Bimodal activity of diurnal flower visitation at high elevation

Abstract Successful pollination in animal‐pollinated plants depends on the temporal overlap between flower presentation and pollinator foraging activity. Variation in the temporal dimension of plant–pollinator networks has been investigated intensely across flowering seasons. However, over the course of a day, the dynamics of plant–pollinator interactions may vary strongly due environmental fluctuations. It is usually assumed there is a unimodal, diurnal, activity pattern, while alternative multimodal types of activity patterns are often neglected and deserve greater investigation. Here, we quantified the daily activity pattern of flower visitors in two different habitats contrasting high elevation meadows versus forests in Southwest China to investigate the role of abiotic conditions in the temporal dynamics of plant–pollinator interactions. We examined diurnal activity patterns for the entire pollinator community. Pollinator groups may differ in their ability to adapt to habitats and abiotic conditions, which might be displayed in their patterns of activity. We hypothesized that (a) pollinator communities show multimodal activity patterns, (b) patterns differ between pollinator groups and habitat types, and (c) abiotic conditions explain observed activity patterns. In total, we collected 4,988 flower visitors belonging to six functional groups. There was a bimodal activity pattern when looking at the entire pollinator community and in five out of six flower visitor groups (exempting solitary bees) regardless of habitat types. Bumblebees, honeybees, dipterans, lepidopterans, and other insects showed activity peaks in the morning and afternoon, whereas solitary bees were most active at midday. Activity of all six pollinator groups increased as solar radiation increased and then decreased after reaching a certain threshold. Our findings suggest that in habitats at higher elevations, a bimodal activity pattern of flower visitation is commonly employed across most pollinator groups that are diurnal foragers. This pattern may be caused by insects avoiding overheating due to elevated temperatures when exposed to high solar radiation at midday.

However, the temporal dimensions defining plant-pollinator interactions can vary from short to long periods (CaraDonna et al., 2021;Schwarz et al., 2020). In a community, both plants and associated pollinators display a seasonal pattern of flowering, versus adult emergence and foraging cycles, respectively. It is expected this causes temporal variations and seasonal turnovers in the pollinator community and plant-pollinator networks (Cane et al., 2005;Herrera, 1988;Pyke et al., 2011;Souza et al., 2018;Rabeling et al., 2019).
Some plant species follow a diurnal rhythm of flowering and that can have a strong influence on plant-pollinator interactions (Fründ et al., 2011), contributing to distinct patterns of foraging activity among different pollinator groups Zoller et al., 2020).
Most pollinators are diurnal such as bees, butterflies, flies, and birds, but some moths, beetles, and bats forage and pollinate along crepuscular-nocturnal rhythms Willmer, 2011). The foraging of bumblebees and honeybees typically follows strong circadian rhythms. Most bee species rely on the circadian clocks to predict sunrise and sunset, as they adopt higher levels of activity during the day (Chittka et al., 2013;Vaudo et al., 2014;Yerushalmi et al., 2006) to make the most efficient use of sufficient sunlight to find preferred flowers for foraging (Bloch et al., 2017).
To our knowledge, most previous research investigating patterns of foraging activities of flower visitors has focused either on a single-few focal plant species or a subset of insect taxa (Gottlieb et al., 2005;Sgolastra et al., 2016;Souza et al., 2017). Past studies of the temporal dynamics of plant-pollinator interactions focused primarily on seasonal and year-to-year variability (Chacoff et al., 2018;Dupont et al., 2009). Therefore, the temporal dimensions of plantpollinator interactions at a community level remain less well studied, particularly at finer resolutions during the course of a floral day (but see Baldock et al., 2011;Schwarz et al., 2021;Zoller et al., 2020).
While the dynamics of plant-pollinator interactions are dependent on internal rhythms (Bloch et al., 2017;Fenske et al., 2018; see above), there are additional and extrinsic parameters, such as the availability of plant resources and/or environmental conditions to determine plant-pollinator interaction dynamics .
Among these extrinsic parameters, abiotic parameters include ambient temperature, solar radiation, relative humidity, and wind speed are the best-studied (Sanderson et al., 2015;Sgolastra et al., 2016;Willmer, 2011). Indeed, visitors adapt their foraging activity to optimize the time of resource collection during the day to maximize their efficiency according to the quality of environmental conditions (Cook et al., 2011;Schäffler & Dötterl, 2011;Stone et al., 1999).
Different groups of pollinators may show different adaptations to the same environmental factors Sgolastra et al., 2016;Tuell & Isaacs, 2010;Vicens & Bosch, 2000). For example, in a study of the pollination of Vaccinium corymbosum, honeybees (Apis mellifera) were most active under conditions of warm temperatures and high solar radiation provided wind velocity and relative humidity were low. In contrast, bumblebees (Bombus impatiens) were the dominant pollinators during periods of poor weather (Tuell & Isaacs, 2010).
Furthermore, it is usually assumed that there is a unimodal pattern of pollinator activities (e.g., Baldock et al., 2011; but Zoller et al. (2020) suggested recently that flower visitors might switch from unimodal to a bimodal within the Arctic Circle. It is thus possible, with more field studies, bimodal or even multimodal patterns of activity in pollination communities may be more common than it is previously anticipated.
As extrinsic abiotic parameters usually vary between or even within habitats, the general habitat context may be a key factor governing spatio-temporal variation in flower visitation patterns (Jha & Vandermeer, 2009). A recent study showed that warmer microhabitats positively affected the richness of flower visitors in alpine communities in Austria (Ohler et al., 2020). Therefore, the pollinatorspecific differences in flower visitation patterns might vary depending on habitat type.
Heterogenous subalpine and alpine environments in the Himalaya-Hengduan Mountains occur at extremely high elevations (e.g., 1,300-6,000 m) along a complex mountain topology, while harboring unusually high plant and insect diversities Xing & Ree, 2017). Previous studies in this region revealed that eusocial bumblebees and the native honeybee (Apis cerana) dominate flower communities that are characterized by generalized pollination systems Zhao et al., 2019).
In such mountain environments, atmospheric temperature decreases while solar radiation increases as elevation increases (Körner, 2007). Therefore, previous studies focused specifically on the effect of temperature and solar radiation (Birrell et al., 2020;Hodkinson, 2005). They found that increasing temperature resulted in increasing flight departures of honeybees up to a certain threshold of solar radiation after which there was a negative response of flight activity to temperature (Burrill & Dietz, 1981). Furthermore, temperature effects can differ between different insect species or their different morphologies (Cena & Clark, 1972). We should expect, therefore, that at high elevations with high degrees of solar radiation both factors must shape the daily patterns of pollinator activity.
As Himalayan mountains are characterized by extreme weather conditions while harboring diverse habitats, they provide ideal settings to test whether diurnal pollinator activities in communities show bi-or multimodal activity patterns under high solar radiation.
To obtain a more generalized understanding, our investigation was conducted in two contrasting habitats. We hypothesized that, under conditions of extreme high solar radiation at high elevations, (a) the visitation patterns are bi-or multimodal but differ for different groups of flower visitors and habitat types, and (b) abiotic factors explain the activities of each pollinator group when combined or addressed independently. We assessed variation in the diurnal activity patterns of different visitor groups across meadow (

| Sampling of plant-insect interactions
We chose three sites with more than 1.0 km distance between any two locations. In each site, we set up two different habitat transects, one in the meadow and a parallel transect in the adjacent forest. Each transect was 100 m in length and 3 m wide. The distance between the two habitat transects exceeded 50 m. In total, six transects were sampled and repeated for 2 years (Table S1). The floral community composition was relatively similar between habitats (Xu et al., unpublished data).
In both study seasons, we observed plant and visitor interactions during daylight hours. Sampling was performed on a total of 14 sunny days (8 days in 2018, 6 days in 2019) excluding days of strong wind, fog, or rainfall. On all sampling days, we collected data in all six transects simultaneously between 08:00 and 20:00 with one person sampling per transect. When an insect arrived at a flower and contacted the plant's reproductive structures, while actively foraging for pollen and/or nectar, we recorded this as one interaction. This included recording each insect observation the date, time, and plant species on which it foraged. Each flower visitor recorded was then collected, stored in a separate 2-ml Eppendorf tube containing 95% ethanol. The specimen was taken to the laboratory for identification. Following identification, specimens were pinned and vouchers deposited at the Kunming Institute of Botany, Chinese Academy of Sciences.
As the dominant foragers were bumblebees (Bombus spp.) and honeybees (A. cerana), they were identified as species (like the bumblebees in Yulong Snow Mountain, Bombus friseanus and Bombus lepidus). Remaining taxa were identified to genus (e.g., Andrena, Musca) or Order (Coleoptera). Insect specimens were segregated into "functional groups," which are defined here as those visitors that interact with the same flowers in a similar manner. We used criteria of the same Order, genus and species, similarity in body size and foraging behavior to identify six functional visitor groups in our study. They are defined here as bumblebee, honeybee, solitary bee (i.e., bees that were not in the genera Bombus or Apis and with a body size less than 8 mm), dipteran, lepidopteran, and other insects (Hemiptera, Coleoptera).

| Environmental abiotic factors
To investigate whether abiotic parameters affect plants and their visitor interactions, we recorded hourly averages of ambient temperature (°C), relative humidity (%), wind velocity (m/s), and solar radiation (μmol·m −2 ·s −1 ) on all sampling days from the local weather station in the meadow-forest transition at one of our field sites, managed by Yunnan Lijiang Forest Ecosystem National Observation and Research Station. We expected that these data were representative across our study sites as all sites were approximately at 3,200 m, were located within approximately 1 km of each other, and were of the same vegetation type (i.e., meadows or forests).

| Statistical analysis
The data were analyzed mainly by a nonparametric kernel density estimation method (Ridout & Linkie, 2009). It was assumed that insect behavior was distributed within a continuous 24-hr time cycle, and behavior events were based on continuous random sampling.
Our sampling was continuous from 8:00 to 20:00 giving us 12 hr each day to record the diurnal pattern of flower visitor abundance.
To test whether the activity pattern varied significantly at different times of the day, we applied kernel density estimation using the overlap package (version 0.3.3; Ridout & Linkie, 2009). The kernel density estimation method does not require any assumptions for the data distribution. It was used to characterize the data distribution from the data set itself (Ridout & Linkie, 2009). First, the time of the day was converted to decimal format (0-1) and then converted to radian format (Ridout & Linkie, 2009). Second, we used the densi-tyPlot function to draw fitted circular kernel distribution curves for each individual group of flower visitor within each habitat type. We then used the overlapPlot function to draw fitted circular kernel distribution overlap curves for different groups of flower visitor within the two habitat types. All kernel plots are based on the pooled data across all observation days.
We then performed an activity patterns overlap analysis to test To determine whether any of the four abiotic factors (ambient temperature, relative humidity, solar radiation, and wind velocity) affected the abundance of flower-visiting insects, we used generalized linear mixed models (GLMMs) implemented in the glmmTMB package (version 1.0.2.1; Bates et al., 2015;Brooks et al., 2017) with negative binomial error distribution and correction for zero-inflation (i.e., the distribution does not predict as much zero as in the data) required to meet model assumptions. We ran a total of seven models, with the abundance of flower visitors and six functional visitor groups as response variables. Each model included habitat type, ambient temperature, ambient temperature 2 , relative humidity, relative humidity 2 , wind velocity, solar radiation, and solar radiation 2 as explanatory fixed factors. Study site and sampling day were included as random factors in each model. We included the quadratic terms of temperature, relative humidity, and solar radiation in the model because high and low values in those variables might have a negative impact on flower visitor activity. The significance of fixed factors was assessed with Type III Wald chi-square tests implemented in the ANOVA function of the car package (version 3.0-10; Fox & Weisberg, 2019). To check if all model assumptions were met, we used the DHARMa package (version 0.3.3.0; Hartig, 2020) and the check_collinearity function of the performance package (version 0.5.0; Lüdecke et al., 2020). In addition, we ran a Pearson correlation among all the abiotic variables across the 2 years, using corr.test function of psych package (version 2.0.7; Revelle, 2020) to test their correlations. After running initial full GLMM models including all four variables, we removed ambient temperature due to high collinearity between temperature and solar radiation (VIF > 5). In fact, solar radiation was highest around midday ( Figure S1d) and correlated strongly with temperature (r = 0.716, p < 0.001; Figure S2). We removed temperature rather than solar radiation even though temperatures at midday were highest (when solar radiation was also at its highest), as they had not reached temperature levels (15.1 ± 1.5°C) known to limit insect species activity (Kühsel & Blüthgen, 2015). Furthermore, due to previous research on the important role of solar radiation on honey bee activity at higher elevations (see Burrill & Dietz, 1981), we decided to retain solar radiation.
Our selection was supported by the fact that models which included temperature instead of solar radiation showed a positive effect only in honeybees; therefore, it showed largely no effect on other insect activity (Table S5). Thus, the final analysis only included relative humidity, solar radiation, and wind velocity as explanatory factors.
Based on the daily bimodal pattern of flower visitor activities, we conducted separate analyses of the abiotic factors associated with each abundance peak with a random factor plotted showing the before noon (before 12:00) and after noon (after 12:00) curves except for the solitary bee group (unimodal pattern within a day).
We used generalized linear mixed models (GLMMs) implemented in the glmmTMB package (version 1.0.2.1; Bates et al., 2015;Brooks et al., 2017) with a negative binomial error distribution. We included the quadratic terms of temperature, relative humidity, and solar radiation in the model because high and low values in those variables might have a negative impact on flower visitor activity. For each model, we also calculated the R 2 , the coefficient of determination, to see which variable explained the most variance (Nakagawa & Schielzeth, 2013). After we ran this model, the environmental variables' significance was corrected for multiple testing and calculated as p-adjusted (Benjamini & Hochberg, 1995;Benjamini & Yekutieli, 2001

| RE SULTS
We documented 4,988 insect visits to 75 plant species noting that 78.3% of the insect specimens were identified to species level, while the remaining 23.5% were identified to the level of Order (Table 1).
The bumblebee was most abundant making up 47.2% of all recorded flower visitations.
As expected, abiotic conditions varied during the course of a day, but the patterns were similar in the two study seasons. In general, the ambient temperature was lowest early in the day, and the warmest period occurred between 2:00 and 4:00 p.m. (Figure S1a).

| Diurnal activity pattern and variation among flower visitor groups
We found a consistent diurnal pattern of flower visitor activity for most groups except for solitary bees. In general, visitation activities reached two peaks. The major peak occurred around 11:00 a.m., and a secondary but smaller peak appeared between 3:00 and 4:00 p.m. There was a marked decrease in activity around mid- In the meadows, the remaining five flower visitor groups showed a similar bimodal activity pattern, but the exact timing of their activity peaks differed slightly among the groups (Figure 1; Figure S3).
The overlap coefficient of activity patterns across flower-visiting community ranged between 0.578 and 0.904 indicating medium to strong similarity among groups. Specifically, bumblebees showed a significantly different diurnal activity pattern compared to the five remaining groups (p < .001; Table S2). In forests, four flower visitor groups (excluding solitary bees and dipterans) showed a similar bimodal activity pattern. The exact timing of their activity peaks varied only slightly among these groups (Figure 1; Figure S4). The overlap in activity patterns varied between 0.738 and 0.941 across the different flower visitor groups indicating strong activity overlaps. Again, bumblebees had a significantly different diurnal pattern of activity compared to honeybees, dipterans, and solitary bees (p < .05; Table S2).

| Effect of four abiotic factors on the abundance of flower visitor
Generally, all members of the flower visitor community together (overall flower visitors) and each separate, flower visitor group showed a significant relationship with solar radiation (p < 0.05; Table 2; Tables S3-S4). In addition, the same flower visitor groups also showed a significant negative relationship with relative humidity (p < .05; Table 2; Table S3). The overall abundance of flower visitors showed a unimodal concave pattern with solar radiation and relative humidity (p = .001 for quadratic term of solar radiation). With increasing solar radiation, the number of visitations increased until it peaked at approximately 450 μmol·m −2 ·s −1 and then declined (Figure 2). With increasing relative humidity, the number of visitations increased until it peaked at about 80% and then declined ( Figure S6). However, wind velocity did not significantly affect the abundance of flower visitors (p > .05 for all terms of wind velocity; Table 2; Table S3). Different flower visitor groups showed different responses to environmental factors. Bumblebee abundance showed a unimodal concave pattern with relative humidity (p < .05). The number of bumblebee visitations increased with increasing relative humidity until it peaked at about 70% and then declined again ( Figure S6). However, the quadratic term of solar radiation for bumblebees (p = .681) and dipterans (p = .304) was not significant, indicating that there was no peak value but a linear increase (Figure 2b,d). In contrast, honeybee, solitary bee, lepidopteran, and other insects' abundances showed a unimodal concave pattern with solar radiation. With increasing solar radiation, the number of visitations increased until it peaked, respectively, at about 370, 350, 280, and 280 μmol·m −2 ·s −1 (Figure 2).
Before noon, all flower visitors including bumblebees and honeybees showed a significantly positive relationship with ambient temperature and solar radiation, but there was a negative relationship with relative humidity (Table 3; Table S6). The abundance of overall visitors showed a bell-shaped curve with ambient temperature, relative humidity, and solar radiation ( Figures S7-S10). Dipterans showed a significantly positive relationship with solar radiation (Table 3; Table S6), with their abundance showing a linear relationship with solar radiation (Figures S7-S10).
After noon, overall flower visitors showed a significantly positive relationship with ambient temperature and solar radiation, but with a negative relationship with relative humidity (Table 3; Table S6). However, dipterans showed a significantly positive relationship with solar radiation and a negative relationship with relative humidity. Dipterans showed no relationship with ambient temperature (Table 3; Table S6). The abundance of overall visitors showed a bellshaped curve with relative humidity and a linear relationship with ambient temperature. The abundance of bumblebees showed a bellshaped curve with ambient temperature and a linear relationship with relative humidity. The abundance of honeybee showed bellshaped with relative humidity and linear relationship with ambient temperature. The abundance of dipterans showed a linear relationship with relative humidity (Figures S7-S10).

TA B L E 1 The total number of flower-visiting insects belonging to six functional groups collected in high elevation communities (meadow and forest) during two flowering seasons on Yulong Snow Mountain, SW China
The activity pattern of the solitary bee group was unimodal, showing a positive relationship with ambient temperature and solar radiation. The solitary bee group also showed a negative relationship with relative humidity (Table 3; Table S6). As wind velocity did not include quadratic terms, so the abundance of flower visitors showed only a linear relationship with wind velocity (Figures S7-S10). In contrast, the abundance of lepidopteran and other insects was not significant for all environmental factors (Table 3; Table S6).
This indicated that solar radiation had the highest fit to explain the variance of flower visitor acitivity.

| Effect of habitat on the abundance and activity of flower visitors
Habitat type influenced significantly the abundance of flower visitors (p < 0.05). The abundances of overall flower visitors (p < 0.001),  (Table 2; Tables S3-S4; Figure S5). However, the abundances of solitary bees (p = 0.805), lepidopterans (p = 0.434), and other insects (p = 0.171) did not vary between these two habitats (

| D ISCUSS I ON
We found a consistent bimodal pattern of foraging activity of diurnal flower visitors across meadow and forest habitats in six high elevation plant communities. We further showed that this general bimodal activity pattern can be explained at least in part by abiotic environmental factors such as solar radiation. Although relative humidity also helped to explain forager activity patterns, the remaining environmental factors were not relevant in our study. The most likely explanation for this was that we selected only sunny days to do sample sites avoiding extreme weather events.
Most of the previous studies of flower visitor activity focused on a single or few insect species and/or their visits to a focal plant. Most of these studies found that their insects showed a unimodal pattern when visiting flowers (Muniz et al., 2013;Sgolastra et al., 2016;Steen, 2017;Totland, 1994

TA B L E 2 (Continued)
peaks of flower visitation (Herrera, 1990). Our community-wide study illustrated a bimodal pattern. Our results are similar to an Arctic flowering community showing a bimodal foraging activity in the course of 1 year (Zoller et al., 2020). In contrast to our findings and the Arctic study, a unimodal activity pattern was also found for all insect groups in ruderal meadow communities in Switzerland . Therefore, our findings supplement our ongoing understanding of the dynamics of plant-pollinator interactions and pollination networks at the community level in high elevation environments. The difference is that we suggest that flower visitors are more likely to produce bimodal and multimodal activity patterns in more extreme regions at far higher high elevations and in polar regions.
Furthermore, the overall bimodal activity pattern in this study was consistent across flower visitor groups except for solitary bees, but the exact time among the flower visitor groups' activity peak was similar (Figure 1). Our study area was dominated by polylectic bumblebees, and their activity patterns showed a bimodal distribution. Bumblebee species are typically covered with dense hair, vibrate their thoracic muscles to regulate body temperatures, and are among the largest floral foraging insects in the cooler-temperate zones (Goulson, 2010). Therefore, they appear well adapted to active foraging in alpine and subalpine habitats representing the most important pollinators in montane regions (Biella, Bogliani, et al., 2017;Egawa & Itino, 2019;Minachilis et al., 2020;Williams et al., 2009).
Unimodal activity of solitary bees might be caused by avoiding competition with other insect groups (including bumblebees) in alpine regions, or these smaller bees may employ the unimodal pattern to increase their body temperatures, so they can continue to forage under temperatures at higher elevations (Willmer & Stone, 2004).
Interestingly, in our study temperature was not a good predictor to explain pollinator activity. This is in contrast with previous studies that found temperature to be the main predictor of polli-  Zoller et al., 2020). Temperatures were relatively low (9.2-21.2°C) at our sites due to their high elevation. The most likely reason why we found this contrasting pattern was the interaction between solar radiation and temperature. The highest solar radiation at noon should increase body temperature of insects at a rapid rate (Cena & Clark, 1972;Corbet et al., 2020). As solar radiation increases the body temperature of an insect at a certain level, it becomes in danger of overheating, so foraging activity must decrease. Several insect species show adaptions to high solar radiation (e.g., UV radiation) at high elevations (Birrell et al., 2020;Hodkinson, 2005).
Consistent with the expectation mentioned above, in our study solar radiation showed a positive relation to insect activity until a certain threshold was reached. This relationship most likely determined the bimodal distribution of visits. At high elevations, solar radiation (e.g., UV radiation) is an important driver of species performance and behavior. In our study, a decrease in activity usually occurred from 1:00 to 2:00 p.m. During this time period, the solar radiation value was higher than 450 μmol·m −2 ·s −1 on 9 out of the 14 study days. Beyond this solar radiation threshold, the activity of members of our flower visitor community decreased with the exception of bumblebees and dipterans (Figure 2), and this explains the decrease in general activity at midday. Our results indicate that high solar radiation may overheat insect bodies causing declines in foraging as in other studies (Barônio & Torezan-Silingardi, 2017;Bellusci & Marques, 2001;Herrera, 1990). The deviation of a bumblebee's lack of a response to peak solar radiation is most likely explained by their ectothermic mode of thermoregulation preventing them from overheating (Heinrich, 1979). Also, sisters in the same colony show multiple flight bouts for collecting floral resources during the same day, both allowing the same colony to continue foraging even during peaks of solar radiation.
Higher solar radiation increases both insect body temperature (Cena & Clark, 1972;Corbet et al., 2020) and the temperature of flowers (Corbet et al., 2020). It has been shown that elevated temperatures, caused by solar radiation, evaporate plant floral nectars making it more difficult for foragers to extract remaining nutrients.
This reduces visitation rates and decreases pollination success lowering rates of sexual reproduction in plants (Corbet et al., 2020;Scaven & Rafferty, 2013). Therefore, temporal patterns of pollinator nectar-feeding and nectar production may coincide. However, a previous experiment with Anemone rivularis conducted in the same area as our study revealed that at midday, when the ambient temperature was highest and the radiation strongest, the internal flower temperature of the plant was lower than the ambient temperature. Such an experiment proves that plants can regulate microenvironments inside their flowers to avoid damage by higher temperatures and radiation (Zhang et al., 2010) possibly limiting their effects on nectar production. Matching of nectar resource availability with insect flower-visiting schedules warrants further investigations in our Himalayan communities.
There was a generally high degree of activity overlap between the meadow and the forest habitats (>80% for most flower visitor groups), but the activity curves differed somewhat between forest and meadow (Table 4). For example, the two flower visitor groups with highest abundance, bumblebees and honeybees, visited flowers during the afternoons in the meadow more often than in the forest (Figure 1b,c). This may be due to the fact that both Apis and Bombus species are generalized floral foragers reducing competition within members of the same groups and between different visitor groups, thus sustaining their high abundance. That is, although the floral resources in alpine meadows are highly frequented, they can still support higher visitation rates compared to forest habitats.
However, the habitat effect proved the opposite for less abundant visitor groups (dipterans, solitary bees, lepidopterans, and other insects). They showed a higher abundance in the forest than in the meadow during afternoons (Figure 1d-g). This asymmetry in habitat transition between different flower visitor groups suggests that alterations in the availability of floral resources may differentially affect habitat and foraging preferences across these flower visitor groups. In the afternoon, bees become more abundant in the meadow, forcing other insects to switch to floral resources in forests to avoiding competition via temporal and/or spatial displacement.

| CON CLUS ION
In conclusion, this appears to be the first study to show temporal foraging activity patterns in pollination communities at their highest (Himalayan) elevations. We found bimodal activities of some flower visitors in this Himalayan environment. We highlighted a consistency in bimodal visitation patterns between meadow and forest habitats, while showing differences in foraging peaks of flower visitor groups between both habitats. This suggests microsite variability. Finally, we suggest that the general activity decline in foraging activity at midday is most likely explained by increased solar radiation at higher elevations. Clearly, further research is Province to H. Wang (YLXL20170001).

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.