Foraging Behavior of Honeybees (Apis Mellifera L.) and Ground Bumblebees (Bombus Terrestris L.) and its Influence on Seed Yield and Oil Quality of Oil Tree Peony Cultivar ‘Fengdan’ (Paeonia Ostii T. Hong et J. X. Zhang)

The experiment was carried out in 2017 at the peony base (N 34°38′30″, E 112°39′43″, 125.5 m above sea level) in the East Garden of the Yibin District, which is one of the largest oil tree peony gardens in Luoyang city. The oil tree peony cultivar planted there is ‘Fengdan’, with an area of more than 106.7 hm². ‘Fengdan’ peony had entered the fruiting period and was eight years old. More than 3,000 plants were used for this experiment.

In March 2017, three equally sized pollination net houses were built parallelly in an east-west direction at the oil tree peony base of Luoyang's East Garden. The net houses were 45 m long from east to west, 8 m wide from north to south, 2.1 m in wall height and 3.2 m in ridge height. The nylon mesh was made of a polyethylene material with a hole 1 mm in diameter. Each house was evenly divided into three small net rooms (15 m×8 m×3.2 m). These rooms corresponded to different pollination methods, namely, honeybee pollination (HP), ground bumblebee (Bombus terrestris L.) pollination (BP), and a blank control (BC, pollination without insects), with three replicates for each treatment. Another 45 m×8 m plot was established to the west of the net houses, which included three replicates of the natural field control (FC). The plants in this treatment were pollinated naturally (Fig. 1). For HP, one hive of honeybees was placed in each net room (approximately 6,000 worker bees). For BP, one hive of ground bumblebees was placed in each net room (approximately 80–200 individuals). The honeybees were provided by the Luoyang Lilou Bee Farm, and the ground bumblebees were provided by the Beijing Agricultural Wings Bee Farm.

Fig. 1

When the first peony flower in a pollination net room opened, honeybees or ground bumblebees were placed inside. Honeybees were placed in the middle of each net room, 120 cm from the ground. Because there are no nectary glands in peony flowers, a sugar feeder (50% sugar water) was placed in each honeybee and bumblebee hive, and a water feeder (drinking water) was placed near the beehive. The ground bumblebee hives were placed in the middle of each net room, 30 cm from the ground, and a water feeder (drinking water) was placed near the beehive. The sugar water and drinking water were regularly replenished and replaced.

To better understand the living habits of honeybees and ground bumblebees for possible application in crop pollination in the future, their foraging behaviors and daily activities were observed.

Between April 14 and 19, 2019, when oil tree peony entered the peak flowering period, the flower visiting behaviors of the bees were observed. Using a stopwatch, the visit interval, single-flower visit time, number of single-flower visits, number of visited flowers per minute and number of stigma contacts by honeybees and ground bumblebees were studied (He, Li, & Zhang, 2012).

The foraging behavior parameters were defined as follows. Visit interval refers to the time it takes for a bee to leave the flower and visit the flower again, including the time when the bee temporarily leaves the flower to carry pollen in the air over the flowers when visiting the same flower and the time from one flower visit to another. Single-flower visit time refers to the total time it takes for a bee that fells on a flower to leave the flower, that is, the sum of the time spent on a single visit and the time spent combing the pollen over the flower when the bee temporarily leaves the flower. The number of single-flower visits refers to the number of visits per flower by bees (He, Li, & Zhang, 2012; He et al., 2019). The number of visited flowers per minute refers to the number of flowers that each bee visited per minute. The stigma contact ratio refers to the percentage of stigma contacts per flower relative to the total number of single-flower visits.

From April 14 to 16 (three consecutive days) 2019, the daily activities of the honeybees and ground humblebees in the pollination net rooms were separately observed. Due to a large population of the honeybees, it was difficult to accurately count the numbers of bees that left the nest, returned to the nest and returned to the nest with pollen during the peak period of daily activities. According to literature (He et al., 2019), their daily activities were observed based on the number of bees that visited flowers. Specifically, in each of the three net rooms with honeybees, two 2 m×2 m sample plots were established, and thirty flowers in full bloom were labeled in each sample plot. From 6:00 to 19:00, the number of bees that visited the flowers in each plot within 10 min every 30 min was recorded. In the three net rooms with ground bumblebees, the numbers of bees that left the nest, returned to the nest and returned to the nest with pollen within 10 min every 30 min were counted (An et al., 2007). In the meantime, the temperature and relative humidity near the nest entrance of the honeybees and ground bumblebees were recorded during different time intervals with a hand-held meteorological instrument (Kestrel 3000).

In early August, the follicles of oil tree peony were harvested (most of the follicles were a golden color). During a single-fruit harvest, the picked fruit was wrapped in newspaper, the fruit follicles of each plant were placed in a single nylon mesh bag, and the follicles of each net were placed in a single woven mesh bag. In the laboratory, the fruit follicles were placed on a ventilated and transparent balcony and matured for approximately fifteen days. After the follicles naturally cracked, the number of fruits per plant, the number and weight of seeds per follicle, and the number and weight of seeds per plant from the different pollination treatments were determined. Finally, the seed yield under different pollination treatments was calculated.

The kernel percentage of peony seeds was determined by manual peeling, and 100 seeds were taken from each sample. This process was repeated three times.

The oil yield of peony seeds was determined by a supercritical CO₂ extractor (HA220-50-06; Huaan, Nantong, China). The extraction conditions were as follows: extraction II: 40°C, 30 MPa; separation I: 40°C, 10 MPa; separation II: 35°C, 5 MPa; sample volume: 200 g; and sample material: peony kernel powder that passed through a sieve of 20–40 mesh per inch (approximately 0.42–0.84 mm).

The fatty acid content of peony seed oil was determined by gas chromatography (Agilent-7890, USA). The specific method followed GB 5009.168-2016 (2016) in the national food safety standards. Peony seed oil fatty acid content was determined by the Agricultural Products Quality Supervision and Inspection Center (Zhengzhou) of the Ministry of Agriculture.

Statistical analysis of the data was performed using SPSS 20.0. Graphs were produced using Origin 2018 software. The Pearson correlation method in SPSS 20.0 software was used to analyze the correlation between the number of single-flower visits and the number of stigma contacts for the honeybees and ground bumblebees on oil tree peony. Measurement data are presented as the mean ± standard error, and categorical data are expressed as percentages. Homogeneity of variances was tested using Levene's method. Independent t tests were used to compare the single-flower visit time, visit interval and number of visited flowers per minute between the honeybees and the ground bumblebees. The number of fruits per plant, number and weight of seeds per follicle, and number and weight of seeds per plant under different pollination treatments were compared using one-way analysis of variance (ANOVA) followed by a Bonferroni test. The kernel percentages, oil yield rates and compositions of seed oil for the different pollination treatments were compared using the nonparametric Kruskal-Wallis H method followed by a Dunn's multiple comparison test.

Honeybees and ground bumblebees showed significant differences in the number of flowers visited per minute (independent-samples t test, t=9.957, F=15.313, P<0.001) (Fig. 2A), single-flower visit time (t=−3.630, F=44.502, P<0.001) (Fig. 2B), flower visit interval (t=−9.072, F=104.792, P<0.001) (Fig. 2C), and number of single-flower visits (t=3.274, F=6.592, P=0.011) (Fig. 2D).

Fig. 2

The number of stigma contacts per flower by honeybees was 0–7, with an average of 0.92±1.32; the stigma contact ratio of honeybees was 7.70% (n=97). Furthermore, honeybees exhibited a significant correlation between the number of single-flower visits and the number of stigma contacts per flower (Pearson r=0.212, P=0.037). The number of stigma contacts per flower by ground bumblebees was 0–5, with an average of 1.84±1.06. The stigma contact ratio of ground bumblebees was 24.11% (n=63) (Fig. 3), and there was a significant correlation between the number of single-flower visits and the number of stigma contacts per flower (Pearson r=0.781, P<0.001). There was a significant difference in the stigma contact ratio between honeybees and ground bumblebees (independent-samples t test, t=−5.939, F=2.254, P <0.01) (Fig. 3).

Fig. 3

The daily activities of honeybees and ground bumblebees are shown in Fig. 4. Honeybees left the nest later than ground bumblebees (after 8:00 vs. after 6:00), and their homing time was earlier than that of ground bumblebees (before 17:30 vs. before 19:00). The peak period of visiting flowers for honeybees was 9:30–11:45, during which the number of visited flowers was approximately 57.72% of the total number of flowers visited during the day. Between 9:30 and 11:45, the temperature was 27–31°C, and the relative humidity was 41%–60% (Fig. 4A). The peak period of visiting flowers for ground bumblebees was 6:00–10:00, during which the number of visited flowers was approximately 59.43% of the total number of flowers visited during the day. During this period, the temperature was 16–27°C, and the relative humidity was 48%–70% (Fig. 4B).

Fig. 4

The four pollination treatments showed significant differences in the number of fruits per plant (one-way ANOVA, df_(3,789)=2.931, P=0.033), number of seeds per follicle (df_(3,3463) =348.960, P<0.01), seed weight per follicle (df_(3,3477)=100.932, P<0.01), number of seeds per plant (df_(3,379)=55.974, P<0.01), and seed weight per plant (df_(3,379)=34.737, P<0.01) (Tab. 1). Compared with the control, bee pollination significantly improved these indices (all P<0.05). Compared with honeybee pollination, ground bumblebee pollination caused a significantly larger number of seeds per follicle, seed weight per follicle, number of seeds per plant, and seed weight per plant (all P<0.05). No significant differences in kernel percentage or the oil yield rate were observed between pollination treatments (Kruskal Wallis H Test, χ²=7.578 P=0.056; χ²=6.765, P=0.080) (Tab. 1).

Palmitic acid, stearic acid, oleic acid, linoleic acid, α-linolenic acid, arachidic acid and arachidonic acid were detected as components of the peony seed oil. The unsaturated fatty acid content in peony seed oil was above 91%, and the relative α-linolenic acid content was as high as 43%. The compositions of peony seed oil for the different pollination treatments are summarized in Tab. 2. There were no significant differences in palmitic acid (χ²=3.167, P=0.367), stearic acid (χ²=6.514, P=0.089), oleic acid (χ²=5.227, P=0.156), linoleic acid (χ²=5.227, P=0.140), α-linolenic acid (χ²=6.649, P=0.084), or unsaturated fatty acid (χ²=4.006, P=0.261) content between the different pollination treatments.