Varroa

Here, we provide a summary description of the BEEHAVE model. A full description is included in Appendix S1 (Supporting information), following the ODD protocol (Overview, Design concepts, Details), a standard format for describing individual‐based models (Grimm et al. 2006, 2010). Additionally, hyperlinks allow the reader to move between the ODD description and corresponding program code. BEEHAVE is available to download at www.beehave-model.net and is implemented in the freely available software platform netlogo (Wilensky 1999). The program code and a user manual are included in Appendices S2 and S3 (Supporting information). Unless explicitly stated otherwise, we always address the modelled colony when using biological terms like ‘colony’, ‘foraging’, etc. in the following description.
The purpose of BEEHAVE is to explore how various stressors, including varroa mites, virus infections, impaired foraging behaviour, changes in landscape structure and dynamics, and pesticides affect, in isolation and combination, the performance and possible decline and failure of single managed colonies of honeybees. The model consists of three integrated modules: the colony, the mite and the foraging model (Fig. 1). An additional, external landscape module (M.A. Becher, V. Grimm, P. Thorbek, P.J. Kennedy & J.L. Osborne, unpublished data) can be used to create input files (Table S2, Supporting information) for the foraging model. We did not add a Nosema module at this point, as the mechanisms of transmission are still not well understood (Higes, Martín‐Hernández & Meana 2010), but we provide some suggestions of how Nosema infections can be addressed by changing some parameters. The colony model describes in‐hive processes, using difference equations to generate the colony structure and dynamics for the brood, in‐hive worker and drone population. The mite model describes the dynamics of a varroa mite population within the honeybee colony. As vectors for viruses, mites affect the mortality of bee pupae and adult bees. Viruses are not implemented as entities but via infection rates of mites and bees (BEEHAVE considers one type of virus at a time). The colony and mite model proceed in daily time steps.
The foraging model is an agent‐based model, which represents foraging at flower patches located in the landscape around the hive. Space is represented implicitly: properties of these flower patches, such as probability of being detected by scouting bees, distance to the hive, or nectar and pollen availability, are either set by the modeller when exploring hypothetical landscapes or extracted from real crop maps using the external landscape module. Landscape dynamics, including changes in location and availability of crop fields of different types, can be taken into account by updating the imported landscape data at every time step of the colony model. The foraging model, which is executed once per day, includes a varying number of foraging trips, depending on the quality of nectar and pollen sources in the landscape, the weather conditions and the size, stores, and demand of the colony. The foraging processes represented thus operate on the time‐scale of minutes.
The structure of the model is a compromise between structural realism (i.e. the ability to represent heterogeneity where it is likely to matter) and computational efficiency and parsimony regarding parameterization and model analysis. Hence, the in‐hive bee population is represented via age cohorts, foraging bees via ‘super‐individuals’ (Rose, Christensen & DeAngelis 1993; Scheffer et al. 1995; Grimm & Railsback 2005), mites via the total number of virus‐free and virus‐carrying mites and viruses via transmission rates between mites and bees. We also considered it essential to explicitly represent environmental factors driving colony dynamics and to link within‐hive dynamics and foraging by representing the seasonally dynamic storage, consumption, demand and collection of nectar and pollen. Including this level of complexity ensures that links and feedback mechanisms between reproduction, brood, food stores and foragers can be successfully captured.

Brood for the experiment was taken from a healthy Apis mellifera colony belonging to the experimental apiary of the Faculty of Veterinary Medicine, University of Belgrade. The colony was Nosema-free, as confirmed by PCR, using methodology described in Stevanovic et al. [9 ], and without any signs of other infections (bacterial, viral, protozoan or fungal) in the past two years. The presence of viruses was checked according to symptoms described earlier [52 ] and Varroa infestation was kept at a low level. Frames with sealed brood were incubated at 34°C ± 1°C and newly emerged worker bees were taken, confined to six cage groups containing 40 bees in each and kept in the incubator [53 ]. In order to provide absolutely equal conditions for all bees from same group, and exclude the impact of all external factors (position in incubator, humidity, temperature, food amount etc.) some modifications (Fig 1) of cages presented in Williams et al. [53 ] were made. There were 40 individuals in each cage, needed for each treatment group (5 replicates for gene expression analyses and 5 for Nosema spore counting for each of the three collection times, plus 10 for mortality recording). Two independent series of experiments with essentially similar results were conducted, so the data were merged. The bees were fed ad libitum with a solution of sucrose (50% w/w). One control group was experimentally infected with N. ceranae spores (I group), the other was not (NI group), but both were fed on pure sugar syrup (without supplement). The remaining four groups were fed with sugar syrup enriched with supplement starting from day 1 (I-BW1 group), 3 (I-BW3 group), 6 (I-BW6 group) and 9 (I-BW9 group) after emerging (Table 2). All groups, except NI, were infected with N. ceranae spores. Small petri dishes (Fig 1) with the same volume of food (12 ml) were replaced daily in all cages. We have monitored the intake and noticed that the whole quantities were consumed. The supplemented sugar solutions were consumed as readily as the non-supplemented. Bees did not regurgitate the food. Dead bees were removed daily and their numbers recorded. In a preliminary investigation in both laboratory and field conditions no obvious harmful effects on bees have been observed.

Technical grade glyphosate (62.27% w/w glyphosate isopropylamine [IPA] salt corresponding to 46.14% w/w glyphosate acid equivalent [a.e.]) and the soluble concentrate formulation of glyphosate (MON 52276) (30.68% glyphosate a.e. as the IPA salt, batch no GLP-0810-19515-A), supplied by Monsanto (St. Louis, MO) were used in the study. All honeybee colonies were obtained from National Bee Unit, FERA, (York, UK) apiaries and were confirmed as having low incidence of adult bee diseases, viruses, and varroa with no clinical signs of brood diseases.

We calculated the infestation rate by varroa mites of each sample. To do so, soap was mixed with warm water in a beaker with ca. 100 bees per sample retrieved from the −80 °C freezer(ThermoFisher Scientific, Frankfurt, Germany). The beaker was shaken and stirred for 10 min to improve the efficiency of dislodging varroa mites from the sample. The sample was then flushed with a large volume of warm water to separate the bees from the mites before pouring the fluid through a sieve (Impexron GmbH Pfullingen, Germany) with a mesh size of 3–4 mm to remove larger debris. To collect the mites, all of which should be washed through the first sieve, a second sieve (aperture 0.3 mm) was placed beneath the first. The mites that remained on the second sieve were counted, as were the washed bees in the sample, and the number of mites per 100 bees was calculated [23 (link),24 (link)].

All statistical analyses and data visualizations were performed in GraphPad Prism 7.0 (GraphPad, La Jolla, San Diego, CA, USA). Normality of data was assessed by use of the Kolmogorov–Smirnov test and homogeneity of variance was determined with Levene’s test. The results were expressed as the number of varroa per 100 bees or the presence or absence of varroa mites, as well as viral presence/absence in each apiary per season. Chi-square tests were used to test for differences in proportions (varroa mite presence/absence; viral prevalence/absence in winter versus summer). The association between viral presence (yes/no) and varroa infestation (number of varroa per 100 bees) were compared using Student’s t-tests with Welch’s correction for unequal variances and Bonferroni correction for multiple testing (12 tests, representing the occurrence of a virus in the winter or the summer). An adjusted alpha level of 0.05 was used for all statistical tests to define significance.

The criteria for declaring a beekeeper as professional was the possession of at least 150 productive colonies before winter. Accordingly, beekeepers with fewer than 150 productive colonies before winter were considered non-professionals. Different forage plants were grouped into two categories. Orchard trees, oil seed rape, maize, sunflower, and cotton were grouped together as agricultural forage sources, and pines/conifers, thyme, heather, and cistus were grouped as natural forage sources. For each beekeeper, a score of +1 was given to each visit declared to a natural habitat (heather, pine or conifers, thyme, and cistus) and −1 to each visit to an agricultural habitat (orchards, OSR, maize, sunflowers, cotton). A positive score indicated a tendency for the beekeeper to choose natural habitats.
The different varroa treatment methods were grouped into four major categories based on previous research categorization [56 (link)]: The “Biotechnical method” includes beekeepers exclusively using hive manipulation techniques without adding substances such as drone removal, hyperthermia, or other biotechnical methods. The “Organic acid” category includes beekeepers treating hives only with natural substances such as formic acid (short- and long-term exposure), lactic acid, oxalic acid (by strip, sublimation, or combined with glycerin), or other commercial mixtures (e.g., Hiveclean^®, Bienenwohl^®, Varromed^®) and Thymol (e.g., Apiguard^®, ApilifeVar^®, Thymovar^®). The “Synthetic acaricide” category was assigned to beekeepers using only chemical treatments such as Tau-Fluvalinate (e.g., Apistan^®), Flumethrin (e.g., Bayvarol^®, Polyvar^®), Amitraz (by strips, e.g., Apivar ^®, Apitraz^®, or by fumigation or aerosol), Coumaphos (by strip, e.g., Perizin^® or strips, e.g., Checkmite+^®) or any other chemical substances. The “Other method” was used to label beekeepers reporting other, unspecified methods for varroa control. Finally, the category “Multiple” was created to represent beekeepers using a combination of at least 2 of the previous categories.