ANIMAL BEHAVIOUR, 2003, 66, 459–467
doi:10.1006/anbe.2003.2253
Genotype–environment interactions in honeybee guarding
behaviour
GREG J. HUNT* ERNESTO GUZMA
u N-NOVOA†‡, JOSE
u L. URIBE-RUBIO†‡ & DANIEL PRIETO-MERLOS‡
*Department of Entomology, Purdue University
†Instituto Nacional de Investigaciónes Forestales y Agropecuarias, Mexico
‡Facultad de Medicina Veterinaria, Universidad Autonoma de Mexico
(Received 12 February 2002; initial acceptance 7 June 2002;
final acceptance 7 November 2002; MS. number: A9285R)
Honeybees have an age-based division of labour that is influenced by genetic variability for the tendency
to perform specific tasks. Individuals in a honeybee colony comprise diverse genotypes and their
interactions can influence task allocation. Colonies from an African race (Africanized honeybees, AHB,
Apis mellifera scutellata Ruttner) usually produce a much stronger defensive response than do European
races of honeybees (EHB), and these races may differ in how individuals are allocated to the tasks of
guarding and stinging. We observed guarding behaviour in colony environments that varied in
proportions of genotypes (AHB, EHB) and population size. In large colonies, AHB showed much greater
guarding persistence (number of days guarding) than EHB; hybrids were intermediate. In another series
of experiments, three families each of AHB and EHB were cofostered in colonies with different AHB: EHB
ratios, then tested in large and small colonies. In colonies of both sizes, colony environment interacted
with both famly and type (AHB or EHB) for propensity to guard. Individuals of both types guarded more
persistently in large colonies, but family and type both interacted with environment. EHB were more
likely to initiate guarding bouts in low-AHB colonies, but persistence did not change with environment.
AHB were insensitive to effects of environment for the tendency to initiate guarding behaviour, but were
more persistent in high-AHB environments. EHB and AHB may differ in how they allocate individuals to
guarding. The positive reinforcement of behaviour that occurs in high-defensive environments and in
large populations could cause a stronger stinging response through alarm pheromone recruitment.

2003 Published by Elsevier Ltd on behalf of The Association for the Study of Animal Behaviour.
Some social insects such as many species of ants and
termites have morphologically specialized soldier castes.
Eusocial bees lack these castes. Instead, nest defence is
performed by workers that are old enough to do so
(reviewed in Wilson 1971). The defence of the honeybee
colony and its resources is necessary for maintaining
colony integrity. Bees defend the nest against both vertebrate (primarily mammalian) and invertebrate intruders,
such as other bees attempting to rob honey. Honeybee
defensive behaviour consists primarily of guarding, stinging and pursuing. A worker bee stinging an animal dies
after losing the stinger. Therefore, there are trade-offs
between stinging to protect the nest and loss of workers.
Correspondence: G. J. Hunt, Department of Entomology, Purdue
University, West Lafayette, IN 47907, U.S.A. (email:
[email protected]). E. Guzmán-Novoa and J. L. Uribe-Rubio are at
Santa Cruz 29B Fracc., Las Hdas., Metepéc 52140, Méx., Mexico. D.
Prieto-Merlos is at Facultad de Medicina Veterinaria, Universidad
Autonoma de México, Ciudad Universitaria, 04510, México, D.F.,
Mexico.
0003–3472/03/$30.00/0

Recruitment of nestmates through alarm pheromones
and visual cues affects the colony response. Guard bees
appear to be an important component in recruiting nestmates to sting at the colony entrance (Butler & Free 1952;
Ribbands 1954; Mauschwitz 1964; Arechavaleta & Hunt,
in press).
A guard bee stands in the colony entrance, approaches
incoming bees and touches them with its antennae. The
guard may stand with its front legs off the ground and
wings spread as if ready to fly (Ribbands 1954; Moore et
al. 1987). After inspecting another bee, the guard sometimes acts aggressively towards the bee, especially if it is a
non-nestmate. The aggressive responses of guards are
similar to the occasional acts of biting and attempts at
stinging by other worker bees towards diseased nestmates
or towards non-nestmates (Drum & Rothenbuhler 1983).
Guards distinguish foreign bees and other intruders from
nestmates by olfactory (Butler & Free 1952; Moore et al.
1987) and visual cues (Butler & Free 1952; Ribbands
1954), but olfaction is of primary importance (Mann &
Breed 1997).
459
2003 Published by Elsevier Ltd on behalf of The Association for the Study of Animal Behaviour.
460
ANIMAL BEHAVIOUR, 66, 3
Guarding is a very specialized behaviour compared
with other honeybee tasks. Only about 15% of the bees in
a colony will ever perform guarding behaviour at the
colony entrance (Moore et al. 1987). However, most
worker bees will perform most of the other tasks required
by the colony in a fairly predictable series, resulting in
temporal division of labour. Because honeybee queens
mate with multiple males, the workers of the colony
comprise a diverse set of genotypes and there is a potential for individuals to partially specialize at specific
tasks that they are genetically predisposed to perform
(Calderone & Page 1989, 1992; Page & Robinson 1990).
Analysis of the behaviour and the allozyme genotypes of
bees in three-subfamily colonies indicated a genetic component for propensity to guard (Robinson & Page 1988).
The persistence in guarding (number of days spent at the
task) of individual workers correlates with the response to
alarm pheromone of the colony from which those individuals originated (Breed et al. 1988; Breed & Rogers
1991). Colonies that had the most persistent guards also
had the most bees emerging from the hive in response to
presentation of alarm pheromone, which is one assay for
the defensive response. An environmental effect on
guarding behaviour has also been demonstrated. European bees from either relatively defensive or gentle colonies are more likely to guard when cofostered in a
colony environment composed of bees from the gentle
source family than when they are cofostered in the
defensive source hive (Breed & Rogers 1991).
So far, observations of guarding behaviour have been
performed only with European races of honeybees. An
African subspecies of honeybee (A. m. scutellata Ruttner)
that was introduced to Brazil in 1956 (Kerr 1967) is well
known for its highly defensive behaviour and has spread
through most of South America and part of North
America in the past 45 years. The descendents of these
bees, called ‘Africanized’ honeybees (AHB), have retained
their highly defensive behaviour and perhaps most of
their African genotype (Hall 1991; Rinderer & Hellmich
1991; Winston 1992). Near the nest, AHB respond more
quickly and with more stings than do EHB to stimuli that
elicit defensive behaviour, such as alarm pheromones,
moving objects and vibrations (Collins et al. 1982, 1987;
Villa 1988). The stinging response of AHB also shows
genetic dominance when measured at the colony level
(Guzmán-Novoa & Page 1994; Guzmán-Novoa et al.
2002). How much of the genome of AHB is African in
origin is unknown, but in our study area, the incidence of
African mitochondrial DNA in feral swarms increased
from near zero in 1990 to 28% in 1991, then reached 96%
by 1996 and remains near this level (Guzmán-Novoa &
Page 1999; unpublished data).
The detection of genetic effects on guarding behaviour
raises the possibility that behavioural interactions
between individuals that differ in their inherent tendency
to guard within a colony could result in interactions
between genotype and colony environment. Observing
interactions between extreme phenotypes such as AHB
and EHB could provide insights into the mechanisms of
task allocation in defensive behaviour. Our goals in these
studies were to analyse genetic effects on guarding per-
sistence, to extend studies of guarding behaviour to AHB
and to study regulation of task allocation in guarding
behaviour by observing interactions between individuals
of different genotypes. We found a higher propensity and
persistence in guarding by AHB and genotype–colony
environment interactions that influenced division of
labour in guarding.
METHODS
We conducted two studies to observe genetic effects on
guarding behaviour and interactions between individuals
of different genotypes. We refer to the two races, AHB and
EHB, as ‘types’ and to the groups of individuals from each
source colony as a ‘family’. Study 1 was designed to
investigate genetic effects on individual guarding persistence (the number of days a bee continued to guard after
being marked while guarding) in the two types of bees
(EHB and AHB), as well as bees from crosses between the
two types. The guarding persistence of bees in full-sized
colonies was observed. Some of the full-sized colonies
contained progeny from queens that were instrumentally
inseminated to provide colonies containing hybrid and
backcross individuals that would enable the analysis of
the inheritance pattern of guarding persistence.
Study 2 was conducted to test for genotype–social
environment interactions on individuals’ propensity to
guard and persistence in guarding. We carried out experiments in both small and large colonies to test for effects
caused by colony size. The colonies had mixed age distributions typical of unmanipulated colonies and contained
two different proportions of EHB to AHB. Three unrelated
sources were used for each of the two types (AHB and
EHB) from open-mated queens, which usually mate with
10–17 drones (Adams et al. 1977). Using six open-mated
queens made it possible to sample many subfamilies to
enable general inferences about differences between AHB
and EHB. The large-colony experiment involved observations of groups of marked bees from the same sources that
were used in small colonies and that were cofostered in
AHB or EHB colonies.
Study 1: Analysis of Inheritance Patterns for
Guarding Persistence
Sources of bees
For all experiments, we obtained AHB stocks from
swarms captured near our study area at the Centre for
Beekeeping Development, near Villa Guerrero, Mexico
(19N, 99W). The AHB sources were tested by morphometrics (Sylvester & Rinderer 1987) and behavioural tests
(Guzmán-Novoa & Page 1993) to ensure that they had
African characteristics and the mitochondrial DNA type
of A. m. scutellata (Nielsen et al. 1999). We used naturally
mated or instrumentally inseminated queens of European
origin as described below.
Experiment 1: genetic effects on number of days that a
bee guards
To observe inheritance patterns in guarding persistence, we counted the number of individuals that guarded
HUNT ET AL.: GENOTYPE–ENVIRONMENT INTERACTIONS
Table 1. Persistence of guarding in 18 full-sized colonies
Number Guards
Maximum
Genotype of colonies marked days guarded
AHB
EHB
F1
BC AFR
BC EUR
3
6
3
2
4
120
250
150
60
250
10
7
8
8
9
% Guarding
on day 6
(N)
10% (12)
0.4% (1)
4% (6)
10% (6)
3% (6)
AHB: Africanized honeybees; EHB: European honeybees; F1: hybrids
produced from crosses between a few of the same AHB and EHB
colonies by single-drone insemination; BC AFR: colony with F1
queen backcrossed to a drone from the AHB parental colony used to
produce the F1; BC EUR: F1 queen backcrossed to a EHB drone.
for 6 days in undisturbed colonies from specific crosses.
We used 18 full-sized colonies that were each housed in
one jumbo-size hive box and one 15-cm deep super
(smaller box) per colony, with heterogeneous age distributions and populations of about 20 000–30 000 bees.
We used colonies with different expected proportions of
African alleles in order to observe genotypic effects on
guarding persistence (see Table 1).
Three EHB colonies had queens that were instrumentally inseminated using multiple drones and came from
either a California or Ontario source. Three others were
open-mated in the U.S.A. or Canada where no AHB were
present. One of the AHB colonies had an open-mated
queen, but the other two had queens that were instrumentally inseminated with semen from multiple drones
from another AHB colony.
We constructed three hybrid colonies and six backcross
colonies, as follows. Three European queens were each
inseminated with semen from single African drones to
produce colonies with hybrid workers. Hybrid queens
were reared from these crosses. Then, two hybrid queens
were crossed to single African drones to produce
African backcross workers and four hybrid queens were
each single-drone-inseminated with European drones to
produce European backcross workers.
About 40 workers displaying guarding behaviour were
marked with white paint in each of the 18 colonies. In all
experiments, we chose the act of approaching and antennating incoming workers at the colony entrance as a
prerequisite for identifying an individual as a guard. We
defined a guarding bout as observing one bee guarding.
On day 6, we made a count to determine how many
marked bees were still guarding in each colony.
Study 2: Genotype–Social Environment
Interactions in Guarding Behaviour
Six colonies of bees with naturally mated queens were
chosen; three of these colonies were Africanized and three
were European. The origins of AHB were determined as in
study 1. The European queens were all mated in areas
without AHB, either in California (referred to as family 4)
or in Canada (families 5 and 6). The Canadian sources
were derived from breeding programmes that used stock
imported from Buckfast Abbey, U.K. Family 4 was originally derived from Italian stock (A. m. ligustica). The
three Africanized sources (families 1, 2 and 3) came from
queens captured from local swarms in different locations.
Experiment 2: guarding
mixed-genotype colonies
behaviour
Marked
Small colonies
EHB
small,
Four colonies were established, each containing 5000
bees. Two colonies were used for each of two mixtures of
AHB to EHB: 25% AHB and 50% AHB. All colonies
received two frames of brood, two frames of pollen plus
nectar and a mated queen unrelated to the experimental
bees. To obtain marked bees from the six families, frames
of brood were placed in screen cages inside of hives of
European bees that were unrelated to the source colonies.
Incubators were not used in this study because significant
pre-eclosion effects on defensive behaviour are unlikely
(Moritz et al. 1987). From 16 to 22 November 2000, the
colonies were established with marked and unmarked
bees (Table 2). Unmarked bees were added in equal
proportions from the same six source colonies. The
unmarked bees were older adults so that the colonies
would have relatively typical age distributions (equal
mixtures of bees collected from outer combs and foragers
trapped from the colony entrance).
Observations began once the youngest marked bees
were 1 week old and the oldest were approaching the age
when they were likely to initiate guarding behaviour. The
colonies were observed on 18 days for at least 0.5 h to
record the presence of colour-marked guards. We chose
18 days to provide ample time to follow the entire
guarding careers of most of the marked individuals, since
most EHB guard for a single day and only a few will guard
Table 2. Average genotypic constitution of experimental colonies in study 2
AHB
in
Unmarked
270 (3 sources) 715 (3 sources) 4000 (25% AHB)
1000 (3 sources) 1000 (3 sources) 3000 (50% AHB)
Large colonies 1696 (2 sources) 1905 (2 sources)
35 000 EHB
1126 (2 sources) 1171 (2 sources)
35 000 AHB
Total
% AHB
population (marked+unmarked)
5000
5000
38 000
37 000
25.4
50
8
94
Low
High
Low
High
Each of the four colony environments was replicated in two colonies. Shown are the average numbers of bees
introduced per colony. Source queens were naturally mated with 10–20 haploid males.
461
462
ANIMAL BEHAVIOUR, 66, 3
for several weeks (Moore et al. 1987). In addition to the
0.5-h observation periods, we examined colonies consecutively for periods throughout the day and identified
marked individuals as guards as in the first study, then
captured them by gently placing an inverted 15-ml plastic
tube over the bee on the landing board of the hive. These
bees were immediately transferred to a refrigerator and
chilled until immobile (2–5 min). We identified each of
these guards by gluing a collared, numbered plastic tag to
the back of the thorax (Graze KG, Weinstadt, Germany).
Then we reintroduced guards to the colonies at the
entrance. Guards were readily accepted back into their
colonies. During each observation period, we recorded
the number of guards for each of the six genotypes. We
also recorded the occurrence of guarding bouts by individual bees tagged with numbers when we observed them
at other times of the day (to determine persistence). The
maximum number of marked bees for each of the six
sources observed guarding at any one time was used for
the analysis of daily numbers of guards (propensity).
Experiment 3: guarding behaviour in large AHB or EHB
colonies
Emerging bees from four of the six sources used in
experiment 2 were available for study in full-sized hives,
two AHB sources (families 1 and 3) and two EHB sources
(families 4 and 6). About 2000 newly emerged bees from
each of the four sources were marked and introduced into
each of four full-sized colonies that already contained
about 30 000 bees each (Table 2). Two host colonies each
of AHB and EHB were used. These colonies were unrelated
to the ones used in the small-colony study. Colony
entrances were extended with plywood platforms to
facilitate observations. Entrances were observed for
30 min in the morning and 30 min in the afternoon each
day for 18 days. Guards also were tagged with numbers as
in experiment 2. The number and mark of each guard bee
was recorded daily.
Statistical Analyses
Each day, we determined the number and proportions
of guards represented by each source family. Some models
of task allocation in honeybees predict that numbers of
individuals performing a task influence the stimulus
environment for recruiting additional individuals to that
task (Page & Robinson 1990). Therefore, we calculated the
daily proportion of guards for each family relative to the
total number of guards for that day. Relative guarding
propensity was defined for each colony as the number of
individual guards from a given source divided by the total
number of guards observed that day, divided by the
proportion of the colony population that consisted of
that genotype. The proportion was based on the numbers
of marked and unmarked bees at the beginning of the
experiment because differential mortality between AHB
and EHB in preliminary experiments was not excessive
(E. Guzmán-Nova & G. J. Hunt, unpublished data). We
performed analysis of variance (ANOVA) on the daily
number of marked individuals guarding for each source
and colony, after weighting the number of guards seen
for a given family to correct for differences in numbers of
marked bees between families and colonies. For example,
the adjusted number of guards (N) for a hypothetical
family X was calculated for colony Y according to the
formula:
N=(G)(IzY families/Ifamily X)(total zIall colonies/total Icolony Y)
where G is the number of guards observed that day for
family X, I is the number of marked bees introduced and
zI is the average number of marked bees introduced. The
adjusted number of guards was calculated for each family
and each day separately for experiments 2 and 3.
A log+1 transformation was performed to produce a
normal distribution before ANOVA. Guarding persistence
was defined as the number of days observed guarding.
These data could not be transformed to normality. Therefore, we used Mann–Whitney U tests to compare the two
types (AHB versus EHB). Data on the number of bees that
were tagged as guards were analysed by chi-square tests
for goodness-of-fit to the expected values based on the
proportion of each family or type in the colony population. We made comparisons of these statistics between
types (AHB and EHB) within each environment (the
ratio of AHB to EHB), and also compared different
environments within each type.
RESULTS
Study 1: Inheritance Pattern for Guarding
Persistence
In large, unmanipulated colonies, AHB guarded for
much longer than did EHB (Table 1). We paint-marked
about 40 guards in each of 18 colonies. Six days after
being marked, 10% of 120 AHB were still guarding but
only one of 250 EHB was doing so. The lack of EHB guards
was not a consequence of our being unable to see them
since guards stand on the landing board in front of the
colony entrance. The guarding persistence of the hybrid
colonies was intermediate between the parental lines
(4%), suggesting additivity. However, the backcross to the
European parental lines was similar to the F1 (3%), and
the backcross to the Africanized parental line was the
same as that of the parental line (10%). The maximum
persistence for AHB guarding behaviour in these colonies
was 10 days, although persistence of some individuals
could have been longer if they had been guarding before
we paint-marked them. We observed one EHB worker
that guarded for at least 7 days.
Study 2: Genotype–Social Environment
Interactions
Genetic effects on propensity to guard
During 18 days of observation, AHB showed a higher
overall propensity to guard when cofostered with EHB
(Figs 1, 2). In small colonies of mixed genotype, we
observed effects of both family (source colony; P<0.0001)
Daily average number of guards
HUNT ET AL.: GENOTYPE–ENVIRONMENT INTERACTIONS
(a)
(b)
Environment:
25% AHB
50% AHB
3
Environment:
EHB
AHB
5
2
4
1.5
3
1
2
0.5
1
0
0
AHB
behaviour
EHB
behaviour
AHB
behaviour
EHB
behaviour
Figure 1. Average daily number of guards for two types of bees
(AHB; EHB) cofostered in different proportions within colonies,
weighted for differences in the number of introduced bees for each
genotype and for differences in numbers of bees introduced to the
four colonies. (a) Small colonies (N=5000 bees each with different
mixtures of EHB and AHB). (b) Large colonies (N=about 35 000
unmarked AHB or EHB and 2300–3600 marked bees).
Small colonies
Relative propensity to guard
Relative propensity to guard
Environment:
EHB
AHB
1.4
1.2
1.2
1
1
1
1
1
404
5
1
5
384
1.004
0.063
2.132
36.494
4.748
0.063
3.111
101.195
F
P
0.6
0.6
0.4
0.4
0.2
0.2
0
0
AHB
behaviour
EHB
behaviour
Type (AHB or EHB)
Environment*
Type*environment
Residual
Family†
Environment
Family*environment
Residual
11.109
0.701
23.600
0.0009
0.4029
<0.0001
11.835
0.789
0.622
<0.0001
0.3748
<0.0001
Log-transformed data for numbers of guards per day, weighted by
the average number of marked bees introduced to the colony per
genotype and by the average number of marked bees added to each
hive.
*Colonies of 5000 bees had environments consisting of either 25%
or 50% AHB.
†Six source colonies were used. Therefore, each family was a
different set of worker genotypes from one mother and multiple
fathers.
Table 4. Large colonies: analyses of variance for type and genotype
effects on the propensity to guard
0.8
0.8
2
1.75
1.5
1.25
1
0.75
0.5
0.25
0
Sum of
squares
Large colonies
Environment:
25% AHB
50% AHB
(b)
df
6
2.5
(a)
Table 3. Small colonies: analyses of variance for type and genotype
effects on the propensity to guard
Environment:
25% AHB
50% AHB
AHB
behaviour
EHB
behaviour
Environment:
EHB
AHB
Type (AHB or EHB)
Environment*
Type*environment
Residual
Family†
Environment
Family*environment
Residual
df
Sum of
squares
1
1
1
296
3
1
3
292
0.319
2.114
0.471
35.625
3.636
2.114
0.917
32.007
F
P
2.650
17.561
3.914
0.1046
<0.0001
0.0488
11.058
19.282
2.788
<0.0001
<0.0001
0.0409
Log-transformed data for numbers of guards per day, weighted by
the average number of marked bees introduced to the colony per
genotype and by the average number of marked bees added to each
hive.
*Environments consisted of either European or Africanized colonies
of approximately 30 000 bees. We added 2200–4200 marked bees
from four source colonies to each hive.
†Six source colonies were used (AHB: families 1, 3; EHB: families 4,
6). Each queen naturally mated with multiple drones.
1.4
1.2
1
0.8
0.6
0.4
0.2
3 1 2
AHB sources
4 5 6
EHB sources
0
3
1
AHB sources
4
6
EHB sources
Figure 2. Genotype–social environment interactions on relative
propensity to guard in large and small colony populations. (a)
Interactions between types of bees (AHB, EHB). (b) Interactions
between genotypes (families). Numbers above bars: number of
families.
and type (AHB or EHB; P=0.009) on the adjusted daily
numbers of guards (Tables 3, 4). AHB performed 30.5% of
the 302 guarding bouts observed in small colonies containing 25% AHB (21 =3.92, P<0.05), and 69.2% of 582
bouts observed in small colonies containing 50% AHB
(21 =84.68, P<0.0001). More total bouts were observed for
both types of bees in the high-AHB environment simply
because there were twice as many marked bees. In the
large-population colonies, we observed a significant genetic component on guarding propensity attributed to
family, but no overall difference between the two bee
types (Table 4).
Guarding propensity is a consequence of both the
tendency to initiate guarding and persistence at guarding.
As a measure of the tendency to initiate guarding, we
analysed the numbers of new guards that we were able to
capture and tag. EHB were more likely to initiate guarding
in low-AHB environments in colonies of both small and
large populations, as inferred from higher than expected
numbers of tagged EHB guards, especially in the large
colonies. More EHB guarded in EHB than in the AHB
463
ANIMAL BEHAVIOUR, 66, 3
Table 5. Number of guard bees tagged in colonies composed of different genotypic ratios and population sizes
Population size
Environment
Small colonies
Low AHB
High AHB
χ2 environment†
Large colonies
χ2 environment†
Low AHB
High AHB
EHB guards
76
71
6.06
216
72
19.91
P
AHB guards
19
91
1.04
147
87
0.72
<0.025
<0.001
P
χ2
1 type*
P
1.33
2.47
NS
NS
NS
<0.025
NS
NS
6.35
0.88
NS
Small colonies (5000 bees): Low AHB=25% AHB; High AHB=50% AHB. Large colonies (about 35 000 bees): Low
AHB=8% AHB; High AHB=94%.
*Chi-square statistic for number of AHB or EHB that were tagged as guards, based on expected proportions
calculated from Table 2.
†Chi-square statistic for number of guards tagged in each environment for each type.
colonies, and the proportion of EHB guarding in the
high-EHB environment was also more than would be
expected (Table 5). This effect was not because one EHB
family was more likely to respond in this way. More than
half of the EHB guards (42 of 76) in the low-AHB environments in small colonies were from family 6, even
though only one-third of tagged EHB were of this
family. However, the two EHB families in large colonies
(families 4 and 6) were equally represented in the EHB
environment.
Interactions in Propensity to Guard
We found interactions for the propensity to guard
between both families and environment and type (AHB
or EHB) and environment (colonies containing different
ratios of AHB to EHB), in both small and large colonies
(Tables 3, 4). In small colonies, the effect of environment
was not significant after accounting for genotypic interactions. In large colonies, both types of bees had more
daily guarding bouts in the high-EHB environment,
resulting in a significant environmental effect.
The interactions between genotypes and environment
were robust. We calculated the proportional representation of each genotype among the guards for each day,
because the number of guards may be regulated by daily
behavioural interactions. The same overall patterns of
changes in relative propensity of guarding in different
environments were observed in both small and large
colonies; European honeybees were more likely to guard
in colonies with high proportions of EHB than they were
in colonies with higher proportions of AHB, but AHB
were overrepresented among guards in colonies with
higher proportions of AHB (Fig. 2a). The same pattern was
also observed for most of the individual progenies of the
six open-mated queens (Fig. 2b). However, the family-5
European source did not fit the typical pattern in small
colonies, and the family-6 European source did not differ
between the two environments in large colonies (Fig. 2b).
Family 1, which was the AHB source with the highest
overall propensity to guard, also was unresponsive to
environmental effects on propensity in both small and
large colonies (Fig. 2b).
Genotypic Interactions in Guarding Persistence
We observed the guarding careers of 777 workers (small
colonies: N=255, large colonies: N=522). Individual
workers of both types guarded for longer periods on
average in the large colonies than they did in the small
colonies (Fig. 3). AHB guarded for more days in the small
colonies than did EHB (Mann–Whitney U test: U=5392.5,
N1 =110, N2 =147, P<0.0001). However, there was a difference between environments; AHB were more persistent
only in small colonies with the higher proportion of AHB
(50%; U=2132, N1 =91, N2 =71, P=0.0002). In this 50%AHB environment, the average AHB persistence was similar to their persistence in the large-colony experiment
(Fig. 3). In the large colonies, AHB again were overall
more persistent than EHB (U=28796.5, N1 =234, N2 =288,
P=0.004), and within the two separate environments, the
difference was seen only in the high-AHB colonies (about
94% AHB; U=1936.5, N1 =87, N2 =72, P<0.0001). However, in contrast to the small-colony study, European bees
guarded for fewer days in the AHB colonies than they
did in EHB colonies (U=6485, N1 =72, N2 =216, P=0.035;
Fig. 3).
(a)
Average number of days
464
6
Environment:
25% AHB
50% AHB
(b)
6
5
5
4
4
3
3
2
2
1
1
0
Environment:
EHB
AHB
0
AHB
behaviour
EHB
behaviour
AHB
behaviour
EHB
behaviour
Figure 3. Average individual persistence of guarding behaviour for
AHB and EHB honeybees measured as the number of days that a bee
guarded. Bars represent the average guarding persistence in days
for the two types in each colony environment. (a) Small colonies.
(b) Large colonies.
HUNT ET AL.: GENOTYPE–ENVIRONMENT INTERACTIONS
Proportion that guarded 1 day
(a)
Small colonies
Environment:
25% AHB
50% AHB
Environment:
EHB
AHB
0.6
0.3
0.5
0.25
0.4
0.2
0.3
0.15
0.2
0.1
0.1
0.05
0
(b)
Proportion that guarded 1 day
Large colonies
AHB
behaviour
EHB
behaviour
0
Environment:
25% AHB
50% AHB
(42)
(7)
(25)
0.5
0.4
0.3
0.2
(9)
(23)
(22)
(25)
(10)
(45)
(22)
0.1
0
EHB
behaviour
Environment:
EHB
AHB
0.7
0.6
AHB
behaviour
1
2
4
5
6
AHB sources EHB sources
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
(82)
(26)
(34)
(46)
(113)
(134)
(24)
(63)
3
1
4
6
AHB sources EHB sources
Figure 4. Proportion of bees that guarded for only a single day in
different colony environments. (a) Proportion of each type (AHB,
EHB). (b) Proportion of each family. Large numbers above bars refer
to families; small numbers in parentheses are the number of bees on
which the proportion is based.
The most striking difference between AHB and EHB was
in the proportion of bees that guarded for just one day. In
small colonies, most individual EHB (58.5%) guarded for
just one day, but only 32.5% of AHB did so (Fig. 4a).
Individual EHB were more persistent guards in large
colonies than in small colonies. Only 18% of the EHB
family 4 guarded for just a single day in large colonies,
but 56% did so in small colonies (Fig. 4b). Both types of
bees and most individual families were more persistent in
environments in which they were relatively more likely
to guard, but AHB showed a greater differential response
to colony environment (Figs 2, 3, 4).
DISCUSSION
Results from the crosses that we observed in 18 full-sized
colonies were consistent with partial genetic dominance
for the number of days that a bee guards. Colonies
containing F1 workers were intermediate between the
parental types, suggesting additivity. However, the backcross to the Africanized line showed guarding persistence
that is typical of that parental type and the backcross to
the European line was close to the F1 phenotype, suggesting partial dominance for higher persistence. In light
of the results of study 2, the observed nonadditivity
could be explained by a genotype–colony environment
interaction.
In study 2, guarding persistence increased with colony
population size. For example, the average number of days
that individuals guarded from family-6 EHB was 2.5 days
in colonies of 5000 bees and 3.2 days in colonies with
about 38 000 bees, and the persistence of individuals
from family-4 EHB increased from 3.5 to 5.3 days between
small- and large-population environments. The average
persistence of AHB also increased in larger colonies.
Interactions between nestmates in small colonies may
have led to accelerated rates of behavioural development,
causing guards to move on to foraging activities earlier
and decreasing the length of their guarding careers
(Winston & Katz 1982; Page & Robinson 1990; Huang &
Robinson 1996). However, this would be expected only if
the small colonies had demographics that were significantly skewed towards younger ages and we established
them with mixed age distributions. Since a genotype–
environment interaction was detected for guarding persistence, the larger populations could have resulted
in reinforced guarding behaviour through increased
interactions between individuals.
Analysis of variance clearly showed genetic effects on
guarding propensity. In general, AHB were overrepresented among the guards. However, this was true only in
colony environments containing higher proportions of
AHB and could be attributed to increased persistence. In
small colonies of study 2, AHB were not significantly
overrepresented in the guards in the 25% AHB colonies
but were in 50% AHB colonies (1.4-fold). In larger hives,
AHB again were not significantly overrepresented (0.96fold) in EHB colonies but were over represented (1.2-fold)
among guards in AHB colonies.
Guzmán-Novoa & Page (1994) found no genetic effects
on guarding behaviour in bees from the same geographical area as our studies. They performed inseminations and
established five-patriline colonies consisting of various
mixtures of European and hybrid (AHB/EHB) workers.
Hybrids were no more likely to guard than Europeans in
samples from colonies containing both 25% and 50%
hybrids. Differences between our results and those of
Guzmán-Novoa & Page could be attributed to differences
in genotypic combinations, demography, colony size or
sample size. Differences in genotypic combinations could
have occurred, for example, if genes influencing guarding
behaviour are additive at the level of the individual, such
that there may have been an undetectable genetic effect
in Guzmán-Novoa & Page’s study. Our experiments used
colonies that differed in the proportion of EHB and AHB
(not EHB and hybrids, as in Guzmán-Novoa & Page 1994)
and involved 18 days of observations. Thus, we had
more power to detect differences in guarding behaviour
associated with differences in genotype.
EHB were more sensitive to environment when it came
to initiating guarding, but AHB were more sensitive to
effects of environment on persistence. EHB were more
likely to initiate guarding bouts in colonies with a high
proportion of European bees. We were able to tag more
EHB guards in these high-EHB environments, regardless
of population size. These results could fit a stimulus
465
466
ANIMAL BEHAVIOUR, 66, 3
response threshold model that assumes that bees influence the stimulus environment of nestmates by performing specific tasks (Robinson & Page 1989; Page &
Robinson 1990). For example, in a colony containing a
low proportion of AHB and few guards, the stimulus for
EHB to initiate guarding might be increased (or an inhibitory factor reduced) relative to a colony in which there
already were a lot of AHB guards at the entrance. Such a
model could also explain why EHB guarded for fewer days
in large AHB colonies than they did in large EHB colonies.
This simple conceptual model, however, cannot account
for our results with AHB, because there was no deviation
from the expected numbers of AHB that initiated guarding behaviour in different environments. Rather, the
higher daily number of guarding bouts by AHB in
the high-AHB environments was a consequence of
differences in persistence (Figs 2, 3).
Breed & Rogers (1991) reported that high-defensive
EHB were more likely to initiate guarding behaviour in
low-defensive colonies than in high-defensive colonies
in two-thirds of their colony pairings, and that lowdefensive bees were more likely to do so in one-third of
these replicates. Thus, EHB guarding behaviour changed
in response to environment, but no genotype–
environment interaction was demonstrated because both
high- and low-defensive EHB had a similar behavioural
response to the altered social environment. Breed &
Rogers’ results are similar to ours in one important way:
EHB in our studies also were more likely to initiate
guarding bouts in colonies containing mostly gentlesource genotypes (i.e. high-EHB environments). However,
our study also demonstrates the existence of genotype–
social environment interactions for guarding propensity,
because AHB were not more likely to initiate guarding in
one environment or the other, yet they showed higher
guarding propensity in high-AHB environments.
Therefore, the genotype–environment interaction for
propensity was a result of differential responses of the two
types of bees for two aspects of guarding: initiation of
guarding and guarding persistence. Overall, AHB guarded
for more days than EHB both in small and large colonies.
This overall difference, however, was caused primarily by
increased persistence of AHB in high-AHB environments.
There were no differences in persistence between bee
types in low-AHB environments of either the small or
large colonies. AHB apparently are more responsive to
stimuli that result in positive feedback for guarding
behaviour. The differential change in persistence between
environments was most evident when we looked at the
number of bees that guarded for just one day. In study 2,
about 60% of EHB in high-AHB environments of small
colonies guarded for just a single day, but only about 30%
of AHB did so. In large EHB colonies, both types of bees
had roughly 28% of the individuals guarding for a single
day in high-EHB environments, but in the high-AHB
colonies, less than 7% of AHB guarded for just one day,
compared to 30% of EHB in the same environment
(Fig. 4).
Few studies have demonstrated genotype–environment
interactions for behavioural traits. One example exists for
honeybee foraging behaviour. Differences in foraging
behaviour between high and low strains selected for the
tendency to collect pollen were more extreme when
cofostered in high-strain colonies than when cofostered
in low-strain colonies (Calderone & Page 1992).
Genotype–environment interactions for honeybee guarding initiation and persistence suggest that EHB are more
sensitive to the stimulus environment for the tendency to
initiate guarding behaviour. However, a positive feedback
mechanism involving interactions between individuals
must be involved in reinforcing guarding behaviour. AHB
were much more sensitive to reinforcement associated
with the presence of individuals of AHB genotype. These
results suggest a difference in the way in which EHB and
AHB allocate individuals to guarding behaviour caused by
a difference in sensitivity to various cues.
Because alarm pheromone can act both as an attractant
and a releaser of an aggressive response, one explanation
for this phenomenon could be greater release of alarm
pheromone (or specific components) by AHB or increased
sensitivity to alarm pheromone as an attractant (or both).
Guards standing in a colony entrance sometimes exert
their stings and release alarm pheromone, which may
reinforce guarding behaviour as well as recruit other bees
to sting without depleting the most defensive genotypes
in the colony. A goal for future research is to determine
the mechanism that reinforces guarding persistence.
Acknowledgments
This research benefited by the support of INIFAP and the
State of Mexico, which established the Centre for Beekeeping Research and Development. Partial support was
provided by Universidad Nacional Autonoma de Mexico
(UNAM). We thank Angelica Gris for help in collecting
behavioural data. This research was supported by grants
from the NIH (R29 GM548580), and the USDA (NRI
35302-10137). This research also benefited from interaction with the Santa Fe Institute Social Insect Working
Group.
References
Adams, J., Rothman, E. D., Kerr, W. E. & Paulino, Z. L. 1977.
Estimation of the number of sex alleles and queen matings from
diploid male frequencies in a population of Apis mellifera. Genetics,
86, 583–596.
Arechavaleta, M. E. & Hunt, G. J. In press. Genotypic variation in
the expression of guarding behavior and the role of guards in the
defensive response of honey bee colonies. Apidologie.
Breed, M. D. & Rogers, K. B. 1991. The behavioral genetics of
colony defense in honeybees: genetic variability for guarding
behavior. Behavioral Genetics, 21, 295–303.
Breed, M. D., Rogers, K. B., Hunley, J. A. & Moore, A. J. 1988. A
correlation between guard behaviour and defensive response in
the honeybee, Apis mellifera. Animal Behaviour, 37, 515–516.
Butler, C. G. & Free, J. B. 1952. The behavior of worker honeybees
at the hive entrance. Behaviour, 4, 262–292.
Calderone, N. W. & Page, R. E., Jr. 1989. Genotypic variability
in age polyethism and task specialization in the honeybee,
Apis mellifera (Hymenoptera: Apidae). Behavioral Ecology and Sociobiology, 22, 17–25.
HUNT ET AL.: GENOTYPE–ENVIRONMENT INTERACTIONS
Calderone, N. W. & Page, R. E., Jr. 1992. Effects of interactions
among genotypically diverse nestmates on task specialization
by foraging honeybees (Apis mellifera). Behavioral Ecology and
Sociobiology, 30, 219–226.
Collins, A. M., Rinderer, T. E., Harbo, J. R. & Bolten, A. B. 1982.
Colony defense by Africanized and European honeybees. Science,
218, 72–74.
Collins, A. M., Rinderer, T. E., Tucker, K. W. & Pesante, D. G.
1987. Response to alarm pheromone by European and Africanized
honeybees. Journal of Apicultural Research, 26, 217–223.
Drum, N. H. & Rothenbuhler, W. C. 1983. Non-stinging aggressive
responses of worker honeybees to hivemates, intruder bees and
bees affected with chronic bee paralysis. Journal of Apicultural
Research, 22, 256–260.
Guzmán-Novoa, E. & Page, R. E., Jr. 1993. Backcrossing Africanized
honeybee queens to European drones reduces colony defensive
behavior. Annals of the Entomological Society of America, 86,
352–355.
Guzmán-Novoa, E. & Page, R. E., Jr. 1994. Genetic dominance and
worker interactions affect honeybee colony defense. Behavioral
Ecology, 5, 91–97.
Guzmán-Novoa, E. & Page, R. E., Jr. 1999. Selective breeding of
honeybees (Hymenoptera: Apidae) in Africanized areas. Journal of
Economic Entomology, 92, 521–525.
Guzmán-Novoa, E., Hunt, G. J., Uribe, J. L., Smith, C. &
Arechavaleta-Velasco, M. E. 2002. Confirmation of QTL effects
and evidence of genetic dominance of honeybee defensive behavior: results of colony and individual behavioral assays. Behavior
Genetics, 32, 95–102.
Hall, H. G. 1991. Genetic characterization of honey bees through
DNA analysis. In: The ‘African’ Honeybee (Ed. by M. Spivak, D. J. C.
Fletcher & M. D. Breed), pp. 45–73. Boulder, Colorado: Westview
Press.
Huang, Z.-Y. & Robinson, G. E. 1996. Regulation of honey bee
division of labor by colony age demography. Behavioral Ecology
and Sociobiology, 39, 147–158.
Kerr, W. E. 1967. The history of the introduction of African bees to
Brazil. South African Bee Journal, 39, 3–5.
Mann, C. A. & Breed, M. D. 1997. Olfaction in guard honeybee
responses to non-nestmates. Annals of the Entomological Society of
America, 90, 844–847.
Mauschwitz, U. W. 1964. Alarm substances and alarm behaviour in
social Hymenoptera. Nature, 204, 324–327.
Moore, A. J., Breed, M. D. & Moor, M. J. 1987. The guard
honeybee: ontogeny and behavioural variability of workers performing a specialized task. Animal Behaviour, 35, 1159–1167.
Moritz, R. F. A., Southwick, E. E. & Harbo, J. R. 1987. Maternal and
pre-eclosional factors affecting alarm behaviour in adult honeybees (Apis mellifera L.). Insectes Sociaux, 34, 298–307.
Nielsen, D. I., Ebert, P. R., Hunt, G. J., Guzman-Novoa, E., Kinnee,
S. A. & Page, R. E., Jr. 1999. Identification of Africanized
honeybees (Hymenoptera:Apidae) incorporating morphometrics
and an improved polymerase chain reaction mitotyping
procedure. Annals of the Entomological Society of America, 92,
167–174.
Page, R. E. Jr & Robinson, G. E. 1990. The genetics of division of
labour in honeybee colonies. Advances in Insect Physiology, 23,
117–169.
Ribbands, C. R. 1954. The defense of the honeybee community.
Proceedings of the Royal Society of London, Series B, 142,
514–524.
Rinderer, T. E. & Hellmich, R. L. 1991. The processes of Africanization. In: The ‘African’ Honeybee (Ed. by M. Spivak, D. J. C. Fletcher
& M. D. Breed), pp. 95–117. Boulder, Colorado: Westview Press.
Robinson, G. E. & Page, R. E., Jr. 1988. Genetic determination of
guarding and undertaking in honey-bee colonies. Nature, 333,
356–358.
Robinson, G. E. & Page, R. E., Jr. 1989. Genetic basis for division of
labor in an insect society. In: The Genetics of Social Evolution (Ed. by
M. D. Breed & R. E. Page, Jr), pp. 61–80. Boulder, Colorado:
Westview Press.
Sylvester, H. A. & Rinderer, T. E. 1987. Fast Africanized bee
identification system (FABIS) manual. American Bee Journal, 127,
511–516.
Villa, J. D. 1988. Defensive behaviour of Africanized and European
honeybees at two elevations in Columbia. Journal of Apicultural
Research, 27, 141–145.
Wilson, E. O. 1971. The Insect Societies. Cambridge, Massachusetts:
Harvard University Press.
Winston, M. L. 1992. The biology and management of Africanized
honeybees. Annual Review of Entomology, 37, 173–193.
Winston, M. L. & Katz, S. J. 1982. Foraging differences between
cross-fostered honeybee workers (Apis mellifera) of European and
Africanized races. Behavioral Ecology and Sociobiology, 10, 125–
129.
467