Flowering, Forage Value, and Insect Pollination in Borage (Borago Officinalis L.) Cultivated in Se Poland

The study was conducted in Lublin (51°16′N, 22°30′E), SE Poland between 2007 and 2009. The study area is characterized by an average annual air temperature of 8.3°C and average annual precipitation of 550.6 mm.

Experimental plots sized 6 m × 3 m were established on grey brown podsolic soil. The plots were fertilised before sowing with Azofoska (Inco S.A., Poland) fertiliser, with the 13:6:17 NPK ratio plus S, Ca, and microelements. The seeds were sown directly into the soil in the middle of April. The spacing between rows was 60 cm. After emergence, the seedlings were thinned to obtain a ca. 40 cm distance between plants in a row. The average density of the plants was 5 · m⁻² plot.

During the period 2007–2009, the flowering onset and termination were recorded. In 2007, the seasonal flowering pattern was examined as well. To this end, six plants were randomly chosen and marked prior to the opening of the first flowers. On the consecutive days of blooming, the number of open flowers on each plant was counted until blooming terminated. The daily flowering patterns were observed on the 16^th and 17^th July 2007 at the peak flowering of the species, and the number of newly opened flowers on 8 stems was recorded every two hours from 8:00 to 20:00 (GMT + 2 hrs). Additionally, the total number of flowers developed per plant and per 1 m² area was determined. The flower life span was observed from the loose bud stage until the petals were shed. These observations were performed on six flowers each year.

The distribution of nectaries in fresh flowers was investigated under an Olympus SZX12 stereoscopic microscope (Tokyo, Japan). Next, the nectaries were prepared for SEM. The flower bases were fixed in 2.5% glutaraldehyde in phosphate buffer (pH 7.4; 0.1 M) at 4°C for twelve hours. Then, the material was washed in phosphate buffer and dehydrated in a graded acetone series. Afterwards, plant material was critical-point dried using liquid CO₂, sputter-coated with gold and examined with TESCAN/VEGA LMU SEM (TESCAN, Brno, Czech Republic) at an accelerating voltage of 30 kV.

To quantify the 24-hr rate of nectar secretion, the flower buds were isolated in the field and the accumulated nectar was collected (Corbet, 2003) using Jabłoński’ pipettes and weighed (in mg). In total, fifty-three nectar samples were collected during this study. A single sample contained nectar collected from ten flowers. The nectar sugar concentration (% wt/wt) was measured with the RL-4 refractometer (PZO, Warsaw Poland). Then, the nectar amount and nectar sugar concentration were used to calculate the total sugar amount (mg) secreted in nectar · flower⁻¹ (Jabłoński, 2002).

The pollen amount available to insects was determined with the modified ether method (Warakomska, 1972). Every year, twelve samples were collected, each consisting of fifty mature stamens. The anthers were removed from randomly chosen flower buds and placed on previously tarred watch glasses. The samples were then placed in an air dryer (SUP-65G Wamed, Poland) at 30°C. After anthers dehiscence pollen grains were washed from the them with pure ether and subsequently with 70% ethanol. The empty anthers were removed from the watch glasses. Next, the pollen samples were dried and weighed on a balance (WA 34 PRL T A14 MERA-KFM, Poland), which allowed the calculation of the mass of air-dried pollen. The results were expressed in mg · flower⁻¹.

Moreover, both the total sugar amount secreted in nectar and the pollen mass · plant⁻¹ and m⁻² of the crop available to insects were estimated. To this end, plant density · m⁻² of a plot and data on the average number of flowers on a plant already formed during a given growing season were used. The pollen grains were mounted in glycerol-gelatine on microscopic slides (n = 10). Each slide was prepared from a mixture sample of pollen collected from 120 flowers. Pollen grains dimensions were determined by the measuring of the length of the polar and equatorial axes using a NIKON ECLIPSE E 600 LM at a magnification of 40 × 15. At least fifty pollen grains were measured on each slide (Andrejev, 1926).

In 2007, the intensity of diurnal insect visits to the borage flowers was monitored. The number of visits was counted for five minutes, three times every two hours, from 8:00 to 20:00 (GMT + 2 hrs). Counting was performed in four marked areas; each sized 1 m², established inside the experimental plot. The procedure was repeated three times during the full bloom of the species in sunny and not windy weather. The observations revealed five insect categories: honey bees, bumblebees, non-Apis and non-Bombus bees, flies (Diptera) and others (butterflies, beetles and thrips). The fruit set and seed set between flowers bagged to exclude insects (n = 399) and those freely pollinated (n = 315) were compared to determine the role of insects as pollinators of borage crops. The pollination treatments were conducted on eight randomly chosen plants. On each plant, six stems with a similar number of flower buds in the same developmental phase were marked. Then, three stems were bagged in tulle isolators (mesh size 1 mm) until flowering cessation, whereas the others stayed unbagged for free pollination. After fruit ripening, the percentage of fruit set in relation to the flowers open on the stems was determined. The seed set was calculated as a percent of the number of seeds harvested from a flower in comparison to the maximum potential number per flower (four seeds per flower). The field pollination experiment was performed in 2007.

All analyses were performed using STATISTICA v.13 (StatSoft, Poland, Kraków). The total number of flowers · plant⁻¹ and the total number of flowers · m⁻² were tested using Kruskal-Wallis non-parametric ANOVA and H-test. Differences in the nectar amount, nectar sugar concentration, total sugar amount in the nectar and the pollen amount between the years of the study were compared using one-way ANOVAs. When significant differences were found, the ANOVAs were followed by the HSD Tukey test at p = 0.05. The Chi-square contingency test was applied for the fruit set and the seed set.

In the climatic conditions of SE Poland, borage blooming started sixty days after sowing in the latter half of June and often lasted until the latter half of August. The detailed dates of the seasonal blooming periods are shown in Table 1. The flowering period on average lasted eight weeks. The time of blooming differed among the years of the study and depended on the weather conditions. Generally, the thermal conditions in all growing seasons were similar and hot weather prevailed - the daily mean temperatures of June, July and August exceeded the multiyear mean values by approx. 1.3ºC. However, the extremely dry weather during the final ten days of June 2007 (total precipitation 3.2 mm) accelerated the blooming process in the plants by ten to eleven days, compared to 2008 and 2009.

During the first week of flowering, the flowers on the borage plants developed at a slow rate, but the process intensified in the second week. Peak flowering was recorded in the third and fourth week of blooming. In the fifth week, the number of open flowers steadily decreased and the termination of blooming started in the sixth week (Fig. 2).

Fig. 2

The observations of the daily flowering pattern (Fig. 3) showed that the greatest number of flowers opened before 8:00 and between 14:00 and 20:00. The lowest portion of newly opened flowers was noted during noon hours. During the peak flowering of the species, the daily number of all open flowers per plant exceeded thirty (Fig. 2). The detailed data on the blooming abundance of borage are shown in Tab. 1. The effect of the growing season on the number of flowers formed per plant and per area unit was not significant (H_{2, 26} = 1.14, p = 0.57 and H_{2, 26} = 0.67, p = 0.71). The life span of a borage flower, on average 22 hours, was similar in each year of the study (H_{2, 18} = 1.62, p = 0.45) (Tab. 1).

Fig. 3

Nectar in a borage flower is produced by an irregular four-lobed nectary ring located at the base of the ovary (Fig. 1bc). The nectary is surrounded by thickened bases of filaments. Five apertures formed in the non-thickened part of the filament base lead to the nectar, and the throat scales in the corolla are located opposite these holes. Nectar is released through nectarostomata in the nectary epidermis (Fig. 1d). The nectar secretion begins at the loose bud stage and lasts throughout the entire life of the flower.

The characteristics of nectar produced by borage flowers in consecutive growing seasons are shown in Tab. 2. All the parameters, i.e. nectar amount, nectar sugar concentration and nectar sugar amount varied among the years of the study (although not statistically; F_{2, 50} = 0.21, p = 0.81; F_{2, 45} = 2.68, p = 0.80, and F_{2, 50} = 2.92, p = 0.06, respectively).

Nectar amount · flower⁻¹ was on average 4.0 ± 2.2 mg (n = 53). The nectar sugar concentration was moderate but ranged widely from 10 to 66%. The mean sugar amount secreted in nectar · flower⁻¹ was 1.2 ± 0.5 mg (n = 53). The estimated production of nectar sugars per plant and per 1 m² of the crop showed no year effect (F_{2, 23} = 1.22, p = 0.31 and F_{2, 50} = 2.34, p = 0.11, respectively). Throughout the growing season, on average 1.1 ± 0.5 g (n = 26) of sugars could be collected by insects from one borage plant. The estimated mean nectar sugar yield · m⁻² of this crop was 5.21 ± 2.34 g (n = 53), i.e., 52.1 kg · ha⁻¹.

Borage flowers are protandrous. Mature anthers open towards the inside of a conical staminal tube consisting of tightly packed anther heads where pollen grains are released and accumulated. When the anthers are empty, the style extends and protrudes beyond the anthers, while the stigma is getting receptive (Fig. 1a). The amounts of pollen · flower⁻¹ differed significantly among the years of the study (F_{2, 30} = 14.77, p = 0.00). The highest average value of 1.3 ± 0.5 (n = 12) mg per flower was found in 2007, whereas the values for the following years of study were 15–23% lower (Tab. 3). The estimated pollen amount · plant⁻¹ showed no effect of the year (F 2, 23 = 1.65, p = 0.22), whereas the pollen amount · 1m⁻² of the crop differed significantly among the years of the study (F_{2, 33} = 4.61, p = 0.02). During the growing season, one borage plant provided 1.1 ± 0.5 g (n = 52) of pollen, i.e. 5.21 ± 2.01 g (n= 36) · m⁻² of the crop.

Borage pollen grains are stephanocolporate (Fig. 4) with the average length of the polar axis (P) of 33.5 mm (range 30.3–38.7 mm) and that of the equatorial axis (E) of 34.0 mm (range 31.3–39.2 mm). The shape index (P/E) is 0.99.

Fig. 4

Hymenopterans were the principal visitors to the borage flowers (Fig. 5). Their visits accounted for 73% of all insect visits observed. Honey bees dominated among them, but bumblebees as well as non-Apis and non-Bombus bees were similarly abundant (Fig. 5). The flowers were visited by a number of dipterans as well. Bees collected both nectar and pollen from the flowers (Fig. 6). Microscopic analysis of grey-beige corbicular pollen loads of honey bees and bumblebees confirmed the presence of numerous borage pollen grains (unpublished data).

Fig. 5

Fig. 6

Insects foraged on the flowers from early morning to evening (Fig. 7). The most intense visitation was observed between 10:00 and 14:00 when the majority of new flowers were open, providing nectar and pollen.

Fig. 7

The presence of insects significantly increased both the fruit set and the seed set in borage (χ² test, df = 1, p < 0.00 in both cases). In freely pollinated flowers, the percent fruit set was 45.1% (n = 315) whereas it was 1.8% (n = 399) when pollinators were excluded. The seed set reached 27.9% (n=1596) in the freely pollinated plants and only 1.1% (n = 1260) in inflorescences bagged to prevent insect access.