Influence of Stirring Parameters on Creaminess of Spring Blossom Honey Measured by Crystal Size, Whiteness Index and Mouthfeel

Spring blossom honey from regions with many rape fields tends to crystalize rapidly after harvesting. The crystallization process needs to be controlled by stirring in order to avoid the formation of coarse crystals and to ensure the creaminess of honey. The aim of this study was to investigate how various parameters of the stirring process influence the creaminess of spring blossom honey in order to give recommendations for beekeeping practices. The creaminess was quantified by measuring the crystal size by microscopic analysis, measuring the whiteness index by color analysis using CIE Lab and by sensory analysis. We investigated the influence of five stirring parameters, including the type of stirring device, honey pretreatment, stirring temperature (14 °C to room temperature), stirring interval (1 to 24 times) and stirring time (1–15 min) on the creaminess of honey. We found that the stirring temperature is the most important factor for honey creaminess. At the optimal temperature of 14 °C, other factors like seed honey, stirring time and stirring interval have only a neglectable effect. If the optimal temperature of 14 °C cannot be maintained, as it may happen in beekeepers’ practice, sieving the honey with a mesh size of 200 µm before stirring, the addition of seed honey prepared with a kitchen food processor, and using a stirring screw and stirring several times per day is recommended.


Introduction
Honeybees forage nectar and honeydew from plants and ripe it to honey in the beehive. The beekeeper harvests the ripe honey as a liquid. Depending on the composition, liquid honey is more or less prone to crystallization. Honey consists predominantly of glucose and fructose [1], which sum up to approximately 70-80% (w/w), typically less than 20% (w/w) water and a number of other di-and trisaccharides in smaller quantities [2] as well as pollen, minerals and small amounts of phytochemicals. Based on the bee's flowering forage, the ratio of the sugars varies [2][3][4], which has a major impact on the honey's tendency to crystallize. Laos et al. [5], as well as Scripca & Amariei [6], found that the most important parameter for crystallization is the fructose/glucose ratio in favor of glucose. Scripca & Amariei [6] emphasize that also the glucose/water ratio is important for the crystallization behavior of honey. A saturated solution of fructose in pure water contains 780 g kg −1 , while a saturated solution of glucose contains only 470 g kg −1 [7]. Thus, particularly glucose-rich honey is a supersaturated solution and crystallizes more rapidly and to a larger extent than fructose-rich honey. When crystallization has initiated, the honey gets opaque, and its viscosity increases significantly.
Most consumers prefer liquid or melt-in-the-mouth creamy honey with small crystals that are not noticeable on the tongue [5,6,[8][9][10][11][12]. There are three possibilities to meet the customers' preferences: delaying or avoiding crystallization, liquefying honey after crystallization or controlling the crystallization process by stirring. In order to avoid viscosity was considered suitable when the 2 walls of the groove touched each other after 5 s ( Figure S1 in supplementary information). After stirring, the honey was filled in jars and stored for several weeks. We determined the creaminess of the honey on different days using the parameters crystal size, whiteness index and mouthfeel in the following way.

Determination of the Crystal Size by Microscopic Analysis and the Whiteness Index
The crystal size was determined as described previously [32] with three replicates per sample. For the determination of the whiteness index, the honey sample was transferred to a petri dish, and the color was determined via a CIE LAB instrument (Spectro-Colour, Type LMG183, Dr. Lange) according to the manufacturer's instructions. The whiteness index was calculated by the following formula [19,25,33,34]: The determination was repeated three times per honey sample.

Determination of the Mouthfeel by Sensory Analysis
Trained testers investigated the mouthfeel of the honey samples in sensory cabins under standardized conditions. The honey samples were offered coded with random numbers in standardized glasses (30 ml). In stirring tests 1 and 2, the testers assessed the samples by ranking their creaminess. In stirring tests 4 and 5, the testers determined the mouthfeel on a scale between 0 and 100, where 0 was defined as velvety creamy, like hazelnut cream "Nudossi," and 100 was defined as coarsely crystalline, like crunchy chocolate cream "Ovomaltine Crunchy Cream," using sensory software RedJade. Beforehand several chocolate creams and hazelnut creams were tested for the definition of scale 0 and scale 100. Due to Corona contact restrictions, sensory testing could not be carried out for stirring test 3.

Influence of the Stirring Device on Creaminess
We hypothesized that the stirring device might have an influence on the creaminess of the treated honey. We stirred equal quantities of the same honey batch manually or with a stirring spiral in a bucket and with a stirring impeller or a stirring screw in a stirring machine ( Figure S2 shows the stirring devices). Furthermore, one aliquot of honey was left unstirred as a control. In this experiment, the temperature was kept at 17.5 ± 1.5 • C. We measured the crystal size and the whiteness index at the end of stirring and at several time points in the following 56 days. Three days after the end of stirring, the unstirred honey, as well as the samples stirred with the spiral or the impeller, had significantly larger crystals than the honey stirred manually or with the stirring screw (ANOVA and Student-Newman-Keuls procedure, α = 0.05; Figure 1a), groups with significantly different crystal sizes are marked with "a" and "b"). After 56 days of storage, the crystal sizes became smaller and equal in all experimental setups so that no significant differences could be found (ANOVA, alpha = 0.05). While the crystal size decreased over time, the whiteness decreased also. At the end of the stirring process, unstirred honey was significantly whiter than the other samples (ANOVA and Student-Newman-Keuls procedure, α = 0.05). Supplementary to this, 56 days after the end of stirring, the unstirred honey and the sample stirred with the stirring impeller were whiter than the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05; Figure 1b). The decrease in whiteness during honey storage is consistent with the findings of Visquert [34] as well as Radtke and Lichtenberg-Kraag [26]. The mouthfeel While the crystal size decreased over time, the whiteness decreased also. At the end of the stirring process, unstirred honey was significantly whiter than the other samples (ANOVA and Student-Newman-Keuls procedure, α = 0.05). Supplementary to this, 56 days after the end of stirring, the unstirred honey and the sample stirred with the stirring impeller were whiter than the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05; Figure 1b). The decrease in whiteness during honey storage is consistent with the findings of Visquert [34] as well as Radtke and Lichtenberg-Kraag [26]. The mouthfeel was tested after 56 days of storage by 54 sensory testers who ranked the samples in creaminess. The statistical analysis detected no significant differences in the mouthfeel of the honey samples (Friedmann Test, α = 0.05; Figure 1c).

Influence of Pretreatment of Honey on Creaminess
Previous data have shown that crystallization can be best controlled when there are no huge crystals in the honey, and its viscosity is low to facilitate homogenization [14]. Prior to stirring, honey is frequently pretreated by sieving and/or mild heating. To investigate the potential effects of such measures, honey was either used untreated or sieved through a conic pointed sieve with a mesh size of 200 µm with or without having short contact with a heating coil of 55 • C ( Figure S3).
Subsequently, the honey was stirred with a stirring spiral at 17.5 ± 1.5 • C two times per day for 15 min. Analysis of the crystal size revealed no significant differences directly after stirring ( Figure 2a). However, the crystal size remained relatively constant over time in the batch that had been sieved and heated, while an increase of crystal size was observed in the other variants. Twenty eight days after stirring end, the honey pretreated with sieve and heating coil had significantly smaller crystals than the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05).
The whiteness index increased while stirring the honey (Figure 2b). Twenty-eight days after stirring ended, there was no significant difference between the pretreatment groups (ANOVA, alpha = 0.05). The whiteness index increased while stirring the honey (Figure 2b). Twenty-eight days after stirring ended, there was no significant difference between the pretreatment groups (ANOVA, alpha = 0.05).
In line with crystal analysis, sensory testing showed that pretreatment with a sieve  In line with crystal analysis, sensory testing showed that pretreatment with a sieve and heating coil led to a significantly creamier mouthfeel than pretreatment only with the sieve or no treatment at all (Friedmann test, alpha = 0.05; Figure 2c).

Influence of the Stirring Temperature and the Addition of Seed Honey on Creaminess
To initiate crystallization, the addition of already crystallized seed honey is frequently recommended in such a way that the fine seed crystals will act as primary crystallization nuclei [14,[24][25][26][27]33,35]. In addition, the seed honey may be pretreated to reduce the size of the crystallization nuclei [24,27,36]. Subramanian et al. [13] described that small air bubbles could provoke nucleation and crystallization. In a preliminary experiment, we investigated different ways of preparing the seed honey and its impact on the creaminess of the stirred honey batches.
For the kitchen food processor method (recommended by a local beekeeper), about 2.5 kg of honey rich in glucose of a creamy consistency were stirred for 5 min with a dough hook attachment followed by a whisk attachment at 150 rpm in a kitchen food processor. As a result, tiny air bubbles with a maximum diameter of about 30 µm were created. Alternatively, the seed honey was prepared using a stirring machine equipped with an impeller in such a way that the impeller was not covered with honey, and small air bubbles were stirred into the honey. These seed honeys were added to fresh honey, which was then stirred at 17.5 ± 1.5 • C two times per day for 15 min. At the end of stirring, honey seeded with the kitchen food processor-treated honey had smaller crystals (59 µm) than honey seeded with the honey prepared by stirring with an impeller (101 µm) or honey stirred without seed honey (146 µm). Thus, the preparation of seeding honey with a kitchen processor gave the best results. This method was used for the subsequent experiments. Dyce [24] recommends storing honey at 14 • C after adding the seed honey. Al-Habsi et al. [36] report that 14 • C is the most frequently used crystallization temperature applied in practice and leads to linear growth of the crystal content of the honey. Here we wanted to investigate the consequence if the stirring temperatures deviate from 14 • C for the crystallization of seeded and non-seeded honey. We stirred the honey with a stirring screw at different temperatures, seeded with honey prepared with the kitchen food processor or without seed honey. We found that all chilled samples had huger crystals right at the end of stirring compared to 14 or 28 days later (Figure 3a), where the crystal size decreased. After 28 days, the honey stirred at room temperature showed significantly (ANOVA and Student-Newman-Keuls, alpha = 0.05) larger crystals than the honey stirred at 14 • C or 18 • C (Figure 3a). The addition of seed honey showed no influence on crystal size for honey stirred at 14 • C, whereas honey stirred at 18 • C had significantly smaller crystals when seeded. These results confirm the findings of Costa et al. [25] and Radtke & Lichtenberg-Kraag [26]. The whiteness is at the end of stirring, as well as 28 days later, significantly higher for honey stirred at 14 • C with seed honey compared to the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05; Figure 3b).
Due to Corona restrictions, sensory testing could not be carried out for this experiment.

Influence of the Stirring Interval on Creaminess
To this end, our results showed that a temperature of 14 • C during stirring is the most critical parameter for improving honey creaminess. Next, we wanted to investigate whether the stirring interval influences the creaminess of honey because, for many hobbyist beekeepers, it is difficult to implement several stirring processes throughout the day. To investigate the stirring process under this condition in more detail, non-seeded honey was stirred at 14 • C for different intervals but with the same total stirring time per day.
At the end of stirring, the crystals were relatively large, and no significant differences (ANOVA, α = 0.05) in the crystal sizes could be detected (Figure 4a). Upon storage, the honey became heavily crystallized, which impeded the separation of intact crystals and thereby determination of the crystal size at later time points. The stirring interval had a very small impact on the whiteness index (Figure 4b). Twenty-three days after stirring ended, the difference in the whiteness index was not significant (ANOVA, alpha = 0.05). Due to Corona contact restrictions, sensory testing could be carried out only with a small sensory panel of five trained testers. The testers determined the mouthfeel on a scale between 0 (velvet creamy) and 100 (coarsely crystalline), as described above. The results show large deviations in the mouthfeel due to the different testers' assessments ( Figure 4c). On none of the days could significant differences in the mouthfeel of the samples be detected (ANOVA, alpha = 0.05). These results indicate that, at a temperature of 14 • C, the stirring interval has no relevant impact on the creaminess of spring blossom honey.
Habsi et al. [36] report that 14 °C is the most frequently used crystallization temperature applied in practice and leads to linear growth of the crystal content of the honey. Here we wanted to investigate the consequence if the stirring temperatures deviate from 14 °C for the crystallization of seeded and non-seeded honey. We stirred the honey with a stirring screw at different temperatures, seeded with honey prepared with the kitchen food processor or without seed honey. We found that all chilled samples had huger crystals right at the end of stirring compared to 14 or 28 days later (Figure 3a), where the crystal size decreased. After 28 days, the honey stirred at room temperature showed significantly (ANOVA and Student-Newman-Keuls, alpha = 0.05) larger crystals than the honey stirred at 14 °C or 18 °C (Figure 3a). The addition of seed honey showed no influence on crystal size for honey stirred at 14 °C, whereas honey stirred at 18 °C had significantly smaller crystals when seeded. These results confirm the findings of Costa et al. [25] and Radtke & Lichtenberg-Kraag [26]. The whiteness is at the end of stirring, as well as 28 days later, significantly higher for honey stirred at 14 °C with seed honey compared to the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05; Figure 3b).  Due to Corona restrictions, sensory testing could not be carried out for this experiment.

Influence of the Stirring Interval on Creaminess
To this end, our results showed that a temperature of 14 °C during stirring is the most critical parameter for improving honey creaminess. Next, we wanted to investigate whether the stirring interval influences the creaminess of honey because, for many hobbyist beekeepers, it is difficult to implement several stirring processes throughout the day. To investigate the stirring process under this condition in more detail, non-seeded honey was stirred at 14 °C for different intervals but with the same total stirring time per day.
At the end of stirring, the crystals were relatively large, and no significant differences (ANOVA, α = 0.05) in the crystal sizes could be detected (Figure 4a). Upon storage, the honey became heavily crystallized, which impeded the separation of intact crystals and thereby determination of the crystal size at later time points. The stirring interval had a very small impact on the whiteness index (Figure 4b). Twenty-three days after stirring

Influence of the Stirring Time on Creaminess
Having established that the stirring interval is of minor importance, we investigated whether the duration of stirring is a critical factor. We stirred the honey samples without seed honey at a temperature of 14 °C.
At the end of stirring, we found a significant influence of the stirring time on the (a) (b) (c)

Influence of the Stirring Time on Creaminess
Having established that the stirring interval is of minor importance, we investigated whether the duration of stirring is a critical factor. We stirred the honey samples without seed honey at a temperature of 14 • C.
At the end of stirring, we found a significant influence of the stirring time on the crystal size. Honey stirred daily twice for 15 min had significantly smaller crystals than the other samples (ANOVA and Student-Newman-Keuls, alpha = 0.05; Figure 5a). Thirty days after the end of stirring, the crystal sizes had declined, and the difference between the stirring groups was diminished. The whiteness index showed little difference between the samples (Figure 5b). At the stirring ended, the sample stirred daily twice for 15 min was the whitest one (ANOVA and Student-Newman-Keuls, α = 0.05). Twenty days later, there was only little impact of the stirring time on the whiteness index (Figure 5b). Due to Corona restrictions, the sensory analysis could be carried out only once with five testers 20 days after the stirring ended. The sensory testers evaluated the mouthfeel of all samples as very creamy. No significant differences could be detected between the kinds of honey with different stirring times (ANOVA, α = 0.05: Figure 5c). This experiment demonstrated that at a temperature of 14 °C, the stirring time has no significant impact on the creaminess of the spring blossom honey.

Discussion
It is well established that the crystallization behavior of honey depends on its composition, mainly the glucose to fructose ratio [9,14,27], which varies greatly depending on the foraging flower and from year to year and is difficult to predict. However, it has been shown that also stirring [24,25] and the storage temperature [26] significantly impact the size of the crystals formed. Moreover, the addition of seed crystals is a common practice to reduce crystal size [27]. Stirring of honey is a complex process in which several factors determine the creaminess of the end product: the stirring device, interval and time as well The whiteness index showed little difference between the samples (Figure 5b). At the stirring ended, the sample stirred daily twice for 15 min was the whitest one (ANOVA and Student-Newman-Keuls, α = 0.05). Twenty days later, there was only little impact of the stirring time on the whiteness index (Figure 5b). Due to Corona restrictions, the sensory analysis could be carried out only once with five testers 20 days after the stirring ended. The sensory testers evaluated the mouthfeel of all samples as very creamy. No significant differences could be detected between the kinds of honey with different stirring times (ANOVA, α = 0.05: Figure 5c).
This experiment demonstrated that at a temperature of 14 • C, the stirring time has no significant impact on the creaminess of the spring blossom honey.

Discussion
It is well established that the crystallization behavior of honey depends on its composition, mainly the glucose to fructose ratio [9,14,27], which varies greatly depending on the foraging flower and from year to year and is difficult to predict. However, it has been shown that also stirring [24,25] and the storage temperature [26] significantly impact the size of the crystals formed. Moreover, the addition of seed crystals is a common practice to reduce crystal size [27]. Stirring of honey is a complex process in which several factors determine the creaminess of the end product: the stirring device, interval and time as well as pretreatment with a sieve and/or heating coil, and the addition of seed honey ( Figure 6). To investigate the impact of these factors in more detail, we performed a series of experiments where we varied the different parameters. Interestingly, the type of stirring device had only a small impact on crystal size, which, moreover, became equal upon storage. Similarly, also pretreatment with a sieve or heating coil affected crystal size, whiteness index and sensory testing only slightly. In sharp contrast, the stirring temperature turned out to be the most important factor, and the smallest crystals were obtained at 14 °C . Interestingly, the stirring time and interval at that temperature had minor effects.
Thus, stirring at the optimal temperature of 14 °C allows other factors to be neglected. Our data also show that in case a temperature of 14 °C during stirring cannot be maintained, seeding with creamy, crystallized honey mixed using a kitchen food processor is recommended. A stirring machine with a screw stirrer, stirring several times per day, as well as pretreatment with a sieve and heating coil, may help to get small crystals.
We were surprised that the crystals became smaller during storage after stirring. We have no physical explanation for this phenomenon. The relation between storage time and crystal size does not seem to be linear but resembles the decay equation. Further research is recommended to investigate this phenomenon in more detail. However, this observation suggests that stirring honey and subsequently keeping it in a cool place for a few weeks before selling might be a helpful practice to obtain a product with the texture preferred by the consumers.

Supplementary Materials:
The following supplementary information can be downloaded at: www.mdpi.com/xxx/s1, Figure S1: Modified version of the "finger test" commonly used by beekeepers. A plastic tongue depressor dipped 2 cm deep into the honey and pulled 10 cm forms two honey walls. The walls touch each other for a distance of approx. 5 cm after 5 seconds; Figure S2: Stirring devices. A Manual stirrer for use in a bucket. B Stirring spiral for use in a bucket. C Stirring impeller outside of the stirring machine. D Stirring screw outside of the stirring machine. E Stirring screw in the stirring machine, Figure S3: Prepreparation of the honey before stirring by conic pointed sieve (mesh size 200 µ m) and heating coil, Table S1: Significance analysis of the findings  To investigate the impact of these factors in more detail, we performed a series of experiments where we varied the different parameters. Interestingly, the type of stirring device had only a small impact on crystal size, which, moreover, became equal upon storage. Similarly, also pretreatment with a sieve or heating coil affected crystal size, whiteness index and sensory testing only slightly. In sharp contrast, the stirring temperature turned out to be the most important factor, and the smallest crystals were obtained at 14 • C. Interestingly, the stirring time and interval at that temperature had minor effects.
Thus, stirring at the optimal temperature of 14 • C allows other factors to be neglected. Our data also show that in case a temperature of 14 • C during stirring cannot be maintained, seeding with creamy, crystallized honey mixed using a kitchen food processor is recommended. A stirring machine with a screw stirrer, stirring several times per day, as well as pretreatment with a sieve and heating coil, may help to get small crystals.
We were surprised that the crystals became smaller during storage after stirring. We have no physical explanation for this phenomenon. The relation between storage time and crystal size does not seem to be linear but resembles the decay equation. Further research is recommended to investigate this phenomenon in more detail. However, this observation suggests that stirring honey and subsequently keeping it in a cool place for a few weeks before selling might be a helpful practice to obtain a product with the texture preferred by the consumers.

Supplementary Materials:
The following supplementary information can be downloaded at: https: //www.mdpi.com/article/10.3390/foods12010048/s1, Figure S1: Modified version of the "finger test" commonly used by beekeepers. A plastic tongue depressor dipped 2 cm deep into the honey and pulled 10 cm forms two honey walls. The walls touch each other for a distance of approx. 5 cm after 5 s; Figure S2: Stirring devices. A Manual stirrer for use in a bucket. B Stirring spiral for use in a bucket. C Stirring impeller outside of the stirring machine. D Stirring screw outside of the stirring machine. E Stirring screw in the stirring machine, Figure S3: Prepreparation of the honey before stirring by conic pointed sieve (mesh size 200 µm) and heating coil, Table S1 Funding: Costs for personnel were covered by Anhalt University of Applied Sciences. The stirring machines and the temperature control system were borrowed by Carl Fritz Imkereitechnik GmbH & Co KG. We acknowledge the support of the German Research Foundation (Deutsche Forschungsgemeinschaft DFG), project number 491460386, and the Open Access Publishing Fund of Anhalt University of Applied Sciences.
Data Availability Statement: The data will be available on request.