Despite growing awareness of the importance of monitoring wild crop pollinators worldwide, there are still few reports, especially in East Asia. Considering ongoing global warming may change the distribution range and diurnal activity of pollinators, it is necessary to describe current geographic and diurnal patterns. We clarified pollinators of Cucurbita maxima Duchesne (Cucurbiales: Cucurbitaceae) in three geographically distinct (>350km, minimum) areas in Japan, focusing on diurnal variation. Apis mellifera L. (Hymenoptera: Apidae) and Halictidae (Hymenoptera) were observed in all of the experimental gardens. Apis cerana japonica Radoszkowski (Hymenoptera: Apidae) were mainly observed in Mie and Kagoshima, while Bombus diversus diversus Smith (Hymenoptera: Apidae) were observed only in Ibaraki. The peak time of flower visits depended both on bee taxa and area and, interestingly, did not necessarily synchronize with the timing of highest pollen loads and the probability of stigma contact. In particular, visits and probability of contacting stigmas of Halictidae tended to increase as time passed, whereas pollen grains on their bodies sharply decreased with time; only a few individuals of Halictidae spp. that visit early can become effective pollinators. There were no differences in yields between supplementary hand- and natural-pollination in all areas, and flower-enclosure experiments using different mesh sizes clarified that small insects that can go across an approximately 4mm mesh may not transport sufficient pollen for fruit set. Our study demonstrated that pollination effectiveness, which is usually regarded as a static value, within a taxon can fluctuate in the space of just several hours. Considering such diurnal patterns can be altered by climate change, we need to carefully monitor the diurnal temporal patterns of pollinators worldwide.

Flower visitor survey

To clarify relative pollinator importance of the flower visitors, considering diurnal and geographical variations, we conducted (1) flower visitor sampling, (2) assessment of the amount of pollen on body surfaces, and (3) the examination of insect behaviors, from May to July 2019. Study plots (25–50m²) were established on experimental fields of C. maxima in Ibaraki, Mie, and Kagoshima, and at each plot, insect sampling was conducted for 5 minutes per hour. 

We captured each insect individual on male and female flowers with a 5ml plastic vial directly or with an insect net (36cm in diameter) by sweeping very slowly and then into the vial. Those vials were immediately put into cooler boxes, to keep the insects calm and to not let pollen grains fall. After the survey, all the vials containing insect individuals were labeled and stored in freezers at -20°C. Since the flowering period and the active time of flower visitors depend on daily temperature and sunrise time, the sampling period was optimized at each site. Specifically, collections were conducted at ten plots from 5:00 to 11:00 on 17 and 27 June, and 9 July in Ibaraki, at three plots from 5:00 to 12:00 on 19 and 25 June and 3 and 5 July in Mie, and at three plots from 6:00 to 11:00 at three plots on 21 and 29 May in Kagoshima. Additionally, extra samples from Kagoshima collected on 8 May were added to increase the replications for pollen load assessment. Specimens were morphologically identified to the lowest taxonomic level where possible at the conservation ecology laboratory at the University of Tsukuba. Identification of Halictidae and Nitidulidae (Coleoptera) was helped by expert taxonomists. Bees and large wasps were identified to species, genus, tribe, or family level. Other insect groups were identified to order level at least.

The amount of pollen load of each insect was estimated, following the method described in detail in Nikkeshi et al. (2019). Briefly, we cut and eliminated hind legs with corbicular pollen loads using scissors and poured 0.4M sucrose solution (1.0–6.0mL, depending on the body size) into each vial with an insect sample. This solution is an isotonic solution for pollen (Nikkeshi 2022) and prevents pollen accumulating at the bottom due to its high viscosity (Nikkeshi et al. 2021; Nikkeshi 2022). After shaking it in order to separate pollen from the insect bodies and uniformize it in the solution, we sampled 10μL of the solution, and counted the number of pollen grains on a microscope slide by microscope. In this process, we counted only squash pollen, which has a remarkably larger size (more than 170㎛) compared with the other plant species in our experimental gardens. We repeated this sampling and counting five times, and calculated the average number of pollen grains per 10μL. Finally, we estimated pollen loads of each insect, the total number of pollen grains adhered to each body surface except for the hind legs, by multiplying the average number of pollen grains per 10uL and the initial solution volume.

To examine the insects’ movement after landing on a flower, we observed female and male flowers for 15 minutes, and recorded insect behaviors considering four categories (i.e., pollen foraging, nectar foraging, wandering, and unknown). In the case of female flowers, we recorded whether the observed insects touched the stigmas. We defined “pollen foraging” as behaviors that involve mouthparts contacting anthers, or collecting pollen by rubbing the legs. “Nectar foraging” involves visitors inserting their heads or proboscis into the nectary at the flower base, and “wandering” involves walking around or standing still on the flower without foraging. In this observation, A. mellifera and A. cerana japonica were categorized as “honeybees”, due to the difficulty of discriminating between two Apis species in the field. This observation was made from 7:00 to 10:30 on 7 days from June 19 to July 8 in Ibaraki (8.25 hours in total) and from 7:30 to 11:30 on May 15 and 17 in Kagoshima (2.5 hours in total). 

Assessment of the contribution of the flower visitors to the squash yield

To verify the smallest body size of flower-visiting insects that contribute to the fruit and/or seed set, we conducted an enclosure experiment manipulating accessibility to flowers depending on the body size of insects. One day before anthesis, female buds were randomly assigned to enclosure treatments: (1) non-woven bag (Daiso Industries, CO., Ltd, Hiroshima, Japan), (2) fine mesh bag (inner length: 4.44 × 2.92mm: “4mm mesh”, Tomoyasu Works, CO., Ltd., Osaka, Japan), (3) coarse mesh bag (inner length: 7.04 × 7.63mm: “9mm mesh”, Tomoyasu Works, CO., Ltd., Osaka, Japan), (4) open pollination, and (5) supplementary hand pollination. Non-woven bags were supposed to admit no insects, fine mesh bags were supposed to admit Halictidae and other tiny insects, and coarse mesh bags were supposed to admit all insects visiting squash flowers, including honeybees and bumblebees. In addition, by comparing treatments (4) and (5), we examined whether fruit and seed productions were limited by the amount of pollen reception of stigmas in the natural condition. A series of experiments was carried out from June 21 to July 8 in Ibaraki (7 replications), from June 6 to 25 in Mie (14 replications), and on May 16 in Kagoshima (5 replications). To confirm effectiveness of bag control treatment (i.e., coarse mesh bag), we walked around near the experimental flowers, and observed that honeybees and Bombus diversus diversus entered the coarse mesh bags.

To protect the experimental flowers from resource competition within the same vine, we eliminated female flowers and buds positioned within five joints from the targeted flowers in Ibaraki, and all female flowers and buds other than the targeted flowers in Mie and Kagoshima. One week after the experimental day, we checked whether the experimental flowers set fruits. After fruit maturation, we compared the number of seeds per fruit and fruit weight between the open (i.e., natural) pollination and the supplementary hand pollination condition.

At a laboratory, to further ensure the accuracy of enclosure treatment, we measured the thorax width of the specimens using a microscope to compare with the inner length of meshes (Sonoda et al. 2022). Since the thorax of Hymenoptera is thicker than the other orders, we also measured the length of the thorax (i.e., the maximum length of the thorax from the back to front). Because the thorax width of Coleoptera tends to be shorter than the abdomen, we measured maximum width of their abdomens. Finally, we tested whether the specimens were able to pass through the mesh physically by hand, and confirmed our experimental setting worked as intended (Table A2).