindependent of the colouris unlikely to be important for many (diurnal) insect pollinators; behavioural tests suggest these insects do not process 'brightness' signals (for review, see van der Kooi et al., 2019a). Perhaps the increased floral reflectance in Australian forest-dwelling species is because of structural differences in flower anatomy that are related

Forum
New Phytologist to mechanical aspects or flower longevity more than the colour signal.
The work by Dalrymple et al. made clear that both biotic and abiotic factors in studies on the evolution of flower colour are important, and should be investigated in floras on other continents. The main strength of the paper of Dalrymple et al. is their impressive taxonomic and geographic coverage (339 species at several habitats). However, not uncommon in such taxonomically vast studies, is that their breadth comes at the expense of detail. As acknowledged by Dalrymple et al., multiple significant results in this study have a low effect size. We also need a more sophisticated map of the ecological network of plant and pollinator species to understand the effects of pollinator diversity and different pollinator species on floral coloration on both local community level and a broader scale. Future research will no doubt illuminate how biotic and abiotic effects contribute to colouring the flora Down Under and in other parts of the world.