1 Cet article a été soumis au journal scientifique Journal of Apicultural Research le 6 juin 2018 (TJAR-2018-
Résumé
La gestion commerciale de colonies d’abeilles domestiques (Apis mellifera L.) est devenue difficile en raison, entre autres, de carences nutritionnelles. Pour pallier le manque de pollen dans l’environnement, certains apiculteurs offrent des suppléments de protéines à leurs colonies. Cependant, leur efficacité varie. Dans cette étude de terrain, deux suppléments protéiques (Global PattiesMD et Ultra BeeMD) ont été offert à 50 colonies en Montérégie, avec accès limité ou non au
pollen. Les colonies supplémentées et restreintes dans leur accès au pollen ont réussi à élever la même quantité de couvain que les colonies témoins. Les abeilles supplémentées contenaient également plus de protéines que les abeilles témoins, mais leur longévité était plus courte, ce qui suggère que les produits testés ne sont pas optimaux. Le supplément contenant du pollen, Global PattiesMD, était le
Abstract
In recent years, honey bee (Apis mellifera L.) colony management has become more difficult due to multiple factors, including nutritional deficiencies. This has incited beekeepers to provide protein supplements to their colonies to make up for the lack of pollen resources. However, the efficiency of these supplements varies. In this field study, we provided two protein supplements (Global Patties and Ultra Bee) to 50 colonies in the Montérégie area, in Quebec, Canada, with either limited or unlimited access to pollen. We found that supplemented colonies limited in pollen collection were able to raise the same amount of brood than control colonies. Bees in supplemented colonies also had a higher protein content compared to control bees, but they displayed shorter lifespan, which casts a doubt on the suitability of these products. The supplement containing pollen, Global Patties, was the most consumed and seemed the most suitable for the colonies.
Introduction
Over the last decade, commercial honey bee (Apis mellifera L.) colonies have suffered high mortality rates in North America and Europe (Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services 2016; Oldroyd 2007; vanEngelsdorp and Meixner 2010), ranging from 29 to 35% from 2007 to 2009, and from 16 to 29% from 2010 to 2015 in Canada (Ministère de l'Agriculture des Pêcheries et de l'Alimentation du Québec 2016). Multiple factors appear to act synergistically to cause this problem, including pesticide exposure, exotic parasites and pathogens, management practices, climate changes and lack of floral resources (Klein et al. 2017; Naug 2009; Oldroyd 2007).
Abundant and diversified floral resources are essential for honey bees to acquire their essential nutrients through consumption of pollen and nectar (Wright, Nicolson, and Shafir 2018). In the current beekeeping industry, pollen nutrition is of special concern for multiple reasons. Firstly, pollen is the only source of proteins, lipids, vitamins and minerals for the honey bee; nectar being mainly a source of carbohydrates. Secondly, unlike nectar and honey, pollen storage within the hive is limited and can deplete rapidly during periods of weather unfavorable to foraging (Schmickl and Crailsheim 2001). Pollen must therefore be available throughout the season in adequate quantity. Finally, as different flowers produce pollen of distinctive nutritional content, flower diversity is fundamental to insure an appropriate and complete bee diet (Roulston and Cane 2000; Roulston, Cane, and Buchmann 2000). However, in intensive agricultural environments or when colonies are used for commercial pollination, these requirements are not always met (Danner et al. 2017; Di Pasquale et al. 2016; Donkersley et al. 2014; Girard, Chagnon, and Fournier 2012; Smart et al. 2016). Poor pollen nutrition negatively affects individual honey bee health as well as colony performances (Brodschneider and Crailsheim 2010; Di Pasquale et al. 2016; Di Pasquale et al. 2013; Scofield and Mattila 2015), rendering bees less resistant to other stressors such as pesticides (Schmehl et al. 2014; Tosi et al. 2017) and pathogens (Alaux et al. 2010; Di Pasquale et al. 2013).
To counter this problem, beekeepers provide protein supplements to their colonies during periods of poor floral abundance or diversity (Brodschneider and Crailsheim 2010). These
supplements are made from ingredients rich in proteins, such as soy or yeast, and may also contain natural pollen (Eccles et al. 2016). It is generally recognised that feeding such artificial diets is beneficial to colonies facing poor foraging conditions or environments (Brodschneider and Crailsheim 2010). However, the efficiency of these products varies depending on their composition, especially when compared to pollen (Alqarni 2006; De Jong et al. 2009; DeGrandi-Hoffman et al. 2016; Peng, D'Antuono, and Manning 2012). Among other things, a minimal protein content, usually of 20-30%, is required for supplements to be functional (Li et al. 2012; Li et al. 2014; Morais et al. 2013). The choice of ingredients can also affect the performance of the product as well. Adding pollen to the diet, for example, can stimulate consumption of the supplement as well as its positive impacts on honey bee health (Alqarni 2006; Manning et al. 2007). On the contrary, soy-based diets are less palatable and overall less efficient (De Jong et al. 2009; DeGrandi-Hoffman et al. 2016; Fleming, Schmehl, and Ellis 2015; Saffari, Kevan, and Atkinson 2010a; Saffari, Kevan, and Atkinson 2010b). Even the surface area of the applied product can be an important factor to consider, as a larger surface reportedly increases consumption and positive impacts on the colony (Avni, Dag, and Shafir 2009). However, despite the surprisingly wide range of homemade and commercial supplements available, basic knowledge as to the essential nutritional requirements of the honey bees are still lacking, especially for lipids, vitamins and minerals (Wright, Nicolson, and Shafir 2018). Thus, an ideal protein supplement formulation, which could effectively replace pollen, has yet to be found (Wright, Nicolson, and Shafir 2018).
Efficacy of different protein supplements has been the subject of relatively few scientific studies, especially under field conditions. Also, comparison between studies is difficult considering the variability of supplements, control treatments, feeding periods, and pollen traps used to simulate dearth. Furthermore, the landscape has a significant impact on pollen and nectar sources having a direct effect on the nutritional status of the colony (Donkersley et al. 2014). Moreover, to our knowledge, the impact of supplemental feeding on commercial colonies has not been investigated yet. As stated by Mattila and Otis (2006), commercially managed colonies accumulate stress during the season, and the impact of supplementation would most probably be different on these colonies than on low-density and relatively
undisturbed research colonies. Therefore, the objectives of this study are to 1) compare the health and strength of commercial honey bee colonies supplemented or not with a protein supplement, 2) compare the consumption and impact on honey bee health of two commercial protein supplements and 3) evaluate the impact of surrounding landscape on the nutritional status of colonies. Our hypotheses were that 1) supplemented colonies would be stronger and more productive, 2) pollen-enriched protein supplement would be more consumed and have a higher positive impact on colony health and 3) colonies based in a more suitable environment (notably less cultivated area) would perform better regardless of the treatment.
Materials and Methods
Sites and colony management
The study was conducted from May to September 2016 in Montérégie, known as the most intensive agricultural region in the province of Québec, Canada, located south-east of the island of Montreal. Fifty colonies of similar strength and genetic origin were provided by the commercial beekeeper Les Ruchers Gauvin Inc. (Saint-Hyacinthe, QC, Canada). Each colony consisted of a single brood chamber Langstroth hive. The colonies were randomly placed on three sites in the vicinity of the towns of Saint-Hyacinthe and La Présentation (Site 1 – 16 colonies: 45°38'43.0"N, 72°55'26.3"W, Site 2 – 16 colonies: 45°42'20.7"N, 73°05'30.3"W, Site 3 – 18 colonies: 45°39'38.8"N, 72°51'42.9"W). Distances between the three sites were 14.5 km (Site 1 – Site 2), 5.1 km (Site 1 – Site 3), and 18.5 km (Site 2 – Site 3), respecting the minimal distance of 4 km between apiaries recommended in Quebec (Centre de référence en agriculture et agroalimentaire du Québec 2011). The hives were managed in accordance with local professional beekeeping practices. Varroa mite treatments were carried out according to the treatment calendar recommended by the Québec’s Ministry of Agriculture, Fisheries, and Food (MAPAQ) (Ministère de l'Agriculture des Pêcheries et de l'Alimentation du Québec 2014). Nosema spp. was also monitored during the experiment.
In 2017, the study was repeated and modified to measure the lifespan of honey bees depending on their experimental group (see below for details). Fifteen colonies of similar strength were used, all placed in the same apiary in St-Hyacinthe (45°38'43.0"N, 72°55'26.3"W). Their management was the same as in 2016.
Experimental design and treatments
In 2016, protein supplementation and access to pollen were manipulated in the 50 experimental colonies. Colonies were either not supplemented in protein (“unsupplemented”), supplemented with a commercial soy-based protein patty containing 15% pollen (distributed by Global Patties, Airdrie, AB, Canada) or supplemented with a commercial plant-based (containing soy but not as a main source of proteins) protein patty containing no pollen (Ultra Bee, distributed by Mann Lake Limited, Hackensack, MN, USA). The protein content of the patties was 17% for Global Patties and 18% for Ultra
Bee. The products tested are commercially available for beekeepers in Québec. The patties were offered ad libitum to the colonies, from May 3rd-5th to August 3rd. Access to pollen was either unrestricted or restricted by the addition of pollen traps under the hives (Bottom Board type). This was done to assess the pollen harvest of colonies and mimic pollen restrictive conditions. Pollen traps were not installed on unsupplemented colonies to avoid malnutrition and colony death, which was not acceptable for the owners of the beekeeping operation. Pollen traps were installed from May 3rd-5th to August 3rd, at which point they were activated every other week and then completely removed by mid-September All the possible combinations of supplementation and access to pollen resulted in five treatments: 1) Control (no supplement, no pollen traps); 2) pollen-enriched supplement (PES); 3) pollen- enriched supplement + pollen traps (PES + traps); 4) pollen-free supplement (PFS); 5) pollen- free supplement + pollen traps (PFS + traps). Ten colonies were attributed to each treatment, for a total of 50 colonies. As stated previously, colonies were placed on three sites and each site included at least two repetitions of each treatment. Treatments within sites were randomly assigned to the hives.
In 2017, only protein supplementation (no supplement, pollen-enriched supplement or pollen-free supplement) was manipulated in the experimental colonies. No pollen traps were installed. Five colonies were randomly assigned to each treatment, hence 15 colonies in total. As before, supplements were given ad libitum, but only for four weeks starting on the 27th of April. After four weeks, brood frames were taken out of the colonies to perform a longevity test in laboratory.
Supplement consumption
In 2016 and 2017, protein supplement consumption was measured every week of the feeding period during the experiment. In 2016, to compare the consumption of the supplements, two patties of about 500 g each (weighed to measure their exact mass) were placed on the top of the brood chamber frames of treated colonies. Every week, new patties were provided, and leftovers were weighed.
Pollen weight
The amount of pollen collected by the colonies was measured by weighing the content of the pollen traps every week (or every other week in August and September).
Sealed brood surface
The sealed brood surface was evaluated every three weeks from May to September 2016. Each frame of the brood chamber containing sealed brood was photographed on both sides without bees. Then, as described by Delaplane, van der Steen, and Guzman-Novoa (2013), a grid was applied on the photographs using the software ImageJ. Each square division of the grid was estimated as 0, 25, 50, 75 or 100% filled with sealed brood. The total per frame was converted in square centimeters and added up for the hive.
Foraging effort
Foraging effort was also evaluated every three weeks from May to September 2016. The entrance of each hive was filmed for 1 minute to count the number of honey bees returning to the hive with and without corbicular pollen loads. Ratio of bees bringing back pollen to the colony was then calculated. The recordings were always made before opening the hives and between 10 AM and 14 PM to maximize activity and minimize differences due to the time of the day (Delaplane, van der Steen, and Guzman-Novoa 2013).
Protein content of adult honey bees
Measuring the protein content of honey bees is a well-known method to compare the quality of different protein diets (De Jong et al. 2009; Li et al. 2012). Five nurse bees per hive were collected every four weeks of the experiment to measure total protein content. Only nurse
bees are used to minimize potential protein differences due to age. Nurse bees were identified by observing a brood frame and picking up bees that tended to the larvae. If no larva were found in the hive, no bees were picked up and a missing value was recorded. Collected bees were placed in a tube and immediately chilled on ice. Within 24 hours, they were stored at - 80°C until further analysis. Upon analysis, five frozen workers were crushed to fine powder using liquid nitrogen, and 200 mg of bee powder was mixed with a phosphate-buffered saline (PBS) buffer solution containing 1% (v/v) phenylmethylsulfonyl fluoride (PMSF) protease inhibitor at the ratio of 1:3. The mixture was then centrifuged at 4°C, 20 000 xg for 10 minutes and the resulting supernatant was used for the Bradford protein assay (Bradford 1976). Samples were done in triplicate. Standard curves were prepared using bovine serum albumin (BSA) and protein absorbance was measured at 595 nm.
Bee lifespan
Lifespan is also a commonly evaluated parameter to test the suitability of different diets in honey bees (Alqarni 2006; Hocherl et al. 2012; Ihle, Baker, and Amdam 2014; Li et al. 2014; Manning et al. 2007; van der Steen 2007). Longevity tests were carried out in 2016 and 2017. In late July 2016, one colony of each supplementation regime without pollen traps was chosen at Site 3, and 90 newly emerged bees per colony were randomly selected and placed in plexiglass cages (13.5 x 12.5 x 17.5 cm). To obtain newly emerged bees, we first removed adult bees from a brood frame, which we then put in a mesh net. The next day, bees present in the net (newly emerged bees) were selected for the test. In 2017, the test was carried out in late May after four weeks of colony supplementation, ensuring that newly emerged bees were fed a supplemented (or unsupplemented) diet as larvae. All of the colonies were used for the test and 60 newly emerged bees per hive were randomly picked and placed in cages, for a total of 300 bees per treatment. In both years, cages contained 30 newly emerged bees and were kept in an incubator at 30°C, 75% humidity, in total darkness (Williams et al. 2013). Bees were fed with 50% (weight/volume) sucrose solution through two 1.5 ml Eppendorf tubes per cage. Feeding tubes were changed daily. Dead bees were counted and removed approximately every 24 hours, until no more bees remained. The trial lasted for 50 days in 2016 and 45 days in 2017.
Nosema and varroa mite infestation
Nosema spp. and Varroa destructor were both monitored during the season. Nosema
infestation levels were calculated at the end of July 2016 using the spore count method described in Pernal and Clay (2015). Varroa mite levels were monitored on May 3rd and July 27th 2016 using the 75% ethanol wash method (Dietemann et al. 2013).
Landscape analyses
To assess the impact of environment on the colonies, landscape was characterized within a radius of 5 km around the apiaries. QGIS software (2.18.14) was used to calculate the ratios of cultivated land and specific crop (La Financière agricole du Québec 2016), wooden area (Ministère des Ressources naturelles et de la Faune du Québec 2000) and urban landscape (DMTI Spatial 2012).
Statistical analyses
Since measurements were taken across time on the same hives, the variables supplement consumption, pollen weight, sealed brood surface, foraging effort, protein content of the honey bees and varroa mite infestation were analysed using repeated measures ANOVA models. However, for Nosema spp. infestation, we used a two-way ANOVA model since measurements were taken at one specific time point. In the repeated measures ANOVA models, the factors site and treatment are between-hives sources of variation, while time is a within-hive source of variation. The best correlation structure between observations taken on the same hive through time was selected based on the Akaike Information Criterion (AIC). Response variables were transformed when necessary to meet the assumption of normality, but the least square means and their corresponding confidence intervals were back- transformed at their original scale to make comparison with other studies easier. Following a significant effect in any ANOVA table, post-hoc multiple comparisons were performed using the Tukey-Kramer method to control the type I error rate. Lifespan was computed using Kaplan-Meier curves of honey bee survival, and a log-rank test was performed to assess for significant differences between curves. Data was right censored if a honey bee died of non- natural cause (crushed or drowned), indicating it could have lived longer than the value
entered. All statistical analyses were performed using the R software (R Core Team 2016) at the significance level of α=5 %.
Results
Supplement consumption
Daily supplement consumption ranged from 0.51 to 270 g per colony. ANOVA revealed significant site x time (F22, 363 = 7.5714, p < 0.0001) and site x treatment (F6, 22 = 4.4267,
p = 0.0044) interactions. For the site x time interaction, multiple comparisons indicated that nine weeks out of 12 showed a significant difference between sites. Up until the end of June, daily consumption was generally highest at Site 3, lowest at Site 1 and Site 2 displayed an intermediate level of consumption. However, after that, consumption at Site 2 decreased, while at Site 1 and especially Site 3 consumption kept rising. During the last week of supplementation, Site 3 colonies consumed, on average, 158 g [95% CI 126.8–196.1 g] of supplement daily, which was 57 g more than colonies at Site 1, and almost 100 g more than colonies at Site 2. As for the site x treatment interaction, as shown in figure 1, the pollen- enriched supplements were generally more consumed than the pollen-free supplements across sites.
Fig. 1: Least square means of weight of supplement consumed per day per hive ± 95% CI
for each treatment and site in 2016. Least square means of weight of supplement consumed per day per hive ± 95% CI for each treatment and site in 2016. For each site, different letters indicate significant differences (Tukey-Kramer, p < 0.05). Means and confidence intervals were back-transformed to the original scale of the data.
Pollen weight
Daily collection of pollen by the traps ranged from 0 to 180 g per colony. A significant site x time interaction (F30, 223 = 2.4892, p = 0.0001) was detected. However, following multiple
comparisons, we found no significant difference between sites except for the last two weeks of the experiment, late August and early September. At that time, the pollen traps at Site 3 weighed more than those at Site 1 and 2, which contained less than 10 g of pollen pellets on average.
Sealed brood surface
Our model revealed a significant site x time interaction (F10, 210 = 3.9585, p = 0.0001) for
sealed brood area. Multiple comparisons indicated differences between sites on July 6th, where Site 2 displayed on average less brood (1961 cm2, 95% CI 1359–2564 cm2 per colony)
95% CI 2973–4114 cm2 per colony for Site 3), and on September 6th, where Site 3 displayed on average more brood (1962 cm2, 95% CI 1392–2533 cm2 per colony) than the other sites (1069 cm2, 95% CI 469–1669 cm2 per colony for Site 1 and 970 cm2, 95% CI 368–1573 cm2 per colony for Site 2).
Foraging effort
Foraging effort, evaluated by the proportion of foragers returning to the hive with corbicular pollen, ranged from 0 to 72%. A significant effect of treatment (F4, 28 = 7.5641, p = 0.0003)
and a site x time interaction (F10, 208 = 4.3091, p < 0.0001) were found. As shown in figure 2,
multiple comparisons confirmed that foraging effort was significantly higher for the PFS + traps treatment, which was around double the value observed for the control and “without traps” treatments. As for the site x time interaction, significant differences between sites were found at 3 sampling dates out of 6, but no clear tendency could be observed.
Fig. 2: Mean proportion of foragers bringing back pollen to the hive ± 95% CI for each
treatment in 2016. Different letters indicate significant differences (Tukey-Kramer, p < 0.05). Means and confidence intervals were back-transformed to the original scale of the data. Since there was no interaction between treatments and time or sites, least square means for the main effect of treatments, over all sites and times, are presented.
Protein content of adult honey bees
Total protein content of the nurse bees ranged from 26.2 to 47.8 mg of protein per gram of