Progress report for LNE23-475R
Project Information
Honey bee colonies experience much greater annual loss than historically, largely because of intense parasitism by the mite Varroa destructor and the viruses that parasite spreads. Annual losses of bee colonies are now exceeding 50%, far more than the 12-15% typically observed before the current crisis began around 2006. Varroa and the viruses are key contributors to those losses.
Early and accurate estimates of mite levels in a hive informs important management decisions, including whether or not and when to treat a colony. The two most common sampling methods, using either powdered sugar or isopropyl alcohol to dislodge varroa from a sample of worker bees, are insensitive to early season mite numbers which delays management decisions until the parasite’s levels are quite large and much tougher to control.
In early spring, when varroa are at their lowest numbers and most difficult to detect, they are 5-10x more abundant on drones than on workers—yet current methods sample workers only. In a “proof of concept” test, we will compare the sensitivity and accuracy of an early-season drone-focused sampling of Varroa to that of the standard alcohol wash method. We expect to learn if/how the new drone-focused sampling provides improved and timely estimates of varroa pressure within hives. In addition, we are measuring the amount of deformed wing virus in drone and worker bees in the sampled hive in order to compare the levels among sexes within hives, and between hives to measure the relationship between mite levels and virus load. Finally, we will develop a sequential sampling plan for hand-sampled drone varroa in order to reduce the amount of sampling needed to obtain estimates of varroa population sizes and threshold levels to trigger management actions. We will determine that it's an easier, earlier, and more accurate way to measure Varroa levels.
In early spring, when varroa are at their lowest numbers and most difficult to detect, they are 5-10x more abundant on drones than on workers—yet current methods sample workers only. In a “proof of concept” test, we will compare the sensitivity and accuracy of drone-focused sampling of Varroa to that of the standard alcohol wash method. We expect to learn if/how the new drone-focused sampling provides improved and timely estimates of varroa pressure and virus levels within hives.
We will determine that it's an easier, earlier, and more accurate way to measure Varroa levels.
Honey bee colonies experience much greater annual loss than historically, largely because of intense parasitism by the mite Varroa destructor and the viruses that parasite spreads. Annual losses of bee colonies exceeding 50%, far more than the 12-15% typically observed before the current crisis began around 2006.
Early and accurate estimates of mite levels in a hive informs important management decisions, including whether or not and when to treat a colony. The two most common sampling methods, using either powdered sugar or isopropyl alcohol to dislodge varroa from a sample of worker bees, are insensitive to early season mite numbers which delays management decisions until the parasite’s levels are quite large and much tougher to control. Varroa control is difficult and not completely effective. This means that treating colonies when varroa numbers are small is essential for obtaining maximal effect, whether that treatment is a colony manipulation such as removal of drone cells or brood-free periods; or a chemical application such as acaricide, oxalic acid or thymol treatments.
Through detailed field observations of honey bee - varroa interactions, we have determined that varroa have a 5-10 fold preference for drone bees and this preference is largest in the spring and early summer. A consequence of this extreme bias is that the standard sampling of worker bees will be missing the majority of the varroa. Alternatively, just as the bank robber Willie Sutton robbed banks because "that is where the money is", we propose sampling the drone bees, because that is where the varroa are. If we focus sampling on drone bees, we will have 5-10 fold more sensitive measures, especially when varroa numbers are low, providing beekeepers an early season advantage in varroa management.
Preliminary tests of drone-focused sampling for varroa have been a success, in terms of sensitivity, but we have not tuned the sampling methods to the needs of beekeepers. Our survey of over 100 beekeepers (with 2 to >30,000 colonies under management) indicated that varroa and the associated viruses were the primary challenge to profitable and sustainable beekeeping. Most of those sampled knew of the importance of sampling for proper varroa control, but most also said that sampling was too time consuming and did not give information when most needed. Therefore, we seek to develop a drone-focused sampling method that is easier and faster than the standard worker sampling methods.
This research will benefit beekeepers, and fruit and vegetable producers in the Northeast through the health of managed pollinators. In the Northeast, more than 8,600 farms maintain 126,000 hives, and produce more than 5 million pounds of honey—in addition to the significant pollination service provided to the fruit and vegetable producers.
Farmers are interested because Varroa is the most significant challenge to all beekeepers and the largest cause of colony loss (2021 NASS report). Improved management information will reduce colony losses, reduce dependence on queen breeders in other regions, increase profits, and enhance the sustainability of beekeeping in the NE Region.
Research
The key question we address here is: “Does a varroa sampling protocol focused on individual sampling of drones perform better than the standard worker-focused methods currently used”? We hypothesize that because varroa have a strong bias for drones as hosts, sampling of drones will be more sensitive, more timely and more informative than worker sampling. We also expect virus sampling of drones to be significantly more informative and sensitive than virus sampling of workers.
In order to assess the sensitivity and the utility of our new varroa sampling strategy centered on sampling of drones, and to compare those results to those of the current worker-focused sampling we will do both types of sampling in each of 30 hives through the entire honey bee foraging season in Maryland (April – September).
Treatments: In the spring of year 1 we will conduct preliminary assessments of varroa numbers on a large number of hives in order to select 30 hives that vary from very low to moderate varroa numbers. Each of the selected hives will undergo drone-focused and standard (worker-focused) sampling monthly, from April through September. Each hive will also be sampled and the abundance of deformed wing virus estimated via qPCR. At the end of each season (September), overall hive health will be assessed through a comprehensive estimation of the number of frames of worker bees and brood, the amount of stored honey and pollen, and the presence of a healthy, egg-laying queen. In the spring, the survivorship of each hive will be noted (primarily intact or dead but also estimating if viable or inviable).
In year 2 we will assess the utility of the sequential sampling plan developed using the data of year 1. Using hives with naturally varying levels of mites from colonies establish with packages the previous summer, we will use drone-focused sequential sampling to estimate varroa infestation levels in each hive and determine when control measures are needed for each hive. Before beginning the sampling, we will modify and fine-tune our methods in aim 1 to reduce sample time and effort, hopefully without significant loss of sensitivity
Methods: Drone-focused sampling. Drone focused sampling involves collection of drone bees only, through hand sampling within the beehive, and placing drones individually into vials containing 50% isopropanol. After several minutes, the vials are agitated and the mites release from the bees and sink to the bottom of the vial, and are then counted. Drones (75 from at least two frames) are sampled individually in order to simulate the process of sequential sampling, to determine the frequency of drones with multiple mites and the numbers of those mites, and to better estimate the key parameters needed for developing a sequential sampling plan (Binn and Nyrop 1992, Heck 2021).
Worker-focused sampling. The currently recommended varroa sampling method involves collection of around 300 worker bees using a small scoop, placing the bees into a jar containing 50% isopropanol, agitation of the jar to dislodge the mites from the bees, and counting the mites that have been washed off. The number of mites per 300 bees allows estimation of the percentage of bees infested.
Virus sampling. Both drones and workers from each sample will be used to estimate the level of deformed wing virus in the hive. The virus levels will be determined by bulk extraction of RNA from 50 bees from each sex and quantitative amplification of target gene sequences of the virus. This will allow us to determine the correlation of varroa numbers in samples and virus load estimates and to compare the deformed wing virus loads through the season on drones and workers. Because our hypothesis is that the drone samples will better estimate the number of varroa in each hive and because mite numbers drive virus transmission, the expectation is that the correlation of drone-focused varroa samples will be greater than those of the worker-focused samples. Because the drone-focused sampling is expected to be approximately 5-fold more sensitive, we also expect the early season drone-focused samples (when varroa are at their lowest levels) to better predict virus levels.
Data Analysis and presentation: For each of the two sampling methods and the virus load measures, the overall infestation rate (varroa per bee) and summary statistics describing variation in the estimates will be calculated. The two estimates from each hive (Drone and worker focused samples) at each sampling date will be compared in an ANOVA framework to test the hypothesis that the methods differ in the number of mites detected and that the drone-focused method detects a greater number of mites at the early-season dates, if not throughout the season. All of the end-of-season health measures will be analyzed along with the varroa (both methods) and virus estimates, to determine if colony health can be explained by earlier varroa and/or virus levels and if so to compare the varroa estimation methods. These end-of-season measures will also allow us to develop draft estimates of threshold levels for varroa infestation levels. These thresholds demarcate the varroa and virus levels that require control measures for long-term survival of the colony and allow us to attach management decisions to the new measures of varroa infestation rates.
Over the winter between years 1 and 2 we will develop a sequential sampling plan for use of the drone-focused sampling method. Using the frameworks described by Heck et al. 2021 and the data gathered in season 1, we will establish the cumulative number of varroa in a sequential sampling of individual drones which when exceeded, an action threshold is reached and treatment would be recommended.
Aim 2 will involve active collaboration with farmers and cooperators to evaluate the utility and feasibility of the sampling method. Workshops will be held with cooperating organizations serving minority and veteran beekeepers, county beekeeping clubs, and summer meetings of larger-scale beekeepers and academics. During these workshops we will teach the drone-focused sampling method, using live demonstration when possible, and describe the logic and protocol for sequential sampling of drones for varroa infestation levels. Participants will be asked if the methods make sense, if the protocol seems like it will be easier than the methods they currently use for varroa sampling, and if the improvements are sufficient to increase the likelihood or frequency of their own sampling of their hives. For those willing to try the method on their own hives, we will ask how long the sampling process took, what their results were, and if they would use this method in the future.
During the final months of year 2 we will be writing 2 publications, one targeting a non-academic audience to be published in Bee Culture and the second will be published in an entomological or apicultural scientific journal, targeting an academic audience.
In both May and June, the earliest two varroa mite sample months, drone sampling identified more colonies with mites than alcohol wash sampling did. In May, the mean number of mites per 30 drones was 0.8 which when scaled to 100 drones, is 2.64 mites per 100 drones. In comparison, the alcohol washes counted 0.44 mites per 100 bees, 6-fold fewer than with the drone sampling. In June that average was 7.27 mites per 100 drones sampled and 0.96 mites per 100 bees in an alcohol wash, a more than 7-fold difference. Clearly the mites are more abundant on drones than on workers.
The number of mites sampled by drone sampling in June was positively related to and predictive of the numbers of mites sampled via alcohol washes in July and August. For the July comparison, the P-value for the regression of June drone samples and July alcohol washes was 0.001 and for August that regression was not significant (p = 0.08), though the regression was still showing a positive relationship between early season drone sampling and later season alcohol washes.
The number of drones found with multiple mites increased as the summer progressed. In May no drones with multiple mites were sampled. In June and July, 17 drones ( ~1% of the total sampled) carried 2,3, or 4 mites and in August 32 drones (3.5% of the total) carried multiple mites.
These results support the original premise of this project; that the varroa mites are much more abundant on drones than on workers and that sampling of drones early in the season will be predictive of mite levels in the colony later in the season. Simalarly, we saw patterns of varroa abundance through the summer that match those typically observed in hives throughout the norther hemisphere; that they begin the summer quite low in abundance and gradually increase into the fall.
The drone sampling was much more difficult and time consuming than the alcohol washes and the relationship betwen early season drone samples and the later season samples had significant variability, reducing the predictive value of the early season samples. Drone sampling took approximately 30 minutes per hive to accomplish because drones were often difficult to find in sufficient numbers. As we develop a sequential sampling protocol for drone sampling the numbers of drones needed in each sample will likely shrink, partly addressing this challenge.