Table 1.
Overview of the analyses as well as the number of bees and cages per treatment of the feeding experiments performed between 2017 and 2019.
Table 2.
Results of the linear regression model (daily food uptake), fractional logit regression model (gut-body weight ratio) and Cox proportional hazard model (survival).
Table 3.
Fructose, glucose, sucrose and melezitose proportion of the feed solution, crop content of honey bees collected from the first feeding experiment, field experiment and the processed melezitose feed.
Fig 1.
Boxplot of food uptake per day and bee in milligram (mg) in all cage experiments.
The three treatment groups are highlighted with different colours. The mean value of daily food uptake per bee fed with control feed (blue) was 20 mg per day (n = 445) and 37 mg per day (n = 388) for bees fed with melezitose feed (yellow) and 70 mg per day (n = 67) for bees fed with melezitose feed from day 10 (green) of all years. The vertical boxplots depict the interquartile range (lower bound/ upper bound of the box correspond to the 25/75% quantile), the median (horizontal line in the box) and outlying observations (points outside the box).
Fig 2.
Boxplot of gut-body weight ratio of honey bees in all cage experiments.
The three treatment groups are highlighted with different colours. The mean value of gut-body weight ratio per bee fed with control feed (blue) was 52% (n = 150) and 60% (n = 150) for bees fed with melezitose feed (yellow) and 56% (n = 30) for bees fed with melezitose feed from day 10 (green). The vertical boxplots depict the interquartile range (lower bound/ upper bound of the box correspond to the 25/75% quantile), the median (horizontal line in the box) and outlying observations (points outside the box).
Fig 3.
Overall survival probability plots in all cage experiments.
The survival rate of the honey bees in the control (blue), in the melezitose group (yellow) and in the bees fed with melezitose from day 10 (green) (Cox regression, Log-rank (4,1740): 628.1, p < 0.001).
Fig 4.
Relative abundance of Lactobacillus kunkeei in the gut microbiota.
The three treatment groups are highlighted with different colours: control (blue), melezitose (not detected at all time points) and melezitose from day 10 (green). Significantly different groups are highlighted by the letters a, b and c (F-statistic: 2.66 on 5 and 102 DF, p = 0.027). For each treatment group and each bee age, 18 individuals were used for analysis. Not detected bacteria are marked with ND. The vertical boxplots depict the interquartile range (lower bound/ upper bound of the box correspond to the 25%/ 75% quantile), the median (horizontal line in the box) and outlying observations (points outside the box).
Fig 5.
Relative abundance of Lactobacillus Firm-4 in the gut microbiota.
The three treatment groups are highlighted with different colours: control (blue), melezitose (yellow) and melezitose from day 10 (green). Significantly different groups are highlighted by the letters a, b, c and d (F-statistic: 4.245 on 5 and 102 DF, p = 0.002). For each treatment group and each bee age, 18 individuals were used for analysis. The vertical boxplots depict the interquartile range (lower bound/ upper bound of the box correspond to the 25%/ 75% quantile), the median (horizontal line in the box) and outlying observations (points outside the box).
Fig 6.
Relative abundance of Lactobacillus Firm-5 in the gut microbiota.
The three treatment groups are highlighted with different colours: control (blue), melezitose (yellow) and melezitose from day 10 (green). Significantly different groups are highlighted by the letters a, b, c and d (F-statistic: 7.048 on 5 and 102 DF, p < 0.001). For each treatment group and bee age, 18 honey bee individuals were used for analysis. The vertical boxplots depict the interquartile range (lower bound/ upper bound of the box correspond to the 25%/ 75% quantile), the median (horizontal line in the box) and outlying observations (points outside the box).