Diversity, Daily Activity Patterns, and Pollination Effectiveness of the Insects Visiting Camellia osmantha, C. vietnamensis, and C. oleifera in South China

Camellia spp., which are self-incompatible plants, are some of the most important woody species producing edible oil in Southeast Asian countries. However, the demand for camellia oil currently exceeds the supply due to low product yields that have resulted from a decrease in pollination services. Although Camellia osmantha, C. vietnamensis, and C. oleifera are cultivated in South China, little is known about the correspondence between pollinator abundance and pollinator services for this plant genus. In this study, the diversity, daily activity patterns, and pollination effectiveness of insects visiting C. osmantha, C. vietnamensis and C. oleifera were investigated. A total of 24 species, belonging to four orders and 11 families, of visiting insects were identified. Apis cerana cerana Fabricius, Vespa bicolor Fabricius, V. velutina Lepeletier, V. ducalis Smith, and Phytomia zonata Fabricius were the dominant pollinators. The daily activity peaks of the five visiting insects were between 10:00 and 14:00, which may have been related to the pattern of floral resource production (particularly nectar). Cross-pollination by insects significantly increased the fruit production rates of C. osmantha, C. vietnamensis, and C. oleifera. Therefore, the wild bees and flies that pollinate wild and cultivated Camellia plants should be protected in South China.


Introduction
The fruit and seed set of many crops rely on pollination through a pollinator, especially by insects. Approximately 35% of global crop production arises from crop species that benefit from insect pollination [1]. As is well known, honey bees are the most ubiquitous and commonly used managed pollinator in the world. However, apart from a few managed bee taxa, the great majority of other pollinators (e.g., flies, moths, butterflies, and beetles) that are free-living or wild also provide an ecosystem service to crops [2,3]. Pollination involves interactions between two predominant groups of organisms: the flowering plants and the vectors of their gametes. Such interactions broadly comprise one of the most varied and widespread of all mutualistic relationships [4]. Generally, plants exhibit a remarkable diversity of floral traits that evolve in response to natural selection by a pollinator [5,6]. were first photographed with a professional single-lens reflex (SLR) camera (EOS 5D Mark II, Canon, Tokyo, Japan). Some species that were not identified in loco were collected using an insect net, and identification of the collected insects was performed using keys in the laboratory [25][26][27]. The total number of insects visiting C. osmantha, C. vietnamensis, and C. oleifera was recorded.
Within each Camellia ssp. field, four trees were randomly observed between 08:00 and 18:00 on 25-26 December 2017 and 10-11 January 2018. Floral visitors, including the number of flowers visited, were recorded hourly and pooled across the two sampling dates, resulting in a total of~120 observation hours for 12 trees of C. osmantha, C. vietnamensis, and C. oleifera.
2.3. Floral Traits of C. osmantha, C. vietnamensis, and C. oleifera To explore the potential effect of floral traits on attraction to pollinator visitation, a total of 30 flowers from 10 trees (three flowers/tree) within each Camellia ssp. field were randomly selected to measure the diameter of the corolla, height, and diameter of the stamens, and height of the pistils with a Vernier caliper at their full-bloom stage ( Figure 1).
Insects 2019, 10, x FOR PEER REVIEW 3 of 11 Tokyo, Japan). Some species that were not identified in loco were collected using an insect net, and identification of the collected insects was performed using keys in the laboratory [25][26][27]. The total number of insects visiting C. osmantha, C. vietnamensis, and C. oleifera was recorded. Within each Camellia ssp. field, four trees were randomly observed between 08:00 and 18:00 on 25-26 December 2017 and 10-11 January 2018. Floral visitors, including the number of flowers visited, were recorded hourly and pooled across the two sampling dates, resulting in a total of ~120 observation hours for 12 trees of C. osmantha, C. vietnamensis, and C. oleifera.

Floral Traits of C. osmantha, C. vietnamensis, and C. oleifera
To explore the potential effect of floral traits on attraction to pollinator visitation, a total of 30 flowers from 10 trees (three flowers/tree) within each Camellia ssp. field were randomly selected to measure the diameter of the corolla, height, and diameter of the stamens, and height of the pistils with a Vernier caliper at their full-bloom stage ( Figure 1).

Nectar Volumes and Sugar Concentrations of C. osmantha, C. vietnamensis, and C. oleifera Flowers
Nectar availability and sugar concentration were determined by sampling flowers at midday from plants within the area of pollinator counts. The nectar volume present in one flower was always too low to determine volume and sugar concentration on a per flower basis. Therefore, volume and concentration measurements were taken on pooled nectar samples from a number of flowers [28]. A total of 5-8 flowers per tree were collected for measurement to determine the dynamic changes of nectar volume and sugar concentration from the 1st (full-bloom stage) to the 5th day (end of blooming). Nectar volume was measured a calibrated micropipette (Rainin, New York, NY, USA) and concentration was measured with a hand saccharimeter (Atago, Tokyo, Japan). Three replications and a total of 111, 57, and 117 flowers were used for C. osmantha, C. vietnamensis, and C. oleifera measurements, respectively.

Effect of Insect Pollination on the Fruit Production Rates of C. osmantha, C. vietnamensis, and C. oleifera
To determine the pollination effectiveness of the insects, five trees of C. osmantha, C. vietnamensis, and C. oleifera, respectively, were selected randomly in the fields. On each of the 15 trees, one branch was enclosed in a white nylon net cage (length× width × height = 100 × 100 × 100 cm, mesh: 0.1 mm) before the onset of flowering to prevent pollination by insects (treatment group). Another branch of similar size at each tree was marked and left uncaged to allow pollination by insects (control group). The number of buds was recorded at all branches to calculate fruit production rates (number of fruits/number of buds × 100%) per branch. The cages were removed after fruit had developed.  Nectar availability and sugar concentration were determined by sampling flowers at midday from plants within the area of pollinator counts. The nectar volume present in one flower was always too low to determine volume and sugar concentration on a per flower basis. Therefore, volume and concentration measurements were taken on pooled nectar samples from a number of flowers [28]. A total of 5-8 flowers per tree were collected for measurement to determine the dynamic changes of nectar volume and sugar concentration from the 1st (full-bloom stage) to the 5th day (end of blooming). Nectar volume was measured a calibrated micropipette (Rainin, New York, NY, USA) and concentration was measured with a hand saccharimeter (Atago, Tokyo, Japan). Three replications and a total of 111, 57, and 117 flowers were used for C. osmantha, C. vietnamensis, and C. oleifera measurements, respectively. To determine the pollination effectiveness of the insects, five trees of C. osmantha, C. vietnamensis, and C. oleifera, respectively, were selected randomly in the fields. On each of the 15 trees, one branch was enclosed in a white nylon net cage (length× width × height = 100 × 100 × 100 cm, mesh: 0.1 mm) before the onset of flowering to prevent pollination by insects (treatment group). Another branch of similar size at each tree was marked and left uncaged to allow pollination by insects (control group). The number of buds was recorded at all branches to calculate fruit production rates (number of fruits/number of buds × 100%) per branch. The cages were removed after fruit had developed.

Statistical Analyses
Statistical analyses were performed using SPSS 16.0 (SPSS Inc., Chicago, IL, USA). Species richness and total flower visitor abundance as well as abundance of dominant flower visitors were compared among C. osmantha, C. vietnamensis, and C. oleifera flowers by one-way analysis of variance (ANOVA), followed by Tukey's honest significant difference (HSD) test for multiple comparisons. The tendency of dominant flower visitors towards flower constancy was evaluated by comparing the percentage of visits of each dominant visiting insect to 1, 2, and ≥3 flower(s) on one Camellia tree by one-way ANOVA, followed by Tukey's HSD test for multiple comparisons. The diameter of the corolla, height and diameter of the stamens, height of the pistils, nectar volume, and sugar concentration were compared among C. osmantha, C. vietnamensis, and C. oleifera flowers by one-way ANOVA, followed by Tukey's HSD test for multiple comparisons. The fruit production rates were compared between the treatment group and the control group for C. osmantha, C. vietnamensis, and C. oleifera separately using a paired samples t-test. Proportional data were subjected to arcsine square root transformation prior to analysis. A level of p < 0.05 was accepted as statistically significant for all statistical analyses.

P. zonata
42.5 ± 9.7 a (198) 15.6 ± 1.5 a (80) 42.0 ± 8.6 a (215) ξ The percentage of visits by one pollinating insect to 1, 2, and ≥3 flower(s) on one Camellia tree relative to the total number of visits. Values (mean ± S.E.) followed by different letters in the same row are significantly different based on Tukey's HSD test at p < 0.05. Numbers in parentheses represent sample sizes.

Effect of Insect Pollination on the Fruit Production Rates of C. osmantha, C. vietnamensis, and C. oleifera
The fruit production rates of C. osmantha trees with branches that were enclosed in a net cage to prevent pollination by insects were significantly lower than those of the control (t = −5.732, df = 4, p = 0.005). There same results were observed for C. vietnamensis (t = −2.561, df = 4, p = 0.043) and C. oleifera (t = −5.012, df = 4, p = 0.007) ( Figure 5). Furthermore, fruit production rates of C. oleifera seems to fall in-between the other two species, if pollinators are present.   Figure 5. Effect of insect pollination on the fruit production rates of Camellia osmantha, C. vietnamensis, and C. oleifera. Asterisks indicate a significant difference between the two groups based on paired samples t-test at p < 0.05.

Discussion
For C. oleifera, a total of 54 species of visiting insects, including 25 hymenopteran species, were recorded in Guangxi, South China [19,20]. Surprisingly little is known about the pollinator assemblages of C. osmantha and C. vietnamensis, although this information is crucial for ecological investigations of reproductive traits. In this study, a total of 24 species of visiting insects were identified in the field investigations, and the dominant insects visiting C. osmantha, C. vietnamensis, and C. oleifera included A. cerana cerana, V. bicolor, V. velutina, V. ducalis, and P. zonata. Although the visiting insects in this study were less abundant than those in previous reports, which may be relative to the geographical site and investigated method, the dominant species of visiting insects were consistent [19,20].
Three out of the five dominant common flower visitors to C. osmantha, C. vietnamensis, and C. oleifera, including the most dominant one, did not show a preference for one species. In contrast, V. bicolor and V. ducalis had an obvious preference for C. oleifera, although the sugar concentration of C. oleifera nectar was significantly lower than that of C. osmantha and C. vietnamensis nectar during flowering. Although nectar volumes in C. oleifera were higher than in C. osmantha, they did not differ from C. vietnamensis ( Figure 4). Therefore, nectar production, which is the main source of energy for visiting insects, does not appear to be the only driver of flower preference.
Similarly, C. vietnamensis produces larger flowers than the other Camellia species (Table 4), but was not visited more often by any flower visitor species. Usually, larger displays of flowers/inflorescences release more floral scents and are more visible than smaller displays, attracting more visiting insects [29]. This may be explained by the theory that floral and inflorescence traits evolved under a trade-off between visual and olfactory cues [30]. In fact, odor is an important cue for visiting insects, which is often influenced by a wide variety of volatile floral scent molecules [31]. These specific floral volatile organic compound mixtures attract specialist pollinators that have evolved an innate preference for them [32]. Whether the floral scent of C. oleifera consists of species-specific volatile organic compounds that attract visiting insects needs further study.
Interestingly, the only dominant flower visitor species showing a clear pattern of flower constancy within a Camellia tree did not discriminate among subspecies (Table 3). In fact, flower constancy within trees was rare among dominant flower visitors, with three out of five species either visiting a single flower and leaving the tree afterwards or visiting a row of flowers consecutively. We considered that flower visitors leave an individual plant if resources are not sufficient in the flower [33] or the perception of scent marks deposited by previous visitors [34].
Many visiting insects show specific daily activity patterns [35]. In the current study, the daily activity of the five visiting insects occurred during the photophase, with clear peak visiting times from 10:00 to 14:00 ( Figure 3). By timing anthesis and/or the presentation of floral rewards to match the activity peaks of their most efficient pollinators, and thus to ensuring the best pollination service, plants might enhance their reproductive output via increased pollen transfer [28]. Therefore, the daily patterns of these visitor activities may be related to the pattern of floral resource production in C. osmantha, C. vietnamensis, and C. oleifera. A previous study showed that the average volume of nectar in flowers exposed to pollinators varied somewhat throughout the day, reaching a maximum around midday [28]. The daily change in nectar may explain the dynamics of the visiting insects' activity in this study. In fact, daily patterns of floral resource presentation, particularly nectar presentation, have been investigated in detail for many species, and these patterns have often been examined in relation to daily variations in pollinator activity [28,36,37]. We believe that flower visitors followed maximum nectar production either by experience or by spending more time on rewarding flowers, thereby increasing the detection probability.
Camellia spp. are self-incompatible and show high fruit production rates after cross-pollination. There is also increasing evidence that cross-pollination by insects increases the fruit production rates of these species [21]. The data indicated that these dominant visiting insects play a very important role in the yield of C. osmantha, C. vietnamensis, and C. oleifera ( Figure 5). In fact, flower thinning had no significant effects on fruit production rates, e.g., in C. osmantha (Table S1). Thus, our results support the view that cross-pollination by insects increases fruit production rates.

Conclusions
In summary, our study examined the five species of insects, namely, A. cerana cerana, V. bicolor, V. velutina, V. ducalis, and P. zonata, visiting C. osmantha, C. vietnamensis, and C. oleifera in Guangxi, South China. The peak visiting times of these insects occurred from 10:00 to 14:00. Fruit production rates significantly increased after cross-pollination by these insects.