The bumblebee Bombus ardens ardens (Hymenoptera: Apidae) visits white clover in orchards before Oriental persimmon blooms

Flowers on the ground of orchards can provide substantial resources for wild pollinators of orchard trees. Few studies, however, have examined the relative importance of groundcover flowers to orchard pollination by analyzing pollen on the body surface of pollinators. Oriental persimmon trees bloom within the longer blooming period of white clover, which is occasionally found as a flowering plant on the ground of persimmon orchards in Japan. The present study compared the insect species assemblage collected on persimmon flowers with that on clover. Before persimmon bloomed, Bombus ardens ardens and Apis cerana japonica were the major visitors of clover flowers. Once persimmon bloomed, the former was the most abundant bee that visited persimmon flowers over the flowering period. Apis mellifera was captured only on clover flowers. We found numerous clover pollen grains on the body surface of bumblebees captured on persimmon flowers, but far fewer persimmon pollen grains on bees that visited clover. These findings show that B. ardens ardens utilized the clover flowers under the orchards before persimmon bloomed.

Pollination service for fruit production is provided by a greater diversity of pollinators, including wild insects, than previously understood (Garibaldi et al. 2013(Garibaldi et al. , 2016. The European honeybee, Apis mellifera L. (Hymenoptera: Apidae), is the representative domesticated pollinator in worldwide fruit production; the species is highly effective due to numerous workers in each hive (Morse 1991). However, introducing honeybees and other managed bees into farms may damage local ecosystems and negatively affect the biodiversity of native wild pollinators (e.g., Huryn 1997). Therefore, enhancing the pollination service performed by diverse wild pollinators seems to be the most sustainable method for several crops (Hoehn et al. 2008;Albrecht et al. 2012); for example, the fruit set of coffee can be predicted by the number of wild pollinating species (Klein et al. 2003). Thus, it is becoming increasingly important to investigate wild pollinators' contribution to fruit production (Bretagnolle & Gaba 2015).
Flowers from native, cultivated, and alien species (Sargent & Ackerly 2008) around fruit trees can provide considerable floral resources to wild pollinators outside of the generally short flowering periods of fruits (Nicholls & Altieri 2013). Therefore, in addition to natural vegetation around orchards (Garibaldi et al. 2011;Norfolk et al. 2016), understory plants in orchards can also serve as important floral resources for wild pollinators (Kammerer et al. 2016). Karamaouna et al. (2019) reported that groundcover with suitable flowering species could be part of a sustainable olive crop management system, providing food for pollinating insects. However, if a trade-off exists between wild bees' flowervisiting frequency to fruit trees and that to groundcover plants, then increasing groundcover plants may cause a decline in pollination success of the fruits. A reliable way to investigate the effect of groundcover plants is to compare the pollen grains attached on the body surface of pollinators captured on the flowers of crops with those captured on surrounding vegetation.
A study by Nikkeshi et al. (2019) revealed that the wild bumblebee Bombus ardens ardens Smith (Hymenoptera: Apidae) is the most important pollinator of Oriental (or Japanese) persimmon, Diospyros kaki Thunb. (Ericales: Ebenaceae), in Hiroshima, Japan. In one of the orchards at the study site, white clover, Trifolium repens L. (Fabales: Fabaceae), was the dominant groundcover vegetation. This legume has been studied as a promising cover crop in orchards for regulating weeds, conserving water, providing nitrogen, and improving tree growth in apple orchards (Granatstein & Mullinix 2008), as well as suppressing aphids in apple orchards and psyllids in pear orchards by providing resources for the pests' natural enemies (Haley & Hogue 1990;Rieux et al. 1999). Leguminous cover crops such as T. repens are also bee-attractant plants (Lee-Mäder et al. 2020). Preliminary observations at our study site showed that a variety of insects including B. ardens ardens visited the flowers of T. repens, suggesting that this cover crop provides floral resources to pollinators of D. kaki.
The main objective of this study was to compare the flower visitors on D. kaki with those on T. repens. The insect species assemblages were identified by capturing flower visitors on both flower species from pre-anthesis until post-anthesis of D. kaki, and the pollen grains on the body surface of insects were then identified and counted to assess which insect species are involved in the pollination of D. kaki. We also discussed whether the insects that had visited the T. repens flowers switched to the floral resources of D. kaki when the trees started blooming.
Flower visitors were captured on T. repens and D. kaki in persimmon orchards at the Institute of Fruit Tree and Tea Science, National Agriculture and Food Research Organization (Hiroshima, Japan) in 2018 (for the procedures, see Appendix S1). No domesticated pollinators, such as A. mellifera, were placed in the orchards. Of the 399 insects captured on T. repens flowers, 396 were identified to the species level based on morphology and, if necessary, confirmed by using CO1 DNA barcoding (Tables S1 and S2). The sequences were deposited in the DNA Data Bank of Japan (DDBJ accession numbers LC500071-LC500094). In total, 63 Apis cerana japonica Radoszkowski (Hymenoptera: Apidae) and 89 A. mellifera were captured on the flowers of T. repens, and 22 and 0, respectively, were captured on those of D. kaki on 17, 21, and 24 May (Table S1). The ratios of the numbers of these honeybee species captured on the two plant species differed significantly (Fisher's exact test, P < 0.0001). Likewise, syrphid flies (Diptera: Syrphidae), which consisted of Eupeodes corollae (Fabricius), Melanostoma mellinum (L.), Sphaerophoria indiana Bigot, and Sphaerophoria macrogaster (Thomson), were collected only from the flowers of T. repens (Table S1). Lasioglossum scitulum (Smith) (Hymenoptera: Halictidae) was the only species in the genus that was captured on the flowers of T. repens, whereas Lasioglossum japonicum (Dalla Torre), Lasioglossum mutilum (Vachal), Lasioglossum occidens (Smith), and Lasioglossum proximatum (Smith) were also captured on the flowers of D. kaki (Table S1). The insect community captured on T. repens flowers on the five collection dates (10, 17, 21, and 24 May and 1 June 2018) was significantly different from that captured on D. kaki flowers on the three collection dates (one-way PERMANOVA nested by collection date; d.f. = 1, F = 6.36, P = 0.019). Non-metric multidimensional scaling illustrated the clear difference between the two insect communities that visited T. repens and D. kaki (Fig. 1). Due to its bell-shaped flowers and hard petals, the availability of D. kaki flowers for insects is restricted by body size and proboscis length (Miura 1982). Bumblebees are reported to prefer bell-shaped flowers, such as blueberry and blue heath (Macfarlane 1992;Kasagi & Kudo 2003). It is noteworthy that no A. mellifera was captured on D. kaki flowers over the flowering period, suggesting that this species was attracted by T. repens more than by D. kaki.
The proportion of the total number of bees in each functional group (at the species level for honeybees) captured on the flowers of T. repens and D. kaki are shown in chronological order in Fig. 2. For bumblebees, 192 individuals captured on both flowers were identified as B. ardens ardens, with one exception, Bombus ignitus Smith (Hymenoptera: Apidae). On 10 May, before D. kaki began to bloom, A. cerana japonica and B. ardens ardens were the major bee species captured on the flowers of T. repens (Fig. 2a). From 17 to 24 May, when D. kaki bloomed, B. ardens ardens was consistently dominant (>60%) among the bees captured on the flowers of D. kaki (Fig. 2b), while this species gradually decreased on the flowers of T. repens (Fig. 2a). On 1 June, when D. kaki had finished blooming, the bumblebee was not captured on the flowers of T. repens, and A. mellifera represented 89.4% of all captured bees. These data suggest that B. ardens ardens did not return to the flowers of T. repens when the flowering period of D. kaki terminated, even though T. repens was still in bloom. This bumblebee might have found other floral resources outside the orchard; Deutzia crenata Siebold et Zucc. (Cornales: Hydrangeaceae), which is potentially visited by bumblebees (Inoue et al. 2008), bloomed simultaneously with D. kaki around the orchards. Alternatively, colonies may have collapsed at the end of the breeding season.
The average number of pollen grains on the body surface of each functional group captured on the flowers of T. repens on 21 and 24 May is shown in Table 1. Pollen grains were counted after cutting away each sample's hind legs to exclude pollen loads. The functional groups (at the species level for bumblebees and honeybees) used for the analysis (A. cerana japonica, A. mellifera, B. ardens ardens, and small bees) had large average numbers (>20,000) of Trifolium pollen grains. A smaller number of Diospyros pollen grains was found on the body surface of B. ardens ardens. In addition to Trifolium and Diospyros pollen, Vicia and Vitis pollen grains (≤3,000 and ≤5,000, respectively) were detected on the body surface of some of the functional groups.
The average number of pollen grains on the body surface of each functional group captured on the flowers of D. kaki from 16 to 24 May was also calculated (Table 1). The numbers of Diospyros pollen grains on each functional group were described previously (Nikkeshi et al. 2019). Considerable numbers (average >8,000) of Trifolium pollen grains were found on the surface of B. ardens ardens, carpenter bees, and small bees. In addition to Diospyros pollen, B. ardens ardens possessed pollen of Actinidia, Trifolium, Triodanis, Vicia, and Vitis, which appeared to be pollen of Actinidia deliciosa  (Table S3). This observation suggests that B. ardens ardens forages on a relatively wide variety of flower resources. In contrast, A. cerana japonica captured on D. kaki flowers possessed pollen of only two genera, Diospyros and Trifolium, implying that it visited the targeted plant species consecutively.
The numbers of pollen grains of Trifolium and Diospyros on the body surface of B. ardens ardens captured Figure 2 Proportions of the total number of bees captured on the flowers of Trifolium repens (a) and those of Diospyros kaki (b) for each functional group. The functional groups are bumblebees (black), carpenter bees (dotted white), medium-sized bees (diagonal lines), small bees (white), the native honeybee Apis cerana japonica (dark gray), and Apis mellifera (light gray). N.D., means no data. on flowers of T. repens (4 individuals) and on flowers of D. kaki (20 individuals) were plotted (Fig. 3). The individuals from T. repens possessed no or small numbers of Diospyros pollen grains, whereas most of the ones from D. kaki possessed large numbers of Trifolium pollen grains. These findings demonstrate that B. ardens ardens had often visited the flowers of T. repens before visiting the flowers of D. kaki at our study site.
Both the number and the proportion of B. ardens ardens among all bees captured on the flowers of T. repens gradually decreased during the observation period, whereas this pollinator was dominant among those on the flowers of D. kaki throughout its flowering period. In addition, by identifying and counting pollen grains, the most prevalent pollen grains on the body surface of this bumblebee proved to be of two genera, Diospyros and Trifolium. Thus, it was conceivable that B. ardens ardens had visited the flowers of T. repens and switched its targeted floral resource to D. kaki. This was supported by the observation that most B. ardens ardens individuals captured on the flowers of D. kaki possessed large numbers of Trifolium pollen grains in addition to Diospyros ones.
In the present study, it was highly likely that B. ardens ardens foraged for T. repens flowers before the anthesis of D. kaki and for D. kaki flowers during its blooming period. The flowers of D. kaki might have been more attractive than those of T. repens for B. ardens ardens. Alternatively, competition with honeybees for T. repens The most probable species for each genus other than Diospyros and Trifolium is Actinidia deliciosa (kiwifruit; planted at the study site), Lonicera japonica (Japanese honeysuckle; weed), Triodanis perfoliata (common Venus' looking-glass; weed), Vicia sativa subsp. nigra (narrow-leaved vetch; weed), and Vitis spp. (Asian common wild grape; weed and grape planted at the study site), based on the list of plant species in bloom in and around the orchard (Table S3). flowers might explain why B. ardens ardens gave more priority to D. kaki as a floral resource. In contrast, A. mellifera continued to forage for T. repens flowers after the anthesis of D. kaki. This difference in foraging strategy between the bee species is consistent with a report that bumblebees were quicker to abandon the degrading food source than were honeybees (Townsend-Mehler et al. 2011). To identify the switching mechanism of B. ardens ardens from T. repens to D. kaki, it would be necessary to compare the following before and after D. kaki anthesis: (i) the quantity and quality of the two floral resources; and (ii) visitation frequency of this bumblebee on both flowers.
Our results indicate that the groundcover T. repens attracts and supports the pollinator B. ardens ardens before the anthesis of D. kaki. Twenty individuals of B. ardens ardens were captured on T. repens flowers before D. kaki bloomed, suggesting that the orchard was attractive as a feeding site even before the anthesis of fruit trees; this bumblebee prefers to nest in forest (including secondary forest) and open lands at the foot of mountains (Katayama 2007;Ushimaru et al. 2008) and very likely lived in such locations around the orchard. However, T. repens flowers might compete with D. kaki flowers for pollinators and reduce pollination of the crop. In fact, negative effects of T. repens flowers on crop pollination in kiwifruit, apple, and cherry orchards have been reported (Clinch 1984;MacRae et al. 2005;Holzschuh et al. 2012). Therefore, further studies should examine whether growing T. repens on the ground of D. kaki orchards has a positive effect on sustaining the local population of B. ardens ardens and any significant effect on pollination of D. kaki by this species.

ACKNOWLEDGMENTS
We are grateful to Dr. Shoko Nakamura (Forest Research and Management Organization) for identifying Bombus spp., Dr. Masayoshi K. Hiraiwa (Institute of Agrobiological Sciences, NARO) for identifying syrphid flies, and Prof. Ryoichi Miyanaga (Shimane University) and Dr. Ryuki Murao (Regional Environmental Planning Co., Ltd.) for verifying our identification of Andrena spp. and Lasioglossum spp. We thank Ms. Satomi Kohara for her technical help. This work was supported financially by the Ministry of Agriculture, Forestry, and Fisheries, Japan, through a research project entitled "Monitoring and enhancement of pollinators for crop production." Rieux R, Simon S, Defrance H (1999) Role of hedgerows and ground cover management on arthropod populations in pear orchards. Agriculture, Ecosystems and Environment 73, 119-127. Sargent RD, Ackerly DD (2008)

SUPPORTING INFORMATION
Additional Supporting Information may be found online in the Supporting Information section at the end of the article.
Appendix S1. Experimental procedures. Table S1. Species compositions of visitors captured on the flowers of Trifolium repens and Diospyros kaki in 2018. Table S2. Identification of 27 individuals (Andrenidae, Colletidae, Eumenidae, Halictidae, Ichneumonidae, Megachilidae, and Syrphidae) captured on the flowers of Trifolium repens. Table S3. Plant species in bloom in and around the Diospyros kaki orchard.