Introduction

The evolution and diversification of floral traits are driven by both biotic and abiotic agents1,2,3. Historically, morphological differences in floral traits have been attributed to pollinator attraction and manipulation4,5. However, such factors as herbivory and environmental conditions may also play a strong role6,7. For example, anthocyanin-based polymorphic petal color may increase fitness during drought or periods of high heat1,8. Among the most well-studied agents is rain, which can wash away pollen grains, reduce pollen viability and dilute nectar9,10,11,12. Plant species that do not produce water-repellent pollen grains commonly exhibit floral traits that protect against pollen degradation due to wetting, suggesting that rain is an important selective force in shaping floral structure10,13,14,15. Ultraviolet-B (UV-B) is known to reduce pollen viability16,17,18,19 and strong UV-B radiation is a typical characteristic of alpine environments; however, the relationship between pollen sensitivity and protective floral forms remains largely unexplored13.

Although rain protection is likely a key driver of floral forms, it may not fully explain the floral evolution of alpine plants for two reasons. First, traits such as flower closure induced by temperature decline (seen in alpine gentians) protect pollen from rain but not from UV-B9,20. Second, traits such as pendulous flowers may be under simultaneous selection by both rain and UV-B21,22. In this study, we seek to clarify the selective pressure of UV-B on the floral traits of alpine plants using in vitro pollen germination experiments. We hypothesized that unprotected pollen grains located on UV-B exposed anthers (Fig. 1 A–F) would be less sensitive than pollen grains protected by flower structures such as bracts or petals (Fig. 1 G–L). We addressed the following specific questions: 1) What is the ability of flower structures to reduce UV-B exposure? 2) What is the difference in viability of protected and unprotected pollen after exposure to UV-B radiation?

Figure 1
figure 1

Flower structures defined as exposing (A–F) and protecting (G–L) the pollen of several representative plant species from the high altitudes of the Qinghai-Tibet Plateau and adjacent areas (Photographs by Y.W.D.).

(A). Gentiana straminea; (B). Codonopsis convolvulacea; (C). Dipsacus asperoides; (D). Sinopodophyllum hexandrum; (E). Rhododendron decorum; (F). Meconopsis racemosa; (G). Adenophora khasiana; (H). Delphinium delavayi; (I). Anisodus luridus; (J). Salvia digitaloides; (K). Rheum alexandrae; (L). Roscoea cautleoides.

Full size image

Results

More than 70% (74.0 ± 2.5%, mean ± S.E.) of UV-B radiation was excluded by flower structures. The blocking ability of the material composing protective or potentially protective structures on both types of flowers was similar (Fig. 2A).

Figure 2
figure 2

The effects of UV-B on in vitro pollen germination.

(A) Proportional reduction in UV-B by floral structures relative to control. (B) Proportional reduction in pollen viability after exposure to UV-B relative to control. The letters indicate that the difference was significant at the P = 0.01 level and the error bars indicate S.E. The number in each bar is the species sample size.

Full size image

Pollen viability, measured by pollen germination, was similar in the control and UV-B treated pollen in 15 of the 42 plant species studied. In the remaining 27 species, pollen viability was reduced significantly after exposure to UV-B light (Table S1). Among these species, pollen from flowers with protective structures experienced a nearly 50% reduction in viability relative to the control, whereas pollen from exposed flowers exhibited only a 20% reduction (Fig. 2B). In general, pollen viability decreased more in pollen from flowers with protective structures and the reduction in viability between protected and exposed pollen was significant (Fig. 2B). We also found a significant positive relationship between more protective flower structures and greater pollen sensitivity to UV-B (χ2 = 8.01, P = 0.011; Table 1).

Table 1 The association between floral protective structure and pollen sensitivity to UV-B
Full size table

Discussion

The UV-B levels on the Qinghai-Tibet Plateau and adjacent highlands are much higher than in many low-altitude places and they are at their highest during the flowering season (May–August)23,24. Thus, UV-B has the potential to inflict considerable damage on these reproductive structures, resulting in decreased pollen viability and lowered fitness of alpine plants16,17,18,19,25. Most of the alpine plants we tested contained pollen that was sensitive to UV-B and pollen that was more sensitive to UV-B was also more likely to belong to species that featured protective floral structures. This observation suggests that UV-B has played a role in shaping the flower structure of alpine plants in the Qinghai-Tibet Plateau and adjacent highlands.

Three caveats should be considered. First, flower structures may protect pollen that is sensitive to UV-B through shading, but it remains unclear how exposed pollen grains protect themselves from damage caused by UV-B. It is possible that these pollen types have an increased pollen wall thickness26 or contain isozymes or antioxidants27. Future research should investigate the mechanisms responsible for UV-B protection in these pollen grains. Second, pollen sensitivity to UV-B might be different in plant species that utilize different pollination systems. Our study only considered pollen from entomophilous plants, which we expect to be more sensitive to UV-B than pollen from anemophilous plants, such as the wind-pollinated Aconitum gymnandrum28, because insect-pollinated species are less exposed. Third, strong UV-B may affect pollen tube growth even after the grains germinate, but our in vitro experiments were limited to quantifying the effect of UV-B on pollen germination16,17. Although the wall of the style and the receptacle tissue surrounding the ovary are likely to form protective barriers against UV-B radiation, this type of protection has not been rigorously tested in alpine plant communities29.

Our work provides compelling evidence that flower structures can protect sensitive pollen from UV-B damage, indicating that UV-B radiation may play an important role in influencing the evolution of floral traits. Adaptable morphology in response to UV-B radiation may become increasingly relevant for these alpine communities as ozone levels decrease and more UV-B reaches the surface of the earth30. The impact of such global changes on floral evolution and/or plant distribution may be particularly evident in extreme alpine environments such as the Qinghai-Tibet Plateau.

Methods

Study species

Our experiments were conducted from June to September 2011 in the Lijiang Forest Ecosystem Research Station (27°00′ N, 100°10′ E, 3,250 m) south of Jade Snow Mountain, Yunnan Province, China. Most of the plant species in this location are native, but some have been introduced from other areas of Yunnan Province. We only included native or introduced plants from high-altitude sites that represent alpine communities.

Effectiveness of flower structures in preventing UV-B

We divided plant species into two categories: plants with floral structures such as bracts or petals that could protect pollen from UV-B exposure and those that lacked such protective structures (Fig. 1). We then measured the effectiveness of flower structures in blocking UV-B.

For protective flowers, to test the proportion of UV-B excluded by flower structures, we used a digital portable UV radiometer (UV-B type, Photoelectric Instrument Factory of Beijing Normal University, Beijing, P. R. China) to examine the intensity of UV-B under and outside the flower parts (e.g., bracts, petals, sepals) that sheltered the anther from UV-B. For exposed flowers, we used the level of UV-B measured near the flowers as the exposure level of UV-B and placed the receptive area of the UV-B meter under the flower parts corresponding to the sheltering structures in the protective flowers to measure the proportion of UV-B that could be blocked. For plant species with large flowers, we measured the UV-B intensity in situ; for small flowers, we removed the flower structure from the plant and placed it immediately on the receptive area of the meter.

In total, we examined 29 plant species, 16 of which we classified as protective of the pollen grains (Table S1). We quantified the effectiveness of flower structures in preventing UV-B using the formula 1 − Nu/No, where Nu and No were the intensity of UV-B under and outside the flower structures, respectively. The difference in the effectiveness of flower structures that protect and structures that expose pollen was analyzed using independent-sample Student's t-tests after performing a power transformation using the Box-Cox method.

Effects of UV-B on pollen viability

We used in vitro pollen germination to determine pollen viability. The pollen grains were harvested from newly opened anthers from at least 10 plants of each species; we used pollen from 42 plant species in 20 families (Table S1).

The anthers were placed in a centrifuge tube with 10 ml of 10% sucrose solution. After being thoroughly shaken to dislodge the pollen grains, the anther tissue was removed. The remaining solution was divided equally between two 9-cm-diameter Petri plates, each with a solid medium containing 15% sucrose, 1.5% bacto-agar, 0.01% H3BO3, 0.03% Ca(NO3)2·4H2O, 0.02% MgSO4·7H2O, 0.01% KNO3 and 0.01% KH2PO431. One plate was designated the control and the other plate was exposed to UV-B light with an intensity of 100 μW/cm2, similar to the intensity of UV-B during one hour of exposure to natural sunlight at the station. All plates were placed in an incubator at 25°C with visible light for eight hours. Upon removal, germination was halted on all plates by adding several drops of killing and preserving solution. This solution consisted of water, glycerine, formaldehyde and glacial acetic acid at a ratio of 72:20:5:3 by volume.

Pollen germination was determined using a light microscope at 100× magnification. At least 1000 pollen grains were scored for each plate. The germination rate for each plant species was analyzed using independent-sample Student's t-tests after power-transformation using the Box-Cox method. The pollen sensitivity was quantified as the proportional reduction in pollen viability relative to the control after exposure to UV-B using the formula 1 − Nu/Nc, where Nu and Nc were the germination rates of the pollen grains exposed to UV-B and of the control grains, respectively. For each species, pollen was classified as sensitive if the in vitro germination was significantly reduced relative to the control. The association between flower structures and pollen sensitivity was tested by chi-square in a cross table.