The First Two Years of the Stationary Hive Project: Abiotic Site Effects

by Francis Drummond¹; Kate Aronstein²; Judy Chen⁴; James Ellis³; Jay Evans⁴; Nancy Ostiguy⁵; Walter Sheppard⁶; Marla Spivak⁷; Kirk Visscher⁸

CAP Updates: 25

• Jointly published in the American Bee Journal and in Bee Culture, April 2012.

¹School of Biology, University of Maine, Orono, ME 04469 USA; ²Honey Bee Research Unit, USDA-ARS-SARC, 2413 E. Hwy 83, Weslaco, TX 78596 USA; ³Department of Entomology & Nematology, Honey Bee Research and Extension Laboratory, P.O. Box 110620, Bldg. 970 Natural Area Drive, University of Florida, Gainesville, FL 32611 USA; ⁴USDA Agricultural Research Service, Bee Research Lab, 10300 Baltimore Blvd., Bldg 476, Rm 100, BARC-east, Beltsville, MD 20705 USA; ⁵Department of Entomology, Pennsylvania State University, 542 Ag Sciences & Industries Building, University Park, PA 16802 USA; ⁶Department of Entomology, Washington State University, Pullman, WA 99164-6382 USA; ⁷Department of Entomology, University of Minnesota, 219 Hodson Hall – 1980 Folwell Ave, Saint Paul MN, 55108 USA; ⁸Department of Entomology, College of Natural and Agricultural Sciences, University of California, Riverside, CA 92521 USA

A stationary hive project was initiated in the spring of 2009 (Spivak 2010a,b). This research project will run through 2013 and consists of two replicate two-year trials (2009-2011 and 2011–2013).  The objective of the study was to conduct a longitudinal study of colonies through a two-year period. In this article we review observations that were recorded in the First Trial of this study (2009-2011), specifically the effects that relate to abiotic factors that characterized each apiary site. 

The experimental design of the project involves stationary apiaries of 30 colonies in each of 7 states (see Spivak 2010a,b for more detail).  In the spring of 2009 seven stationary apiaries consisting of ca. 30 colonies per site were initiated from 3- lb packages obtained from various commercial honey bee producers across the southern and western U.S.  New hive hardware was used in all apiaries. A single assigned manufacturers batch of Pierco black plastic waxed foundation was used exclusively at all apiaries and a pre-installment wax sample was collected for pesticide analysis. At all sites colonies were re-queened either at colony establishment or shortly after establishment with Italian race queens from a similar genetic background obtained from Koehnen  & Sons, Inc.  All colonies were fed sugar syrup and pollen substitute (MegaBee^®) when necessary and were managed typical for the region, except there were NO efforts to manage pests, parasites, or pathogens. Sampling of colony strength, disease incidence, pest and parasite loads, and select environmental factors was carried out at each apiary through the duration of the honey bee foraging season.

Fig. 1. Minnesota apiary in summer 2009 (left) and Maine apiary in winter 2009(right).

The following measures were taken at each of the apiaries in 2009, 2010, and are now being collected for the 2011 Trial. Monthly pollen sampling for estimating pesticide exposure was conducted by deploying traps on 5 colonies in each apiary for 1-2 days two times a month. Dr. Eitzer, an analytical chemist at the Connecticut Agricultural Experiment Station, analyzed the pollen for many of the common pesticides using a modified QuEChERS procedure. Wax comb samples were taken from colonies when they were first noticed to have died out. Colony strength and condition was sampled in all colonies monthly by assessing presence of queen, presence of eggs and young larvae, egg laying pattern, supersedure rates based upon marking of queens, and brood and adult population size was sampled by estimating percentage of total frame area occupied per colony.

Pest occurrence such as small hive beetle (Aethina tumida) and wax moth (Galleria mellonella) were recorded monthly and parasite loads were estimated by dissection of bees for tracheal mites (Acarapis woodi) and sampling approximately 300 bees per colony for Varroa mite (Varroa destructor) levels via sugar rolls in 2009 and 2010. Disease agent sampling was also conducted monthly. One hundred forager honey bees were collected per colony for dissection to quantify Nosema spore levels and determination of Nosema species (apis or ceranae).  Symptoms of European (Melissococcus plutonius) and American foulbrood (Paenibacillus larvae ssp. larvae) and Chalkbrood (Ascosphaera apis) were recorded in each colony per monthly sample. A honey bee sample of at least 100 bees per colony were collected and molecular markers were used for qualitative and quantitative PCR to obtain disease presence and/or incidence of Nosema ceranae and the following viruses: deformed wing virus (DWV), Kashmir bee virus (KBV), Israeli acute paralysis virus (IAPV), and black queen cell virus (BQCV).

Landscape and climate features relevant to honey bee foraging were assessed for each apiary site in all trials. This involved spatial analysis of landscape coverages in a two-mile radius from each apiary. The geographic area of land-use/habitat types was calculated using Google Earth^® or other suitable digital spatial coverages. A database was constructed for each apiary so that the percentage of surface waters, riparian habitat, urban, suburban/residential, forest, wetland, meadow, agricultural pasture, orchard, or vegetable/field crops could be estimated via spatial analysis software such as “freemaptools”.  Climate zones were recorded for each apiary and weather experienced at each site was represented by summaries of local weather conditions during the year (monthly mean, max, and min air temperatures and precipitation) measured at or near each apiary.

The initial design for this longitudinal study was to initiate a trial consisting of seven apiaries in 2009, follow the colonies through at least two years and then initiate a Second Trial in 2011 that would be followed for two years. However, in 2009, California had to drop out of the study, but was able to setup a trial in 2010. In order to provide a basis of comparison, a Second Trial was also set up in 2010 in Maine with a 15-colony apiary located in a different site than the 2009 Maine apiary. The reduced 2010 Trial was again initiated with packages from commercial bee production operations on brand new equipment and foundation, exactly as was done in 2009. All colonies at these two sites were re-queened with Italian race queens obtained from Koehnen  & Sons, Inc.

The FOCUS of this article is a review of the preliminary findings with colony losses, hive strength, and queen status that were associated with the abiotic factors characterizing each apiary site.

Summary of Our Results

Colony loss rate. Colony losses, measured as percent cumulative colony mortality, were different among apiaries at the end of each of the first two years for the 2009 Trial (ranging from 3 (MN) – 40 (PA)% in 2009 and 31 (MN) to 100 (PA)% in 2010), but not for the 2010 Trial where the apiaries in ME and CA at the end of 2010 had loss rates of 20 and 25%, respectively, and by the spring of 2011 both apiaries had loss rates of 100%.  

Our observations suggest that colony loss is quite variable and not necessarily similar in effect across the U.S. and across different types of beekeeping operations. Apiaries in the northern tier states of MN, ME, and WA were characterized by colony loss rates which were highest during overwintering, while a more continuous rate of colony losses occurred over the entire year for the southern tier states of PA, TX, and FL. In the 2010 Trial the apiary in ME (a new apiary setup in 2010 in a different location) is characterized by primarily overwintering mortality, while colony loss in CA is similar to the southern tier apiaries in exhibiting a more continuous loss during the year.

Another pattern in colony losses that was observed for the initial 2009 Trial was that the proportion of colony loss that occurred during the first growing season was highly related to the level of mortality over the subsequent winter. This is consistent with a causal effect that is characterized by “carryover”, such as parasites, pathogens, pesticide contamination of the hive, or cumulative stress. This carryover effect could not be detected in 2010/2011 since all apiaries setup in 2009, except FL, experienced 100% loss. 

Supersedure rates. High supersedure rates are often an indication of colony stress, poor egg laying by the queen, or poor colony health and ultimately a subsequent high likelihood of colony loss. Supersedure rates, averaged over all colonies in each apiary, ranged from 0.14 to 0.31 supersedures / colony during the first year of the 2009 Trial, 0.18 to 0.61 during the second year of the 2009 Trial, and 0.08 to 0.38 in the first year of the 2010 Trial).  Conducting an analysis at the colony level within apiaries we found that in the 2009 Trial there were strong effects of apiary location on both supersedure rate and colony loss rates during the study. There was no interaction between apiary site and supersedure rate affecting colony loss rates. This suggests that these effects of supersedure on colony loss showed a similar trend across all sites. The colony loss rate being statistically explained supersedure might be due to reduced colony populations in those colonies that have high supersedure rates. We found that this was the case in PA and ME, the higher the supersedure rate in a colony the lower the colony brood population size. However, this was not the case for the other states where there appeared to be no relationship between colony-level supersedure rate and colony brood population size.

A similar relationship appears to be the case for colony losses and supersedure rates in the 2010 Trial. Differences in supersedure rates between ME and CA existed, with the ME apiary having a low average supersedure rate of 0.08 and CA having a high rate of 0.38. Yet the colony loss rates in these apiaries were quite similar as mentioned above. An apiary site effect was apparent in colony population levels of brood and workers (CA being much lower than ME), but there was no relationship between the supersedure level for each colony and its colony brood population level as was the case for PA and ME in the 2009 Trial. 

Climate and Weather effects. The stationary hive project has apiaries assigned to 5 broad climate zones of the continental U.S. These are: 1) semi-arid steppe climate (WA), 2) Mediterranean climate (CA), 3) humid sub-tropical climate (FL, TX), 4) humid continental warm summer climate (PA), and 5) humid continental cool summer climate (MN, ME). Weather data was collected at or near all apiaries in 2009 – 2011. The reason for this was to assess abiotic conditions during the conduct of the study. Weather parameters that were assessed were maximum and minimum air temperatures, mean precipitation, a derived index of stress (maximum air temperature relative to the seasonal precipitation), and a composite new variable explaining most of the variation among all of the weather variables collected (first eigenvalue of a principal components analysis (PCA)).  These weather parameters were used to determine if weather / climate effects were correlated with colony loss and supersedure rates over the first foraging season, over the winter, over the second foraging season, and over the entire duration of the study for each trial.  Of the more than 40 analyses conducted (many combinations of weather factors crossed with different lengths of monitoring times), there was only one marginally significant relationship that was observed between weather factors and apiary site-level colony loss rates and one relationship between weather factors and supersedure rates. Maximum mean air temperatures in the second year (2010) of the 2009 Trial, were positively correlated with colony loss rates. When both the 2009 and 2010 Trials were combined into one analysis, supersedure rates were correlated with maximum temperature during foraging. These relationships are only weak, but they suggest that hotter climates might, in a given year, might result in higher colony losses or poorer colony population levels via supersedure, than cooler climates.  Hot summer daily temperatures may be an index of heat stress. Such heat stress may not only be directly operating on colony health, but may have an indirect effect of colony losses and colony health due to effects on the quality of bee pasture. It will be interesting if these same relationships are borne out in the 2011 Trial (we will have a much greater sample size of apiaries to test this than what currently exists (apiaries = 8) in the 2009 and 2010 Trials).

Landscape effects. The characteristics of the apiary landscape for an area of a 2-mile radius from the apiary (Fig. 2) did show an interesting relationship with colony loss rates. The landscape was characterized by land use as seen in Table 1, an example of the Florida apiary. The classification was aggregated to six categories in order to have relevant common categories across all apiary sites. These categories were: 1) forest, 2) old field /scrub shrub, 3) pasture, 4) wetlands, 5) urban / suburban, and 6) intensive agriculture. We found a strong positive correlation between the proportion of landscape in intensive agricultural production and the proportion of colony losses for the 2009 Trial through the spring of 2010. When both Trials (2009 (n=5) and 2010 (n=2)) are examined over a twelve-month period in relation to percent of foraging area in agricultural production, we found a significant year effect and a significant proportion agricultural land effect, but no year by agricultural land interaction. Therefore, this relationship appears consistent for both years for different apiaries with the result of proportion of land area in intensive agriculture explaining a meaningful amount of the variation in colony losses. It is unknown why the proportion of landscape in agriculture is related to colony losses. There are several factors that might be associated with intensive agricultural production, pesticides being the most obvious, but others could be quantity and quality of bee pasture, water availability, and intra-colony competition.

Fig. 2. Landscape within a 2-mile foraging radius of the Texas apiary, 2009-2011.

Table 1. Land use classification patterns in a two-mile radius of the Florida apiary.

Land Cover Description	Acres	Percent
Coniferous Pine	14.6	0.47%
Emergent Aquatic Veg	8.9	0.29%
Field Crops	543.3	17.62%
Forest Regeneration	83.9	2.72%
Freshwater Marshes	201	6.52%
Horse Farm	99	3.21%
Improved Pastures	1166.6	37.83%
Institutional	6.1	0.20%
Lakes	9.4	0.30%
Mixed Crops	216.5	7.02%
Mixed Scrub-shrub Wetland	25.3	0.82%
Mixed Upland Nonforested	1.4	0.05%
Mixed Wetland Hardwoods	15.6	0.51%
Reservoirs	0.4	0.01%
Residentual. Low desity	18.4	0.60%
Residentual. Med. desity	51.5	1.67%
Streams and Waterways	3.3	0.11%
Unimproved Pastures	2.7	0.09%
Upland hardwood Forest	9.3	0.30%
Upland Mixed Coniferous/Hardwood	457	14.82%
Wet Prairies	12.8	0.42%
Wetland Forested Mixed	15.3	0.50%
Woodland Pastures	121.9	3.95%

Pesticide exposure. Pollen trapping was conducted 2009-2011. The chemical analyses were conducted by Dr. Brian Eitzer at the Connecticut Agricultural Experiment Station. Five colonies per apiary were fitted with pollen traps and trapped twice monthly for a 3-day period. Overall, the results show variation in both time and apiary location.  This is not unexpected as different pesticides are used at different times of the year at a single location, and, different pesticides are used for different crops grown at different locations.   Over the First and Second Trials (2009-2011) 44 different pesticide or pesticide metabolites have been observed (18 insecticides, 14 fungicides, 10 herbicides, 2 metabolites).  Over the two-year period of the First Trial, the numbers of different pesticide residues in pollen was 10 and that was found in a single composite sample (composited over apiary and season). Residue concentrations also had great variation.  Most residues observed are less than 10 PPB, a fair number are in the 10-100 PPB range, and a few are greater than 100 PPB.  It should be noted that the compositing process does affect the data; high concentration individual samples are averaged with lower concentration samples, and a low concentration individual sample could drop below the detection limits.  However, compositing is necessary given the number of samples and cost of analysis.  Wax samples from the apiaries have also been collected from recent dead-out colonies throughout the season and all surviving colonies at the end of the fall of each year. These wax samples are currently undergoing analysis.

We have investigated the relationship between residues in trapped pollen and colony loss rate and supersedure events. In order to do this we used several measures of pollen contamination: 1) total number of pesticides over the season in each apiary, 2) concentration (ppb) of all pesticides over the season, 3) concentration (ppb) of all insecticides, 4) concentration (ppb) of all miticides, and 5) concentration (ppb) of all neonicotinoid insecticides. We did not find any relationship with any of our measures of pesticide contamination and colony loss rate at the apiary level for either 2009 or 2010 in the 2009 Trial (the sample size for the 2010 Trial is only 2 apiaries and the CA and ME apiaries had almost identical mean pesticide residues. We did see a marginal relationship with the number of total pesticides (diversity) in trapped pollen at an apiary and the integral of supersedure rates (area under the plotted curve of supersedure events) for 2009. We did not see a similar trend for the same apiaries in 2010, although colony loss had deleted PA from the analysis and TX had very few colonies surviving in 2010. It is surprising that we do see a trend in 2009, since most of the residues are at extremely low levels. Another question that can be asked with the trapped contaminated pollen is whether the proportion of foraging area in agriculture is related to the contamination of pollen trapped at a give apiary. We conducted this analysis for 2009 and 2010 using the pollen collected over two years in the 2009 Trial. We found no relationship for 2009 and 2010 between proportion of foraging area in agriculture and total number of pesticides, concentration of all pesticides, concentration of all miticides, concentration of all insecticides, and concentration of neonicotinoid insecticides. There were also no relationships between any of the measures of pollen-pesticide contamination and the proportion of land in urban / suburban land use. The 2011 Trial will provide a larger sample size when included with the 2009 Trial to further investigate these trends since these analyses have very low power with only 5-6 apiaries per analysis.

Summary and conclusions.

            We have reported on just the abiotic apiary site effects for the First Trial of the CAPS Stationary Hive Project. Perhaps we have stimulated more new questions than we answered. What we have found is that geography of the apiary affects colony losses. Now many readers may suggest that this is an obvious finding, but what is not obvious…is why? Yes, honey bee colony stress levels may increase as one moves from the southern U.S. to the northern U.S., but we argue that not all stressors are expected to be higher in the South. For instance, overwintering stress would be hypothesized to be much greater in the North.  

We have also conducted analyses on the effects of Varroa and tracheal mites, Nosema ceranae, and viruses on colony losses in this study. We are not reporting on this part of the study, but we not only found strong support in some preliminary analyses for the role of Varroa, Nosema, and IAPV in colony losses, BUT some of these biotic causes of colony loss were also moderated by apiary-site. This suggests that the factors discussed in this article may be key in determining the relative risk of colonies dying out due to pathogens and parasites.

            So, in summary, what are these apiary-site factors that might influence colony losses? First we found that our apiaries experiencing the hotter daily maximum temperatures tended to have higher colony losses and supersedure rates over the two-year period, although this was a weak trend at best. However, as far back as 1869, high levels of colony losses have been attributed to extreme high temperatures during the summer (anon. 1869). Also, associated with an apiary site is the habitat surrounding the apiary. Our First Trial, along with the Second Trial setup in 2010, suggest that intensive agricultural landscapes may affect colony losses. As the percent of intensive agricultural land area increased within a 2-mile radius of an apiary, colony losses within an apiary also increased. On the one hand, agricultural landscapes can be very beneficial to foraging honey bees, but more and more evidence suggests that this is not the case for native bees (Kremen et al. 2004) and we suggest that this also might not be the case for honey bees. Now, an obvious hypothesis regarding the deleterious effects of increasing land area of agricultural landscapes might be increased pesticide exposure. Although, it is the case that as the proportion of agricultural land increases in an area, the proportion of non-agricultural, less disturbed habitats decreases and so what is the mechanism? We did, however, find that pesticide contamination of pollen brought back to the hive in our apiaries did correlate weakly with increased mean supersedure rates. Now the measure that we found to correlate with supersedure rate was the number of total pesticides (and metabolites) found in the contaminated pollen. There is a high degree of attention currently on interactive lethal and sub-lethal effects of pesticide exposure to honey bees, much of it supported by the CAPS project. Our findings with supersedure provide evidence that subtle effects such as queen failure found in more controlled field and laboratory studies (Ellis 2010) can be detected in large-scale typical bee yard-level studies.

            In CONCLUSION, what can we say about abiotic conditions associated with an apiary-site that might affect colony loss. We CAN say that we have strong evidence to suggest that apiary-site effects do exist. However, with our study many of the site-effects appear to be correlated with one another. Thus we have identified several potential abiotic effects that might increase colony losses, but teasing out which factors are the key factors may have to be the focus of a study specifically designed to test independent hypotheses while controlling for other factors, if this is possible at the scale of foraging honey bees.  We think that the power of the CAPS Stationary Hive Project will be to identify factors that consistently result in colony losses over two large-scale trials (2009 and 2011) and then to identify which factors might interact with one another to produce synergistic colony losses. These will be the dangerous combinations of potential causative factors that need to be studied more intensively in the future.

Acknowledgements

We would like to acknowledge the assistance of Dr. Brian Eitzer at the Connecticut Agriculture Experiment Station for pesticide analysis of pollen samples in this study. In addition much thanks in the apiary and laboratory go to Jennifer Lund (Maine), Beth Kahkonen (Washington), Mark Dykes (Florida), Art Cavazos (Texas), Kerry Lynott (Pennsylvania), Mike Goblirsch (Minnesota). Special thanks go out to Dr. Keith Delaplane (Georgia), who helped keep this project on track.

References Cited

• Anonymous, 1869. Report of the Commissioner of Agriculture for the year 1868. U. S. Government Printing Office, Washington, D. C. Pp. 272-281.
• Ellis, M.D. 2010. Pesticides applied to crops and honey bee toxicity. Amer. Bee J. 150(5): 485-486.
• Kremen, C., N.M. Williams, R.L. Bugg, J.P. Fay, and R.W. Thorp. 2004. The area requirements of an ecosystem service: crop pollination by native bee communities in California. Ecology Letters 7: 1109-1119.
• Spivak, M. 2010a. Honey bee “medical records”: The stationary apiary monitoring project. Amer. Bee J. 149(3): 271-274.
• Spivak, M. 2010b. Honey bee “medial records”: The stationary apiary project. Managed Pollinator CAP Update. Bee Culture 150(3): 270-274.