Spatial organization and division of labour in the bumblebee Bombus impatiens

Abstract

Individuals in many types of animal groups show both reproductive and task-related division of labour. In some social insect species, such division of labour may be related to the spatial organization of workers inside the nest. We examined colonies of bumblebees and found that (1) 11-13% of workers maintained small spatial fidelity zones inside the nest, and all workers tended to remain at a specific distance from the colony centre independent of their age; (2) smaller individuals maintained smaller spatial zones and tended to be closer to the centre; and (3) individuals that were more likely to perform the in-nest task of larval feeding tended to remain in the centre of the nest, whereas foragers were more often found on the periphery of the nest when not foraging. Individuals that performed other tasks did not maintain a predictable distance to the centre, and there was no evidence that spatial preferences changed over time. Instead, spatial patterns may result from inherent differences between individuals in terms of activity level, and may be a self-organized sorting mechanism that influences division of labour among workers.

Full text

Spatial organization and division of labour in the bumblebee Bombus impatiens
Jennifer M. Jandt
*
, Anna Dornhaus
Department of Ecology & Evolutionary Biology, University of Arizona
article info
Article history:
Received 11 March 2008
Initial acceptance 19 June 2008
Final acceptance 21 November 2008
Published online 10 January 2009
MS. number: A08-00164R
Keywords:
Bombus impatiens
bumblebee
division of labour
self-organization
spatial assortment
Individuals in many types of animal groups show both reproductive and task-related division of labour.
In some social insect species, such division of labour may be related to the spatial organization of
workers inside the nest. We examined colonies of bumblebees and found that (1) 11–13% of workers
maintained small spatial fidelity zones inside the nest, and all workers tended to remain at a specific
distance from the colony centre independent of their age; (2) smaller individuals maintained smaller
spatial zones and tended to be closer to the centre; and (3) individuals that were more likely to perform
the in-nest task of larval feeding tended to remain in the centre of the nest, whereas foragers were more
often found on the periphery of the nest when not foraging. Individuals that performed other tasks did
not maintain a predictable distance to the centre, and there was no evidence that spatial preferences
changed over time. Instead, spatial patterns may result from inherent differences between individuals in
terms of activity level, and may be a self-organized sorting mechanism that influences division of labour
among workers.
The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
Many animal groups can be described as complex biological
systems whose components are capable of dividing tasks among
one another in the absence of a central leader, a process described
as self-organization (Beshers & Fewell 2001; Camazine et al. 2001).
Division of labour occurs when individuals specialize on different
colony tasks, such as brood care, foraging or colony defence
(reviewed in Beshers & Fewell 2001). While different mechanisms
that may create a division of labour have been identified, it is
unclear whether there is a more basic property of social groups that
underlies organization across species. In social insects, for example,
individuals within some groups organize labour through domi-
nance interactions (e.g. paper wasp foundresses: West-Eberhard
1969), whereas others divide labour by age, such that there is
a sequential order of tasks correlated with an age-dependent
hormonal change within each individual (e.g. honeybees: Seeley
1982; Robinson 1987, 1992; Robinson et al. 1989). In other groups,
however, task allocation may be based on worker size: the larger
‘soldiers’ defend the colony while the smaller workers perform the
remaining tasks (e.g. army ants: Powell & Franks 2006; termites:
Roux & Korb 2004).
In bumblebees, dominance (measured as ovarian development)
seems to have little effect on division of labour among workers in
a queenright colony (Bombus terrestris:Duchateau & Velthuis 1989;
Geva et al. 2005). There is also little evidence of an age-sequential
pattern of task switching (B. impatiens and B. bimaculatus:Cameron
& Robinson 1990), although younger bees may be more likely to
care for brood (B. griseocollis:Cameron 1989;B. bifarius:O’Donnell
et al. 2000;B. terrestris:Yerushalmi et al. 2006). Instead, it is widely
accepted that in bumblebees, division of labour may depend more
on worker body size (Brian 1952; Free 1955; Goulson et al. 2002;
Goulson 2003; Yerushalmi et al. 2006). Bumblebees show a wide
range of body sizes among workers, and larger individuals are often
better at foraging and more likely to forage (Alford 1978; Goulson
et al. 2002; Spaethe & Weidenmu
¨ller 2002; Foster et al. 2004;
Worden et al. 2005), whereas smaller individuals are more often
observed remaining inside the nest throughout their lifetime (Brian
1952; Goulson et al. 2002; Heinrich 2004; Yerushalmi et al. 2006).
However, there is little evidence for further division of labour based
on body size for those specific tasks found inside the nest, such as
incubating brood or fanning (Foster et al. 2004; Weidenmu
¨ller
2004). Therefore there must be some other mechanism that fine-
tunes the division of labour in bumblebees. We propose that the
spatial sorting of individuals inside the nest may, to some degree,
play a role.
Spatial sorting of individuals within a group correlates with
division of labour in other types of social insect (ants: Odonto-
machus brunneus:Powell & Tschinkel 1999;Temnothorax uni-
fasciatus:Sendova-Franks & Franks 1995;Temnothorax albipennis:
Backen et al. 2000;Pheidole dentata:Wilson 1976; wasps: Ropalidia
marginata:Robson et al. 2000; bees: Apis mellifera:Seeley 1982).
Individuals within a social insect colony that tend to return to
*Correspondence: J. M. Jandt, Department of Ecology & Evolutionary Biology,
P.O. Box 210088, University of Arizona, Tucson, AZ 85721, U.S.A.
E-mail address: jandt@email.arizona.edu (J.M. Jandt).
Contents lists available at ScienceDirect
Animal Behaviour
journal homepage: www.elsevier.com/locate/yanbe
0003-3472/$38.00 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd.
doi:10.1016/j.anbehav.2008.11.019
Animal Behaviour 77 (2009) 641–651

particular areas of the nest are said to maintain ‘spatial fidelity
zones’ (Sendova-Franks & Franks 1995). By remaining in
nonrandom areas of the nest, workers may be able to minimize the
distance moved between tasks, thereby increasing overall effi-
ciency at the colony level (Wilson 1976; Seeley 1982).
Unlike nests of honeybees (Camazine 1991), ants (Franks &
Sendova-Franks 1992) or paper wasps (Karsai & Pe
´nzes 1998),
bumblebee nests are not arranged in a predictable pattern
(Cameron 1989). Egg clumps (3–4 eggs each) are laid in all
regions of the nest. Therefore, the centre of the distribution of
workers in the colony cannot be arranged around a single ‘brood
pile’ or brood area, as it is in other species. Because of the small
nest area and these interspersed nest characteristics, in-nest
space use and its relationship to division of labour has not been
investigated in bumblebees (Cameron 1989). We propose,
however, that bumblebee colonies allow us a unique opportunity
to compare individual space use with other species whose nests
show more predictable patterns of organization. For example, if
spatial organization arises because individuals are orienting
around particular nest characteristics, as may be the case in
honeybees (Seeley 1982), we may expect bumblebees to divide
space into small, concentrated zones throughout the nest. Alter-
natively, if spatial segregation is limited by the size of the nest
(Cameron 1989), we should not expect to see any spatial
patterning emerge, nor should we observe correlations between
space use and task. Finally, if inherent individual differences affect
space use, as in Temnothorax ants (Backen et al. 2000), given the
lack of evidence for age polyethism in bumblebees, we would
expect spatial patterns that differ between individuals to be
consistent over time.
We performed an in-depth analysis of the spatial arrangement
of individuals within bumblebee colonies (B. impatiens) to examine
how these spatial patterns relate to division of labour. Our goals
were to determine (1) whether individual bumblebees maintain
spatial fidelity zones inside the nest, (2) whether use of spatial
zones can be predicted by age or body size of individuals and (3)
how space use correlates with division of labour (i.e. whether tasks
are zone specific).
METHODS
We purchased four queenright colonies of Bombus impatiens
through Koppert Biological Systems (Romulus, MI, U.S.A.). Colony A
was set up on 7 August 2006, while colonies B, C and D were
established on 28 August 2006. Each colony was placed inside
a wooden nestbox (13 13 8 cm) that was immediately adjacent
to an outer box (13 13 8 cm), which was connected to
a foraging arena (70 6040 cm; for colony B: 20 20 8 cm)
via 2.5 cm (ID) Tygon tubing. Both boxes contained a 3 cm layer of
Feline Pine
Ò
cat litter (Nature’s Earth Products, Inc., West Palm
Beach, FL, U.S.A.) to reduce moisture build-up inside the colony. All
boxes were covered with glass to provide constant viewing of in-
nest behaviour (bumblebees habituate quickly to light in their
nest). All colonies were kept under a 10:14 h light:dark cycle. While
the lights were on, we did not observe bees flying inside the nest or
behaving in a way that would suggest that the glass cover affected
in-nest worker behaviour.
Bees had access to the foraging arena at all times. In the arena,
they were provided with sugar solution (Bee-Happy, Koppert Bio-
logical Systems) and pollen (obtained from Apis mellifera colonies
and ground to a powder) ad libitum. Food was replenished daily
after spatial observations for that day were completed to ensure
that treatment was consistent across colonies and that spatial
arrangements of individuals were collected in the absence of a rush
of activity towards new food.
All individuals in each colony were marked on the thorax with
a colour and number-coded plastic tag (Queen Marking Numbers,
Mann Lake Ltd, Hackensack, MN, U.S.A.). Newly emerged adult bees
were marked on the same day, and their date of emergence (eclo-
sion) recorded. Marking took place after all data collection was
completed for that day.
All colonies adjusted to the laboratory conditions except for
colony C. On day 8 of data collection, workers began constructing
honeypots and laying male eggs in the feeding arena, thus leaving
the original nest. Many workers never returned to the nestbox
where the queen remained. Because of this behaviour, there is
reason to suspect that spatial arrangement of the few remaining
individuals inside the nest may not be truly representative of what
would be observed in a ‘normal’ colony. Therefore, we do not
present results from colony C here.
Data were collected for the entire duration that the queen was
present (colony A: 24 days; colony B: 32 days; colony D: 49 days)
and on all individuals that emerged during that time (colony A: 94
bees; colony B: 90 bees; colony D: 154 bees).
Recording Positions of Individual Workers
Colonies were allowed 2 weeks after set-up to adjust to labo-
ratory conditions before data collection began. Spatial data were
collected daily between 0700 and 1000 hours, 4–6 times per week.
A2525 cm grid divided into 100 square grid cells was placed
over the glass top of the nestbox so that the Xand Ycoordinates of
each individual, including the queen, could be recorded. Bees
foraging in the arena during data collection were recorded as such.
Not all bees were recorded every day, as some were hidden under
the nest substrate.
To account for changes in nest structure, once a week we
recorded the dominant nest character present in each grid cell:
brood (including egg and larval clumps), pupae, honeypots or
nothing (i.e. off nest substrate). The grid cells in the centre of two
sides of the nestbox were designated as ‘window’ cells because
a screened opening for airflow was located on the wall in these
areas.
Comparing Spatial Zone Size to Random Expectation
The centroid of the coordinate recordings for each individual
bee was calculated (i.e. the averages for the Xand Ycoor-
dinates SD). As a measure of the size of the area used by each bee,
we multiplied double the standard deviations of those Xand Y
averages, yielding the area of a rectangle around the centroid of
each bee’s distribution. This will be called ‘spatial zone size’. We
used this method as opposed to calculating a polygon or convex
hull because it takes into account the frequency with which
different areas are used (Fig. 1).
We used a Monte Carlo simulation to determine the expected
spatial zone size for each bee to test the null hypothesis that all bees
have the same spatial preferences in the nest. We therefore
calculated expected spatial zone sizes under the assumption that
every day all bees chose their locations randomly from the overall
distribution of bees on the nest. This way we could determine
whether each bee’s actual spatial zone size was larger or smaller
than her expected spatial zone size. To calculate a focal bee’s
expected spatial zone size, a random set of coordinates were picked
from individuals recorded on each of the same days as the focal bee.
The standard deviations of those randomly selected Xand Ycoor-
dinates were multiplied to calculate a random spatial zone size. For
each bee, this process was repeated until 1000 random spatial zone
sizes were calculated. Bees whose actual spatial zone size was
smaller than 95% of all of the random spatial zone sizes were
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651642

considered to have zones that were significantly smaller than
expected (P<0.05). On the other hand, individuals whose zones
were larger than 95% of all the expected spatial zone sizes were
considered to have zones that were significantly larger than
random (P<0.05; see Fig. 1). By calculating spatial zones in this
way, we may have, in some cases, overestimated the observed size,
and consequently, underestimated the total proportion of spatial
zones smaller than random within each colony. This conservative
method provides additional assurance in the results on those
individuals whose zones we had determined were smaller than
random.
Calculating Distance to In-nest Reference Points
We also investigated space use by calculating an individual’s
distance to particular reference points inside the nestbox on each
day: (1) the colony centre, defined as the centre of the current
distribution of workers, (2) the location of the queen, (3) the nest
entrance (a fixed point) and (4) the bee’s previous day’s coordinate.
The colony centre was recalculated for each day by averaging all X
and Ycoordinates of all individuals recorded on that day. It is thus
a measure of the centre of the overall distribution of individuals,
not of the physical nest structure. The location of the queen was
determined by her Xand Ycoordinates recorded on that day. The
nest entrance was a fixed point in the nestbox where individuals
could enter. Finally, we measured an individual’s movement over
time by comparing its location on a given day to that from the
previous day. Only data recorded on individuals that were inside
the nest on 2 consecutive days were included in this analysis. We
also determined what nest character (i.e. brood or honeypots, etc.)
each individual was on when she was recorded each day.
Recording Task Performance
Task data were collected daily between 1030 and 1300 hours
after food in the arena was replenished. All visible individuals
inside the nest, including the queen, were spot-checked, and the
task they were performing was recorded. In-nest active tasks
included larval feeding (manipulating wax in larval clumps and
feeding), incubating (pressing thorax and abdomen against pupal
cells), constructing honeypots (manipulating wax around honey-
pots), guarding (standing near the entrance or window and
antennating other individuals that approach), fanning (quietly
buzzing wings inside the nest without attempting to fly) and
honeypot probing (inserting head into a honeypot, either to deposit
or retrieve food). If bees were found in the foraging arena during
data collection, they were considered to be foraging. For each bee
that was observed five times or more in an active task, we calcu-
lated the proportion of time spent performing each of the seven
tasks listed above. We used Spearman rank correlations to deter-
mine the probability of performing a task relative to space use. If
individuals were motionless while task data were collected, they
were recorded as ‘inactive’. We calculated the proportion of time
that individuals with five or more observations (active and inactive)
remained inactive inside the nest. Because inactivity is not a task,
the relationship between probability of inactivity and space use
was analysed separately using a Spearman rank correlation test.
Rare activities such as excavating, undertaking, digging and
aggressive interactions were not included in the analyses
(for detailed ethogram of bumblebee worker tasks, see Cameron
1989). Approximately 80% or more of the individuals within
a colony were located each day.
Measuring Body Size
Head and thorax widths of all bees were measured using digital
callipers to the nearest 0.1 mm. Both measurements were analysed
and showed similar results. Thorax width showed the greater range
in values. Therefore, because of the consistency observed between
head and thorax measurements, we only report results regarding
thorax width here.
Calculating Age Effects
To determine how space use changed over time, we performed
a linear regression analysis on all three colonies while blocking for
the individual worker (average worker life span XSD: colony A:
35 14 days; colony B: 37 20 days; colony D: 37 18 days). To
determine whether different age cohorts were found on different
nest characters or were more likely to perform a specific task, we
performed an ANOVA on each colony. In these analyses, we blocked
for individual worker, treated the nest character or task as the
discrete factor and age as a continuous response variable. When
analysing whether different age cohorts participated in certain
tasks, we included inactivity. This way we could determine
whether inactive individuals were of a particular age class as well.
RESULTS
Do Individual Bumblebees Maintain Spatial Fidelity Zones
Inside the Nest?
In each colony, individual worker bees varied in the area of the
nest that they used. Across colonies, 12.10 1. 01% ðXSDÞof
workers maintained spatial fidelity zones that were significantly
smaller than random (colony A: 11.70%; colony B: 11.36%; colony C:
13.25%). On the other hand, 8.47 1.86% maintained zones that
were significantly larger than random (colony A: 6.38%; colony B:
9.09%; colony C: 9.93%). All together, 20.57 2.55% of bees within
each colony were not randomly choosing positions from the overall
worker distribution. Note that expected ‘random’ zone size was
calculated from the actual overall worker distribution and not
(a) Small zone (b) Large zone
N = 26
SDX = 2.38 cm; SDY = 2.38 cm
Spatial zone size = 20.2 cm2
Spatial sig. = 4.0%
N = 29
SDX = 3.25 cm; SDY = 3.75 cm
Spatial zone size = 48.8 cm2
Spatial sig. = 100%
Figure 1. Calculating spatial zone size from raw data. Frequency of observations for
one bee in a grid cell is represented with grey shading. The darker the box, the more
often the individual was recorded in that grid cell. Here we show examples of indi-
viduals whose zone sizes were (a) significantly smaller and (b) significantly larger than
random. Black dots represent the worker’s centroid standard deviations (SD).
Dashed boxes represent the calculated spatial zone size. N’s represent the number of
in-nest spatial observations recorded on the individual; spatial significance denotes
the percentage of random calculated spatial distributions that were smaller than an
individual’s actual zone size.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651 643

under the assumption that there was a homogenous, even distri-
bution of workers across the whole box.
In addition, we found significant differences between individ-
uals with regard to the distance maintained from the colony centre
(ANOVA: P<0.05 for all colonies; Table 1). In two of the three
colonies, there were significant differences between individuals
with regard to the distance maintained from the nest entrance and
the queen (colonies B and D: P<0.05; colony A: P>0.05; Table 1).
Bees also differed in the distances they moved each day from the
previous day’s coordinate in two of the three colonies (colonies A
and D: P<0.05; colony B: P>0.05; Table 1). When bees that used
small zones or large zones were examined separately, there were
no differences within those groups in distance maintained from the
colony centre (Table 1). In two of three colonies, however, bees that
used small zones differed from one another in how far they
remained from the nest entrance (colonies A and D: P<0.05;
colony B: P>0.05; Table 1).
Individual bumblebees’ zone sizes increased with their average
distance to the colony centre (Spearman rank correlation: P<0.001
for all colonies; Fig. 2,Table 2) and average distance to the queen
(P<0.001 for all colonies; Table 2). Individuals that stayed away
from the nest entrance also maintained larger spatial zone sizes in
colonies B and D (P<0.001) but not in colony A (P>0.05; Table 2).
Finally, we compared spatial zone size with the distance that an
individual moved from day to day. Individuals with large spatial
zone sizes, on average, were found further from their previous day’s
coordinate than individuals with small spatial zone sizes
(Spearman rank correlation: P<0.001 for all colonies; Table 2).
There were also positive correlations in all three colonies between
the distance from the previous day’s coordinate and the average
distance that an individual maintained from the colony centre
(P<0.001 for all colonies; Table 2). While these variables were not
completely independent of one another, these results illustrate that
workers that maintained large spatial zone sizes moved widely
around the nest. This means that large zone sizes are not a by-
product of gathering multiple data points on individuals that move
only short distances from day to day. Instead, bees that maintained
large zones were more likely to move greater distances across the
nest space from day to day, while remaining far from the colony
centre. This finding suggests that these bees were probably
primarily using the periphery of the nest.
There was no consistent relationship across all three colonies
between spatial zone size and proportion of time spent on brood
(larvae þeggs), pupae, honeypots or window areas (Table 3). In
colony A, smaller zones were located more often around brood
clumps (P¼0.001), whereas in colonies B and D, smaller zones
were located around pupal clumps (P0.01 for both colonies).
Individuals with larger spatial zone size inside the nestbox were
more likely to be found off the nest substrate (P<0.001 for all
colonies; Table 3). This result suggests that individuals with larger
Table 1
Differences between space use of individual bumblebee workers inside the nest
Variable Colony A Colony B Colony D
FdfPFdfP FdfP
All bees
Centre 1.35 1,93 0.02 2.13 1,91 <0.001 2.39 1,153 <0.001
Queen 1.02 1,91 0.43 1.40 1,91 0.009 1.41 1,15 3 0.001
Entrance 1.12 1,93 0.21 1.29 1,91 0.04 1.60 1,153 <0.001
Previous day 1.44 1,70 0.02 1.25 1,76 0.09 2.10 1,118 <0.001
‘Small zone’
Centre 0.56 1,10 0.84 1.61 1,10 0.11 1.23 1,23 0.22
Queen 1.10 1,10 0.37 2.26 1,10 0.02 0.63 1,23 0.91
Entrance 2.30 1,10 0.02 1.02 1,10 0.43 2.95 1,23 <0.001
Previous day 1.51 1,10 0.16 1.01 1,10 0.45 1.32 1,23 0.16
‘Large zone’
Centre 0.29 1,6 0.94 0.60 1,9 0.79 0.85 1,17 0.64
Queen 0.30 1,6 0.93 0.62 1,9 0.78 0.69 1,17 0.81
Entrance 0.52 1,6 0.79 0.72 1,9 0.69 0.77 1,17 0.73
Previous day 0.44 1,6 0.85 1.09 1,9 0.38 1.36 1,17 0.17
Separate ANOVAs were conducted to compare (1) all bees in the colony, (2) only
those bees whose zone sizes were significantly smaller than random and (3) only
those bees whose zone sizes were significantly larger than random. ‘Centre’ is the
distance of a bee’s position from the centre of the worker distribution in the nest,
‘Queen’ is the distance of worker positions from the queen, and ‘Entrance’ is the
distance of worker position from the entrance of the nestbox. ‘Previous day’ reflects
how workers move around in the nest: it measures the distance of a bee’s position
from that same bee’s position on the previous day. Significant differences between
individuals are highlighted in bold.
Colony A
Colony B
Mean distance to colony centre (cm)
Colony D
S
p
atial zone size (cm2)
0
rS = 0.83, N = 94, P < 0.001
rS = 0.90, N = 90, P < 0.001
rS = 0.88, N = 154, P < 0.001
15 30 45 60
5
3.75
2.5
6.25
3.75
2.5
1.25
6.25
6.25
5
3.75
2.5
1.22
5
Figure 2. Spatial zone size and the mean distance to the colony centre for all three
colonies. Black dots represent individuals whose spatial zone size was significantly
smaller than random, white dots represent individuals whose spatial zone size was
significantly larger than random, and grey dots represent individuals that maintained
a distribution not significantly differing in size from a random distribution. Stars
represent the queen. We used Spearman rank correlations to calculate r
S
and Pvalues.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651644

zones were preferentially located at the edge of the nest, as
opposed to being randomly dispersed throughout the nest,
consistent with the pattern described above.
Can the Use of Spatial Zones Be Predicted by Age or Body
Size of Individuals?
While there were differences between individuals with regard to
their position in the nest, there was little evidence to suggest that
individuals predictably adjust their position as they age. We found
different effects in each of the three colonies. Only in colony A was
there evidence that individuals gradually moved away from the
colony centre over time (linear regression: colony A: R
2
¼0.13;
individual differences: F
1,63
¼1.36,P¼0.04; age effects: F
1,1
¼16.64,
P<0.001; colony B: R
2
¼0.12; individual: F
1,50
¼1.74, P¼0.002;
age: F
1,1
¼0.92, P¼0.34; colony D: R
2
¼0.14; individual:
F
1,90
¼2.31, P<0.001; age: F
1,1
¼0.48, P¼0.49). In colony A, there
was no evidence that individuals adjusted their position relative to
the queen over time (R
2
¼0.13; individual: F
1,61
¼1.09, P¼0.31;
age: F
1,1
¼3.42, P¼0.07), whereas in colony B, individuals gradually
moved closer to the queen (R
2
¼0.10; individual: F
1,50
¼1.28,
P¼0.10; age: F
1,1
¼12.26, P<0.001), and in colony D they moved
further away (R
2
¼0.09; individual: F
1,90
¼1.28, P¼0.05; age:
F
1,1
¼5.60, P¼0.02). There was no evidence of individuals changing
their distance from the nest entrance with age in any of the colonies
(colony A: R
2
¼0.11; individual: F
1,63
¼1.36, P¼0.04; age:
F
1,1
¼2.42, P¼0.12; colony B: R
2
¼0.07; individual: F
1,50
¼0.92,
P¼0.64; age: F
1,1
¼0.49, P¼0.48; colony D: R
2
¼0.09; individual:
F
1,90
¼1.43, P¼0.007; age: F
1,1
¼0.42, P¼0.52) or the distance
moved from the previous day’s coordinate (colony A: R
2
¼0.20;
individual: F
1,45
¼1.7 3 , P¼0.004; age: F
1,1
¼3.64, P¼0.06; colony
B: R
2
¼0.16; individual: F
1,40
¼1.45, P¼0.04; age: F
1,1
¼0.09,
P¼0.76; colony D: R
2
¼0.18; individual: F
1,65
¼1.89,P<0.0 01;age:
F
1,1
¼0.09, P¼0.77).
In all colonies, younger bees tended to reside near brood clumps
whereas older bees were more likely to be found near the window
or in the arena (colony A: individual differences: F
1,80
¼5.35,
P<0.001; nest character differences: F
1,6
¼36.75, P<0.001;
colony B: individual: F
1,7 3
¼10.93, P<0.001; nest: F
1,6
¼35.64,
P<0.001; colony D: individual: F
1,109
¼12.29, P<0.001; nest:
F
1,6
¼47.75, P <0.001). Because of the patchiness of nest charac-
teristics inside the nest, the results of these analyses do not
necessarily imply that individuals showed a tendency to move
‘outward’ as they grew older. Instead, bees may have been moving
to a different nest characteristic while remaining within their zone,
or the nest characteristics within a bee’s zone may have changed as
larvae pupated, emerged and left a new honeypot behind.
Coloniesvaried with regard tothe distribution of worker body size
(Kruskal–Wallis test: H
2
¼24.29, P<0.001; thorax width: colony A:
median ¼4.34 mm, interquartile range (IQR) ¼0.62; colony B:
median ¼4.43 mm, IQR ¼0.89; colony D: median ¼4.11 mm,
IQR ¼0.72). Even so, in two of three colonies, larger bees were
likely to use more space (Bonferroni corrected alpha ¼0.01;
Spearman rank regression: colonies B and D: P<0.001; colony
A: P¼0.30; Fig. 3,Table 4 ), and stay further from the colony
centre (colonies B and D: P<0.001; colony A: P¼0.15) and the
nest entrance (colonies B and D: P<0.005; colony A: P¼0.11).
Larger individuals were also more likely to stay away from the
queen in all three colonies (P0.01). Larger bees were not more
likely to move further from their previous day’s location than
smaller bees in two of three colonies (colonies A and B: P0.02;
colony D: P<0.001), illustrating that body size alone does not
necessarily predict how much an individual will move around
inside the nest.
How Does Space Use Correlate with Division of Labour?
In all three colonies, individuals that were more likely to feed
larvae were more often found closer to the colony centre (Bonfer-
roni corrected alpha ¼0.007; Spearman rank correlation: P0.006
for all colonies; Table 5). In two of three colonies, these workers
maintained smaller zones (colonies B and D: P0.007; colony A:
P¼0.02; Table 5,Fig. 4). On the other hand, foragers maintained
larger zones (P0.003 for all colonies; Fig. 5) and were found
further from the colony centre (P0.001 for all colonies). There
were no consistent relationships across colonies between space use
and specialization in any other task (see Table 5). In all three
colonies, inactive bees remained further from the colony centre
(critical alpha ¼0.05; P0.03 for all colonies). In two of three
colonies, they tended to use more space inside the nest (colonies B
and D: P0.02; colony A: P¼0.08; Table 5).
In all three colonies, we found that tasks were performed by
bees of different ages (colony A: F
1,7
¼3.65, P¼0.001; colony B:
F
1,7
¼4.64, P<0.001; colony D: F
1,7
¼6.05, P<0.001). Younger
bees were more likely to feed larvae and, in colonies B and D,
construct honeypots (Fig. 6). Older bees were more likely to forage
or be inactive in colonies B and D as well (Fig. 6). No other
consistent age effects were observed among other tasks.
We pooled individuals from all three colonies and categorized
them by body size to determine whether different-sized individuals
were likely to perform certain tasks. We performed an ANOVA on
each task and used a Tukey’s post hoc test to compare between
body size categories. We did not include those bees whose thorax
width was less than 3 mm or greater than 5.5 mm in the ANOVA, as
there were too few bees recorded in those categories; however,
their data are still presented in Fig. 7. Our analyses showed that
larval feeders and incubators were often the smallest bees (larval
feeders: P¼0.001; incubators: P¼0.003), whereas guards, fanners
or foragers were larger (guards: P¼0.008; fanners: P¼0.03;
Table 2
Space use compared to distance from reference points inside the nest
Variables Colony A Colony B Colony D
r
S
NP r
S
NP r
S
NP
Spatial zone
size
Centre 0.83 94 <0.001 0.90 90 <0.001 0.88 154 <0.001
Queen 0.58 91 <0.001 0.65 90 <0.0 01 0.68 154 <0.001
Entrance 0.12 94 0.27 0.56 90 <0.001 0.48 154 <0.001
Previous day 0.70 70 <0.001 0.68 73 <0.001 0.75 119 <0.001
Centre Previous day 0.66 70 <0.001 0.61 73 <0.001 0.70 119 <0.001
Spearman rank correlations were used to compare spatial zone size to four different
reference points; the Bonferroni corrected alpha value of 0.0125 was used as
significance criterion. Significant correlations are highlighted in bold. Spearman
rank correlation compared distance to the colony centre with distance moved from
the previous day separately; critical alpha ¼0.05.
Table 3
Spatial zone size relative to nest character
Spatial zone
size vs
Colony A Colony B Colony D
r
S
NP r
S
NP r
S
NP
Brood 0.34 94 0.001 0.15 90 0.15 0.11 154 0.16
Pupae 0.07 94 0.49 0.27 90 0.01 0.29 154 <0.001
Honeypot 0.18 94 0.08 0.07 90 0.53 0.21 154 0.01
Window 0.24 94 0.02 0.02 90 0.84 0.22 154 0.007
Nothing 0.50 94 <0.001 0.62 90 <0.001 0.59 154 <0.001
Spearman rank correlation results of worker spatial zone size relating to the
proportion of observations in which workers were found on a particular nest
character. Because we compared spatial zone size to five different nest characters
within each colony, we used the Bonferroni corrected alpha value of 0.01 as our
significance criterion. Significant regressions are highlighted in bold.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651 645

foragers: P¼0.002; Fig. 7). There was no evidence that construct-
ing or probing honeypots were performed by a certain size bee
(P>0.05 for both tasks). There was some evidence to suggest that
the extraordinarily large bees were more likely to be inactive, but
our sample size for that group of bee was too small to draw any
definitive conclusions on this.
DISCUSSION
We have shown that Bombus impatiens workers distribute
themselves in a nonrandom way inside the nest, and that, on
average, 12% of workers maintain restricted spatial zones. All
workers maintain a consistent distance from the colony centreover
their entire lifetime. Up until this point, there was only anecdotal
evidence that bumblebees arrange themselves inside the nest
nonrandomly during the growth phase of the colony cycle (van
Doorn & Heringa 1986). This is the first study to provide rigorous
statistical analysis that this is indeed the case. Space use in
bumblebees is related to the performance of some in-nest tasks
(such as larval feeding). Our finding that spatial patterns did not
change over time is consistent with what we had predicted for
a group whose workers show little age polyethism (B. griseocollis:
Cameron 1989;B. bifarius:O’Donnell et al. 2000), although we
Colony A
Colony B
Spatial zone size (cm2)
Colony D
Thorax width (mm)
2.5 3 3.5 4 4.5 5 5.5 6
rs = 0.11, N = 93, P = 0.30
rs = 0.43, N = 89, P < 0.001
rs = 0.32, N = 154, P < 0.001
0
45
30
15
60
30
15
0
60
60
45
30
15
0
45
Figure 3. Worker body size (measured as thorax width) and spatial zone size. Black
dots represent individuals whose spatial zone size was significantly smaller than
random, white dots represent individuals whose spatial zone size was significantly
larger than random, and grey dots represent individuals that maintained a distribution
not significantly differing in size from a random distribution. We used Spearman rank
correlations to calculate r
S
and Pvalues.
Table 4
Worker body size (thorax width) and space use
Body size vs Colony A Colony B Colony D
r
s
NP r
s
NP r
s
NP
SZS 0.11 93 0.30 0.43 89 <0.0 01 0.32 154 <0.001
Centre 0.15 93 0.15 0.39 89 <0.001 0.32 154 <0.001
Queen 0.27 90 0.01 0.40 89 <0.001 0.33 154 <0.001
Entrance 0.17 93 0.11 0.37 89 <0.001 0.23 154 0.004
Previous day 0.01 69 0.93 0.27 73 0.02 0.37 119 <0.001
Spearman rank correlations were used to compare body size to five different spatial
statistics within each colony; the Bonferroni corrected alpha value of 0.01 was used
as our significance criterion. Significant regressions are highlighted in bold. ‘SZS’ is
spatial zone size as defined in the text.
Table 5
Space use and task performance
Colony A Colony B Colony D
r
S
NP r
S
NP r
S
NP
Larval feeding SZS 0.26 82 0.02 0.29 84 0.007 0.33 108 0.001
Centre 0.30 82 0.006 0.39 84 <0.001 0.43 108 <0.001
Queen 0.06 80 0.62 0.41 84 <0.001 0.36 108 <0.001
Entrance 0.04 82 0.71 0.14 84 0.21 0.13 108 0 .18
Incubating SZS 0.12 82 0.30 0.17 84 0.13 0.08 108 0.42
Centre 0.17 82 0.12 0.18 84 0.11 0.02 108 0.84
Queen 0.03 80 0.81 0.20 84 0.07 0.10 108 0.31
Entrance 0.04 82 0.73 0.08 84 0.47 0.16 108 0.10
Construct
honeypot
SZS 0.19 82 0.09 0.03 84 0.81 0.12 108 0.21
Centre 0.17 82 0.13 0.00 84 0.97 0.11 108 0.27
Queen 0.24 80 0.03 0.10 84 0.37 0.16 108 0.11
Entrance 0.00 82 0.99 0.05 84 0.67 0.16 108 0.10
Guarding SZS 0.12 82 0.28 0.30 84 0.0 06 0.17 108 0.07
Centre 0.13 82 0.23 0.33 84 0.002 0.21 108 0.03
Queen 0.01 80 0.90 0.30 84 0.005 0.23 108 0.02
Entrance 0.02 82 0.87 0.21 84 0.05 0.28 108 0.003
Fanning SZS 0.19 82 0.10 0.04 84 0.71 0.05 108 0.60
Centre 0.16 82 0.16 0.04 84 0.75 0.04 108 0.70
Queen 0.05 80 0.68 0.07 84 0.53 0.08 108 0.39
Entrance 0.02 82 0.87 0.08 84 0.50 0.05 108 0.62
Probing
a honeypot
SZS 0.08 82 0.47 0.05 84 0.69 0.03 108 0.76
Centre 0.04 82 0.72 0.02 84 0.87 0.09 108 0.37
Queen 0.06 80 0.60 0.07 84 0.52 0 .16 108 0 .10
Entrance 0.06 82 0.62 0.13 84 0.23 0.03 108 0.76
Foraging SZS 0.39 82 <0.001 0.40 84 <0.001 0.28 108 0.003
Centre 0.39 82 <0.001 0.44 84 <0.001 0.32 108 0.001
Queen 0.16 80 0.16 0.28 84 0.009 0.29 108 0.002
Entrance 0.05 82 0.65 0.11 84 0.33 0.07 108 0.46
Inactive SZS 0.18 92 0.08 0.26 87 0.02 0.27 145 0.001
Centre 0.22 92 0.03 0.25 87 0.02 0.24 145 0.003
Queen 0.22 89 0.04 0.11 87 0.32 0.31 145 <0.001
Entrance 0.06 92 0.61 0.18 87 0.10 0.35 145 <0.001
Spearman rank correlations were used to compare worker space use to proportion of
time performing one of seven active tasks within each colony; the Bonferroni cor-
rected alpha value of 0.007 was used as the significance criterion. Proportion of
inactivity was calculated on a different data set, so the significance criterionfor these
analyses remained at alpha ¼0.05. Significant correlations are highlighted in bold.
‘SZS’ is spatial zone size as defined in the text.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651646

found that age may influence task choice to some extent. The
spatial pattern that we observed in bumblebees is similar to what
has been observed in Temnothorax ants (Sendova-Franks & Franks
1995; Backen et al. 2000), whose nests are organized in concentric
patterns radiating from the centre of the nest (Franks & Sendova-
Franks 1992). Spatial sorting in bumblebees, as in Temnothorax,is
neither dependent on the organization of brood in the nest (Sen-
dova-Franks & Franks 1994; Backen et al. 2000), nor constrained by
the small space used by a bumblebee colony, as was previously
suggested (Cameron 1989). Bees close to the centre of the nest in
our study tended to have small zones, be of small body size and
feed larvae, while bees at the periphery tended to have large body
size and were more likely to forage. We found no evidence that
spatial zones were related to performing any other in-nest task.
Space Use and Division of Labour
Bees that fed larvae were more likely to be in the centre of the
nest, whereas foragers were more often found on the periphery.
This result is similar to what has been observed in ants (T. albi-
pennis:Sendova-Franks & Franks 1995;O. brunneus:Powell &
Tschinkel 1999) and paper wasps (R. marginata:Robson et al. 2000),
in that workers that took care of brood were more likely to be found
in the centre of the colony. What is more surprising about our
Colony A
0
0.2
0.4
0.6
0.8
1
Colony B
Proportion of time performing larval feeding
0
0.2
0.4
0.6
0.8
1
Colony D
S
p
atial zone size (cm2)
0
0.2
0.4
0.6
0.8
1
rS = −0.26, N = 82, P = 0.02
rS = −0.29, N = 84, P = 0.007
rS = −0.33, N = 108, P = 0.001
15 30 45 60
Figure 4. Spatial zone size and the proportion of time that an individual was observed
feeding larvae inside the nest. Black dots represent individuals whose spatial zone size
was significantly smaller than random, white dots represent individuals whose spatial
zone size was significantly larger than random, and grey dots represent individuals
that maintained a distribution not significantly differing in size from a random
distribution. Stars represent the queen for each colony. We used Spearman rank
correlations to calculate r
S
and Pvalues.
Colony A
0
0.2
0.4
0.6
0.8
1
Colony B
Proportion of time foraging
0
0.2
0.4
0.6
0.8
1
Colony D
S
p
atial zone size (cm2)
0
0.2
0.4
0.6
0.8
1
rS = 0.39, N = 82, P < 0.001
rS = 0.40, N = 84, P < 0.001
rS = 0.28, N = 108, P = 0.003
15 30 45 60
Figure 5. Spatial zone size and the proportion of time an individual foraged. Black dots
represent individuals whose spatial zone size was significantly smaller than random,
white dots represent individuals whose spatial zone size was significantly larger than
random, and grey dots represent individuals who maintained a distribution not
significantly differing in size from a random distribution. Stars represent the queen for
each colony. We used Spearman rank correlations to calculate r
S
and Pvalues.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651 647

results is that, unlike in ants, bumblebee brood was not necessarily
organized around the centre of the colony. Brood and honeypots
(and by extension, task stimuli) were dispersed throughout the
nest. In one of our colonies, small zones were concentrated around
young brood, whereas in the other two colonies, they were
concentrated around pupae. Bees with larger zones, however, are
more likely to encounter many of the tasks associated with multiple
nest characters (and task stimuli) within their zone. Even though
we found only some evidence of an association between space use
and task specialization, Wilson (1976) and Seeley (1982) suggested
that by remaining in nonrandom areas of the nest, workers in
a social insect colony can minimize the distance that they move
between tasks. Among bumblebee species, the next important step
will be to determine whether spatial organization indeed reduces
switching costs and how ubiquitous such spatial organization is
across species.
Age Effects
Younger B. impatiens workers were more likely to care for
brood or construct honeypots, whereas older workers were more
likely to forage. While this trend may be consistent across colonies
and species (see also B. griseocollis:Cameron 1989), there does not
seem to be a discrete division of labour based on age, like that in
honeybees (Seeley 1982). Previous work on bumblebee division of
labour has shown that, in general, younger bees are more likely to
perform ‘in-nest’ tasks, whereas older bees are more likely to
forage (Brian 1952; Free 1955; Yerushalmi et al. 2006). These age
effects are not strict, however, since many bees remain inside the
nest throughout their entire lives (Brian 1952; Free 1955; Yer-
ushalmi et al. 2006). Bumblebees do not divide labour strictly by
age (B. griseocollis:Cameron 1989;B. bifarius:O’Donnell et al.
2000), nor do they switch between in-nest tasks in a particular
order like honeybees do (Seeley 1982), so it is perhaps not
surprising that space use inside the nest also does not change as
bees get older.
Body Size Effects
Smaller bees, more often in the centre of the nest, were more
likely to feed larvae or incubate brood. On the other hand, larger
bees, more often on the periphery when inside the nest, were more
likely to guard, fan or forage. Previous studies on bumblebees have
shown that small bees are more likely to remain inside the nest,
whereas larger bees are more likely to go out and forage (Brian
1952; Alford 1978; Goulson et al. 2002; Yerushalmi et al. 2006).
Foster et al. (2004) found no evidence to suggest that B. bifarius
workers of a particular size were more or less likely to perform an
in-nest task at the end of the colony cycle. Our study is the first to
show quantitative evidence that during the ergonomic phase of the
colony cycle, some in-nest worker tasks are divided by body size
(Fig. 7).
It is possible that body size, spatial zone and task are all linked in
bumblebees. As in Temnothorax ants, space use in bumblebees may
be determined by an individual’s activity level (Sendova-Franks &
Franks 1994; Backen et al. 2000). Larger individuals may be more
active, use more area inside the nest, and end up engaging in tasks
that are more often performed in those areas, such as guarding. On
the other hand, smaller individuals may be less active and be more
likely to remain in a concentrated location (such as a brood clump)
where they engage in larval feeding or incubating. If this is indeed
the case, the spatial sorting may be an important link between body
size and division of labour in bumblebees.
Inactive Bees
In other social insect species, a large percentage of workers are
often observed remaining motionless inside the nest (40% of Apis
mellifera honeybees: Lindauer 1978; 55% of Temnothorax allardycei
ants: Cole 1986). These inactive workers are often hypothesized to
be ‘reserve’ workers, ready and capable to perform a task when the
colony experiences a sudden disturbance (Seeley 1995; Ho
¨lldobler
& Wilson 1998; Evans 2006). When a bumblebee colony is in need,
workers that are already engaged in a task are more likely to switch
to a different task than inactive workers are to become active
(B. flavifrons,B. melanopygus,B. occidentalis:Cartar 1992), sug-
gesting little need for reserves. In our study, observations were
conducted during the daylight hours; therefore, inactivity of the
large B. impatiens was probably not due to the circadian resting
pattern of foragers (Yerushalmi et al. 2006). It is possible that our ad
libitum feeding method affected the normal foraging rates, and our
enclosed experimental design reduced forager mortality, leaving
Colony A
0
10
20
30
40
50
Colony B
Age (days)
0
10
20
30
40
50
Colony D
Larval feeding
Incubating
Guarding
Fanning
Probing a honeypot
Foraging
Inactive
0
10
20
30
40
50
62
10 67
3
15
24 44 46
59
17
56
7
13 24 35 44
86 18 87
7
635
27
76
a
a
a
a a
a
a
a
ab
ab
ab
ab
b
ab bab
ab bb
ab
ab
ab b
b
Constructing honeypots
Figure 6. Box plots showing the age range (in days) of individuals from each colony
that performed each task. Boxes represent medians (solid line) 25th percentile
range. Whiskers denote 10th–90th percentiles, while dots denote outliers. Sample size
for each task is listed inside each box. ANOVA results show that different ages per-
formed different tasks in all three colonies (colony A: F
1,7
¼3.65, P¼0.001; colony B:
F
1,7
¼4.64, P<0.001; colony D: F
1,7
¼6.05, P<0.001). Letters denote Tukey’s post hoc
comparison results.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651648

Feeding larvae
0
0.2
0.4
0.6
0.8
1
Incubating
0
0.2
0.4
0.6
0.8
1
Constructing
honeypots
0
0.2
0.4
0.6
0.8
1
Guarding
0
0.2
0.4
0.6
0.8
1
Fanning
0
0.2
0.4
0.6
0.8
1
Probing
honeypot
0
0.2
0.4
0.6
0.8
1
Foraging
0
0.2
0.4
0.6
0.8
1
Thorax width (mm)
2.5–3 3–3.5 3.5–4 4–4.5 4.5–5 5–5.5 5.5–6
Inactive
0
0.2
0.4
0.6
0.8
1
F1,4 = 4.60, P = 0.001
F1,4 = 4.21, P = 0.003
F1,4 = 1.51, P = 0.20
F1,4 = 3.54, P = 0.008
F1,4 = 2.71, P = 0.03
F1,4 = 0.36, P = 0.84
F1,4 = 4.32, P = 0.002
N = 2 1847998723 1
aac ab
b bc
a
ab
bbab
a
a
ab ab
b
ab a
ab bab
a
aab b
ab
N = 2 25 58 114 94 23 3
F1,4 = 0.90, P = 0.47
Figure 7. Box plots showing the probability that bumblebees of a particular size range would perform each task or remain inactive in the nest. Boxes represent medians
(solid line) 25th percentile range. Whiskers denote 10th–90th percentiles, while dots denote outliers. Sample sizes for each size range are listed above the graph. ANOVA results
(listed to the right of the graph) do not include the smallest (2.5–3 mm) or largest (5.5–6 mm) size ranges as the sample sizes were too small. Letters above each box denote Tukey’s
post hoc comparison results. Probability of performing a task was calculated from the frequency that individuals performed one of seven active tasks, whereas inactivity was
calculated from the frequency that individuals were inactive or active when they were observed. Only bees observed five or more times were included in either analysis. Data from
all three colonies were pooled for analyses.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651 649

unemployed foragers to become ‘lazy’ inside the nest (Sladen 1912).
On the other hand, inactive bees were not as likely to remain in the
nest periphery as workers that sometimes foraged (Table 5). In
B. terrestris, smaller bees are less active when the inside of the nest
is illuminated (Yerushalmi et al. 2006). However, inactive B. impa-
tiens workers were also not significantly smaller than other bees in
the colony (Fig. 7). Therefore, although somewhat fewer foragers
may have been active by using this experimental set-up, the
phenomenon of inactive bees in the nest is not likely to be a labo-
ratory artefact.
Conclusion
Nonrandom spatial assortment of individuals has been
described in a variety of social groups (e.g. fish shoals: Hoare et al.
2000; humpback whales: Weinrich et al. 2006; peafowl leks: Loyau
et al. 2007). In order for these patterns to be considered self-
organized, individuals must not use an external cue or template to
orient themselves (Camazine et al. 2001). This is the first study to
show that individuals within a bumblebee colony maintain spatial
assortment inside the nest, and that these spatial patterns are not
necessarily dependent upon the location of nest characteristics.
Furthermore, this spatial assortment may either influence division
of labour among workers or be the result of it. For example, as in
Temnothorax ants (Sendova-Franks & Franks 1994; Backen et al.
2000), differences between individual activity levels may predis-
pose certain workers to use larger or smaller areas of the nest
where they are likely to encounter different task stimuli. On the
other hand, spatial organization may be a by-product of workers
‘foraging-for-work’ (Franks & Tofts 1994), and the spatial organi-
zation that arises could be the result of each individual’s attraction
to a particular task across different areas of the nest. Manipulative
experiments are needed before determining which hypothesis is
more likely. Either way, we conclude that the spatial organization
shown by Bombus impatiens workers may be important in the
division of labour within a colony.
Acknowledgments
We thank Eden Huang, Nicolas Skye Robbins, Amanda Barth and
Wendy Isner for their help in data collection and assistance with
bumblebee maintenance. Margaret Couvillon, Tuan Cao, Amelie
Schmolke, Michele Lanan and Emily Jones provided feedback on
the manuscript. Research supported through the College of Science,
Department of Ecology & Evolutionary Biology, University of
Arizona, U.S.A.
References
Alford, D. V. 1978. The Life of the Bumblebee. London: Davis-Poynter.
Backen, S. J., Sendova-Franks, A. B. & Franks, N. R. 2000. Testing the limits of
social resilience in ant colonies. Behavioral Ecology and Sociobiology,48,
125–131.
Beshers, S. N. & Fewell, J. H. 2001. Models of division of labor in social insects.
Annual Review of Entomology,46, 413–440.
Brian, A. D. 1952. Division of labour and foraging in Bombus agrorum Fabricius.
Journal of Animal Ecology,21, 223–240.
Camazine, S. 1991. Self-organizing pattern formation on the combs of honey bee
colonies. Behavioral Ecologyand Sociobiology,28, 61–76. doi:10.1007/BF00172140.
Camazine, S., Deneubourg, J.-L., Franks, N. R., Sneyd, J., Theraulaz, G. &
Bonabeau, E. 2001. Self-organization in Biological Systems. Princeton, New
Jersey: Princeton University Press.
Cameron, S. A. 1989. Temporal patterns of division of labor among workers in the
primitively eusocial bumble bee, Bombus griseocollis (Hymenoptera: Apidae).
Ethology,80, 137–151.
Cameron, S. A. & Robinson, G. E. 1990. Juvenile hormone does not affect division of
labor in bumble bee colonies (Hymenoptera: Apidae). Annals of the Entomo-
logical Society of America,83, 626–631.
Cartar, R. V. 1992. Adjustment of foraging effort and task switching in energy-
manipulated wild bumblebee colonies. Animal Behaviour,44, 75–87.
Cole, B. J. 1986. The social behavior of Leptothorax allardycei (Hymenoptera, For-
micidae): time budgets and the evolution of worker reproduction. Behavioral
Ecology and Sociobiology,18, 165–173.
van Doorn, A. & Heringa, J. 1986. The ontogeny of a dominance hierarchy in
colonies of the bumblebee Bombus terrestris (Hymenoptera, Apidae). Insectes
Sociaux,33, 3–25. doi:10.1007/BF02224031.
Duchateau, M. J. & Velthuis, H. H. W.1989. Ovarian development and egg laying in
workers of Bombus terrestris.Entomologia Experimentalis et Applicata,51,
199–213. doi:10.1007/BF00347883.
Evans, T. A. 2006. Foraging and building in subterranean termites: task switchers or
reserve labourers? Insectes Sociaux,53, 56–64. doi:10.1007/s00040-005-
0835-8.
Foster, R. L., Brunskill, A., Verdirame, D. & O’Donnell, S. 2004. Reproductive
physiology, dominance interactions, and division of labour among bumble bee
workers. Physiological Entomology,29, 327–334. doi:10.1111/j.0307-6962.2004.
00388.x.
Franks, N. R. & Sendova-Franks, A. B. 1992. Brood sorting by ants: distributing the
workload over the work-surface. Behavioral Ecology and Sociobiology,30,
109–123. doi:10.1007/BF00173947.
Franks, N. R. & Tofts, C. 1994. Foraging for work: how tasks allocate workers.
Animal Behaviour,48, 470–472. doi:10.1006/anbe.1994.1261.
Free, J. B. 1955. The division of labour within bumblebee colonies. Insectes Sociaux,
2, 195–212. doi:10.1007/BF02224381.
Geva, S., Hartfelder, K. & Bloch, G. 2005. Reproductive division of labor, domi-
nance, and ecdysteroid levels in hemolymph and ovary of the bumble bee
Bombus terrestris.Journal of Insect Physiology,51, 811–823. doi:10.1016/
j.jinsphys.2005.03.009.
Goulson, D. 2003. Bumblebees: Behavior and Ecology. Oxford: Oxford University
Press.
Goulson, D., Peat, J., Stout, J. C., Tucker, J., Darvill, B., Derwent, L. C. &
Hughes, W. O. 2002. Can alloethism in workers of the bumblebee, Bombus
terrestris, be explained in terms of foraging efficiency? Animal Behaviour,64,
123–130. doi:10.1006/anbe.2002.3041.
Heinrich, B. 2004. Bumblebee Economics. Cambridge, Massachusetts: Harvard
University Press.
Hoare, D. J., Ruxton, G. D., Godin, J.-G. J. & Krause, J. 2000. The social organization
of free-ranging fish shoals. Oikos,89, 546–554.
Ho
¨lldobler, B. & Wilson, E. O. 1998. The Ants. Cambridge, Massachusetts: Harvard
University Press.
Karsai, I. & Pe
´nzes, Z. 1998. Nest shapes in paper wasps: can the variability of forms
be deduced from the same construction algorithm? Proceedings of the Royal
Society of London, Series B,265, 1261–1268. doi:10.1098/rspb.1998.0428.
Lindauer, M. 1978. Communication among Social Bees. Cambridge, Massachusetts:
Harvard University Press.
Loyau, A., Jalme, M. S. & Sorci, G. 2007. Non-defendable resources affect peafowl
lek organization: a male removal experiment. Behavioural Processes,74, 64–70.
doi:10.1016/j.beproc.2006.09.008.
O’Donnell, S., Reichardt, M. & Foster, R. 2000. Individual and colony factors in
bumble bee division of labor (Bombus bifarius nearcticus Handl; Hymenoptera,
Apidae). Insectes Sociaux,47,164–170. doi:10.1007/PL00001696.
Powell, S. & Franks, N. R. 2006. Ecology and the evolution of worker morphological
diversity: a comparative analysis with Eciton army ants. Functional Ecology,20,
1105–1114. doi:10.1111/j.1365-2435.2006.01184.x.
Powell, S. & Tschinkel, W. R. 1999. Ritualized conflict in Odontomachus brunneus
and the generation of interaction-based task allocation: a new organizational
mechanism in ants. Animal Behaviour,58, 965–972.
Robinson, G. E. 1987. Regulation of honey bee age polyethism by juvenile hormone.
Behavioral Ecology and Sociobiology,20, 329–338. doi:10.1007/BF00300679.
Robinson, G. E. 1992. Regulation of division of labor in insect societies. Annual
Review of Entomology,37, 637–665.
Robinson, G. E., Page, R. E., Strambi, C. & Strambi, A. 1989. Hormonal and genetic
control of behavioral integration in honey bee colonies. Science,246, 109–112.
doi:10.1126/science.246.4926.109.
Robson, S. K. A., Bean, K., Hansen, J., Norling, K., Rowe, R. J. & White, D. 2000.
Social and spatial organization in colonies of a primitively eusocial wasp,
Ropalidia revolutionalis (de Saussure) (Hymenoptera: Vespidae). Australian
Journal of Entomology,39, 20–24.
Roux, E. A. & Korb, J. 2004. Evolution of eusociality and the soldier caste in
termites: a validation of the intrinsic benefit hypothesis. Journal of Evolutionary
Biology,17, 869–875. doi:10.1111/j.1420-9101.2004.00727.x.
Seeley, T. D.1982. Adaptive significance of the age polyethism schedule in honeybee
colonies. Behavioral Ecology and Sociobiology,11, 287–293. doi:10.1007/
BF00299306.
Seeley, T. 1995. The Wisdom of the Hive. Cambridge, Massachusetts: Harvard
University Press.
Sendova-Franks, A. B. & Franks, N. R. 1994. Social resilience in individual worker
ants and its role in division of labour. Proceedings of the Royal Society of London,
Series B,256, 305–309. doi:10.1098/rspb.1994.0085.
Sendova-Franks, A. B. & Franks, N. R. 1995. Spatial relationships within nests of
the ant Leptothorax unifasciatus (Latr.) and their implications for the division of
labour. Animal Behaviour,50, 121–136. doi:10.1006/anbe.1995.0226.
Sladen, F. W. L. 1912 . The Humble-Bee. London: Macmillan.
Spaethe, J. & Weidenmu
¨ller, A. 2002. Size variation and foraging rate in
bumblebees (Bombus terrestris). Insectes Sociaux,49, 142–146. doi:10.1007/
s00040-002-8293-z.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651650

Weidenmu
¨ller, A. 2004. The control of nest climate in bumblebee (Bombus ter-
restris) colonies: interindividual variability and self reinforcement in fanning
response. Behavioral Ecology,15, 120–128. doi:10.1093/beheco/arg101.
Weinrich, M. T., Rosenbaum, H., Baker, C. S., Blackmer, A. L. & Whitehead, H.
2006. The influence of maternal lineages on social affiliations among humpback
whales (Megaptera novaeangliae) on their feeding grounds in the southern gulf
of Maine. Journal of Heredity,97, 226–234. doi:10.1093/jhered/esj018.
West-Eberhard, M. J. 1969. The social biology of polistine wasps. Miscellaneous
Publications Museum of Zoology, University of Michigan,140, 1–101.
Wilson, E. O. 1976. Behavioral discretization and the number of castes in an ant
species. Behavioral Ecology and Sociobiology,1, 141–154. doi:10.1007/
BF00299195.
Worden, B. D., Skemp, A. K. & Papaj, D. R. 2005. Learning in two contexts: the
effects of interference and body size in bumblebees. Journal of Experimental
Biology,208, 2045–2053. doi:10.1242/jeb.01582.
Yerushalmi, S., Bodenhaimer, S. & Bloch, G. 2006. Developmentally determined
attenuation in circadian rhythms links chronobiology to social organization in
bees. Journal of Experimental Biology,209, 1044–1051. doi:10.1242/jeb.02125.
J.M. Jandt, A. Dornhaus / Animal Behaviour 77 (2009) 641–651 651