© INRA/DIB-AGIB/EDP Sciences, 2009

1. INTRODUCTION

Temperature is an important factor affecting larval and pupal development of insects (Nylin and Gotthard, 1998). Increased temperatures typically result in higher growth rates, higher respiration rates, and shorter development times, as well as influence the adult body size (Büns and Ratte, 1991; Sibly and Atkinson, 1994; Petz et al., 2004). Also mortality rates are affected by temperature with extreme temperatures having lethal effects (Howe, 1967). Control of brood temperature is therefore considered as an important evolutionary advantage of many eusocial insect colonies to optimize the rearing conditions of the brood. Specific behavioral adaptations of the workers including active heating and cooling or transport of larvae to cooler or warmer nest regions ensure optimal temperature conditions for the brood (Steiner, 1929; Heinrich, 1993). The honeybee, Apis mellifera, has been shown to have a most precise temperature regulation in the brood nest and the brood temperature ranges within narrow limits between 32 °C and 36 °C with a mean of 34.5 °C (Himmer, 1927; Kronenberg and Heller, 1982). This temperature homeostasis is on the one hand achieved by active heating of workers through clustering and/or the simultaneous activation of their thoracic muscles. On the other hand evaporation of water through fanning is used for cooling (Lindauer, 1954; Southwick, 1983; Harrison, 1987; Esch et al., 1991; Kleinhenz et al., 2003). Indeed the constant brood nest temperature is important because major deviations cause malformations of the emerging adults (Himmer, 1927; Groh et al., 2004). However, even small temperature fluctuations in individual brood cells within the physiological limits regulated by the workers have substantial significance for adult workers later in life. The brood temperature affects many traits of adult bees, including learning abilities, outdoor activities and the pace of temporal polyethism (Tautz et al., 2003; Groh et al., 2004; Becher et al., 2009). Since brood temperature interferes with the cognitive abilities of the bees and their task specialization, colony temperature is likely to be a major driver of colonial organization and allocation of workers to certain tasks. However, in order to quantify the effect of brood temperature on colony organization, it is essential to accurately measure the temperature profile in individual cells throughout the larval and pupal development and compare this with the behavioral phenotype of the adult worker. Although experiments have been conducted, using a large set of incubators each set to a different temperature, this approach does not allow for assessing the natural variance for each individual cell. Clearly it would be much more enlightening if we could obtain this data in situ in the colony to assess the natural temperature variance in brood cells. We developed a precise thermo-device which allows for monitoring the temperature profile across the brood comb at both a high spatial and temporal resolution while providing minimal disturbance to the colony.

Figure 1 256 resistor NTC resistor sensors were placed in a grid of 16 rows by 16 columns on a circuit board (A). The temperature sensors project through the front panel, just reaching the cell wall junctions of the test comb. The perspex box can easily be placed in any standard hive (B).

2. METHODS AND RESULTS

2.1. General description

256 resistor sensors were arranged on a 15 × 15 cm area consecutively delivering temperature data from the bottom of the cells in a test comb (Fig. 1A). Each sensor touched the comb in the centre of three adjacent cells, so altogether temperature data from 768 brood cells could be recorded. The temperature sensors project through the front panel of a perspex box which holds the electronic circuit board (Fig. 1B). A test comb is arranged in front of this box so the sensors just reach the bottom of the cells in the test comb and are inaccessible to the bees. The sensor board is connected to a personal computer, which executes the addressing of the resistors and reads in the received data.

2.2. Sensor board

For the temperature measurements we used an array of 256 thermistors with a negative temperature coefficient (“NTC” resistors; SEMI 833 ET, Hygrosens^® Instruments). The resistance of a NTC thermistor drops non-linearly when the temperature rises. These resistors are low-cost items originally developed for the use in clinical thermometers. The small diameter (1.5 mm), a short reaction time (0.7 s), a high sensitivity, and a high long-term stability make them particularly useful for our purpose.

Figure 2 Simplified circuit diagram: 256 NTC sensors, each provided with a standard diode (1N4148) are placed on a 16 × 16 grid with rows and columns are chosen via four 8-channel multiplexers (MAX 4051). Addressing of the channels is accomplished by a PC controlled I/O-board (Quancom: PCI TTL-I/O 32), resulting signals are amplified (MAX 4166) and read in by an A/D-converter.

The 256 NTC-sensors were placed in a grid of 16 rows by 16 columns on a circuit board (Fig. 2). Each sensor was placed in the cell wall junction of three adjacent cells and did not insert into the cell lumen. This way an area of 768 cells could be monitored on the comb. The sensors were activated via four 8-channel multiplexers (MAX 4051, Maxim^® Integrated Products). The addressing of the multiplexers was accomplished by a PC controlled I/O-board (PCI TTL-I/O 32, Quancom^®). Standard diodes (1N4148) prevented the addressing of more than one sensor. The resulting analog signals were amplified (MAX 4166, Maxim^® Integrated Products) converted to digital values (A/D-converter produced by Point Electronic, Halle, Germany) and stored in a text file.

Figure 3 (A) Recorded temperature data are analyzed and displayed via a visualization software. Shown here is a snapshot of the temperature distribution on June 16, 2006, 2:57 a.m. with brood being present in the upper part of the comb (red and green area). For the cells No. 212 and No. 284 the temperature profile is given in Figure 4. The software also provides fundamental statistical values (mean temperature, standard deviation, minimum and maximum temperature, number of cells in a given temperature range). (B) The standard deviation of cell temperatures from 16.-24.06.2006. Temperature fluctuations are strongly reduced in the upper part of the comb, where the broodnest was located (blue and green area).

2.3. Data record and processing

The sensors were consecutively addressed, with three measurements within three seconds for each sensor. The median of these three values was used for further analysis. The temperature in each cell was computed as the mean of the two nearest sensors, with the closer one weighted double. The data recording was conducted in an endless loop that could be continued for weeks and was graphically displayed by a software tool in Delphi/Pascal (source code available on request). This tool shows the graphic presentation of all monitored cells for any time step in false color (Fig. 3A), including parameters such as mean temperature, standard deviation, minimum and maximum temperature and number of cells in a given temperature range. The software tool together with a data file is provided as electronic-only material on the Apidologie website (http://www.apidologie.org).

2.4. Empirical test of the instrument

We tested the instrument in a hive with the test comb, three additional frames and about 3000 workers bees. The colony was placed in the laboratory at room temperature (25 °C) with a flight entrance to the outside. The queen was confined on the comb area over the sensor array with a queen excluder (through which workers can pass but not the queen) to ensure egg-laying on the test comb. Worker bees had access to the queen and the brood for feeding and tending, while the queen could not move to the other frames until a sufficiently large brood nest was established. Temperature data were recorded as described above.

Figure 4 Two examples for individual temperature profiles of a warm and a cool broodcell during the pupal stage. The warm cell No. 212 was situated close to the centre of the broodnest, whereas the cooler cell No. 284 was located at the periphery (compare with Fig. 3A).

2.5. Results

We found a negative correlation between the absolute temperature and the temperature fluctuations in the brood nest. A higher temperature leads to a highly significantly reduced variance over time (Pearson test: P = 0.0001, r = − 0.64, N = 30) (Fig. 5), indicating that temperatures are more constant in the centre of the brood nest, where the highest temperatures occur (Fig. 3, Fig. 4). As a consequence, the negative correlation of mean brood cell temperatures with the distance from the brood nest centre is highly significant (Pearson test: P = 0.0005, r = − 0.60, N = 30).

The mean temperature measurements in 30 brood cells during the three days of the egg phase was 32.7 °C, ranging from 30.5 °C to 34.0 °C. During larval development, the mean temperature was 33.3 °C (minimum larval cell temperature: 30.0 °C, maximum temperature: 34.8 °C), and for the pupal phase we measured a mean temperature of 33.2 °C (minimum: 31.1 °C, maximum 34.6 °C). The coldest cells (mean temperatures < 30 °C) as well as the highest temperature fluctuations (standard deviation greater than ± 1 °C) were found in the broodless area (Fig. 3B).

2.6. Comparing front and back side temperatures

To detect the temperature differences between the front and back sides of cells, we used two instruments measuring the temperature distribution in an empty comb from both sides at the same time. The instruments were placed in an artificially generated temperature gradient without any bees. After a stable temperature distribution was reached, the temperatures on the warm side of the comb ranged from 29 °C to 34 °C, whereas the temperatures on the other side were at the average 1.4 °C lower (Fig. 6). We thus underestimate the actual cell temperature by 1.4 °C.

Figure 5 The standard deviation of the temperature plotted against mean developmental temperature of those cells, where definitely brood was present. (Pearson test: P = 0.0001, r = − 0.64, N = 30).

Figure 6 Comparison of front- and back side temperatures in an empty comb. We used two instruments placed in a temperature gradient to measure the temperatures in one comb simultaneously on both sides. Each data point represents the temperatures of a pair of two opposing sensors, averaged over two hours.

3. DISCUSSION

First measurements of temperature in social insects were conducted with mercury thermometers (Newport, 1837; Himmer, 1927; Andrews, 1929), yielding only highest and lowest values with special maximum-minimum thermometers. The first continuous temperature measurements became possible with the use of thermocouples or thermistors yet both were usually only applied in a small number so that they did not show a global temperature pattern across a comb (e.g. Cameron, 1985 in Bombus; Southwick and Heldmaier, 1987; Fahrenholz et al., 1989 in Apis; Hozumi et al., 2005 in Polybia). More recently, infrared thermography has been used for temperature measurements in honeybees. Indeed this technique delivers a spatial temperature distribution and has been repeatedly used to monitor temperature data of individuals (Stabentheiner and Schmaranzer, 1987; Kastberger and Stachl, 2003; Kleinhenz et al., 2003). However, information from inside the colony can hardly be gathered, nor does this method allow to measure temperatures within individual cells of a comb.

In comparison with brood nest temperatures reported in older literature (about 34.5 °C; Hess, 1926; Himmer, 1927; Dunham, 1931), the temperatures measured in our study were about 1.4 °C lower which is exactly the difference obtained in our control experiment without bees, where we heated the air on one side of the comb and measured the other. The average development temperature of 33.1 °C is thus due to the construction of the instrument: bees only have access to the front side of the comb but not to the back side which holds the electronic apparatus. The insulative layer of worker bees on the back side of the comb was missing which causes the lower than expected temperatures. Since thermistors poking into the cell lumen would interfere with larval and pupal development, the actual temperatures inside the cells and in the developing larvae cannot be directly recorded. Nevertheless, temperature distributions and fluctuations over time can be accurately monitored by the termistors on the back side of the combs, because the temperature difference of 1.4 °C is linear over the expected range of temperatures observed in the hive (Fig. 6).

We found higher and more constant temperatures close to the centre of the comb. Temperatures in the periphery of the brood nest were not only lower, but they showed a much stronger variance. As brood temperature results mainly from the activity of adult bees, the temperature profile reflects the distribution of the workers on the comb. The high number of workers near the centre of the brood nest raised the temperature and reduced the temperature fluctuations. A lower temperature in the periphery of the brood nest was also found by Rosenkranz and Engels (1994). For Carniolan honeybees, they measured mean temperatures in capped cells of 35.2 °C to 35.4 °C in the centre and of 33.5 °C to 34.5 °C in the peripheral areas.

The thermo-device we present here provides the possibility to constantly measure the temperature distribution under near natural conditions on the comb with a high spatial resolution. Difficulties due to lower temperatures on the back side of the test comb might be overcome by allowing the bees to enter both sides of the comb. Contrary to previous studies, where honeybee pupae had been raised in incubators to analyze the influence of brood temperature on the adults (Tautz et al., 2003; Groh et al., 2004), we now are able to record the complete temperature history of any individual from the egg stage to the emergence of the adult worker. This will not only open the way to understand the impact of temperature on the development of a honeybee, but also the role it will play as an adult in the colony and hence for overall colony organization of honeybees.

Acknowledgments

For technical advices and help in programming we would like to thank Gunther Tschuch, Sven Black-Hand Ewald and Felix Lehmann. This study was funded by the Deutsche Forschungsgemeinschaft with a grant given to RFAM.

References

• Andrews E.A. (1929) Populations of Ant Mounds, Q. Rev. Biol. 4, 248–257 [CrossRef].
• Becher M.A., Scharpenberg H., Moritz R.F.A. (2009) Pupal developmental temperature and behavioral specialization of honeybee workers (Apis mellifera L.), J. Comp. Physiol. A, DOI:10.1007/s00359-009-0442-7.
• Büns M., Ratte H.T. (1991) The combined effects of temperature and food consumption on body weight, egg production and developmental time in Chaoborus crystallinus De Geer (Diptera: Chaoboridae), Oecologia 88, 470–476.
• Cameron S.A. (1985) Brood care by male bumble bees, Proc. Natl Acad. Sci. USA 82, 6371–6373 [CrossRef] [PubMed].
• Dunham W.E. (1931) Hive temperatures for each hour of a day, Ohio J. Sci. 31, 181–188.
• Esch H., Goller F., Heinrich B. (1991) How do bees shiver? Naturwissenschaften 78, 325–328 [CrossRef].
• Fahrenholz L., Lamprecht I., Schricker B. (1989) Thermal investigations of a honey bee colony, thermoregulation of the hive during summer and winter and heat production of members of different bee castes, J. Comp. Physiol. B 159, 551–560 [CrossRef].
• Groh C., Tautz J., Rössler W. (2004) Synaptic organization in the adult honey bee brain is influenced by brood-temperature control during pupal development, Proc. Natl Acad. Sci. USA 101, 4268–4273 [CrossRef] [PubMed].
• Harrison J.M. (1987) Roles of individual honeybee workers and drones in colonial thermogenesis, J. Exp. Biol. 129, 53–61 [PubMed].
• Heinrich B. (1993) The Hot-Blooded Insects, Harvard University Press, Mass, p. 606.
• Himmer A. (1927) Ein Beitrag zur Kenntnis des Wärmehaushaltes im Nestbau sozialer Hautflügler, Z. Vgl. Physiol. 5, 375–389.
• Hess W.R. (1926) Die Temperaturregulierung im Bienenvolk, Z. Vgl. Physiol. 4, 465–487 [CrossRef].
• Howe R.W. (1967) Temperature Effects on Embryonic Development in Insects, Annu. Rev. Entomol. 12, 15–42 [CrossRef] [PubMed].
• Hozumi S., Yamane S., Miyano S., Mateus S., Zucchi R. (2005) Diel changes of temperature in the nests of two Polybia species, P. paulista and P. occidentalis (Hymenoptera, Vespidae) in the subtropical climate, J. Ethol. 23, 153–159 [CrossRef].
• Kastberger G., Stachl R. (2003) Infrared imaging technology and biological applications, Behav. Res. Methods, Instrum. Comput. 35, 429–439.
• Kleinhenz M., Bujok B., Fuchs S., Tautz J. (2003) Hot bees in empty broodnest cells: heating from within, J. Exp. Biol. 206, 4217–4231 [CrossRef] [PubMed].
• Kronenberg F., Heller H.C. (1982) Colonial thermoregulation in honey bees (Apis mellifera), J. Comp. Physiol. B 148, 65–76 [CrossRef].
• Lindauer M. (1954) Temperaturregulierung und Wasserhaushalt im Bienenstaat, J. Comp. Physiol. A 36, 391–432.
• Newport G. (1837) On the Temperature of Insects, and Its Connexion with the Functions of Respiration and Circulation in This Class of Invertebrated Animals, Philos. T. Roy. Soc. B 127, 259–338 [CrossRef].
• Nylin S., Gotthard K. (1998) Plasticity in Life-History Traits, Annu. Rev. Entomol. 43, 63–83 [CrossRef] [PubMed].
• Omholt S.W., Amdam G.V. (2004) Epigenetic Regulation of Aging in Honeybee Workers, Sci. Aging Knowl. Environ. 26, pe28.
• Petz M., Stabentheiner A., Crailsheim C. (2004) Respiration of individual honeybee larvae in relation to age and ambient temperature, J. Comp. Physiol. B 174, 511–518.
• Rosenkranz P., Engels W. (1994) Genetic and environmental influences on the duration of preimaginal worker development in eastern (Apis cerana) and western (Apis mellifera) honey bees in relation to Varroatosis, Rev. Bras. Genet. 17, 383–391.
• Scherba G. (1962) Mound Temperatures of the Ant Formica Ulkei Emery, Am. Midl. Naturalist 67, 373–385 [CrossRef].
• Sibly R.M., Atkinson D. (1994) How rearing temperature affects optimal adult size in ectotherms, Funct. Ecol. 8, 486–493 [CrossRef].
• Southwick E.E. (1983) The honey bee cluster as a homeothermic superorganism, Comp. Biochem. Physiol. A 75, 641–645 [CrossRef].
• Southwick E.E., Heldmaier G. (1987) Temperature control in honey-bee colonies, Bioscience 37, 395–399 [CrossRef].
• Stabentheiner A., Schmaranzer S. (1987) Thermographic determination of body temperatures in honey bees and hornets: calibration and applications, Thermology 2, 563–572.
• Steiner A. (1929) Temperaturuntersuchungen in Ameisennestern mit Erdkuppeln, im Nest von Formica exsecta Nyl. und in Nestern unter Steinen, Z. Vgl. Physiol. 9, 1–66.
• Tautz J., Maier S., Groh C., Rössler W., Brockmann A. (2003) Behavioral performance in adult honey bees is influenced by the temperature experienced during their pupal development, Proc. Natl Acad. Sci. USA 100, 7343–7347 [CrossRef] [PubMed].

Online material

Download PDF file.

Description of the Software-Tool

1.) Save the files „GraphicComb_Demo.exe“ and „Data.txt“ in the same folder and execute „GraphicComb_Demo.exe“. Choose the input file („Data.txt“) and load the data.

2.) Mean temperature, standard deviation of the temperature, minimal and maximal temperature as well as the according sensors are shown

3.) Use the the buttons (≪ | < | > | ≫) or enter the time step and click „Go“

4.) Check the boxes for the information you are interested in and click „Refresh“

5.) Insert the upper and lower temperatures to count the cells within a temperature range

6.) Check the „show legend“ box, „Refresh“ and „JPG“ to create a picture of the current time step

7.) Click „outfile“ to create a new text file with the mean temperatures etc. for each time step. Start the calculation from time step 1! Calculation is finished, when the „Out of dataset¡‘

8.) Choose the high and low temperatures assigned to red and blue or click „B/W“ for a black and white display. Do not forget to „Refresh“ again

All Figures

	Figure 1 256 resistor NTC resistor sensors were placed in a grid of 16 rows by 16 columns on a circuit board (A). The temperature sensors project through the front panel, just reaching the cell wall junctions of the test comb. The perspex box can easily be placed in any standard hive (B).
In the text

	Figure 2 Simplified circuit diagram: 256 NTC sensors, each provided with a standard diode (1N4148) are placed on a 16 × 16 grid with rows and columns are chosen via four 8-channel multiplexers (MAX 4051). Addressing of the channels is accomplished by a PC controlled I/O-board (Quancom: PCI TTL-I/O 32), resulting signals are amplified (MAX 4166) and read in by an A/D-converter.
In the text

Figure 3 (A) Recorded temperature data are analyzed and displayed via a visualization software. Shown here is a snapshot of the temperature distribution on June 16, 2006, 2:57 a.m. with brood being present in the upper part of the comb (red and green area). For the cells No. 212 and No. 284 the temperature profile is given in Figure 4. The software also provides fundamental statistical values (mean temperature, standard deviation, minimum and maximum temperature, number of cells in a given temperature range). (B) The standard deviation of cell temperatures from 16.-24.06.2006. Temperature fluctuations are strongly reduced in the upper part of the comb, where the broodnest was located (blue and green area).

In the text

	Figure 4 Two examples for individual temperature profiles of a warm and a cool broodcell during the pupal stage. The warm cell No. 212 was situated close to the centre of the broodnest, whereas the cooler cell No. 284 was located at the periphery (compare with Fig. 3A).
In the text

	Figure 5 The standard deviation of the temperature plotted against mean developmental temperature of those cells, where definitely brood was present. (Pearson test: P = 0.0001, r = − 0.64, N = 30).
In the text

	Figure 6 Comparison of front- and back side temperatures in an empty comb. We used two instruments placed in a temperature gradient to measure the temperatures in one comb simultaneously on both sides. Each data point represents the temperatures of a pair of two opposing sensors, averaged over two hours.
In the text

In the text

In the text

In the text

In the text

In the text

In the text

In the text

In the text