Lethal effects of tea‐oil Camellia on honeybee larvae due to pollen toxicity

Flowers in the tea family (Theaceae) are bowl-shaped, with numerous anthers presenting large amounts of freely accessible pollen simultaneously. Species that flower mainly in winter and early spring are pollinated by bees and/or birds. In the 1950s, beekeepers in China observed larval mortality if honeybees (Apis cerana) had foraged on the flowers of Camellia oleifera (Zhu, 1957). Although early studies proposed that the lethal effect of C. oleifera on honeybee larvae might be caused by toxic nectar and/or pollen (Kang and Fan, 1991), later studies have focused on nectar sugar components and suggested the oligosaccharides in nectar as essential toxic components for Apis mellifera (Li et al., 2022). Our hand pollinations showed that both fruit and seed production were pollinator limited in this winter-flowering shrub. Field observations on three populations in Guizhou, Guangxi and Hubei Provinces showed that five potential pollinators were white-eyes Zosterops japonicus, two species of wasp, Vespa mandarinia and Vespa velutina nigrithorax, and one species of Andrena bee and the honeybee Apis cerana (Figure 1A–D). In three sites wasps showed the highest mean visitation frequency (Wald χ2 = 28.98, P < 0.001; Table S1), accounting for 50.8% of the total visits, while the other three types of visitors (bird, solitary bee and social bee) did not differ (Wald χ2 = 1.916, P = 0.384; Figure S1a). All visitors foraged mainly for nectar in C. oleifera although both types of bees collected pollen. Honeybees mainly collected nectar (Figure S1b; 89.9% of visits in Wuhan and 55.5% in Guilin; G-test = 144.11, P < 0.001; G = 3.95, P = 0.047, respectively) rather than pollen. To compare pollination effectiveness between the bee and the bird, we caged flowers to exclude bird visits and bagged flowers to exclude both bird and bee visits (Figure 1B). Compared with open-pollinated flowers, fruit and seed set of caged flowers decreased significantly (Wald χ2 = 85.449, P < 0.001; Wald χ2 = 51.13, P < 0.001, respectively) by 56.28% and 27.61% (Figure 1F), indicating that bird visits greatly enhanced reproductive success. Bagged flowers did not set fruits. We conclude that C. oleifera could not be self-pollinated and fruit/seed set are pollinator limited without apomixis (Figure 1G). To see whether the two types of bees positively collected pollen and to compare their pollen transfer efficiency with that of the wasp species, we quantified pollen removal and deposition in a single visit. Our results showed that pollen removal (Wald χ2 = 0.175, P = 0.676), pollen deposition (Wald χ2 = 0.605, P = 0.437) and the ratio of pollen deposition to pollen removal (Wald χ2 = 0.181, P = 0.913) did not differ significantly among three visitors V. velutina, Andrena sp., and A. cerana (Table S1). The honeybee did not remove or collect more pollen grains than the wasp and solitary bee, consistent with our observation that honeybees rarely groomed pollen of C. oleifera into their pollen loads. Foraging behavior of floral visitors on winter-flowering tea-oil camellia (Camellia oleifera), comparisons of fruit/seed production under different pollination treatments, content of theasaponin among six tissues and cumulative survival of honeybee larvae among four diet treatments with pollen and six diet treatments with theasaponin (A) The warbling white-eye (Zosterops japonicus) collecting nectar; note the exposed pollen grains on the beak (pink arrow). (B) The wasp Vespa velutina collecting nectar from a caged flower from which bird visitations are excluded. (C) A solitary bee crawling on the anthers to collect pollen grains into its pollen baskets on the hind legs (pink arrow); note the stigma in the middle of the stamen cluster, positioned slightly lower than the stamens (green arrow). (D) Chinese honeybee occasionally carrying pollen of C. oleifera in the two corbiculae (pink arrow). (E) Developing capsules in April. Comparisons of fruit and seed set (mean ± SE) under four pollination treatments with cages (F) and two bagged treatments (G), E+bag indicates emasculation (anthers removed) and bagged. (H) Comparisons of the content of theasaponin (mean ± SE) in six tissues among three sites. Different letters indicate significant differences at P < 0.05 under the generalized linear models. Comparisons of survival proportion of honeybee larvae among four diet treatments with pollen (1 mg/larva) and six diet treatments with theasaponin. Photographs (I–L) show the status of honeybee larvae survival for 5 d on diets with yeast (I), Brassica napus pollen (J), 50:50 B. napus pollen and Camellia oleifera pollen (K), or pure C. oleifera pollen (L), respectively. Different letters indicate significant differences at P < 0.05 under the Log Rank Test among four diet treatments with C. oleifera pollen or none (M) and among diet treatments with six concentrations of theasaponin (N). Triterpenoid theasaponins (TS), secondary metabolites characteristic of Theaceae species, occurring in various tissues especially in the fruits of C. oleifera as well as in leaves (Guo et al., 2018), are known for their bitter taste and antimicrobial and insecticidal properties. Here, we hypothesize that the primary cause of bee larvae mortality could be toxic components in tea-oil pollen rather than sugars in floral nectar. To detect the TS content in C. oleifera, we collected six tissues (stem, leaf, petal, fruit, pollen, and nectar) from at least 15 flowering individuals at each site. HPLC analyses showed that the TS content was significantly higher (Wald χ2 = 611.75, P < 0.001) in pollen (292.65 ± 11.69 mg/g) than in fruits (160.53 ± 10.93) and other tissues, and not detectable in the nectar across the three sites (Figures 1H, S1c). The content of proteins (Wald χ2 = 2.706, P = 0.258) and lipids (Wald χ2 = 0.524, P = 0.77) in pollen did not differ among the three sampled sites (Table S1). The ratio of protein to lipid content (P/L) in pollen was 1.05–1.39, lower than the P/L ratio of 5:1 that is considered to be preferred for collecting by bees. Our investigation of nectar properties in the three sites indicated that C. oleifera usually produced larger volumes (333.33 ± 48.28 μL) and relatively lower sugar concentrations (21.84 ± 0.01%) than bee-flowers which is >30% in general. Nectar sugar component analysis showed that sucrose accounted for >70% (70.19 ± 0.06%), and that fructose and glucose accounted for the rest, consistent with Camellia nectar characteristics of sucrose as the major component, a type that passerines prefer (Sun et al., 2017). To test the possible lethal effect of pollen toxicity on honeybee larvae, we conducted two diet treatments during larval development, adding C. oleifera pollen or seven graded concentrations of TS into the larval diet. The pollen-feeding toxicity tests showed that the proportion surviving Brassica napus pollen treatment did not differ (Wald χ2 = 0.203, P = 0.652) from the proportion surviving yeast treatment, indicating that B. napus pollen used as a control diet had no obvious lethal effect on honeybee larvae. However, C. oleifera pollen induced 16.7% mortality on the 2nd d of toxicity tests (Figure 1M). On the 4th d, the survival on 50:50 B. napus and C. oleifera pollen first decreased to 70.8%, and that on pure C. oleifera pollen decreased to 58.3%. On the 6th d, the last treatment, survival on 50:50 B. napus and C. oleifera (45.8%) and C. oleifera (29.1%) pollen was significantly lower than that on yeast (91.7%; Wald χ2 = 18.785, P < 0.001, Wald χ2 = 11.171, P = 0.001, respectively) or B. napus pollen (87.5%; Wald χ2 = 16.797, P < 0.001, Wald χ2 = 8.675, P < 0.001, respectively) treatments (Figure 1I–L). These results indicate that the pollen of C. oleifera had a significant lethal effect on honeybee larvae. The TS toxicity tests showed that the proportion of honeybee larva survival decreased with the higher concentration of TS (Figure 1N). Theasaponin at 35 and 50 mg/mL, induced 80.3% and 95.9% mortality, respectively, on the 2nd d indicating that the high toxin concentrations had strong lethal effects. On the 6th d, survival at 5 mg/mL (approximately 1.71% of its natural concentration in pollen) was 53.8%, significantly lower than survival (79.2%) in the dimethyl sulfoxide treatment (DMSO; cell culture grade from Solarbio; see Jones and Agrawal, 2016) (Wald χ2 = 9.623, P = 0.002) and in the 0.1 mg/mL TS treatment (75.0%) (Wald χ2 = 9.844, P = 0.002), indicating that TS in pollen had a lethal effect on honeybee larvae at a low concentration. Why do numerous flowers, especially bird-pollinated ones, present freely accessible pollen? This trait is notably prevalent in species where pollen grains in numerous anthers are often simultaneously presented (Thomson et al., 2000), as in the Cactaceae, Malvaceae, and Theaceae. Unlike species with gradual pollen presentation, in which small pollen doses limit pollen removal by frequent pollinator visits, species with simultaneous pollen presentation usually donate large amounts of pollen to recruit infrequent visitors (Thomson et al., 2000; Song et al., 2019). Although this pollen presentation theory evidently represents a widespread adaptive strategy involving various plant lineages (Thomson et al., 2000), the disadvantageous effect of simultaneously presenting a large pollen mass, the risk of loss from pollen thieves, remains to be addressed. The chemical defense of nectar and pollen has been considered to play a pivotal role in filtering pollinators and florivores (Irwin et al., 2010). There is growing evidence that secondary metabolites in pollen are more diverse and often at much higher concentrations than those in nectar (Palmer-Young et al., 2019). The theasaponin-containing pollen could deter overexploitation by honeybees; little of the freely accessible pollen passively placed on the bee body was groomed into the corbiculae, suggesting some chemical protection from pollen collectors. Evolutionary transitions from bee- to bird-pollinated flowers have involved multiple modifications of floral morphology and nectar properties (Thomson and Wilson, 2008; Trunz et al., 2020). Here we have shown the role of pollen toxicity in the tea-oil camellia, revealing a new "anti-bee" pollen trait in a bird flower. This work was supported by the National Natural Science Foundation of China (Grant Nos. 32030071, 31730012). We thank Jun-Han Li and local staff in the conservation areas for their help in the field, Hua Yan and Xiao-Chen Yang for their guidance and help in bee toxicity experiment, Jian Wan for advice of chemical component analysis and Sarah Corbet for helpful suggestions for an early version of the manuscript. The authors declare no conflict of interest. S.-Q.H. designed the study. C.Z. and Y.L.L. performed the study and collected data, C.Z. analyzed the data, C.Z., H.H.F. and S.-Q.H. drafted the manuscript. All authors approved the final version of the manuscript and agreed to be held accountable for the content therein. Additional Supporting Information may be found online in the supporting information tab for this article: http://onlinelibrary.wiley.com/doi/10.1111/jipb.13731/suppinfo Figure S1. Comparisons of the total visit frequency, foraging behavior of honeybees and the total contents of theasaponin (TS) Table S1. List of floral traits, visiting frequency and pollination efficiency Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.