Journal of Comparative Physiology B

, Volume 174, Issue 7, pp 511–518

Respiration of individual honeybee larvae in relation to age and ambient temperature

Authors

  • Markus Petz
    • Institut für ZoologieUniversität Graz
    • Institut für ZoologieUniversität Graz
  • Karl Crailsheim
    • Institut für ZoologieUniversität Graz
Original Paper

DOI: 10.1007/s00360-004-0439-z

Cite this article as:
Petz, M., Stabentheiner, A. & Crailsheim, K. J Comp Physiol B (2004) 174: 511. doi:10.1007/s00360-004-0439-z
  • 162 Views

Abstract

The CO2 production of individual larvae of Apis mellifera carnica, which were incubated within their cells at a natural air humidity of 60–80%, was determined by an open-flow gas analyzer in relation to larval age and ambient temperature. In larvae incubated at 34 °C the amount of CO2 produced appeared to fall only moderately from 3.89±1.57 µl mg−1 h−1 in 0.5-day-old larvae to 2.98±0.57 µl mg−1 h−1 in 3.5-day-old larvae. The decline was steeper up to an age of 5.5 days (0.95±1.15 µl mg−1 h−1). Our measurements show that the respiration and energy turnover of larvae younger than about 80 h is considerably lower (up to 35%) than expected from extrapolations of data determined in older larvae. The temperature dependency of CO2 production was determined in 3.5-day-old larvae, which were incubated at temperatures varying from 18 to 38 °C in steps of 4 °C. The larvae generated 0.48±0.03 µl mg−1 h−1 CO2 at 18 °C, and 3.97±0.50 µl mg−1 h−1 CO2 at 38 °C. The temperature-dependent respiration rate was fitted to a logistic curve. We found that the inflection point of this curve (32.5 °C) is below the normal brood nest temperature (33–36 °C). The average Q10 was 3.13, which is higher than in freshly emerged resting honeybees but similar to adult bees. This strong temperature dependency enables the bees to speed up brood development by achieving high temperatures. On the other hand, the results suggest that the strong temperature dependency forces the bees to maintain thermal homeostasis of the brood nest to avoid delayed brood development during periods of low temperature.

Keywords

ApisHoneybee larvaRespirationCO2 productionTemperature

Abbreviations

m

body mass

R

rate of development or respiration

TI

inflexion point of a logistic (sigmoid) curve

TL

lethal temperature

TO

temperature of optimum (maximum) development

Introduction

Honeybee colonies can be seen as superorganisms that reproduce by swarming (Moritz and Southwick 1992). Colonies with large numbers of individuals are considered to produce more swarms and to be more resistant to environmental fluctuations, thus making rapid brood production crucial for successful colony reproduction (Bodenheimer 1937). This is even more important in temperate regions, where the time frame to mature and swarm during the warm season is especially limited. After swarming, the task of raising offspring has to be accomplished as quickly as possible in order to achieve a sufficient colony strength to enable the bees to gather enough energy reserves and to rear enough workers to survive the following winter. Evolution, therefore, has optimized honeybee larvae for a rapid development. According to Wheeler and Wheeler (1979), Apis mellifera development periods are among the shortest in social hymenopterans (3 days as eggs, 5.5 days as larvae, 3 days as prepupae, 8–9 days as pupae for workers, with queens developing even faster). From hatching to diapause, the larvae grow from about 0.15 to 150 mg in less than 6 days (Wang 1965).

The energetic requirements of larval growth have been investigated several times. Melampy and Willis (1939) measured the oxygen uptake and carbon dioxide output of worker and queen larvae, using the Barcroft-Warburg technique. Allen (1959) determined the oxygen consumption of worker and drone larvae in Warburg respirometers. Shuel and Dixon (1959) used a modified Barcroft-Warburg constant-volume respirometry method for measurements on single larvae. They made a comparison of oxygen consumption of small larvae feeding on worker jelly, royal jelly and older larvae’s pollen-jelly mixture. Schmolz and Lamprecht (2000) measured larval heat production by microcalorimetry. Mackasmiel and Fell (2000) measured the CO2 production of single honeybee eggs at different ages and temperatures. In all the studies on respiration of bee larvae mentioned above, the larvae were exposed to an unnatural environment. They were incubated in dry air and not fed at all (Melampy and Willis 1939) or fed on sugar solution (Allen 1959). In order to maintain an incubation time of 1 h, several small larvae were incubated at once. In the present study, we made use of the benefits of a flow-through gas analyzer. The high sensitivity of the device allowed measurements to be made on single and much smaller larvae than previously possible. The animals could be left within their “natural environment”, their cell, feeding on naturally provided food jelly.

Honeybees maintain thermal homeostasis of their brood by heating actively with their wing muscles, clustering for insulation, or cooling by vaporizing water and wing fanning when necessary (Lindauer 1954; Esch 1960; Heinrich 1981, 1993; Schmaranzer et al. 1987; Bujok et al. 2002; Kleinhenz et al. 2003). Different mean temperatures of the brood area were reported depending on the developmental stage of the brood and their position within the nest. Air temperatures near the brood can range from ~32–36 °C (Büdel 1955; Ritter and Koeniger 1977; Ritter 1982; Kleinhenz et al. 2003). Temperature fluctuates by only ±1–1.5 °C in areas in the center of brood combs, whereas temperature variations in peripheral areas are greater (Büdel 1955; Kleinhenz et al. 2003). The narrow temperature window found in the brood area can be assumed to be the optimal temperature for a high rate of development, up to which the energetic costs are still affordable. The bees maintain the high rate of development of their brood partly with great energy investment (Kronenberg and Heller 1982; Ritter 1982; Fahrenholz et al. 1989). The accurate regulation of the brood temperature suggests a strong dependence of larval energy turnover on temperature. To elucidate this question, we determined the temperature dependence of larval respiration (CO2 production).

Materials and methods

Animals

In the summer months of 2001 (July–September) and 2002 (July), two honeybee colonies (Apis mellifera carnica Pollmann) of about 15,000 individuals at the Department for Zoology, University of Graz, were used to obtain honeybee larvae of different ages. When weather conditions were poor for foraging, and in mid-September, the bees were fed sugar solution directly into the hive. The solution consisted of 500 ml of 1:1 (sugar to water) solution for 11 days in mid-July 2001, and 1,000 ml of 1:1.5 solution for 6 days in mid-September 2001. To get cohorts of similarly sized and similarly aged larvae, the queen was confined for 4 h in an area of 15×15 cm on an empty part of comb within a wooden frame that held a piece of queen-excluder lattice, which let worker bees pass but not the queen because of her bigger size. After caging, the queen was prohibited from accessing that comb area until the next day using queen excluders. The larvae were expected to hatch after 72 h within a 4-h period (based on literature data). In order to obtain newly emerged larvae, the development of the cells was examined after 72 and 76 h. However, the first lot of larvae appeared to hatch about 10 h later that expected. The time of emergence was then estimated by comparing the live weight of the young larvae with data from earlier publications on growth of honeybee larvae (Wang 1965). This way, larvae that were 0.5–5.5 days old were obtained on 5 successive days until the oldest group of larvae had freshly capped cells.

For each incubation period, the comb was taken from the hive and a single cell containing a larva was cut out of it. Attention was taken not to stress the larva by temperature changes or over-handling. Food jelly and residues of larvae were removed carefully from the walls of adjacent cells.

CO2 measurements

Carbon dioxide production of the larvae was determined in a differential flow-through infrared-light-absorption CO2 analyzing system (URAS 14, computer controlled with digital read-out; Hartmann and Braun, ABB) with an accuracy of 1 ppm CO2. In order to maximize the system sensitivity (<0.2 ppm), air was taken from outside the laboratory window, and before it entered the URAS for the first time, the air passed through two containers to dampen fluctuations in CO2 content, one (10 l) before and the other one (5 l) after the pump and mass flow controllers (Brooks 5850 S, 0–100 and 0–1000 ml/min). The air was dried by passing it through two Peltier-driven cool traps (~7.5 °C) before it entered the URAS reference and measurement tubes, where it was heated to 60 °C. The cell containing the larva was placed in a 2-ml tube (measurement chamber) which was immersed in a water bath (Julabo F33) for temperature control to the nearest 0.1 °C. The tube was the front part of a plastic syringe; the rear end had been sealed by a rubber stopper.

The cell opening pointed towards the outlet of the tube. The caps of capped cells were opened prior to measurement to ensure an air exchange comparable to unsealed cells. In the first series of measurements the incubation temperature and relative humidity (RH) were adjusted to resemble that of a hive, with RH between 60 and 80% and a temperature of 34 °C. Humidity was adjusted by saturating the air with water vapor by passing it through two water flasks immersed in a water bath prior to the measurement chamber. The water-bath temperature was adjusted to the dew-point temperature, which corresponded to the desired RH at the temperature inside the measurement chamber. In the second series of measurements, we varied the incubation temperature from 18–38 °C in steps of 4 °C but again adjusted RH to 60–80%. These experiments were made with larvae of medium age (about 2.5 days old) that had shown the highest growth rate in the first measurement series, with six individuals at each temperature. Earlier investigations (data not shown) showed that the respiration rate did not differ between hive CO2 levels (approximately 6,000 ppm; Seeley 1974) and outdoor air (380 ppm, SD=11.9, n=27; E. Stabentheiner, personal communication). Thus we chose to use outdoor air instead of an artificial mixture. The airflow in the system was set at 10–150 ml/min, depending on the size of the larvae. The air pressure in the chamber was approximately 6.67 kPa (50 mm Hg) above atmospheric pressure. Volumes (µl) of CO2 production reported in this paper refer to standard conditions (0 °C, 101.32 kPa). The CO2 production was recorded at intervals of 1 s.

An incubation time of about 45 min per larva was chosen because the respiration rate started to decrease in large larvae when they were left without nutrition beyond this period. The mean CO2 value of the final 20 min was measured for further analysis. The system was calibrated before and after each larva, and the data were corrected for any drift during the incubation period. The larvae can double their weight every 3 h. Therefore, in order to keep a narrow time frame for groups of similarly aged larvae, we measured a maximum of four larvae per day.

For determining the live weight at the end of the experiments, food-jelly residues were removed by submerging the larvae in 0.5 M glucose solution, which was used instead of distilled water in order to prevent osmotic bursting of the larval skin. After the washing, the surfaces of the larvae were dried by blotting them carefully on filter paper.

CO2 production and larval mass were determined in larvae aged 0.5, 1.5, 2.5, 3.5, 4.5 and 5.5 days with 8–13 individuals in each group.

Analyses

For relations of development times to temperature, as well as the relation of larval respiration rates to temperature, the logistic (sigmoid) function is applicable (Ratte 1984):
$$ R = a + \frac{b} {{{\text{1}} + {\text{e}}^{{c - dT}} }} + {\left\{ {\frac{{\text{e}}} {{T - T_{{\text{L}}} }}} \right\}} $$
(1)
where R is the rate of development or respiration and T is the temperature and TL is the lethal temperature in °C. The parameters a, b, c and d are constant. This function denotes the sigmoid part A–B of the curve in Fig. 1. The region around the inflexion point TI can be approximated by a straight line. The normal environmental temperature of most insect larvae is situated there. We fitted the temperature relation curve (CO2 produced per unit mass, related to temperature; 1 µl mg−1 h−1=44.615 nmol mg−1 h−1), to this logistic function. This is not an exponential function, so the Q10 is not constant but must be calculated separately for each point of the curve. We calculated the Q10 at a given temperature by using the CO2 production rate from Eq. (1) at that temperature and 10 °C above it. Beyond point B in Fig. 1, the slope of the development rate curve approaches zero at the temperature TO, the optimal temperature for development, which is the peak point of the development rate. Above this point, a decline due to metabolic insufficiencies begins. Usually, the upper thermal limit TL lies close to TO. The part of the equation in brackets could be used to describe the whole temperature relation including peak and decline (death).
Fig. 1

Generalized plot of the temperature dependency of the respiration rate in developing insects. The steepness of the curve rises until the inflexion point TI, then falls to zero in TO (bold line). The strong decline beyond this point is the result of metabolic insufficiencies (thin line). TL is the lethal temperature. The part AB of the curve can be approximated by the logistic part of Eq. (1) (part outside brackets, see text). The whole curve ATL can be approximated by including the part in brackets in Eq. (1)

Results

Weather and oviposition

The weather conditions during the observation periods varied considerably. Temperatures at noon ranged from 20–25 °C in July 2001, from 25–32° C in August 2001, and from 20–27°C in September 2001. In July 2001, most of the time, the sky was cloudy or overcast, with short periods of rainfall. This “bad weather” did not allow the bees to forage at full strength. During the observation period in July 2002, the noon temperature was 27–31°C with clear skies, allowing the bees to forage at full strength. During the 4 h of caging, the queens produced between 23–100 eggs (average was 63). During the cool periods, the larvae emerged about 80 h after oviposition.

Larval growth

Figure 2 shows the mean masses (m) of the larvae we used for measuring the dependency of CO2 production on age. To get a usable age estimate of freshly hatched larvae, we weighed a representative sample of them and estimated the age from the tables provided by Wang (1965). The weight of 12-h-old larvae was 0.36±0.08 mg (mean±SD). They grew nearly exponentially until 4.5 days (131.44±18.7 mg). This relation can be described by the function m=0.18e0.063×age (mg; R2=0.994). The mass of newly capped larvae was 159.66±12.91 mg. The relative growth rate was 4.05 day−1 (4.05-fold in 1 day) between 12 and 36 h (Fig. 2). Afterwards, the growth rate reached its peak of 6.93 day−1 between 36 and 60 h. After 60 h, the relative growth rate declined to 3.7 day−1 and to 3.48 day−1 on the fourth and fifth days, respectively. On the last day of larval development, we observed relatively little growth (1.21 day−1).
Fig. 2

Relative growth of honeybee larvae for a previous 24-h period (mt/mt−24 h). The plot shows a comparison of larval growth data with earlier work by Stabe (1930) and Wang (1965)

The larvae used in the temperature dependency measurements weighed 10.05±2.43 mg. There was no significant difference in the mean masses among the groups measured at the six temperatures (Kruskal-Wallis test: P>0.05).

Relation of CO2 production to larval age

Mass-specific CO2 production (µl CO2 per mg of larval tissue per hour) related to the age of the larvae is plotted in Fig. 3. It was 3.89±1.57 µl mg−1 h−1 in 12-h-old larvae, and decreased to 0.95±1.15 µl mg−1 h−1 in 132-h-old larvae. A linear regression analysis resulted in the function RCO2=(−0.0229×age)+4.2496 (R2=0.902), a polynomic (quadratic) regression resulted in RCO2=(−0.0001×age2)−(0.008×age)+3.887 (R2=0.925). Over the investigated life span (12–132 h), a honeybee larva produced a total of about 10.922 ml (487.285 µmol) CO2.
Fig. 3

CO2 production rate on a unit-mass basis in relation to larval age and larval mass (inset). Vertical bars SD, n=9, 9, 13, 9, 8, 8 from left to right. X denotes the mean CO2 value at 34 °C from the temperature dependency measurements (n=6; Fig. 5). For comparison, we inserted data from Melampy and Willis (1939) and Allen (1959), which were measured at 35 and 32 °C, respectively. From our CO2 measurements in relation to temperature (Fig. 5), we were able to project the values to 34 °C. The RQ from Melampy and Willis (1939) and the mass data by Wang (1965) were used to estimate the CO2 output from the O2 uptake Allen (1959) measured. Dashed lines are extrapolations from linear regressions of the data by Melampy and Willis (1939) and Allen (1959) and from our own data of the three oldest groups of larvae

When scaling the respiration rate of the whole animal to its live body mass (m), the curve generally follows the equation R=a×mb. The coefficient b is commonly called the mass scale exponent. Excluding the oldest group (freshly capped cells) from the analysis because growth had already stopped, the equation was RCO2=3.361×m0.9009 (Fig. 4).
Fig. 4

Dependency of CO2 production on the body mass. The CO2 production of larvae 0.5–4.5 days old (12–108 h) was taken for the regression. The 5.5-day-old group (132 h) were left out as the larvae were already capped and had stopped growing. The regression equation is RCO2=3.361×m0.9009 (R2=0.977; n=48)

Relation of CO2 production to temperature

The rate of CO2 production per unit mass (µl mg−1 h−1) in 60-h-old bee larvae at 18 °C was 0.48±0.03 µl mg−1 h−1 (Fig. 5). CO2 production per unit mass rose approximately exponentially until 34 °C as expected (0.83±0.05, 1.19±0.15, 2.05±0.16 and 3.09±0.29 µl mg−1 h−1 at 22, 26, 30 and 34 °C, respectively). The fit results, including the values at 38 °C (3.97±0.50 µl mg−1 h−1), were RCO2=0.1114×1.0998T (R2=0.951). At 38 °C, the CO2 production per unit mass deviated a little from an exponential curve (Fig. 5). This points to the sigmoid approach of the temperature relation function in Eq. (1), which fitted the data better (R2=0.966), with the following parameters (T given in °C and RCO2 in µl mg−1 h−1):
Fig. 5

CO2 production rate of 60-h-old honeybee larvae in relation to temperature. The production of CO2 per unit mass per hour fits the sigmoid function part of Eq. (1) from 18–38 °C (R2=0.966). The normal brood nest temperature in honeybees (~34–36 °C) is situated in the upper region of our investigated temperature range. Vertical bars SD, n=6 for each of the points. The Q10 derived from the fitted function is an arched curve (range 2.49–3.40). The average Q10 for the temperature range of 18–38 °C (23–33 °C mid-temperature) is 3.13 (SD=0.261, n=101; calculated from the logistic interpolation curve in steps of 0.1 °C). TI Temperature of the inflexion point. The curve for larval O2 consumption was estimated according to an RQ of 1.42 for 60-h-old larvae (Melampy and Willis 1939; Table 1). The bees nearly exclusively metabolize sugars. Therefore, their RQ=1 (Rothe and Nachtigall 1989), and the two curves (CO2, O2) are identical. Data for additional curves from (1) Mackasmiel and Fell (2000), (2) Stabentheiner et al. (2003), (3) Rothe and Nachtigall (1989), and (4) calculated from calorimetric measurements of Schmolz et al. (2002)

$$ R_{{{\text{CO}}_{{\text{2}}} }} = 0.345 + \frac{{4.715}} {{{\text{1}} + {\text{e}}^{{7.257 - 0.223T}} }} $$
(2)
This specific function is plotted in Fig. 5. The inflexion point TI is at 32.54 °C. The Q10 of this function was 3.05 at a mid-temperature of 23 °C (28:18 °C), 3.40 at 26.6°C (31.6:21.6 °C) and 2.49 at 33 °C (38:28 °C). The average Q10 from 23–33 °C mid-temperature (total temperature range: 18–38°C) was 3.13 (Fig. 5).

Discussion

The larvae did not hatch in the time frame predicted in the literature. This may have been due to the rather cool weather in the initial weeks. Egg development in Apis mellifera normally lasts about 72 h. When lowering the brood nest temperature by even a small amount, a noticeable increase in the time of development can be observed. For example, a reduction of 0.5 °C in nest temperature delays egg development by 4.4 h (6.1%; Harbo and Bolten 1981). The further larval weight development resembled that reported by Wang (1965) and Stabe (1930) in the main characteristics though there were some differences in detail (Fig. 2). Our larvae weighed slightly less than those of Stabe in most age groups (average −13.5%, varying from −71.8 to +14.3%), while they were heavier (average +13.6%, −2.6 to +45.8%) than those of Wang.

With the open-flow CO2 analyzer, we were able to quantify the CO2 production even of very small, single honeybee larvae while maintaining environmental conditions appropriate for them. We found a decrease in the CO2 production per unit weight with increasing larval age (Fig. 3). This way, for their own thermal comfort, the young and middle-aged larvae contribute disproportionately to heat production, though their total contribution is modest because of their small mass. In larvae older than 72 h, similar curves can be found in the works of Melampy and Willis (1939) and Allen (1959). It must be noted, however, that Allen (1959) determined the consumption of oxygen but not the production of carbon dioxide. Thus we had to estimate the CO2 production from the O2 data by taking the respiratory quotients (RQ) determined for larvae of a different age by Melampy and Willis (1939; Table 1). Our temperature dependency measurements (Fig. 5) enabled us to project results to 34 °C, allowing comparison of measurements. Figure 3 demonstrates the necessity of measuring the respiration of small larvae. Melampy and Willis (1939) started with 2.5-day-old (72 h) larvae and Allen (1959) pooled larvae below 20 mg (younger than ~72 h) into a single group. Our measurements revealed a much weaker age dependence in the younger larvae (<~84 h, Fig. 3), which is supported by the finding that a simple quadratic function fitted the data better (R2=0.925) than a linear regression (R2=0.902). Therefore, the CO2 production rate of the old larvae (3.5–5.5 days) must not be extrapolated to the younger ones, because their CO2 production rate is considerably lower (up to 35%) than indicated by an extrapolation from the older larvae. A similarly weak age dependence (“plateau”) was reported for larval heat production in 2- to 4-day-old larvae by Schmolz and Lamprecht (2000). Even lower is the weight-specific CO2-production rate of honeybee eggs (1.7 µl mg−1 h−1 at 34 °C; Mackasmiel and Fell 2000).
Table 1

Number and mass of larvae. These data were used to estimate RCO2 values from RO2 measurements performed by Allen (1959; see Fig. 3) and RO2 values of larvae (Fig. 5)

Age (h)

n

Mean mass (mg)

SD (mg)

RQa

12

9

0.36

0.08

36

9

1.47

0.18

60

13

10.2

2.02

1.42

84

9

37.73

7.97

1.23

108

8

131.44

18.7

1.23

132

8

159.66

12.91

1.13

aMelampy and Willis (1939)

The remarkably high mass scale exponent, b=0.9009, reveals a high dependency of respiration on body mass rather than body surface. According to Wieser (1986), this has to be expected in the growth phase. In Schistocerca americana grasshoppers, Grenlee and Harrison (2004) reported a mass scale exponent of b=0.73. Up to the third instar, however, their data suggest a higher exponent, which is closer to the value of our honeybee larvae.

The temperature dependency of CO2 production fits a logistic function (Fig. 5) rather than a simple exponential function. This kind of relationship for development rates as well as respiration rates has been observed in various insects, as summarized by Ratte (1984). Because of the high respiratory quotient of the fast-growing 60-h-old larvae we used for the present measurements (RQ=1.42; Melampy and Willis 1939, see Table 1), the estimated curve of O2 consumption is below the curve of CO2 production (Fig. 5). Up to a temperature of about 25 °C, the respiration (and energy turnover) of middle-aged larvae are very similar to freshly emerged young bees (Stabentheiner et al. 2003) and adult bees (Rothe and Nachtigall 1989; Schmolz et al. 2002). The temperature dependency of larval CO2 production in the range 18–38 °C is greater (mean Q10=3.13) than in freshly emerged resting young bees (mean Q10=1.72, Stabentheiner et al. 2003; compare Fig. 5). Based on the high RQ of 60-h-old larvae (Table 1), the difference is smaller for the Q10 of oxygen consumption. The Q10 value of 1.9 for mitochondrial respiration of flight muscle tissue in adult bees (Leonhard and Crailsheim 1999) resembles that of young bees rather than that of middle-aged larvae. The temperature dependency of the O2 consumption of resting adult bees (Rothe and Nachtigall 1989) was measured to be lower than in our middle-aged larvae (i.e., is the same as in young bees) up to a temperature of 30 °C but resembles that of the larvae at 35 °C. In sleeping adult bees, the heat production rate (Schmolz et al. 2002; calorimetric measurement) increases with temperature (20–35 °C) at a rate similar to that of larval respiration, i.e., it resembles the larvae rather than the young bees. The temperature dependency of the CO2 production of eggs (28–36 °C; Mackasmiel and Fell 2000) resembles that of middle-aged larvae, though the mass-specific gas exchange is lower, which is similar to freshly emerged bees (Fig. 5).

We assume the respiration rate of A. mellifera larvae to be indicative of their development rate. Therefore, the strong temperature dependency of larval metabolism allows the bees to speed up brood development particularly by regulating the brood’s temperature at a high level (ca. 33–36 °C; Hess 1926; Himmer 1932; Büdel 1955; Kronenberg and Heller 1982; Ritter 1982; Fahrenholz et al. 1989; Kleinhenz et al. 2003; our own unpublished measurements). On the other hand, this strong temperature dependency probably forces the bees to regulate the brood temperature within narrow limits (maintain thermal homeostasis) in order to avoid delayed brood development during cold weather. More predictable and faster development cycles could have presented an evolutionary pressure towards temperature regulation. Investigations on bees that were temperature-treated during the pupal stage revealed that the bees’ development is indeed strongly adapted to high temperatures. The bees show shorter food dances and less odor-learning ability if the pupae are reared at 32 °C (Tautz et al. 2003).

At 38 °C, the curve is still quite steep (Fig. 5), which suggests that the absolute thermal limit for larval development is significantly above this temperature, at least for periods of a few hours as in the present investigation. It is not yet clear, however, why the bees do not increase their brood temperature to higher values in order to accelerate brood development further. The measurements of Mackasmiel and Fell (2000) on eggs suggest the temperature of maximum development (TO in Fig. 1) to be lower in eggs than in middle-aged larvae (Fig. 5). However, because the measurement time was much longer for the eggs (up to 14 h) than for our larvae (45 min), a final determination will depend on tests of the longevity of larvae at high temperatures (>36 °C). It is also not yet clear whether the sensitivity to high temperatures changes during larval and pupal development.

Hepburn et al. (1983) reported that the ductility of beeswax is optimally related to honeybee colony temperature. The tensile strength of both white comb wax and brown comb with pupal silk decreases above 35 °C (Hepburn and Kurstjens 1988). Besides the increased energetic costs for thermoregulation and possible long-term effects of high temperatures on larval development, the deterioration of the mechanical properties of beeswax and comb with increasing temperature (Hepburn 1986) may be an additional (though probably not the only) factor which prevents a further increase in brood temperature and thus in development speed in honeybees.

When insects are exposed to fluctuating temperatures, their development time is usually not the same as with constant temperatures (Ratte 1984). This is a result of the non-linearity of the temperature relation function. Symmetric temperature changes around a certain mean value cause asymmetric rate changes that result in different development times when integrating the rate of development over time. The change in development speed with temperature is highest at the inflection point TI of the logistic curve (Fig. 1). Temperature fluctuations below TI accelerate the development compared to a development at a constant mean temperature. Fluctuations above TI result in slower development than expected from constant temperature (Ratte 1984). The breeding temperature of A. mellifera is usually regulated above the TI of 32.54 °C (Ritter and Koeniger 1977; Kronenberg and Heller 1982; Ritter 1982; Fahrenholz et al. 1989; Kleinhenz et al. 2003; our own unpublished results). The temperature of individual capped brood cells was measured to fluctuate by about ±0.8 to ±1.2 °C within a range of about 32.5–36 °C (Kleinhenz et al. 2003). The temperature of uncapped brood is often somewhat lower (Kronenberg and Heller 1982). From a simplified model calculation (iterative, growth rate estimated from weight, with a cut-off at 150 mg), we estimated that by applying a ±3 °C cyclic fluctuation (cycle duration=1 h) to a mean temperature of 33 or 34 °C, the period for a larva to gain a mass of 150 mg might be prolonged by 0.3 or 0.87 h (0.2 or 0.7%), respectively. A cyclic fluctuation of ±1.5 °C, which resembles the observed values (Kleinhenz et al. 2003), delays development by only 0.1 or 0.23 h (0.08 or 0.2%) at the two temperatures, respectively. Therefore, in honeybee larvae, which usually develop at temperatures above the inflexion point of the logistic function, an even greater accuracy in brood-temperature regulation would not advance growth significantly.

Acknowledgements

We are greatly indebted to Stefan K. Hetz (Lehrstuhl für Tierphysiologie, Humboldt-Universität zu Berlin) for many tips and self-sacrificing technical help, and to E. Stabentheiner for measurements of air CO2 content in Graz. The research was supported by the Austrian Fonds zur Förderung der Wissenschaftlichen Forschung (FWF). The authors declare that the experiments performed in the production of this article comply with the current laws of the Republic of Austria.

Copyright information

© Springer-Verlag 2004