REAL-TIME MONITORING OF MPCA HAZARDS TO HONEY BEES IN A MICROBIAL CONTAINMENT FLIGHT CHAMBER

J.J. Bromenshenk^a, B. Lighthart^b, R.L. McGraw II^a, M. R. Loeser^a, B. Birch^a, and P.J. Bowman^a

SUMMARY

Honey bees (Apis mellifera L., Hymenoptera: Apidae) are important producers of honey, wax, and other products. Bees also serve as supplemental pollinators of most entomophilous crops. Therefore, the United States Environmental Protection Agency requires testing to assess the hazards of chemical and biological pesticides to this non-target species. Unlike chemical pesticides, microbial pest control agents (MPCAs) are self-propagating and dispersive. Our challenge is to assess the hazards of MPCAs to bee colony health and productivity while safeguarding the external environment. Our objectives are to provide adequate containment during testing, devices to continuously evaluate several pathogenic endpoints, and methods that provide estimates of data uncertainty and assist data interpretation and analysis.

We test nucleus colonies in miniature wind tunnels (WT) inside a negative-pressure environmental containment module (ECM). An IBM-compatible computer system, called the DAC, provides real-time data acquisition and controls the ECM. The DAC continuously monitors the environmental systems and measures colony responses such as changes in activity, biomass, temperature, relative humidity, air flow, and sound emissions. Data output from the DAC links to an ecotoxicological simulation model for honey bee populations and to an artificial neural network. Eventually, these computer tools should yield a better understanding of the effects of MPCAs on colony population dynamics. They also provide predictive control and a means of calibrating the responses of colonies maintained in indoor chambers to those kept in the natural environment.

Key words: Insecta, Apis mellifera, microbial pathogens, isolation chambers

INTRODUCTION

Chemical insecticides can produce immediate, acute kills of bees at the place of exposure, often from pesticide drift that affect foragers in the field (Johansen and Mayer, 1990). By contrast, microbial insecticides usually do not induce immediate death. Several weeks may pass before adult bees exhibit disease symptoms produced by pathogens such as fungi. Thus, foragers have many opportunities to bring MPCAs back to the hive, where the pathogens may be dispersed among hive bees, brood, food stores and wax.

The EPA's Interim Protocol (Burgett, 1989) for testing the effects of microbial pesticides on honey bees recognizes these complicating factors. It emphasizes a need to consider the MPCA mode of action, type of exposure, and host stadium (i.e., immature or adult life stages). The Protocol recommends three testing Tiers. Tier 1 tests use isolated subsets of larvae or worker bees (i.e., caged bees). These assays employ classical bee pathological and toxicological tests (summarized in Johansen and Mayer, 1990). Tier 2 assays address colony effects such as alterations of queen behavior, brood rearing, or colony homeostasis. Tier 3 trials consist of exposing foragers from field colonies to microbial pathogens applied to blooming crops at recommended application rates.

Burgett (1989) notes that Tier 1 tests rely on small groups of larvae or worker bees. Therefore, the results may not produce standardized LD₅₀, LD₉₀, or LT₅₀ values nor adequately reflect colony effects. For Tier 2 tests, he recommends the use of nucleus colonies maintained for short periods (< 20 d) in a closed (non-flight incubator) environment or for longer periods (42-63 d) in an open-flight status. He does not address issues such as: i. how best to measure colony responses, ii. how to relate laboratory test results to those obtained from field tests, and iii. how to use test results for assessing hazards as part of a risk assessment.

Because MPCAs are readily dispersed and self-propagating, any testing must guard against environmental release of these materials, especially exotic or genetically engineered pathogens. Initial testing of new organisms should be done indoors under stringent biological containment to prevent accidental release from the test facility and to avoid cross-contamination between test treatments.

To address these issues, we initiated research that employs real-time monitoring of bee colony responses to microbial pesticides. Specific tasks include: i. identifying and assessing colony measurement endpoints that delineate MPCA hazards, including measures of the physiology of the colony social group; ii. testing and developing Positive Microbial Control Test Agents for two classes of bee pathogens (e.g., bacterial, fungal); iii. evaluating the effect of secondary factors such as poor diet, temperature stress, or parasitism on colony responses to select MPCAs; iv. comparing the effects of registered MPCAs on nucleus-colonies to positive control pathogens and to published evaluations of effects based on tests using caged bees, and v. refining and validating PC BEEPOP, our honey bee ecotoxicological model and database. The model provides a tool for designing laboratory and field tests to assess ecological risks of MPCAs. It also evaluates whether test results are due to the pathogen, biotic or abiotic stressors, or natural variability.

We employ methods that focus on the colony, not the individual bees, as the appropriate response unit. Ultimately, honey bee susceptibility to an MPCA must be defined in terms of the potential for altering colony health and productivity. Test exposures to the pesticide should reflect those likely to occur at recommended field application rates. Effects to any life stage must be extended to the reproduction, growth, and production of the colony as a whole.

MATERIALS AND METHODS

This approach required considerable research and development to produce the specialized equipment, facilities, and test methods needed to accomplish our objectives. Because this is a new endeavor, this article summarizes our current procedures. It does not present extensive test results, which are just becoming available.

Environmental Containment Module (ECM). Under a cooperative research program with Dr. Lighthart at the Corvallis Environmental Research Laboratory, we developed an Environmental Containment Module (ECM). The ECM has a computer data acquisition and control system (DAC). It allows indoor testing of nucleus colonies of bees and MPCAs.

We fully describe this facility in the ECM and DAC User Manuals. B. Lighthart, the EPA project officer (Environmental Research Laboratory, U.S. EPA, Corvallis, OR) can provide copies of these reports. The ECM is a portable, negative pressure, HEPA-filtered, walk-in isolation chamber. Variable speed fans control air inflow and outflow. Filtration efficacy is 99.97% D.O.P. at 3 microns. Water filter filtration removes particles as small as 0.2 micron. The chamber itself is a 2.2m x 2.8m x 2.2 clear plastic bubble suspended from a stainless steel frame.

Bees are isolated by the ECM from the external environment (i.e., the room), while protecting the outside environment from bees and pathogens inside the ECM bubble. The ECM's isolation chamber sits inside a portable PVC framework supporting banks of fluorescent lights. The lights are fitted with near-solar, wide-spectrum Vita-Light® bulbs and energy-efficient electronic ballasts. These ballasts have a high flicker-frequency (>20k/sec). Space blanket reflectors diffuse the lights. The high flicker frequency and diffuse light reduce attraction of bees toward the lights.

Data Acquisition System (DAC). We built and programmed an analog/digital data collection/controller system to monitor and manage environmental conditions (temperature, light, humidity) in the chamber. DAC functions include providing real-time monitoring of colony dynamics (e.g., changes in hive weight, food consumption, microclimate). The DAC connects to an IBM-compatible computer. The system is constructed using a system of modular data acquisition and control devices (e.g., ABus cards, Alpha Products, Fairfield, CN 06430).

The DAC continuously monitors 32 input sensors and controls 16 outputs (e.g., lights, a humidifier, fans, and heaters). It can produce fixed or random sequences of environmental conditions (e.g., increasing daylight, alternating cool nights and warm days, systematic or random variations in temperature and humidity). The DAC provides on screen, disk-file, and printer charts for all parameters. It has alarms for system malfunctions. Data output is sent directly to electronic spreadsheets or linked to either an artificial neural network or our honey bee population model, PC BEEPOP.

Artificial Neural Networks and PC BEEPOP. We use a Windows-based artificial neural network program (e.g., BrainMaker Pro 2.03, California Scientific Software, Sierra Madre, CA) to process our real-time colony response data. Artificial neural networks (ANN) are self-learning, pattern-recognizing software programs that can integrate and process complex data sets.

In addition, we modified our simulation model to link the model to the data output files from both the ANN and the DAC. We also added a link that enters ECM temperature and photoperiod data through the input menu of the model's weather file. Another interface menu allows a user to change the coefficients of the basic equations in the model without re-writing the underlying source code.

Mini-Hives and Wind Tunnels. Six to nine mini-hives can fit onto steel racks inside the ECM. Each hive contains a mini-colony with a queen, brood, about 4,000 adult bees, and five 1/2 sized combs. A 3m long wind tunnel (WT) connects each hive to the feeders (filled with syrup and water). Pollen substitute is injected into the hive with a large syringe.

Squirrel-cage fans force HEPA-filtered air through a PVC manifold to each WT. The filtered air flows from the direction of the feeders toward the hive. The HEPA-filtered air delivery system to the WTs reduces the likelihood of cross-contamination between treatments.

Mini-hives and feeders are constructed of clear acrylic and fitted with seals to make them air-tight. WTs attach to hives and feeders via flexible PVC tubes and valves. The valves allow closure of the tubes for attachment and removal from each WT. A roll of clear acetate covers the floor of each WT. Pulling the acetate through a slot in the floor and into a sealed bag collects dead bees, bee feces, and any debris.

A HEPA-filtered blower provides directional air streams for driving bees out of the feeders and hives. After driving the bees into the WT, a valve is closed to prevent bee re-entry. The blower is then reversed to draw a negative vacuum. This allows removal of the feeder or hive for cleaning or replacement.

Real-time Monitoring Equipment and Periodic Sampling. Electronic sensors continuously monitor conditions inside each hive (e.g., temperature, relative humidity, and air flow). Sensors also monitor changes in the weight of the entire unit. We make periodic observations of colony sound emissions by inserting a small lapel microphone into each hive through a port in the hive cover. A portable computer equipped with a sound card makes direct recordings of colony noise. This computer is then used to analyze the digital sound files.

Respiratory quotient (RQ) or the respiratory exchange ratio (R) also is found for each colony. Atmospheric air is drawn through hives using a vacuum source (negative pressure open-circuit biocalorimetric method). A digital flowmeter calibrated using an NBS standard measures dry excurrent air flow. We adjust the air flow to lower any background levels of CO₂ that would interfere with determination of RQ. Also, we reduce and equalize the colony's honey and pollen stores to minimize CO₂from these sources. The fractional concentrations of oxygen and carbon dioxide in the incurrent and excurrent air streams are then measured using Applied Electrochemistry gas analyzers (S-3A and CD-3A). Using appropriate equations (Withers, 1977), we calculate oxygen consumption and carbon dioxide production of the colony. We obtain values for RQ or R from the ratio of carbon dioxide produced to oxygen consumed.

Water, sucrose solution, and pollen or pollen substitute are provided ad libitum. We weigh these provisions before and after feeding the bees. We estimate consumption rates from the amount remaining when replenishing the feeders. Food and water are placed in a feeder at one end of the WT. The hive is at the other end of the WT.

We inspect the bees through the clear acrylic hive bodies. Between observations, a plastic-bubble insulation blanket covers each hive. The foil-coated insulation material reduces heat transfer through the acrylic hive walls and shields the colony from external light. By removing the cover, we can see the general condition of the bees and food stores.

A telescoping bottom provides access to a set of false floors and screens. The removable bottom allows removal of dead bees, hive debris, and any pathogens for testing and quantification. Removing a telescoping cover permits withdrawal of the frames to examine each comb and to measure amounts of brood and food stores. This also simplifies examining combs for evidence of MPCA induced pathologies or toxicology, such as dead larvae or pupae.

We monitor mortality of adult bees and immature stages at regular intervals. We visually examine dead bees and test for the presence and viability of pathogenic microbes. Our procedure consists of surface sterilization of bee bodies followed by plating of whole bees and of intestinal contents. Host tissues are examined for: i. evidence of the presence of the MPCA and its viability, ii. evidence of a pathological response by the host, and iii. indication of infection and multiplication of the MPCA.

We periodically assess brood mortality by counting the number of cells exhibiting diseased larvae. We also remove the caps from a subset of pupal cells to be sure that the bees have not capped dead larvae.

Application of the MPCA. We administer the microbial agent via spiked food or by contact application. To mimic exposures to pesticide drift or direct spray application, we apply the MPCA in a wetting agent using an artist's air brush. Our methods follow those of S. Jaronski (Mycotech Inc., Butte, MT). Negative controls include the administration of inactivated MPCAs-- heat treated, ethylene oxide (ETO) fumigated, or gamma-irradiated. Application of the carrier used to deliver the MPCAs provides an additional negative control.

We use indigenous bee pathogens as positive controls. For a bacterial control, we are investigating the use of American Foulbrood (AFB), Bacillus larvae, as the best characterized bacterial pathogen of honey bees. Chalk Brood (CB), Ascosphaera apis, serves as a common fungal pathogen. To produce a reliable positive control pathogen, we are infecting the bees with these microbes and testing their effects on colonies in the WTs.

These pathogens are endemic to U.S. bee colonies. Therefore, we sample bees (whole bees and intestinal contents), brood, bee bread (stored pollen or pollen substitute), and honey for these materials--before, during, and at the conclusion of testing. We also test for the presence of the MPCA being tested. We use a minimum of five bees, bee bread from five cells, honey from five cells, and two pieces of pollen patty.

Ranking Colonies by Hygienic Behavior. Colonies used to assess the potential hazards of MPCAs must typify colonies exposed to applications of these pest control agents. Hazard tests based on highly vulnerable colonies may produce results that are too conservative. Using resistant colonies may fail to identify a hazard. Rothenbuhler (1964) and Gilliam et al. (1988) showed that the hygienic behavior of worker bees modifies susceptibility to microbial diseases. According to Rothenbuhler (1964), a recessive gene for uncapping and another for removal of larvae controls hygienic behavior. Gilliam et al. (1988) and Tabor (1989) recommended using frozen brood to rank breeder queens for disease resistance. They cut chunks of capped brood from combs and froze them overnight. They then replaced the frozen brood comb in and monitored the time taken by the bees to remove the freeze-killed brood.

We follow much the same procedure, but kill the brood in situ with liquid nitrogen. By using this procedure, we can rank queens and colonies based on the probable level of resistance. Resistance categories range from highly susceptible, to moderate, to resistant.

Selection of Colonies for Testing. Because it is nearly impossible to start bee colonies indoors, we establish and maintain nucleus colonies in an outdoor apiary. We started these colonies on disease-free equipment and inspected them frequently for symptoms of disease or parasitism by mites. Detailed "health" and management records are kept for each colony. For MPCA testing, we transfer randomly selected colonies to the acrylic hives.

Before selection, we again visually inspect colonies and test them for any microbial diseases or mites. Generally, we equalize and standardize the size of the units before testing. Often, we reduce the size of the colony, because by mid-summer these nucleus colonies may contain two or more pounds of bees. We eliminate older forager bees from the population. Older bees generally do not forage well in indoor chambers. We select for younger bees by using only adult bees that emerge from brood frames kept in incubators or by moving hives to another stand during the day so that the returning foragers cannot find the hive. In either case, the colony is left with younger bees.

Diet. The feeders supply ad libitum sterile, distilled-deionized water, protein (pollen or pollen substitute) (Vandenberg, 1987), and sucrose syrup scented with Anise oil.

Secondary Stress Factors. Factors such as climate and nutrition may modify expression of effects. The DAC controls a humidifier and an air conditioner for the ECM. It can be programmed to produce extended cool, moist conditions or dry periods to test the interaction of these climatic conditions on infection by pathogens. Similarly, a substandard diet (such as one short in protein) can be used to test the effects of nutritional stress on colony responses to MPCAs.

Duration/Rates Tested. Short term range finding and screening tests take less than 20 days. More rigorous testing including positive controls, MPCAs of interest, and secondary factors encompass several brood cycles (42 - 62 days). Recommended field dosage rates for some MPCAs may be outside the range of observable pathogenic effects (Vandenberg and Shimanuki, 1986; Vandenberg et al., 1986). Therefore, two methods (consistent with Burgett, 1989) seem appropriate for most testing: i. range finding dosages using serial dilutions of the MPCA with ca. 4-6 orders of magnitude, and ii. application rates that simulate expected field dosages.

Replication. For routine testing, we employ a minimum of three replications per MPCA dosage and appropriate numbers of controls. For most tests, we minimally employ a negative control, a positive control, and one to five treatment levels. Different colonies provide replication.

Sterilization. We use an insecticide fogger and an appropriate sterilant such as Clydox (a two-part solution of chlorine dioxide) to sterilize the chamber. Bleach solutions, alcohol, and ultra-violet light sterilize hive bodies, WTs, and other chamber components. We monitor the effectiveness of sterilization by swabbing surfaces followed by plating on appropriate substrates, incubating, and then examining the plates for the presence of pathogens of concern.

Statistical Design and Methods. For most tests we use a randomized complete block (Sokal and Rohlf, 1981) or Latin square design (Steele and Torrie, 1960). We analyze the chamber data using: i. a two-way anova with replication (e.g., comparison of a negative control and a single treatment level of a positive control, or of controls and a MPCA at a field application rate, or ii. a two-way anova without replication. In the latter case, we use repeated testing over a time.

Burgett (1989) recommends that cumulative percent mortality data be subjected to angular transformation (Sokal and Rohlf, 1982). The transformed data is then assessed by least squares regression of the data on time (days) for each dose/method/product combination (SAS, 1982). Analysis of variance and confidence limits for the effects of each MPCA, dose, and method are used to compare LD₅₀ , LD₉₀, LT₅₀, and regression slopes.

Quality Assurance. We intend to produce cost-effective, reliable, and appropriate methods for testing the effects of MPCAs on colony performance and structure. These methods must meet rigorous criteria for containment of pathogens and quality assurance. We follow a well-documented quality control/quality assurance plan for all tests. The EPA audits the work annually.

RESULTS

A portable ECM avoids the need to set up air ventilation/filtration systems for entire rooms, which are costly and require considerable space dedicated to a single purpose. It can be set up in a warehouse or almost anywhere with adequate space and a reliable heating/cooling system. When not in use, the ECM is dismantled and stored in a closet.

The DAC keeps the ECM environment under stable conditions. It can be programmed to for daily changes in daylength and to simulate dawn and twilight. Relative humidity remains at ± 2.5 for 35%. Day and night temperatures stay within ± 3C of the target value (Figure 1). The DAC also can systematically or randomly vary environmental conditions. Highest temperatures occurred at the top of the ECM nearest the lights. Temperatures at individual positions in the ECM typically varied less than ± 1.5C.

Our changes to PC BEEPOP ease input of chamber data. In addition, the modifications provide the user with a simple means of changing the model's structure to adjust its estimates to match the colony responses observed in the chamber. Initial model predictions show that the population dynamics of the control colonies should stabilize and then slowly diminish due to age-induced changes in the queen's egg-laying (Figure 2).

During our preliminary trials, bees foraged at food dispensers, gathered, and hoarded food inside the WTs. Bee locomotion to and from the feeders was stimulated and made directional by air movement. Absence of air flow and high air velocities suppressed movement. Colonies survived in the ECM with low levels of adult bee mortality. Bees lived for periods of eight weeks to more than four months. Queen activity appeared consistent for the time of year; although egg-laying rates were low.

Initial real-time monitoring data suggests that brood nest temperatures may change depending on queen condition and possibly because of infection by pathogens. We are still evaluating the data for other response endpoints.

Preliminary trials with B. bassiana, a registered fungal MPCA, demonstrated that the pathogen could be recovered from a few of the dead bees. These bees were taken from colonies dosed by spray application (i.e., air-brush application). Bees sampled at the end of the trials were generally free of viable B. bassiana spores. Some samples contained a single bee with viable spores (out of 30).

Colonies exposed to A. apis, as a positive control, displayed variable response. Although some colonies cast out chalk brood mummies, we were unable to recover viable A. apis spores from bee cadavers.

Ranking of colonies based on hygienic behavior proved to be inconsistent using the methods of Gilliam et al. (1988) and Tabor (1989). The disturbance caused by excising and replacing pieces of brood comb often induced housekeeping activities. These activities resulted in all of the frozen brood being removed and discarded from all of the hives within a few hours. Therefore, we developed a less invasive method of killing brood. We press a thin-walled metal tube into a comb containing capped brood. Pouring 80 ml of liquid nitrogen into the tube freezes the pupae in less than two minutes. There were no visible signs of disturbance other than a slight ring-shaped depression in the comb. This procedure is easy, fast, and reproducible. Differences among colonies are evident in the time taken to open cells and remove dead pupae and in the numbers of pupae eliminated.

DISCUSSION

Generally, the ECM, DAC, WTs, and other components of the Microbial Containment Facility operated as expected. We spent considerable time and effort developing, testing, and completing the DAC and the sensors. The RH sensors proved to be the most troublesome, requiring specialized signal conditioning circuits. Commercial interface software proved to be limited in scope, so much so that we are currently writing our own interface software.

Conditioning or training of bees to scented water and food sources is required to get bees to forage in the WTs. Air movement provides odor gradients that appear to help the bees orient to the food supplies and reduces moisture condensation in the acrylic hives.

Inducing queen egg-laying during the fall and winter continues to be a problem. We are currently trying various methods of increasing queen egg-laying in the ECM. Kefuss (1978) and Kefuss and Nye (1970) recommend a stepwise lengthening of photoperiod. Gilliam et al. (1988) provided supplemental protein (pollen or pollen substitute) to stimulate egg-laying and brood rearing.

Our preliminary data suggests that brood nest temperature may be a useful assessment endpoint. Changes in temperature and relative humidity may occur if a disease alters colony homeostasis as well as the ability of the colony to regulate temperature, queen egg-laying, and colony brood production change. The presence of brood usually is accompanied by maintenance of a high and stable core temperature (Southwick, 1987; Southwick and Mugaas, 1971).

Biomass (weight) of the hive provides an indication of population size and/or hoarding and if the measures are sufficiently sensitive can show flight activity (Buchmann and Thoenes, 1990). Our pressure transducers monitor daily changes in total colony biomass but are not sensitive enough to distinguish low levels of foraging activity. Recently, we developed an infra-red, bi-directional counting system to profile bee movements.

Our sampling of acoustic emissions is in a preliminary phase. Dietlein (1985) pioneered continuous monitoring of sound emissions in colonies. Sound output by healthy colonies in terms of amplitude and duration appears to follow a diurnal and seasonal pattern. The diurnal rhythm may be modified by factors such as nectar flow, while the long-term patterns reflect general colony activity, possibly including brood rearing (Dietlein, 1985). Whether a diseased colony produces different sounds from healthy colonies is unknown, but most beekeepers can recognize a colony that has recently lost the queen by the sound it produces.

Recent advances in computer equipment and software provide opportunities for direct digital recording using a computer and far more comprehensive analysis of sound emissions than reported by Dietlein. Current computer and software technologies provide the ability to compare the energy (power) of the spectrum and the distribution of the energy as a function of frequency.

Respiratory Exchange Ratio (RQ) is a useful general indicator of the nature of metabolism of individual organisms or, as in our case, a group of organisms. Changes in RQ suggest changes in metabolic state as might be expected if hives of bees were experiencing some nutritional problem stemming, for example, from disease or changes in metabolic substrates. Baseline data (Southwick and Mugaas, 1971) show that the RQ of bee hives is within the normal range of such values reported from individual organisms. RQ has proven useful for monitoring the impacts of tracheal mites (pers. communication, 1996, D.J. Eyre, Orillia, Ontario, CN).

The bee literature, both peer reviewed and popular, often cite external factors such as weather or the nutritional quality of food sources as contributing factors to susceptibility to various diseases. For example, A. apis reportedly is a problem to colonies in the spring during humid periods. However, in Tucson this disease often appears during hot, dry periods in the summer (pers. com., 1994, M. Gilliam, USDA, ARS, Tucson, AZ). Similarly, Shimanuki (USDA, ARS, Beltsville, MD, pers. com., 1994) believes that parasitic mites may act as vectors for microbial diseases. Because of these reports, our 1995 trials of B. bassiana included treatments using colonies that had a history of susceptibility to A. apis, as evidenced by our disease management records, or that had parasitic mites, Varroa jacobsoni. We are still assessing the effects of these secondary stressors.

Our studies lay the groundwork for protocols to estimate the hazards to honey bees posed by the application of MPCA formulations during normal agricultural or horticultural use. Because field tests are expensive and because many microbial insecticides cannot be released into the environment until thoroughly tested, tests using bee colonies in indoor chambers must be relied upon to provide much of the needed information. We prefer using a whole colony approach rather than caged bees because MPCAs present several hazards to bees, including: i. alterations of larval development, survival, and morphology (e.g., teratologies); ii. behavioral changes (individuals and colony); iii. sublethal effects; and iv. chronic and acute toxicity to adult or immature stages. Any of these perturbations can affect colony survival, development, and productivity.

We have focused on measurement endpoints (e.g., colony homeostasis and classical pathogenic/toxicological tests) that provide values that can be readily quantified and related to risk assessment endpoints-those characteristics that if affected could adversely affect bees and ecosystems. With respect to honey bees, the ecological value to be protected is pollination.

DISCLAIMER

Development of the indoor chamber and the associated data acquisition and control system, as well as the honey bee population dynamics model, was funded in part by the United States Environmental Protection Agency. The viewpoints described in this report are those of the authors and do not necessarily reflect those of the EPA; no official endorsement should be inferred. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.

REFERENCES

Buchmann, S.L., and Thoenes, S.C. (1990) The electronic scale honey bee colony as a management and research tool. Bee Science 1:40-47.

Burgett, M. (1989) Interim protocol for testing the effects of microbial pathogens on the honey bee, Apis mellifera L. (Hymenoptera: Apidae). EPA 600/3-89/053, pp. 1-19.

Dietlein, D.G. (1985) A method for remote monitoring of activity of honeybee colonies by sound analysis. Jour. of Apic. of Research. 24:176-183.

Gilliam, M., Taber III, S., Lorenze, B.J., and Prest, D.B. (1988) Factors affecting development of chalkbrood disease in colonies of honey bees, Apis mellifera, fed pollen contaminated with Ascosphaera apis. Jour. of Invert. Path. 52:314-325.

Johansen, C.A. and Mayer, D.F. (1990) Pollinator Protection: A Bee and Pesticide Handbook. Cheshire, Connecticut, USA. Wicwas Press.

Kefus, J.A. (1978) Influence of photoperiod on the behaviour and brood-rearing activities of honeybees in a flight room. Jour. of Apic. Res. 17:137-151.

Kefuss, J.A., and Nye, W.P. (1970) The influence of photoperiod on the flight activity of honeybees. Jour. of Apic. Res. 9:133-139.

Rothenbuhler, W.C. (1964) Behaviour genetics of nest cleaning in honey bees. I. Responses of four inbred lines to disease-killed brood. Anim. Behav. 12:578-583.

SAS (1982) Regression procedure and general linear models procedure. In: Statistical analysis system user's guide: statistics. SAS Institute, Cary, NC.

Sokal, R.R. and Rohlf, F.J. (1981) Biometry: The Principles and Practice of Statistics in Biological Research. 2nd Ed, New York. W.H. Freeman and Company.

Southwick, E.E. (1987) Cooperative metabolism in honey bees: an alternative to antifreeze and hibernation. Journal of Thermal Biology. 12:155-158.

Southwick, E.E. and Mugaas, J.N. (1971) A hypothetical homeotherm: The Honeybee hive. Comp. Biochem. Physiol. 40A:935-944.

Steel, R.G.D., and Torrie, J.H. (1960) Principles and Procedures in Statistics. McGraw Hill, New York.

Taber, S. (1989) Backyard breeding. Gleanings in Bee Culture. 117:330-331.

Vandenberg, J.D. (1987) Safety testing of microbial control agents versus adult worker honey bees. Abs. Ann. meeting Soc. Invert. Path., pp. 74-75.

Vandenberg, J.D. and Shimanuki, H. (1986) Two commercial preparations of the beta exotoxin of Bacillus thuringiensis influence the mortality of caged adult honey bees, Apis mellifera (Hymenoptera: Apidae). Environ. Entomol. 15:166-69.

Vandenberg, J.D., Hamm, J.J. and Shimanuki, H. (1986) Vairimorpha sp. spores do not reduce the longevity of caged adult honey bees, Apis mellifera (Hymenoptera: Apidae). Environ. Entomol. 15:207-09.

Withers, P.C. (1977) Measurement of V_O2, V_CO2 and evaporative water loss with a flow-through mask. J. Appl. Physiol. 42:120-123.

Figure 1. Temperatures at various positions inside the ECM. Highest temperatures occur nearest the top of the chamber.

Figure 2. PC BEEPOP estimates of colony population dynamics under chamber conditions. The simulation is for a colony with a queen capable of laying 500 eggs per day. From top to bottom of the Figure, the traces represent numbers of adult bees, forager bees, hive bees, pupae, larvae, and eggs.