Hosted by the College of Agricultural and Environmental Sciences at the University of Georgia
Resources page banner

CAP Team Articles

Honey Bee “Medical Records”: The Stationary Apiary Monitoring Project
Jointly published in the American Bee Journal and in Bee Culture,
March 2010

Marla Spivak, University of Minnesota

Marla Spivak Why are so many honey bee colonies dying across the U.S.?  Wouldn’t it be great if we could download medical records for every honey bee colony that died over the last 25 years, or even the last four years, to piece together the primary factors or patterns of factors that correspond with colony death?  

Thanks to funding from the USDA-CAP project, a number of researchers are compiling the first of four years worth of medical records on 420 colonies located across seven states.  Our objective is to determine the role of pests, pathogens and pesticides in causing death in these stationary honey bee colonies.  We chose to study colonies that are not transported for pollination and honey production because researchers involved in USDA-ARS Area Wide projects, funded by a different branch of the USDA, are studying the health of migratory colonies. This study will provide unprecedented and very important data on colony health within stationary apiaries across the U.S.

In April and May 2009 we established 30 colonies in each of seven apiaries located in Washington, California, Texas, Florida, Maine, Pennsylvania, and Minnesota.  The apiary sites were situated in a variety of urban, suburban and rural landscapes including areas surrounded by agricultural fields, organic farms, wooded areas and nature reserves.  The cooperating researchers in each state are:  Steve Sheppard, Washington State Univ.; Kirk Visscher, Univ. California-Riverside; Kate Aronstein, USDA-ARS Weslaco; Jamie Ellis, Univ. Florida; Frank Drummond, Univ. Maine; Nancy Ostiguy, Penn State Univ.; and Marla Spivak, Univ. Minnesota.  Brian Eitzer from the CT Ag. Exp. Stn. is in charge of analyzing pesticide exposure from pollen samples collected by the honey bees in each of the states. Anne Averill from Univ. Massachusetts is studying the health of bumblebees that are nesting naturally in the area surrounding the apiaries. 

Each of us purchased package bees from our closest or region-specific distributor:  the WA, CA and MN packages came from two suppliers in CA; the FL, ME, and PA packages came from two suppliers in GA; and the TX packages came from TX. This was done for two reasons. First, it is not possible to have packages shipped from a single supplier to all of the seven states involved in this study. Second, we wanted to know the initial disease and mite levels in packages bees from different sources. The 210 colonies were hived in new wood boxes, and we used new wax-coated plastic foundation purchased from Pierco® Beekeeping Equipment.  In May we replaced the queens in each of the packages with queens of Italian descent purchased from a single operation in northern CA (C. F. Koehnen and Sons, Inc.) to establish relatively uniform genetics among the colonies. We are not treating any of the experimental colonies for diseases or mites.  Each of us are using management practices typical to our different climate and resource conditions.  We initially fed the packages sugar syrup and pollen substitute (MegaBee®) as needed, and later provided supplementary syrup to some colonies in fall to bring them up to weight for winter, but otherwise all colonies have been left to develop on their own. Beginning when the packages were hived and continuing throughout the duration of the bee season in each state, we have been collecting an enormous amount of data from each colony every month.  We will follow the colonies until they die, which unfortunately may not take too long.  In 2011, we will start up again with a new set of 210 colonies, and will use one or more different queen sources.

As an example of what is involved in this experiment, I describe one of the monthly data collection trips to our stationary apiary in Minnesota on July 28. Each of us in the seven locations conducted similar collection trips once a month starting in April or May.  Most of our colonies by that time had grown to occupy three deep brood chambers. In addition, some of the colonies were provided honey supers as needed (Figure 1). 

Fig. 1
The stationary apiary located at Carver Nature Reserve in Minnesota at the end of July 2009 during our monthly data collecting trip. Variation in colony strength is evident but most colonies were strong and apparently healthy by the end of summer. The colony in two-deep brood chambers, next to the large colony with supers, is one of five randomly chosen colonies in the apiary fitted with a pollen trap so we could sample pesticide residues in incoming pollen.

The collection trip began the day before when Mike Goblirsch, one of my students who took primary responsibility for this project, prepared and labeled 150 collection vials, copied data sheets and assembled all other necessary equipment.  We met at 8:00am, loaded our gear into the University truck, stopped to put dry ice in three different coolers and drove 45 miles to our site located at the Carver Nature Reserve, west of the Twin Cities.  This is traditionally a great location for honey production and we thought would give our colonies the best possible resource conditions.

When we arrived on site, Mike reviewed the protocol with us (Figure 2): 

Fig. 2
Mike Goblirsch, graduate student in charge of organizing the data collection trips, reviewing the sample collection procedure at the start of the long day.  Clockwise, starting from the far left: Mike Goblirsch, Gary Reuter, Encarna Garrido (visiting from Spain), Katie Lee and Betsy Ranum.   

1) Collect a sample of 40 foragers returning to the colony to test for Nosema load and species (Nosema ceranae or N. apis) and put the vials immediately on dry ice; 2) Collect samples of 40 foragers, 40 nurse bees and 20 drones to test for viruses and place these vials in separate coolers of dry ice; 3) Collect a sample of 50 bees from the inner cover into a vial of alcohol to sample for tracheal mites; 4) Collect 300 bees to sample Varroa using the powdered sugar method to dislodge and count the mites (at least we could return these bees to each colony);  5) Determine presence or absence of small hive beetles;  6) record observed disease symptoms in colony; 7) Estimate the number of adult bees on every frame in every box and the amount of sealed brood on each frame using a standard procedure; 8) Check for the presence of the marked queen, and mark any new queens resulting from supercedure; and 9) Collect pollen from the traps placed on 5 of the 30 colonies to analyze for pesticide residues. 

We worked in pairs; one person conducted the adult bee and brood area estimates, while the other took samples and recorded data. We worked slowly and methodically, taking extra care not to damage the queens.  By 4:00pm we were hot, sticky and giddy. I won’t tell you about the stupid songs we started singing, including one about 99 bottles of beer on the wall, and another about finding a peanut.

Preliminary findings from this large-scale experiment were presented at the American Bee Research Conference, held in conjunction with the American Beekeeping Federation meeting in Orlando, FL, January 14-15.  While much of the data are still being analyzed, some interesting information is emerging from the data at hand.  

Package bees:

The prevalence of parasites, pests and pathogens varied widely among the original bees that came in the packages.  Although the package bees were replaced with workers from the new queens by June, it is important to know the health status of the bees that were hived at the beginning of the experiment.

For the most part, the bees in the packages had low or no detectable Varroa and tracheal mites, except for those from TX that had relatively high levels ofVarroa (Table 1). All samples were positive for N. ceranae and negative for N. apis.  The Nosema levels varied greatly among packages, being highest in MN and non-detectable in TX and PA.  All samples tested positive for at least 3 viruses: Black Queen Cell Virus (BQCV), Deformed Wing Virus (DWV) and Sacbrood virus (SBV), while Israeli Acute Paralysis Virus (IAPV) was detected in only a few bees in some locations.  

Table 1. Initial measures (May 2009) of parasites, pests and pathogens from package bees used to initiate colonies for the Stationary Apiary Project. The packages in MN and WA were purchased from two locations in CA.  Packages in TX were purchased from TX, and those in FL, PA and ME were purchased from two locations in GA.  Data are not yet included from the apiary in CA. NA refers to data not yet available. 

Site n Varroa mites

(mites /100 bees)

Tracheal mites

(% bees infested )

Small Hive Beetle

(presence / absence)

Nosema ceranae

(spores in millions / bee)

Virus prevalence (positive samples)
MN 32 0 0.2 ± 0.9 No 3.4 ± 2.3 87% 56% 2% 16%
WA 30 0.1 ± 0.2 0.3 ± 1.1 No 1.5  ±1.0 58% 61% 7% 77%
TX 30 2.8 ± 6.0 0 No 0.03 ± 0.06 24% 96% 0 0
FL 30 1.5  ±1.2 1.7 ± 3.1 No 0.9 ± 2.4 20% 83% 0 20%
PA 30 NA 0.5 ± 1.6 No 0.03 ± 0.07 67% 74% 1% 10%
ME 30 0.5  ±1.9 0 No 0.4 ± 0.5 29% 55% 0 22%

Colony Growth:

Colony populations of adult bees and brood increased at different rates in each state (Figure 3). 

Fig. 3
Monthly averages of adult bees (top graph) and sealed brood (bottom graph) in the colonies in six of the seven stationary apiary locations. Each apiary was initiated with 30 colonies in spring.  When all the data is compiled, we will relate colony strength to the various pest, pathogen and pesticide levels that were measured throughout the testing period. We are using this comparative approach to determine if specific causes for colony losses can be found and to evaluate the possible regional differences in the pressures that honey bees face. 

The colonies in MN became the most populous by July with an average of 22,000 bees and 10,000 cells of sealed worker brood. Population growth of colonies in ME was probably hindered by cold, wet weather conditions, and in TX by very hot and dry weather.  We will be factoring landscape and climate conditions into our analysis when we analyze all the data together.  


Pollen was sampled with traps one day per week from five hives in each of the stationary apiaries. Initially samples were combined to generate a monthly composite sample for each apiary. Brian Eitzer in CT used a technique known as QuEChERS (for Quick, Easy, Cheap, Effective, Rugged and Safe) to analyze the samples using high performance liquid chromatography/mass spectrometry.  This technique allows over 140 different pesticides to be analyzed in the parts per billion (ppb) concentration range.

To date 29 of the monthly composite samples have been analyzed.  Within these 29 samples, residues of 32 different pesticides or pesticide metabolites have been observed including: 14 insecticides plus one insecticide metabolite, 9 fungicides and 8 herbicides.  The average composite pollen sample had an average of 4.1 pesticide residues detected.  The concentration of residues when detected was mostly in the low PPB range (1< to 30 ppb) but some residues were substantially higher.

The results indicate that honey bees at the stationary apiaries are being exposed to varying amounts of pesticides within and across locations.  This variability of pesticide exposure will be further examined as we continue to monitor these hives over the next several years.    

Queen Supercedures:

Throughout the summer, there were a number of colonies that superceded the queens.  In MN, 16 of the 30 colonies replaced their queens, 8 of them twice.  In PA, 12 colonies superceded queens, in Maine there were 7 colonies that superceded, 2 of them twice, and in both TX and FL, 6 colonies superceded queens. The high number of queen replacements did not appear to affect the development of the colonies in MN; they still became quite populous. Beekeepers have been claiming that many queens are superceded.  Our findings show that queen replacement is something that deserves more research attention. 

Status of native bumble bees?

Anne Averill will be monitoring native bee populations that forage near the stationary apiaries for known pathogens of both honey bees and bumble bees.  It has previously been shown that honey bee viruses as well as Nosema ceranae can infect bumble bees.   We will be able to compare pathogen presence and/or titer data with similar data collected from the stationary honey bee hives at each of the apiaries. In addition, pollen collected from the bumble bees sampled will be analyzed by Brian Eitzer to determine if bumble bees and honey bees are being exposed to the same pesticide composition. 

End of the First Summer:

By the end of the summer, TX and FL had the highest levels of Varroa, and were the only states with small hive beetles (Table 2).  The tracheal mite levels in the Maine colonies, which were subject to the coldest and wettest weather conditions, were quite high (27%) in July but dropped to very low levels by August.  The levels of Nosema dropped to low levels in all locations, and only N. ceranae was detected throughout the season. The virus data is being analyzed, but early results show high virus prevalence in all locations analyzed.  The apiaries in TX and PA suffered the largest number of colony deaths, with 14 and 18 colonies left of the original 30, respectively.  We will be collecting data on the remaining colonies through 2010, or as long as they survive.  

Table 2. Late season measures (Sept-Nov 2009) of parasites, pests, and pathogens from the remaining colonies in the Stationary Apiary Project. Data are not yet included from the apiary in CA and on virus prevalence. NA refers to data not yet available. 

Site n Super-ceded colonies Varroa mites  (Sept- Nov)

(mites /100 bees)

Tracheal mites  (Aug)

(% bees infested)

Small Hive Beetle

(presence / absence)

Nosema ceranae (Aug)

(spores in millions / bee)

MN 29 16 2.5 ± 3.4 0.2 ± 0.9 No 0.02 ± 0.5
WA 29 NA 0.5  ± 0.7 0.5 ± 2.1 No 0.03 ± 0.1
TX 14 12 17.9 ±15.7 4.1 ± 5.5 Yes 0
FL 23 6 7.1 ± 6.6 2.6 ± 3.8 Yes 0.2 ± 0.7
PA 18 6 NA 1.5 ± 2.4 No 0.25 ± 0.24
ME 20 7 0.9 ± 2.3 0.6 ± 2.2 No 0.26 ± 0.43

What can we conclude?

What was the cause of death of the colonies that were lost? Once our medical records are compiled for this first experimental set-up, we expect to be able to relate colony health (or lack thereof) to the various pest, pathogen and pesticide levels that were measured throughout the testing period.  In addition, information about weather and available nutrition will be included.  We plan to use this comparative approach to try and determine if specific causes for colony losses can be found and to evaluate the possible regional differences in the pressures that honey bees face. In 2011, we will start the entire experiment up again with a new batch of 210 colonies and new queen sources.

For now, watching the experiment unfold and documenting the variation in the strength and health among colonies in different climatic zones is fascinating in and of itself.  It has given me a renewed appreciation for the diversity (or lack of diversity) in the landscapes that our bees experience across the U.S.  It has stimulated me to think more about regionally adapted bee stocks.  Mostly, the pathogens, parasites and pesticides our bees face on a daily basis gives me pause for thought, and concerns me greatly.

The acronym CAP stands for Coordinated Agricultural Project. This research project is a good example of how multi-institutional funding is realized.  In addition to learning about the factors that contribute to colony loss in stationary apiaries, we are learning how to better coordinate our efforts across research institutions.  Our goal is to facilitate bee health, best management practices, and productive research collaborations.   This is truly a win-win effort.