Flower surface is warmer in center than at edges in alpine plants: evidence from Qinghai-Tibetan Plateau

Abstract

Although flower temperature plays an important role in plant reproduction, how it varies spatially on the flower surface is unclear, especially in alpine plants. To characterize spatial variation in flower surface temperature, we examined thermal images of flowers of 18 species along an altitudinal transect from 3200 to 4000 m on Lenglong Mountain on the north-eastern Qinghai-Tibetan Plateau. The surface temperature varied considerably within a flower or floral unit in all plants under sunlight, and was about 1 °C with a maximum of 11 °C higher in the center than at the edges. Solar radiation and flower shape significantly affected the temperature range and standard deviation and the ratio of flower center to edge temperature. The spatial variability of temperature increased with flower size. Flowers in the Asteraceae had higher surface temperatures, greater spatial variability of temperature, and consistently higher and more stable temperatures in the center than at the edge. The ratio of flower center to edge temperature increased with altitude in most species. Heat buildup at the flower center is likely to be widespread in alpine plants; further studies are needed to explore its ecological and evolutional roles.

摘要

高山植物花中央的温度比边缘高：来自青藏高原的证据

花表面温度对植物繁殖有重要作用，但花表面温度的空间分布规律仍不清楚，尤其是在高山植物中。为探讨花表面温度空间分布，我们分析了青藏高原东北部冷龙岭海拔3200至4000 m的18种植物的热成像照片，发现在直射辐射环境下花或花序表面温度空间变化很大，大多数花或花序的中央温度比边缘高出约1 °C，最大可达11 °C。太阳辐射和花的形状显著影响花表面温度的空间分布，如温度的空间变化范围、空间变化的标准差以及花中心与边缘部分的温度比值。花冠越大，表面温度的空间变异性越高。菊科植物表面花温及其空间变异性都比较高，花中央温度比边缘温度也比其它观测的物种更高更稳定。大多数物种花中央与边缘部分温度的比值随海拔上升而增大。这些结果表明，在高山植物中花或花序中央区域的热积累可能很普遍，其生态和进化意义有待进一步探索。

INTRODUCTION

Plants are in general considered to be poikilothermic organisms, and cannot maintain a constant body temperature independent of their thermal environment (e.g. Körner 2006). There is no concept of ‘body temperature’ as used for homeotherms that can be used to assess plant temperature. The temperature of an individual plant, as a sessile poikilotherm, varies among parts or organs, depending largely on microenvironment (Guo et al. 2022; Little et al. 2016).

In comparison with that of a leaf, the surface temperature of a single flower or a floral unit, such as the involucrate inflorescence common in the Asteraceae, tends to be spatially more variable because of the complicated morphology and thermal structure in flower organs. For example, the average temperature difference within a flower organ ranged from about 5 °C in most of 116 species observed (Harrap et al. 2020) to as high as 10 °C in some composite plants (Dietrich and Körner 2014). The surface temperature is often higher in the center of a flower or floral unit than at its periphery (Harrap et al. 2017). Surface temperature is potentially a key indicator of the ecological physiology of flowers from the perspective of the energy budget, but there is still insufficient information to characterize the surface temperature of flower organs, especially those in alpine plants that experience highly variable thermal environments (van der Kooi et al. 2019).

The uneven thermal distribution on a flower surface results from both internal physiological metabolism and thermal environment. Biologically, flower surface temperature is highly related to the morphological and anatomic characteristics (Jones 2013). A flower can maximize its capture of heat from solar radiation through adjusting its shape, size, color orientation and diurnal movement (van der Kooi et al. 2019). Changes in flower temperature can be also due to physiological activities such as transpiration and respiration; e.g. flowers in the Araceae can produce heat through respiration. Environmentally, flower surface temperature is affected by shortwave radiation, surrounding temperature and humidity, air dynamics and soil water availability (Rejšková et al. 2010); however, how biological and environmental factors regulate it is still unclear.

Spatial heterogeneity of flower surface temperature is conducive to reproduction in various ways. Firstly, the temperature of the flower surface and its spatial heterogeneity are assumed to affect plant reproduction via blooming, pollination, pollen tube elongation, fertilization and ovule development (Körner 2021). Secondly, a heterogeneous thermal pattern on the flower surface can help pollinators to locate or identify their nectar resources; e.g. bees can learn and detect temperature differences of >2 °C (Hammer et al. 2009). The contribution of flower temperature to reproductive success would be critical in alpine plants, which often experience both very low and highly variable temperatures (Arnold et al. 2009; Dietrich and Körner 2014; Fabbro and Körner 2004; Garcia et al. 2021). However, there is still insufficient information with which to assess flower thermal ecology in alpine plants. Such information is necessary for understanding the evolution and ecology of alpine plants, in particular under future global warming.

We supposed that the thermal environment within a flower or a floral unit will change with altitude, as radiation, temperature, humidity and wind speed change with altitude (Körner 2021). Here, therefore, we looked for general patterns in the spatial distribution of flower or floral unit surface temperature, and how biotic and abiotic factors affect them. We hypothesized that temperature would be higher in the flower center, where heat-sensitive male and female organs are located. Moreover, a warmer flower center may be critical in alpine plants for which temperature often limits for physiological activities.

MATERIALS AND METHODS

Study area and plant materials

We conducted all measurements during July and August 2020 on Lenglong Mountain at Haibei (37°37ʹ N, 101°19ʹ E, 3215 m a.s.l.), on the north-eastern Qinghai-Tibetan Plateau. In this area, the mean annual temperature during 1980–2001 was −1.2 °C, and the mean annual precipitation was about 560 mm, about half occurring during July to September (Li et al. 2004). Diurnal relative humidity (RH) fluctuated widely, being highest (90%) in early morning and lowest in late afternoon (30%). High RH along with low air temperature causes frequent fogs in the early morning. Frosts and even snow can occur even in July, the hottest month. The mean annual wind speed was about 1.7 m s⁻¹. Daily wind speed was lowest at sunrise and highest typically around late afternoon (Li et al. 2008).

Flowers of 18 species in 14 genera were common and had sufficient individuals for observation during the late growing season. We chose flowers that were available during July to August, insect-pollinated and above a certain size (corolla diameter >1 cm) to avoid difficulties in observing small flowers. We chose Argentina anserina, Aster diplostephioides, Dasiphora fruticosa, Delphinium caeruleum, Gentiana aristata, Gentiana lawrencei, Gentiana straminea, Gentiana veitchiorum, Gentianopsis paludosa, Geranium wilfordii, Ligularia virgaurea, Lomatogonium carinthiacum, Oxytropis ochrocephala, Potentilla saundersiana, Ranunculus tanguticus, Saxifraga sinomontana, Saxifraga pseudohirculus and Taraxacum mongolicum.

We recorded the scientific name, petal color, floral shape, corolla diameter, flower height above ground, floral orientation and altitude of all species (Supplementary Table S1). We categorized flower shapes depending on the effect on floral temperature (Kevan 1989; van der Kooi et al. 2019): flat, bowl (including upward- and sideward-facing), bell, globular, capitate and tubular. We classified petal color by eye as blue, purple, white or yellow. Most flowers faced upward, except in D. caeruleum, in which the corolla faces horizontally.

Thermal imaging and data processing

We took thermal images of fresh, healthy flowers without visible damage under sunlight with a thermal imager (Tix660, 7–14 nm, Fluke Inc., USA), perpendicular to the plane of the flower. All flowers were blooming in the open. Next, we shaded the plant with a neutral cloth for at least 5 min (Dietrich and Körner 2014) and then took thermal images of the same flower or one beneath it to obtain the surface temperature under conditions of homogeneous diffuse light and lower radiation input.

We processed the thermal imaging data in Smartview v. 4.0 software (Fluke, Inc.). The emissivity of a flower was uniformly set to 0.98 (Harrap and Rands 2021; Rubio et al. 1997). The ambient temperature and RH measured at the same time were used in the calculations.

All surface temperature analysis was based on data extracted from the thermal images along a line from the center to the edge of each of 3–5 flowers or floral units per species. We used one line per petal of flowers with up to five petals, and five or more at equal orientations around the circumference of composite floral units such as those of A. diplostephioides. The relative position of each pixel along a line was determined between 0 at the flower center and 1 at the edge of the petal.

Observation of microclimate

When taking thermal images, we monitored the temperature, RH and dew point with an S580-EX datalogger (Shenzhen Huatu Company, China), photosynthetic photon flux density (PPFD) with a 190R light sensor (LI-COR, USA), and wind speed and temperature with a WWFWZY-1 anemometer (Beijing Tianjianhuayi Company, China) every 1–3 s. Environment background information of thermal imaging photos is summarized in Supplementary Fig. S5.

Statistics and analysis

To compare spatial variation of surface temperature among measurements, we scaled the temperature data between a range of 0 and 1 with min–max normalization.

To examine the influence of flower shape and color on temperature distribution, we calculated maximum and minimum temperatures and their mean and locations along each line. We defined the surface temperature of the center as the mean of the values from locations 0 to 0.1, and that of the edge as the mean of the values from locations 0.9 to 1.0. We used the ratio of center temperature to edge temperature to evaluate the temperature difference between the center and the edge. To assess the spatial characteristic of the surface temperature, we evaluated the range (highest to lowest) and standard deviation (SD) of all data points along each line.

We used Wilcoxon’s test to compare maximum and minimum temperature locations within a flower or floral unit in sunlight and shade. We used the maximum likelihood-like approach of Box and Cox to select a suitable data transform for normality with the ‘powerTransform’ function of the car package (Fox and Weisberg 2019) in R software (R Core Team 2020), then the Shapiro–Wilk test to test the normality of data. We used principal component analysis (PCA) to explore the relationship among environmental factors, flower diameter, flower height (above ground) and temperature indicators with the ‘princomp’ function of the factoextra package (Kassambara and Mundt 2020). We used a generalized linear model with stepwise feature selection to explore the effects of the main factors on normalized temperature indicators with the stats package, and variance decomposition to explore explainable factors with the ‘varpart’ function in the vegan package (Oksanen et al. 2022). We used the Kruskal–Wallis multiple comparison test (‘kruskalmc’ and ‘multcompLetters’ function at P = 0.01) in the pgirmes and multcompview packages to compare multiple floral shapes (Giraudoux 2022; Graves et al. 2019). We tested the relationships among indicators, flower diameter and altitude with the ‘stat_cor’ function with Spearman coefficients in the ggpubr package (Kassambara 2022).

RESULTS

Spatial pattern of flower surface temperature

Surface temperature was spatially heterogeneous within a flower or floral unit under natural sunlight (Fig. 1). In most cases, it decreased from the center to the edge (Figs 1 and 2), steeply in all capitate plants (A. diplostephioides, T. mongolicum and L. virgaurea). Under sunlight condition, the temperature difference between center and edge ranged from 0 to 11 °C depending on species and thermal environment, and was about 1 °C in most flowers or floral units (Supplementary Fig. S1). Under shade, the surface temperature was much more homogeneous (Figs 1 and 2). It tended to be higher in the center, but decreased toward the edge much less than under sunlight, and the difference of floral center and edge was much less than under sunlight (Supplementary Fig. S1).

To characterize the spatial variation patterns of surface temperature, we examined the relative locations of the highest and lowest temperatures along each line (Figs 3 and 4). Under sunlight, the lowest temperature was close to the edge and the highest was close to the center. Under shade, their positions were much less distinct.

Association of surface temperature with abiotic and biotic factors

PCA analysis

Principal components 1 and 2 (PC1 and PC2) together explained about 58% of the variance in the surface temperature of the whole data set under sunlight, and around 64% under shade (Fig. 5). Although this is less than two-thirds of the variance, PCA provides insights in the data structure: The surface temperature (mean, minimum, maximum, floral center and edge) were strongly associated with air temperature, PPFD and RH. In contrast, the spatial variability indicators, such as range, SD, CV (coefficient of variation) and ratio of flower center to edge temperature, were more strongly associated with flower diameter, flower height and the location of the maximum and minimum surface temperatures along the lines.

Abiotic factors

Under sunlight, the range and SD of surface temperature were partly explained by PPFD and air temperature, respectively, accounting for about 12% and 5% of variance, but the ratio had no correlation with environmental factors (Fig. 6; Supplementary Table S2). Under shade, PPFD explained separately 3% and 2% variance of the range and SD, and RH explained 2% variance of the ratio of flower center to edge temperature.

Biotic factors

Flower shape explained a larger amount of variation of flower surface temperature than abiotic factors in both sunlight and shade, notably about 33% of the range and SD and 50% of the variance of the ratio of flower center to edge temperature under sunlight, and around 40% of the range, SD and the ratio under shade (Fig. 6). Under both sunlight and shade, the range, SD and ratio of capitate flower center to edge temperature were all significantly higher than those of other flowers (Fig. 7).

Flower diameter accounted for 28%–38% of variation of floral surface temperature range, SD and ratio under shade (Fig. 6). Larger flowers had a greater range, SD and ratio under both sunlight (R = 0.53, 0.49 and 0.47, respectively, all P < 0.001) and shade (R = 0.6, 0.59 and 0.52, all P < 0.001; Fig. 8).

Flower color appeared to only explain 8% of the ratio (Fig. 6). There was no significant difference between colors in effect on temperature indicators (Supplementary Fig. S2).

Flower height explained around 26% of variation of flower surface temperature under shade (Fig. 6). The range, SD and ratio increased with flower height, such correlation was much tighter under shade condition (R = 0.5, 0.48 and 0.52, all P < 0.001; Supplementary Fig. S3) than under sunlight (R = 0.35, 0.34 and 0.37, all P < 0.001; Supplementary Fig. S3).

Surface temperature variability with altitude

Although altitudes explained less than 6% of the variance in range, SD and ratio under sunlight (Fig. 6), the range and SD of surface temperature increased marginally with altitude, but the ratio had no change with altitude under sunlight (Fig. 9). These indicators had no correlation with altitude under shade.

The spatial variability of flower surface temperature in relation to altitude varied widely among species (Supplementary Fig. S4). The ratio of flower center to edge temperature tended to increase with altitude in 11 of the 18 species observed (Supplementary Fig. S4). The ratio increased significantly with altitude in A. diplostephioides, G. veitchiorum and T. mongolicum.

DISCUSSION

Reproductive success depends on flower warmth (Distefano et al. 2018; Ida and Totland 2014; Seymour 1997; Stanton and Galen 1989). The spatial pattern of flower surface temperature is likely to be an important cue for pollinators (Dyer et al. 2006; Harrap et al. 2017); however, little information is available about spatial changes in it. We characterized spatial patterns of the surface temperature of flowers with different shape, colors and size under sunlight and shade. Our findings suggest that a warmer center with a cooler edge is widespread.

Influence of environment on spatial variation of flower surface temperature

Flower temperatures increase with radiation (e.g. Dietrich and Körner 2014; Rejšková et al. 2010). This study shows that direct solar radiation also increases the spatial variability of the surface temperature, as both the range and SD of temperature were higher under sunlight than under shade, and increased with PPFD (Fig. 5).

The more contrastive spatial pattern of flower surface temperature under higher solar radiation may have important ecological implications. Field observations and experiments suggest that the spatial thermal pattern of flowers can be a cue for pollinators (Harrap et al. 2017, 2020). The thermal pattern differs among species, as observed here and in another study (Harrap et al. 2017), and is likely to allow some pollinators to distinguish flower species. We expected that the spatial heterogeneity of flower surface temperature would be more distinct under direct or high radiation. However, too high radiation could decrease activities by some pollinators to avoid body overheating (Xu et al. 2021).

Other environmental factors may contribute slightly to flower surface temperature. RH and wind speed influence convection and transpiration, altering plant temperature. However, we found weak relationships of both with floral temperature and its heterogeneity, maybe as a result of the smaller density of or smaller stomata in petals than in leaves (Patiño and Grace 2002; Roddy et al. 2016; Zhang et al. 2018) and the cold environment. However, evaporation through flowers can still be important in regulating flower temperature, as seen in the sepals of tropical species (Patiño and Grace 2002) and the cuticles of flowers (Roddy 2015; Roddy et al. 2016). Other floral traits may also be involved in floral transpiration, such as petal veins (Zhang et al. 2018).

Effect of flower traits on spatial variability of surface temperature

Flower shape

Flower traits, such as shape, color, size orientation and movement, affect floral temperature through physical or biochemical processes (van der Kooi et al. 2019). We found that flower shape is a major determinant of surface temperature and its spatial pattern. In particular, the range and SD of the surface temperature of capitate flowers were significantly higher than those of other flowers, with a range as high as 11 °C (Fig. 7), as reported in previous studies. In the capitate flowers, the central area of the head is composed of densely arranged disc florets, whereas the periphery is composed of long and narrow ray florets. The flower head uses narrow gaps to capture and trap heat and prevent it from escaping through its thicker boundary layer. We propose that the higher temperature in the flower center in Asteraceae species contributes to their greater diversity and abundance at higher than lower altitudes. For example, T. mongolicum had not only a higher ratio of center to edge temperature but also a steep slope of the ratio against altitude. This example suggests that flowers of Asteraceae species may retain a slightly warmer center temperature as altitude changes, since the temperature of petals follows environmental temperature closely.

Flower size

Previous studies found no consistent relationship between flower size and temperature magnitude, but flower size had positive correlation with the difference of flower and leaf temperature (Dietrich and Körner 2014). Here, we found that larger flowers tended to have larger spatial variability of surface temperature (Fig. 8), maybe because they can intercept more heat from their environment or they have more complex morphologies and thus greater surface heterogeneity. We suspect that flowers with more spatially variable surface temperatures favor pollinators, as temperature is a cue for foraging behavior (Dyer et al. 2006).

Flower color

Flower color is likely an important ecological trait that warrants further exploration (Binkenstein et al. 2016; Wang et al. 2021). Darker flowers produced more nectar, likely by using the absorbed heat to increase their metabolic rate (Kevan and Baker 1983). Flower color can influence temperature by determining light absorption and reflectance (Ida and Totland 2014; Jewell et al. 1994; Molgaard 1989). However, evidence for its relationship with surface temperature is inconsistent (Shrestha et al. 2018).

Flower color can influence the spatial variability of surface temperature through spatially differential energy exchanges caused by differences in surface or internal structure, pigment distribution (Koski et al. 2020) and even interaction with microbes (Rebolleda‐Gómez et al. 2019). However, we cannot yet generalize the relation between flower color and the spatial pattern of surface temperature. Here, however, flower color had no significant effect on the spatial variability of surface temperature when all data were pooled (Supplementary Table S2). When we grouped flowers by color to examine spatial variability in surface temperature, there was no significant difference between colors (Supplementary Fig. S2). Since the analysis was based on human eyesight and limited combinations of flower colors with other traits, it is difficult to draw reliable conclusion from these results. In addition, ultraviolet absorption may affect flower surface temperature (van der Kooi et al. 2019). Among all 18 species observed here, 7 show UV patterns on their petals (data not shown). Floral UV patterns are created by the local distribution of UV-absorbing or UV-reflecting pigments and are adapted to the visual system pollinators to increase the attractiveness (Chen et al. 2020; Papiorek et al. 2016). Subsequent research will be conducted to elucidate the relationship between UV patterns and surface temperature, including the underlying mechanisms and the degree of influence.

CONCLUSIONS

Flower temperature is a critical factor in sexual reproduction, especially in highly variable or extreme environments such as alpine habitats. Our study provides evidence that flower temperature is generally higher in the center, particularly under sunlight. The strong influence of flower shape and morphology on the accumulation of heat within flowers suggests that thermal regulation through morphology contributes to reproductive success and fitness. However, more quantitative evidence is needed to support this claim. Furthermore, our study shows that solar radiation is essential for higher thermal accumulation and stability in the flower center. As shortwave radiation increases, the spatial pattern of floral surface temperature may change, or the ratio of flower center to edge temperature may increase. Further research is needed to explore how these eco-physiological and genetic traits contribute to the success of alpine plants in their extreme thermal habitats.

Supplementary Material

Supplementary material is available at Journal of Plant Ecology online.

Table S1: Information of plant materials.

Table S2: Explanation of the effects of floral traits and environmental factors on the variability of floral surface temperature under sunlight and shaded conditions, respectively.

Figure S1: Distribution of temperature difference between floral center (T_center) and edge (T_edge) within a flower or floral unit under sunlight and artificial shade condition.

Figure S2: Distribution of the range (top row) and standard deviation (middle row) of the surface temperature, and the ratio of floral center to edge temperature (bottom row) for different floral colors under natural sunlight (left column) and shaded (right column) conditions.

Figure S3: The range (top row) and standard deviation (middle row) of the surface temperature, and the ratio of floral center to edge temperature (bottom row) in relation to flower height under natural sunlight (left column) and shaded (right column) conditions.

Figure S4: The ratio of floral center to edge temperature on each transect line for each individual species under natural sunlight conditions.

Figure S5: Distribution of measurement environment data of thermal imaging photos.

Funding

This work was financially supported by the National Natural Science Foundation of China (91837312) and the Second Tibetan Plateau Scientific Expedition and Research (STEP) program (2019QZKK0106).

Acknowledgements

We thank Haibei National Alpine Grassland Ecosystem Research Station for access to the experiment site. We thank Dr Mani Shrestha and Dr Ren Zongxin for advices on experiment design, and Kang Huixing, Ke Xinran, Guo tong, Zheng Tianyu, Yu yuan, Liu Wei and Qin Wuying from Peking University for help on field work and suggestions.

Conflict of interest statement. The authors declare that they have no conflict of interest.

REFERENCES