Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 May 26.
Published in final edited form as: Ecol Entomol. 2010 Apr 26;35(4):424–435. doi: 10.1111/j.1365-2311.2010.01198.x

Ontogeny of worker body size distribution in bumble bee (Bombus impatiens) colonies

Margaret J Couvillon 1,2,*, Jennifer M Jandt 1,*, Nhi Duong 3, Anna Dornhaus 1
PMCID: PMC4444232  NIHMSID: NIHMS689731  PMID: 26023250

Abstract

  1. Bumble bees exhibit worker size polymorphisms; highly related workers within a colony may vary up to 10-fold in body mass. As size variation is an important life history feature in bumble bees, the distribution of body sizes within the colony and how it fluctuates over the colony cycle were analysed.

  2. Ten commercially purchased colonies of Bombus impatiens (Cresson) were reared in ad libitum conditions. The size of all workers present and newly emerging workers (callows) was recorded each week.

  3. The average size of bumble bee workers did not change with colony age, but variation in body size tended to decrease over time. The average size of callows did not change with population size, but did tend to decrease with colony age. In all measures, there was considerable variation among colonies.

  4. Colonies of B. impatiens usually produced workers with normally distributed body sizes throughout the colony life cycle. Unlike most polymorphic ants, there was no increase in worker body size with colony age or colony size. This provides the first, quantitative data on the ontogeny of bumble bee worker size distribution. The potential adaptive significance of this size variation is discussed.

Keywords: Bombus impatiens, bumble bees, polymorphism, size distribution, social insects

Introduction

In complex biological systems, individual components often interact to ensure the survival and the reproduction of the group (Maynard Smith & Szathmáry, 1995). These individual components may be genetically identical, such as the cells that comprise a multi-cellular organism, or genetically dissimilar, as in the symbiotic relationship between eukaryotes and their mitochondria (Maynard Smith & Szathmáry, 1995). In complex systems, the whole may be greater than the sum of the parts, as different properties emerge from the interactions that would not be present at the individual level (Page & Mitchell, 1998; Sendova-Franks & Franks, 1999; Camazine et al., 2001). Social insect colonies are considered complex biological systems (Bonabeau et al., 1999; Fewell, 2003). Their within-colony genetic relatedness varies because colonies may be headed by either singly-mated or multiply-mated queens, which results in highly related or slightly related workers, respectively. The components that make up this system (i.e. members of a colony) may be broadly divided into the reproductive (queen/males) and non-reproductive (worker) castes (Hölldobler & Wilson, 2008), analogous to the reproductive and somatic cells within a multi-cellular organism.

Within the non-reproductive worker caste, there sometimes also exists further differentiation, analogous to the differentiated somatic cells of different tissue types. This differentiation within the non-reproductive worker caste may happen in several ways. In some insects, such as the honey bee (Apis mellifera), workers are similar in size and shape but differ in physiological attributes, like brain organisation, hormone levels and gene expression levels related to age (Robinson, 1987; Farris et al., 2001; Ben-Shahar et al., 2002; Whitfield et al., 2003), all of which contribute to different behaviors (Lindauer, 1953; Seeley, 1982; Robinson et al., 1994; Fjerdingstad & Crozier, 2006). In other insects, like some ants (Family: Formicidae), workers differ in more obvious morphological ways (Noirot & Pasteels, 1987; Hölldobler & Wilson, 1990; Stern & Foster, 1997), such as the dome-shaped head of some workers of the ant Cephalotes (Baroni Urbani, 1998; Powell, 2008) or the differently-sized workers within a colony of leaf cutter ants (Hölldobler & Wilson, 1990). All of the differentiations above contribute to variation in behaviours, which ultimately lead to division of labour (Beshers & Fewell, 2001) and to the functioning of the colony as a complex system. Morphological differences within the worker caste are especially interesting because they are produced by nutritional differences in the larval stage and fixed at eclosion, as most social insects are holometabolous (Couvillon & Dornhaus, 2009). Therefore, in order to examine morphological variation, it is imperative to study it at the colony level.

Bumble bee (Bombus spp.) workers, unlike other eusocial bees, exhibit a high degree of intra-colony morphological differentiation. That is, offspring of a singly-mated queen (in spite of a relatedness of 0.75) might differ by as much as 10-fold in their mass (Cumber, 1949; Plowright & Jay, 1968; Goulson, 2003). This size polymorphism is linked to bumble bee division of labour: smaller workers tend to feed and incubate brood and larger workers tend to forage, guard, and fan (Richards, 1946; Goulson et al., 2002; Jandt & Dornhaus, 2009). Previous work has reported the distribution of bumble bee worker sizes for the species Bombus terrestris (Goulson et al., 2002; Goulson, 2003). However, these descriptions were either derived from the distribution of all of the workers produced across the year, measured at the end of the colony cycle, or were collected on manipulated colonies at the end of the colony cycle (Couvillon & Dornhaus, 2010). Consequently, it is unknown whether different-sized workers are actually present at the same time in the colony. Nor does this end-of-cycle measurement describe whether different-sized workers are produced as a cohort at the same time or merely accrue over time as a consequence of changes in nutrition across the colony cycle. In other words, does the worker body size distribution within a bumble bee colony change over time? Previous authors have suggested that worker body size does increase over time in some bumble bee species (B. auricomus, B. griseocollis, B. fervidus, and B. perplexus) (Sladen, 1912; Frison, 1927; Knee & Medler, 1965; Plowright & Jay, 1968); however, these data were anecdotal in nature and sometimes included examples to the contrary (B. terricola and B. ternarius) (Goulson, 2003). Clearly a detailed analysis of the size distribution over time is needed.

Here, the size distribution of workers of the bumble bee species Bombus impatiens was analysed. Between-colony variation, the effect of colony size, and the shape of the distribution were quantified. Importantly, these features were tracked over time and as population size (number of workers in the colony) changed. If the production of large workers is limited by the resources available to the colony, then under ad libitum feeding conditions, all workers produced should have a large body size, and body size may also increase as the population of workers increases.

Materials and methods

Study organism

Between September 2005 and August 2006, 10 queenright colonies of B. impatiens (Koppert Biological Systems, Romulus, Michigan) were obtained, and each was placed inside a wooden nest box with a Plexiglas cover to provide constant viewing of in-nest behaviour (see Jandt & Dornhaus, 2009 for details). Colonies were kept at room temperature and on a light:dark 8:16 h cycle. Although foraging behaviour has strong diurnal rhythms, brood care has weak or no diurnal rhythms (Yerushalmi et al., 2006). Thus, larval feeding is performed around the clock and the effects of our laboratory light regime should neither have strongly influenced larval feeding behaviour in the nest nor our observations of the worker body size variation throughout colony development.

At the start of the experiment, colonies had 68 ± 27 workers. As the colonies used in our study were of commercial origin, the initial set of workers may have been comprised of mixed genetic ancestry. However, data collection began 2 weeks after colony set-up, a period after which bees that emerge are those raised by workers present. Bees had access to a foraging arena at all times, where they were provided with a sugary solution (‘Bee-Happy’, Koppert Biological Systems) and pollen ad libitum. When the colonies first arrived, all individuals were marked by gluing a numbered plastic tag (‘Opalithplättchen’) to the thorax. Afterwards, all newly emerging bees were marked and recorded on the date of eclosion. Thorax widths of workers, a standard measurement of body size in bumble bees (Goulson, 2003), were obtained post mortem using digital calipers to the nearest 0.01 mm. Any bee that died during the experiment was stored and measured as well. Rarely, we were unable to account for a bee, in which case it could not be included in the analysis; however, this was recorded and taken into account when estimating weekly population size. Large workers were readily distinguished from newly emerging gynes (i.e. virgin queens) as there tends to be little overlap between the size of workers and gynes in this type of bumble bee (i.e. pollen-storers) (Goulson, 2003).

At week 1 of data collection, the queen’s age was unknown; as such, data collection may have begun at slightly different times in the life-cycle for each colony. A bumble bee colony life-cycle includes a ‘growth phase’, when worker population size increases, and a ‘senescent phase’, when population size decreases (Cameron, 1989). Most colonies were early in their cycle, as all but three subsequently experienced a population size increase (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Average worker body size did not significantly change with colony age in Bombus impatiens, but larger colonies had on average smaller workers. Black filled circles represent the population size of workers in the colony; open circles represent the peak in colony population size; grey circles represent the average (±SE) body size of all workers present in the colony each week. Regression lines are present when the relationship between average worker size and colony age was significant (Bonferroni’s corrected α = 0.005 was used). Vertical lines represent the point at which the queen was no longer present in the colony.

These colonies were also used for other non-manipulative, observational studies. Therefore, the duration of time that data were collected on body size distribution varied between colonies. However, throughout these analyses reported here, no live worker bees were removed from the colonies. In some cases, the queen was removed once the colony began producing new reproductives, a signal that the colony was nearing the end of its annual cycle and entering into the senescent phase (Cameron, 1989). The point at which the colony became queenless is denoted in the figures.

Analyses

Statistics were done using Minitab (Student Version 14) and JMP (Version 7). Unless otherwise stated, averages are reported as mean ± standard error (X ± SE). When analysing all 10 colonies separately, we used a Bonferroni corrected α = 0.005 (P = 0.05, k = 10, P/k = 0.005) as our significance criterion. As each bee was individually marked and recorded daily, we knew which bees were alive and present in the colony each week. The degree to which the distribution of all adult worker body sizes per week was normal was analysed using a Shapiro–Wilk test.

Multiple linear regressions (MLR), with a block for colony-level effects, were used to determine whether average worker body size changed across colony age and/or different population sizes. The size of newly emerged bees (callows) per week was analysed separately. To understand within-colony patterns, linear regressions were calculated for each colony to determine whether average size of workers present in the colony and the size of newly emerging bees increased or decreased across colony age and/or different population sizes. To determine whether average size of all workers across colony age was better explained with a curvilinear fit, results from a second order regression were compared with those from the linear analyses. Furthermore, to determine whether there was a difference of worker body sizes present during the growth or senescent phase, we used linear regression to compare body size over time for each phase separately.

Finally, the standard deviation, degree of skewness, and degree of kurtosis were calculated to better understand the shape of the size distribution per week for each colony. The degree of skewness determines the presence and direction of a distribution’s tail, while kurtosis determines how steeply the distribution curve is peaked. Multiple linear regressions, with a block for colony-level effects, were used to determine whether the shape of the distribution (standard deviation, skewness, and kurtosis) changed across colony age and/or different population sizes. Where there were significant interactions between colony level effects and age or population size, separate linear regressions were calculated for each colony to identify within-colony patterns.

Results

Analyses of all bees in the colony

There were significant differences between colonies in average size of workers in the colony (F9,892 = 12.80, P < 0.001). Eight out of 10 colonies produced a normal worker body size distribution, as measured across the entire colony cycle (Table 1). This normality was maintained in 7 of the 10 colonies when worker sizes were examined weekly (Table 1). Colony populations were not manipulated, although steep declines in population size during the senescence phase of some colonies resulted in an overall significant population size decrease with colony age (see Fig. 1, black circles; MLR: r2 = 0.82; colony differences: F1,9 = 19.67; P < 0.001; colony age: F1,1 = 53.02; P < 0.001; colony × age: F1,9 = 8.34; P < 0.001). As a result of the strong differences between colonies, the effects of population size were analysed separately from colony age.

Table 1.

P-value results from Shapiro–Wilk tests of the null hypothesis that colonies produce a normal distribution of worker body sizes in the colony overall (‘All’), and maintain a normal distribution of worker body sizes within the colony each week.

Colony All Wk 1 Wk 2 Wk 3 Wk 4 Wk 5 Wk 6 Wk 7 Wk 8 Wk 9 Wk 10
F.05.01 0.251 0.183 0.247 0.260 0.221 0.321 0.405 0.511 0.548 1.00 1.00
F.05.02 0.827 0.152 0.210 0.339 0.365 0.696 0.968 0.871 0.989 0.920 0.920
F.05.03 0.049 0.698 0.231 0.110 0.209 0.454 0.488 0.462 0.028 0.062 0.885
S.06.01 0.043 0.575 0.324 0.169 0.029 0.026 0.012 0.121 0.168 0.333 0.305
S.06.02 0.0028 0.076 0.136 0.162 0.162 0.0004 0.006 0.004 0.006 0.013 0.013
S.06.03 0.0026 0.012 0.002 0.001 0.002 0.949 0.987 0.977 n/a n/a n/a
F.06.01 0.926 0.886 0.893 0.888 0.753 0.965 0.659 n/a n/a n/a n/a
F.06.02 0.0003 0.703 0.096 0.0006 0.0001 0.0001 0.0003 0.0005 0.009 0.061 n/a
F.06.03 0.463 0.259 0.640 0.507 0.657 0.634 0.675 0.423 0.384 0.706 n/a
F.06.04 0.807 0.647 0.691 0.573 0.899 0.899 0.857 0.698 0.773 0.168 0.438
Open in a new tab

P-values represent the probability that the distribution of worker body sizes was not normal. All 10 colonies were analysed separately; therefore, Bonferroni’s corrected α = 0.005 was used as the significance criterion. Significant P-values are in bold. Seven of the 10 colonies demonstrated an overall normal body size distribution and maintained such a distribution through the colony cycle. ‘N/A’ represents weeks after colony death until week 10 for four colonies.

Average body size of the workers present in the colony did not consistently increase or decrease with colony age (Fig. 1, grey circles; MLR: r2 = 0.11; colony differences: F1,9 = 45.75; P < 0.001; colony age: F1,1 = 1.96; P = 0.16; colony × age: F1,9 = 3.40; P < 0.001), yet different trends were observed across colonies. One colony showed an increase in body size (F.06.03: r2 = 0.02, F1,736 = 15.92, P < 0.001) and one showed a decrease in body size over time (F.05.02: r2 = 0.04, F1,344 = 15.17, P < 0.001). There was no change in average worker body size over time in the other eight colonies (P ≥ 0.05 for all other linear regressions). As there was no obvious monotonic trend, we repeated these analyses to determine if the data were more likely to follow a curvilinear fit (using quadratic regressions). In this case, there was again no significant relationship of worker body size with colony age in most (6 out of 10) colonies (F.05.02: r2 = 0.04, P = 0.001; S.06.01: r2 = 0.10, P < 0.001; F.06.02: r2 = 0.03, P < 0.001; F.06.03: r2 = 0.02, P < 0.001; P > 0.03 for all other colonies; see Fig. 1). In those colonies where the curvilinear fit was significant, the trend showed that body size tended to be lowest in the middle. As the curvilinear fit was no better at describing the data than the linear, we will focus on the linear regression in our discussion, as it is easier to interpret.

Changes in worker body size with colony age were also analysed separately during the growth and senescent phases of the colony cycle for each colony. The growth phase included the weeks until the peak in the population size (Fig. 1) and the senescent phase included the weeks after the peak in population size. There were no consistent trends across colonies with regards to the average body size during either phase (Table 2). In two colonies (S.06.01 and F.06.02), there was a decrease in average body size during the growth phase. In one colony, there was an increase in body size during the senescent phase (F.06.03), whereas in another there was a decrease (F.05.03). There were no significant trends in either phase in the other six colonies (Table 2).

Table 2.

Regression results for the relationship between worker body size and colony age (Tage) for each colony during their growth and senescence phases. Growth phase was defined as the time from week 1 until the peak in population size (represented in Fig. 1) and the senescence phase as the time that followed the peak in population size.

Growth phase Senescence phase
Colony N Wks Tage R2 P N Wks Tage R2 P
F.05.01 3 − 1.05 3.2% 0.300 8 0.96 2% 0.345
F.05.02 4 − 0.91 0.6% 0.364 7 − 2.39 2.9% 0.018
F.05.03 4 − 0.10 0.0% 0.919 7 − 3.01 5.2% 0.003
S.06.01 7 − 3.61 6.0% <0.001 4 2.07 7.3% 0.043
S.06.02 9 − 0.92 0.3% 0.356 3 − 0.00 0.0% 1.00
S.06.03 2 − 0.61 0.4% 0.545 5 1.04 1.1% 0.301
F.06.01 2 − 0.23 0.0% 0.822 4 0.12 0.0% 0.902
F.06.02 5 − 4.09 3.8% <0.001 4 2.33 1.7% 0.020
F.06.03 2 − 0.38 0.1% 0.703 7 3.41 2.1% 0.001
F.06.04 3 − 1.44 0.6% 0.151 8 − 0.07 0.0% 0.945
Open in a new tab

N Wks represents the number of weeks included in the analyses. All 10 colonies were analysed separately; therefore, Bonferroni’s corrected α = 0.005 was used as the significance criterion. Significant regression results are in bold.

Overall, average worker body size was lower with higher population size (Fig. 2; r2 = 0.89; colony differences: F1,9 = 45.41; P < 0.001; population size: F1,1 = 19.48; P < 0.001; colony × population size: F1,9 = 16.98; P < 0.001). However, when colonies were analysed individually, this was not consistent across colonies. There was a negative relationship between population size and average body size in three colonies (S.06.01: r2 = 0.92, F1,9 = 104.44, P < 0.001; S.06.02: r2 = 0.69, F1,10 = 22.54, P = 0.001; F.06.03: r2 = 0.97, F1,7 = 225.9, P < 0.001), a positive relationship in one colony (F.05.03: r2 = 0.76, F1,9 = 28.65, P < 0.001), and no relationship in the remaining six colonies (F.05.01: r2 = 0.11, F1,10 = 1.22, P = 0.30; F.05.02: r2 = 0.13, F1,9 = 1.38, P = 0.27; S.06.03: r2 = 0.38, F1,5 = 3.02, P = 0.14; F.06.01: r2 = 0.25, F1,4 = 1.36, P = 0.31; F.06.02: r2 = 0.68, F1,7 = 14.63, P = 0.006; F.06.04: r2 = 0.20, F1,9 = 2.25, P = 0.17).

Fig. 2.

Fig. 2

Open in a new tab

Average worker body size was overall negatively correlated with population size. However, this pattern was not particularly strong: different trends occurred across colonies. Regression lines are present when the relationship between average worker size and population size was significant for that colony (Bonferroni’s corrected α = 0.005 was used).

Analyses of newly emerging bees (callows)

Overall, the size of callows produced within a colony decreased over time, although this pattern was not consistent across colonies (Fig. 3; MLR: r2 = 0.20; colony differences: F1,9 = 5.05, P < 0.001; colony age: F1,1 = 4.62, P = 0.03; colony × age: F1,9 = 2.14, P = 0.03). There was a significant decrease in the average size of callows with colony age in only one colony when colonies were analysed separately (Fig. 3; S.06.01: r2 = 0.22, F1,41 = 9.34, P = 0.004), but no significant change in callow size in the other eight colonies (P > 0.02 for all remaining colonies). Additionally, there was no evidence that population size affected the average callow size (Fig. 4; r2 = 0.35; colony differences: F1,9 = 1.30; P = 0.27; population size: F1,1 = 0.10; P = 0.75; colony × age: F1,9 = 0.43; P = 0.91).

Fig. 3.

Fig. 3

Open in a new tab

Average body size of callows (newly emerging workers) decreased overall with colony age in Bombus impatiens. Black circles represent the population of all workers in the colony; grey circles represent the average (±SE) callow body size each week. Regression lines are present when the relationship between average callow size and colony age was significant for that colony (Bonferroni’s corrected α = 0.005 was used). Vertical lines represent the point at which the queen was no longer present in the colony.

Fig. 4.

Fig. 4

Open in a new tab

There was no relationship between average (±SE) body size of callows and population size, either overall or when colonies were analysed separately. Each point represents the average callow size for one colony in 1 week.

The shape of the worker size distribution

The degree of worker size polymorphism, measured here as the standard deviation of thorax sizes, varied across colonies and showed a general decrease with colony age (Fig. 5; MLR: r2 = 0.61, colony differences: F1,9 = 10.14, P < 0.001; colony age: F1,1 = 4.13, P = 0.05; colony × age: F1,9 = 0.91, P = 0.52); however, this decrease was only significant in 4 out of the 10 colonies (F.05.02: P = 0.006; F.05.03: P = 0.03; S.06.02: P < 0.001; F.06.03: P < 0.001; for all other colonies: P > 0.25). On the other hand, the standard deviation increased with population size (Fig. 5; r2 = 0.72; colony differences: F1,9 = 14.96, P < 0.001; population size: F1,1 = 7.44, P = 0.008; colony × population size: F1,9 = 3.46, P = 0.001). Again, however, when colonies were analysed separately, this positive relationship was only observed in two colonies (S.06.01: P = 0.001; F.06.03: P = 0.003); one colony exhibited a negative relationship (S.06.02: P = 0.001) and the other seven colonies showed no relationship using the Bonferroni-corrected α = 0.005 (F.05.01: P = 0.88; F.05.02: P = 0.04; F.05.03: P = 0.006; S.06.03: P = 0.06; F.06.01: P = 0.48; F.06.02: P = 0.02; F.06.04: P = 0.03).

Fig. 5.

Fig. 5

Open in a new tab

The shape of the worker size distribution (in terms of standard deviation, skewness, and kurtosis) with colony age (left column) and with population (right column). Each symbol/colour combination represents a different colony.

The degree of skewness also varied across colonies and became more negative over time (Figs 5 and 6; r2 = 0.83; colony differences: F1,9 = 28.49, P < 0.001; colony age: F1,1 = 7.45, P = 0.008; colony × age: F1,9 = 11.88, P < 0.001), although this negative relationship was only observed in one colony when analysed separately (see Fig. 6: F.05.01: P = 0.003). In one colony there was a positive relationship (S.06.01: P = 0.004) and in the remaining eight colonies, there was no relationship (F.05.02: P = 0.77; F.05.03: P = 0.15; S.06.02: P = 0.51; S.06.03: P = 0.08; F.06.01: P = 0.04; F.06.02: P = 0.43; F.06.03: P = 0.03; F.06.04: P = 0.03). The distribution tended to be more positively skewed in larger population sizes (r2 = 0.76; colony differences: F1,9 = 16.14, P < 0.001; population size: F1,1 = 4.89, P = 0.03; colony × population size: F1,9 = 6.34, P < 0.001), although this held true for only two colonies (F.06.01: P = 0.002; F.06.03: P = 0.005). One colony showed a negative relationship (F.06.04: P = 0.001) and the remaining seven colonies showed no relationship at all (F.05.01: P = 0.03; F.05.02: P = 0.02; F.05.03: P = 0.19; S.06.01: P = 0.28; S.06.02: P = 0.69; S.06.03: P = 0.007; F.06.02: P = 0.74).

Fig. 6.

Fig. 6

Open in a new tab

Thorax width frequency distribution across representative colonies with varying patterns of change in skew (decrease: F.05.01; no change: F.05.03; or increase: S.06.01 in skew).

Finally, there was no evidence to suggest that the degree of kurtosis changed over time (r2 = 0.35; colony differences: F1,9 = 1.97, P = 0.06; colony age: F1,1 = 1.56, P = 0.22; colony × age: F1,9 = 2.41, P = 0.02) or as the population size increased (r2 = 0.46; colony differences: F1,9 = 2.23, P = 0.03; population size: F1,1 = 0.95, P = 0.33; colony × population size: F1,9 = 4.29, P < 0.001).

Discussion

In the present study it was shown that the average size of worker bumble bees present in the colony, while different between colonies, does not consistently increase or decrease as the colony ages from its growth to its senescent phases. Overall, the worker sizes in a colony form a normal distribution and colonies exhibit significant variation in worker body size despite a standardised feeding regime. Although colony size was not manipulated to be independent from colony age, we found a statistical decrease in average worker size with increasing population size. However, this pattern was not consistent across colonies and different sizes of workers are produced throughout the colony cycle. Interestingly, the amount of variation among workers seemed to decrease with colony age, rather than increase, as is often the case for worker-polymorphic ant colonies. Similarly to ant colonies, however, variation may increase with population size. We conclude that worker size polymorphism in bumble bees is not necessarily a result of different-sized bees being produced at different stages in the colony life cycle, nor is food limitation a necessary factor in creating size polymorphism. This study is, to our knowledge, the first to describe this ontogeny of the worker body size distribution in a bumble bee colony.

Variation among workers in body size was maintained even though colonies were fed sugar and pollen ad libitum. This demonstrates that the maintenance of worker body size polymorphism is independent of variation in or limitation of available resources at the colony level, suggesting it may be adaptive. Colonies may be under selection to maintain worker polymorphism in B. impatiens. Alternatively, lack of effective organisational mechanisms to ensure equal care for all larvae may lead to worker polymorphism in bumble bees (Couvillon & Dornhaus, 2009).

It is also shown here that small colony size does not limit production of workers of a particular size; the average size of workers produced even decreased with colony population size in some colonies. This shows that production of small workers is not a result of an insufficient foraging force. Younger and larger colonies tended to show a higher degree of polymorphism among workers (higher standard deviation of average worker body size). These results are surprising as they run contrary to what is usually found in ants. Worker size polymorphism evolved many times within Formicidae (Davidson, 1978), and it has been frequently found that the smaller-sized workers of a new colony give way to the larger, and sometimes more polymorphic, workers of a mature colony (Brian, 1952; Oster & Wilson, 1978; Herbers, 1980; Porter & Tschinkel, 1986). This was observed in ants of the genus Pogonomyrmex badius, for example, where not only do majors solely emerge late in the colony cycle, the average minor worker size also increases (Tschinkel, 1998). In contrast, no evidence was found that B. impatiens produce larger workers as the colonies grows or ages. Colonies may even contain more polymorphic workers when they are younger. In ants, there is a positive correlation between colony size and worker size in Myrmecia (Gray, 1971), Pogonomyrmex (Porter & Tschinkel, 1986), Solenopsis (Markin et al., 1973; Tschinkel, 1988), and Atta (Wilson, 1983). Lastly, while worker size distribution in some ants might become more positively skewed over time (Wood & Tschinkel, 1981; Tschinkel, 1988; Wheeler, 1991), we found here that in bumble bees the distribution may become more negatively skewed as the colony ages.

Bumble bee colonies are annually founded by a queen that is typically singly-mated (Schmid-Hempel & Schmid-Hempel, 2000). The ultimate size that an adult worker reaches is not only determined by genetic factors, but by how much food is given to the developing larvae by adult nurses (Michener, 1974; Alford, 1975; Couvillon & Dornhaus, 2009). A normal colony cycle typically runs for several months, with the queen initially performing all colony tasks (nursing, laying, and foraging) until a small cohort of workers emerges (Sladen, 1912; Goulson, 2003). It is sometimes assumed that the first worker cohorts are significantly smaller than the bees produced later in the colony cycle, as resource abundance and population (i.e. available foragers) increase from spring to summer (Knee & Medler, 1965). Afterwards, the queen devotes herself to laying and the colony experiences a population explosion, where worker numbers will go from a few to possibly several hundred (Sladen, 1912; Cameron, 1989). Previous authors have suggested that worker body size increases over time in some bumble bee species (B. auricomus, B. griseocollis, B. fervidus, and B. perplexus) (Sladen, 1912; Frison, 1927; Knee & Medler, 1965; Plowright & Jay, 1968); however, these data were anecdotal in nature and sometimes included examples to the contrary (B. terricola, and B. ternarius) (Goulson, 2003).

How might bumble bees maintain body size polymorphism within individual colonies? Worker size polymorphism in bumble bees has been described previously (Knee & Medler, 1965; Plowright & Jay, 1968); however, most of this did not quantify size distributions. More recently, others have examined the size variation within (Goulson, 2003; Peat et al., 2005b) and between (Peat et al., 2005a) species, but this was less on the ontogeny of the distribution and more on the role of size polymorphism in the division of labour. Previous work has suggested that size variation in ants might be a result of limited food availability (Herbers, 1980; Rissing, 1987), as colonies, when resources are not limited, produce overall larger workers and fewer small ones. In contrast, we show here that in bumble bees, worker size variation is still present with unlimited resources. It is likely that the in-nest spatial organisation of bumble bee workers and larvae is the proximate mechanism maintaining variation in worker body size. This is because workers active as nurses tend to be found more in the centre of the nest, and larvae developing there receive more food, and thus develop into larger workers, whereas larvae in the periphery are fed less and develop into smaller adults (Couvillon & Dornhaus, 2009; Jandt & Dornhaus, 2009). The spatial organisation of workers directly leads to variation in feeding rate over the nest surface and thus to worker size polymorphism (Couvillon & Dornhaus, 2009).

Why might bumble bees maintain a normal distribution of worker body sizes within individual colonies, instead of the highly skewed distribution usually seen in ants? Larger workers tend to be foragers and guards, and smaller workers tend to be nurses in bumble bee colonies (Jandt & Dornhaus, 2009), and one may expect that worker bees perform the tasks they are most suited for. Indeed, larger worker bees have been shown to be adept at foraging-related activities: they can bring back more nectar per time (Goulson et al., 2002; Spaethe & Weidenmuller, 2002), are superior thermoregulators (Free & Butler, 1959; Heinrich, 1979), can see (Spaethe & Chittka, 2003) and smell (Spaethe et al., 2007) better, and might learn better and remember longer (Worden et al., 2005, although see Raine et al., 2006; Raine & Chittka, 2008), all of which supports the hypothesis that size polymorphism is an adaptation for division of labour. There is, however, also evidence to suggest that larger bees are also better nurses, a task usually done by smaller workers (Cnaani & Hefetz, 1994), which is at variance with the hypothesis that size polymorphism is an adaptation for division of labour. However, if the large bees are simply higher-quality workers, then we would predict under ad libitum food conditions that colonies should rear disproportionately more large bees. In that case, we would expect to see a decrease in the variation of thorax widths when resources are not limited, because colonies are producing less small workers or an increase in the distribution skew, because colonies are rearing larger workers. We show here that worker size variation persists in B. impatiens even in the most resource-rich environments and the distribution skew is more likely to decrease, if it changes at all. Therefore, although the adaptive value of the production of bumble bee worker body size variation itself for division of labour remains unresolved, our results suggest that the maintenance of this variation may either be adaptive at the colony level for some yet undefined reason (Couvillon & Dornhaus, 2010) or it could simply be an epiphenomenon of unequal larval feeding within the nest. It is thus possible that the adaptive function of size polymorphism in bumble bees is quite different from that in ants.

Are there other adaptive explanations for size polymorphism that are independent of division of labour? It has been suggested that intra-colony variation might allow colonies to adapt more quickly to environmental conditions, as larger bees might forage better in cooler temperatures and smaller bees in warmer temperatures (Peat et al., 2005a). However, although this has been applied to inter-specific variation in other organisms (Bergmann, 1847; Stillwell et al., 2007), it remains unsupported in bumble bees (Couvillon et al., 2010; Peat et al., 2005a). Rather, perhaps size polymorphism in bumble bees might result from the trade-off of cheap, lower-quality workers and expensive, higher-quality workers. Future work should focus on cost-benefit ratio of different sized bees. Perhaps colonies with smaller bees actually consume less energy overall, although they are less efficient at collecting and processing that energy. Finally, workers actually perform a variety of tasks inside the nest in addition to feeding and warming the brood; however, these tasks are traditionally not included in division of labour analysis. It would be interesting to investigate the efficiency of smaller workers performing these other tasks to give a better understanding of how and why bumble bees produce polymorphic workers.

References

  1. Alford DV. Bumblebees. London, UK: Davis-Poynter; 1975. [Google Scholar]
  2. Baroni Urbani C. The number of castes in ants, where major is smaller than minor and queens wear the shield of the soldiers. Insectes Sociaux. 1998;45:315–333. [Google Scholar]
  3. Ben-Shahar Y, Robichon A, Sokolowski MB, Robinson GE. Influence of gene action across different time scales on behavior. Science. 2002;296:741–744. doi: 10.1126/science.1069911. [DOI] [PubMed] [Google Scholar]
  4. Bergmann C. Ueber die verhältnisse der wärmeökonomie der thiere zu ihrer grösse. Gottinger Studien. 1847;3:595–708. [Google Scholar]
  5. Beshers SN, Fewell JH. Models of division of labor in social insects. Annual Review of Entomology. 2001;46:413–440. doi: 10.1146/annurev.ento.46.1.413. [DOI] [PubMed] [Google Scholar]
  6. Bonabeau E, Doringo M, Theraulaz G. Swarm Intelligence: From Natural to Artificial Systems. New York, New York: Oxford University Press; 1999. [Google Scholar]
  7. Brian AD. Division of labour and foraging in Bombus-agrorum Fabricius. Journal of Animal Ecology. 1952;21:223–240. [Google Scholar]
  8. Camazine S, Deneubourg J-L, Franks N, Sneyd J, Theraulaz G, Bonabeau E. Self-Organization in Biological Systems. Princeton, New Jersey: Princeton University Press; 2001. [Google Scholar]
  9. Cameron SA. Temporal patterns of division of labor among workers in the primitively eusocial bumble bee, Bombus griseocollis (Hymenoptera: Apidae) Ethology. 1989;80:137–151. [Google Scholar]
  10. Cnaani J, Hefetz A. The Effect of workers size frequency-distribution on colony development in Bombus-terrestris. Insectes Sociaux. 1994;41:301–307. [Google Scholar]
  11. Couvillon MJ, Dornhaus A. Location, location, location: larvae position inside the nest is correlated with adult body size in worker bumble bees (Bombus impatiens) Proceedings of the Royal Society B. 2009;276:2411–2418. doi: 10.1098/rspb.2009.0172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Couvillon MJ, Dornhaus A. Small worker bumble bees (Bombus impatiens) are hardier against starvation than their larger sisters. Insectes Sociaux. 2010;57:193–197. doi: 10.1007/s00040-010-0064-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Couvillon MJ, Fitzpatrick G, Dornhaus A. Ambient air temperature does not predict body size of foragers in bumble bees (Bombus impatiens) Psyche. 2010 doi: 10.1155/2010/536430. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Cumber RA. The biology of humble-bees, with special reference to the production of the worker caste. Transactions of the Royal Entomological Society of London. 1949;100:1–45. [Google Scholar]
  15. Davidson DW. Size variability in the worker caste of a social insect (Veromessor pergandei Mayr) as a function of the competitive environment. The American Naturalist. 1978;112:523–532. [Google Scholar]
  16. Farris SM, Robinson GE, Fahrbach SE. Experience-and age-related outgrowth of intrinsic neurons in the mushroom bodies of the adult worker honeybee. Journal of Neuroscience. 2001;21:6395–6404. doi: 10.1523/JNEUROSCI.21-16-06395.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fewell JH. Social insect networks. Science. 2003;301:1867–1870. doi: 10.1126/science.1088945. [DOI] [PubMed] [Google Scholar]
  18. Fjerdingstad EJ, Crozier RH. The Evolution of worker caste diversity in social insects. The American Naturalist. 2006;167:390–400. doi: 10.1086/499545. [DOI] [PubMed] [Google Scholar]
  19. Free JB, Butler CG. Bumblebees. London, UK: Collins; 1959. [Google Scholar]
  20. Frison TH. The development of the castes of bumblebees (Bremidae: Hym.) Annals of the Entomological Society of America. 1927;20:156–178. [Google Scholar]
  21. Goulson D. Bumblebees - Behaviour and Ecology. Oxford, UK: Oxford University Press; 2003. [Google Scholar]
  22. Goulson D, Peat J, Stout JC, Tucker J, Darvill B, Derwent LC, et al. Can alloethism in workers of the bumblebee, Bombus terrestris, be explained in terms of foraging efficiency? Animal Behaviour. 2002;64:123–130. [Google Scholar]
  23. Gray B. A morphometric study of the ant species, Myrmecia dispar. Insectes Sociaux. 1971;18:95–110. [Google Scholar]
  24. Heinrich B. Bumblebee Economics. Cambridge, Massachusetts: Harvard University Press; 1979. [Google Scholar]
  25. Herbers JM. On caste ratios in ant colonies: population responses to changing environments. Evolution. 1980;34:575–585. doi: 10.1111/j.1558-5646.1980.tb04845.x. [DOI] [PubMed] [Google Scholar]
  26. Hölldobler B, Wilson EO. The Ants. Cambridge, Massachusetts: Belknap Press; 1990. [Google Scholar]
  27. Hölldobler B, Wilson EO. The Super organism: The Beauty, Elegance, and Strangeness of Insect Societies. New York, New York: W.W. Norton & Co.; 2008. [Google Scholar]
  28. Jandt J, Dornhaus A. Spatial organization and division of labor in the bumble bee, Bombus impatiens. Animal Behaviour. 2009;77:641–651. [Google Scholar]
  29. Knee WJ, Medler JT. Seasonal size increase of bumblebee workers (Hymenoptera - Bombus) Canadian Entomologist. 1965;97:1149–1155. [Google Scholar]
  30. Lindauer M. Division of labour in the honeybee colony. Bee World. 1953;34:63–90. [Google Scholar]
  31. Markin GP, Dillier JH, Collins HL. Growth and development of colonies of the red imported fire ant, Solenopsis invicta. Annals of the Entomological Society of America. 1973;66:803–808. [Google Scholar]
  32. Maynard Smith J, Szathmáry E. The Major Transitions in Evolution. Oxford, UK: Oxford University Press; 1995. [Google Scholar]
  33. Michener CD. The Social Behaviour of the Bees. Cambridge, Massachusetts: Harvard University Press; 1974. [Google Scholar]
  34. Noirot C, Pasteels JM. Ontogenetic development and evolution of the worker caste in termites. Cellular and Molecular Life Sciences. 1987;43:851–952. [Google Scholar]
  35. Oster GF, Wilson EO. Caste and Ecology in the Social Insects. Princeton, New Jersey: Princeton University Press; 1978. [PubMed] [Google Scholar]
  36. Page RE, Mitchell SD. Self-organization and the evolution of division of labor. Apidologie. 1998;29:171–190. [Google Scholar]
  37. Peat J, Darvill B, Ellis J, Goulson D. Effects of climate on intra- and interspecific size variation in bumble-bees. Functional Ecology. 2005a;19:145–151. [Google Scholar]
  38. Peat J, Tucker J, Goulson D. Does intraspecific size variation in bumblebees allow colonies to efficiently exploit different flowers? Ecological Entomology. 2005b;30:176–181. [Google Scholar]
  39. Plowright RC, Jay SC. Caste differentiation in bumblebees (Bombus Latr - Hym). 1. Determination of female size. Insectes Sociaux. 1968;15:171–192. [Google Scholar]
  40. Porter SD, Tschinkel WR. Adaptive value of nanitic workers in newly founded red imported fire ant colonies (Hymenoptera: Formicidae) Annals of the Entomological Society of America. 1986;79:723–726. [Google Scholar]
  41. Powell S. Ecological specialization and the evolution of a specialized caste in Cephalotes ants. Functional Ecology. 2008;22:902–911. [Google Scholar]
  42. Raine NE, Chittka L. The correlation of learning speed and natural foraging success in bumble-bees. Proceedings of the Royal Society B. 2008;275:803–808. doi: 10.1098/rspb.2007.1652. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Raine NE, Ings TC, Ramos-Rodriguez O, Chittka L. Intercolony variation in learning performance of a wild British bumble bee population (Hymenoptera: Apidae: Bombus terrestris audax) Entomologia Generalis. 2006;28:241–256. [Google Scholar]
  44. Richards OW. Observations on Bombus agrorum (Fabricius) (Hymen Bombindae) Proceedings of the Royal Entomological Society of London. 1946;21:66–71. [Google Scholar]
  45. Rissing SW. Annual cycles in worker size of the seedharvester ant Veromessor pergandei (Hymenoptera: Formicidae) Behavioral Ecology and Sociobiology. 1987;20:117–124. [Google Scholar]
  46. Robinson GE. Regulation of honey bee age polyethism by juvenile hormone. Behavioral Ecology and Sociobiology. 1987;20:329–338. [Google Scholar]
  47. Robinson GE, Page RE, Huang ZY. Temporal polyethism in social insects is a developmental process. Animal Behaviour. 1994;48:467–469. [Google Scholar]
  48. Schmid-Hempel R, Schmid-Hempel P. Female mating frequencies in Bombus spp. from central Europe. Insectes Sociaux. 2000;47:36–41. [Google Scholar]
  49. Seeley TD. Adaptive significance of the age polyethism schedule in honeybee colonies. Behavioural Ecology and Sociobiology. 1982;11:287–293. [Google Scholar]
  50. Sendova-Franks AB, Franks NR. Self-assembly, self-organization and division of labour. Philosophical Transactions of the Royal Society B: Biological Sciences. 1999;354:1395–1405. [Google Scholar]
  51. Sladen FWL. The Humble-Bee. London, UK: Macmillan and Co.; 1912. [Google Scholar]
  52. Spaethe J, Chittka L. Interindividual variation of eye optics and single object resolution in bumblebees. Journal of Experimental Biology. 2003;206:3447–3453. doi: 10.1242/jeb.00570. [DOI] [PubMed] [Google Scholar]
  53. Spaethe J, Weidenmuller A. Size variation and foraging rate in bumblebees (Bombus terrestris) Insectes Sociaux. 2002;49:142–146. [Google Scholar]
  54. Spaethe J, Brockmann A, Halbig C, Tautz J. Size determines antennal sensitivity and behavioral threshold to odors in bumblebee workers. Naturwissenschaften. 2007;94:733–739. doi: 10.1007/s00114-007-0251-1. [DOI] [PubMed] [Google Scholar]
  55. Stern DL, Foster WA. The evolution of sociality in aphids: a clone’s-eye view. In: Choe JC, Crespi BJ, editors. Social Behavior in Insects and Arachnids. Cambridge, UK: Cambridge University Press; 1997. pp. 150–165. [Google Scholar]
  56. Stillwell RAC, Morse GAE, Fox CA. Geographic variation in body size and sexual size dimorphism of a seed-feeding beetle. The American Naturalist. 2007;170:358–369. doi: 10.1086/520118. [DOI] [PubMed] [Google Scholar]
  57. Tschinkel WR. Colony growth and the ontogeny of worker polymorphism in the fire ant, Solenopsis invicta. Behavioural Ecology and Sociobiology. 1988;22:103–115. [Google Scholar]
  58. Tschinkel WR. Sociometry and sociogenesis of colonies of the harvester ant, Pogonomyrmex badius: worker characteristics in relation to colony size and season. Insectes Sociaux. 1998;45:385–410. [Google Scholar]
  59. Wheeler DE. The developmental basis of worker caste polymorphism in ants. The American Naturalist. 1991;138:1218–1238. [Google Scholar]
  60. Whitfield CW, Cziko A-M, Robinson GE. Gene expression profiles in the brain predict behavior in individual honey bees. Science. 2003;302:296–299. doi: 10.1126/science.1086807. [DOI] [PubMed] [Google Scholar]
  61. Wilson EO. Caste and division of labor in leaf-cutter ants (Hymenoptera: Formicidae: Atta): III. Ergonomic resiliency in foraging by A. cephalotes. Behavioral Ecology and Sociobiology. 1983;14:47–54. [Google Scholar]
  62. Wood L, Tschinkel W. Quantification and modification of worker size variation in the fire ant, Solenopsis invicta. Insectes Sociaux. 1981;28:117–128. [Google Scholar]
  63. Worden BD, Skemp AK, Papaj DR. Learning in two contexts: the effects of interference and body size in bumblebees. Journal of Experimental Biology. 2005;208:2045–2053. doi: 10.1242/jeb.01582. [DOI] [PubMed] [Google Scholar]
  64. Yerushalmi S, Bodenhaimer S, Bloch G. Developmentally determined attenuation in circadian rhythms links chronobiology to social organization in bees. Journal of Experimental Biology. 2006;209:1044–1051. doi: 10.1242/jeb.02125. [DOI] [PubMed] [Google Scholar]

RESOURCES