Behavioral Ecology Vol. 11 No. 3: 326-333
© 2000 International Society for Behavioral Ecology
Adaptive shifts in honey bee (Apis mellifera L.) guarding behavior support predictions of the acceptance threshold model
Department of Animal and Plant Sciences, University of Sheffield, Sheffield, S10 2TN, UK
Address correspondence to S. G. Downs, Laboratory of Apiculture and Social Insects, Department of Animal and Plant Sciences, University of Sheffield, Sheffield S10 2TN, UK. E-mail: bop97sgd{at}sheffield.ac.uk .
Received 5 July 1999; revised 1 October 1999; accepted 10 October 1999.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The acceptance threshold model predicts that in a fluctuating environment a recognition system should be adaptive rather than fixed. In particular, discriminating individuals, such as guards at a nest entrance, should be less permissive to conspecifics when both the frequency of non-nest-mate contact and the cost of accepting non-nest mates is high. We tested these predictions by studying honey bee guarding during a period in which nectar conditions changed from dearth to abundance. Initially, during nectar dearth, individual guards accepted 80% of introduced nest mates and 25% of non-nest mates. As nectar conditions improved, both the intensity of robbing and guarding and the cost of non-nest-mate acceptance declined. In response, individual guards became more permissive to nest mates and non-nest mates until eventually an "accept-all" threshold occurred—all nest mates and non-nest mates were accepted. These data are consistent with a shifting acceptance threshold and provide the first field data to support the model. A simple linear relationship occurred between the number of guards and the number of fights, 9:1, observed at the hive entrance, suggesting that guarding may be regulated by intruder intensity or otherwise regulated in an adaptive manner.
Key words: acceptance thresholds, conspecific recognition, honey bees, odor convergence.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Recognition occurs in many contexts and in most organisms and ranges from the recognition of self to the recognition of a suitable mating partner (for review, see Sherman et al., 1997
). In the social insects, recognition by workers occurs in two
main contexts, intracolony and intercolony
(Ratnieks, 1991). The former
occurs as part of the discriminatory rearing of reproductives to enhance
worker inclusive fitness (for review, see
Bourke and Franks, 1995;
Crozier and Pamilo, 1996;
Ratnieks and Reeve, 1992).
Intercolony recognition involves the discrimination of nest mates from
non-nest mates, including conspecific non-nest mates. Nest-mate versus
non-nest-mate recognition is important because many social insects can harm
other colonies of their own species. Harm occurs in various ways such as
competition for territory, cannibalism, slave making, and robbing food. For
example, African weaver ant (Oecophylla longinoda) colonies compete
with each other for territories, with one colony eventually excluding the
other
(Hölldobler
and Wilson, 1978). Such ant wars have become cannibalistic in
Formica polyctena, which kill ants from other colonies to use as food
(Driessen et al., 1984).
Intercolony cannibalism also occurs in the Japanese paper wasps Polistes
chinensis antennalis (Kasuya et al.,
1980). Territorial raids appear to have led to slave making in a
few ant species. Colonies of Myrmecocystus mimicus raid the nests of
conspecifics and capture brood, which they adopt as nest mates
(Hölldobler,
1981). Honey bee (Apis mellifera) workers steal honey
from conspecific colonies during times of nectar shortage
(Free, 1977).
Insect nests have guards who deter entry by both conspecific and
allospecific intruders (honey bees: Free,
1977; ants:
Hölldobler
and Wilson, 1990; termites:
Wilson, 1971). Recognizing
allospecific intruders will usually be simpler than recognizing conspecific
intruders because conspecifics are expected to be more similar to each other.
In a perfect recognition system a guard would accept all nest mates and reject
all non-nest mates (Sherman et al.,
1997). To achieve this a guard could learn the phenotype of her
nest mates and use this as a recognition template
(Lacy and Sherman, 1983). On
encountering individuals at the nest entrance, the guard compares the template
with the phenotype of the entering individual [the latter is known as the
recognition "cue" (Getz,
1982)]. Discriminating between nest mates and non-nest mates may
not, however, be this simple if nest mates and non-nest mates have overlapping
recognition cues (Getz, 1981,
1991;
Lacy and Sherman, 1983). Also,
depending on the nature of odors used in colony recognition, nest mates will
likely have a range of phenotypes (Getz,
1982; Wilson,
1971). This will make discrimination less straightforward because
a guard cannot simply reject all those individuals that do not match the
template because she will reject some nest mates. To reduce nest-mate
rejection a guard may not require an exact match between template and cue.
However, allowing too much dissimilarity between template and cue will result
in more non-nest mates being accepted.
In a theoretical study, Reeve
(1989) investigated factors
that affect how conspecifics behave toward each other in a recognition context
in order to maximize their inclusive fitness. Reeve defined the acceptance
threshold as the maximum amount of dissimilarity between template and cue that
a guard would tolerate without rejecting the entering individual (see
Figure 5a). Above the
threshold, conspecifics are classified as non-nest mates and rejected, and
below the threshold, they are classified as nest mates and accepted. Rather
than having identical odors, nest mates and non-nest mates are generally
thought to have a distribution of odors
(Getz, 1982). Assuming that
nest mates and non-nest mates have overlapping recognition cues,
discrimination errors are inevitable, and two types of errors can be made:
reject nest mates (rejection error) and accept non-nest mates (acceptance
error) (Reeve, 1989). The
probabilities of these two errors will be negatively correlated. By adopting a
sufficiently permissive acceptance threshold, a guard can accept all nest
mates, thereby reducing the rejection error to zero. However, a higher
proportion of non-nest mates will be accepted, increasing the acceptance
error. Conversely, a restrictive acceptance threshold will reduce the
proportion of non-nest mates accepted but will also cause increased rejection
of nest mates. If the threshold can be adjusted, what is the optimal threshold
for guards to adopt? Reeve showed that the optimal threshold varies with
factors such as the cost of accepting non-nest mates, the benefits of
accepting nest mates, and the frequencies with which nest mates and non-nest
mates are encountered. For example, when the frequency or cost of intrusion by
non-nest mates is high, guards should adopt a restrictive threshold. As the
frequency or cost of intrusion declines, the threshold should become more
permissive, thereby rejecting fewer nest mates, assuming that rejection of
nest mates has a cost.
|
In the honey bee (Apis mellifera) workers steal honey from other
colonies when nectar is in shortage
(Seeley, 1985). Singly, a
robber bee steals only a small amount of honey per trip, which is
insignificant when considered against the many kilograms that are typically
stored (Seeley, 1995).
However, the cost can be magnified due to recruitment using the waggle dance
(von Frisch, 1967), which
allows large numbers of workers to be directed to the new food source.
Unchecked robbing can lead to the loss of all honey stores and death of the
victim colony (Winston, 1987).
Guard workers stationed at the nest entrance prevent robbing by checking
entering bees and excluding non-nest mates
(Butler and Free, 1952;
Free, 1954). When a guard
encounters another bee, it licks and antennates her head and body, suggesting
that cuticular odors are involved in conspecific discrimination
(Breed, 1983;
Kalmus and Ribbands, 1952).
There is evidence that such odors are derived from beeswax
(Breed, 1983; Breed et al.,
1988,
1995) and floral or honey
scents (Ribbands, 1953;
Saleh-Mghir, 1992), giving
each colony a more or less distinct colony odor. The latter suggests that
conspecifics may share odors due to shared floral odors and that nest mates
may have variable rather than identical odors because foragers visit different
flower species.
Recognition by honey bee guards in an ideal model system to study adaptive
shifts in acceptance threshold, thereby testing theoretical predictions
(Reeve, 1989), because nectar
availability (Seeley, 1995)
and robbing change dramatically over time depending on the local flora. As a
result, honey bee colonies are subject to constantly changing costs and
benefits of accepting non-nest mates and nest mates. Based on Reeve
(1989), we predicted that the
acceptance threshold should become more permissive and that levels of guarding
should decline as nectar becomes more abundant. We tested these predictions by
making a field study of the acceptance behavior of honey bee guards to nest
mate and non-nest mate foragers across a 30-day period, during which nectar
conditions changed from poor to excellent. Guarding declined and acceptance
became much more permissive as nectar conditions improved, strongly supporting
the model's predictions.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study organism
Honey bees (Apis mellifera) studied were of mixed European race, as used in U.S. commercial beekeeping. The colonies were queenright and housed in standard Langstroth hives. The colonies were all of similar size and had approximately 20,000-30,000 workers and 4-6 frames of brood. Initially, each hive had two hive bodies (one deep and one medium), giving a total volume of approximately 70 1. Additional hive bodies (two deep bodies) for honey storage were added on day 67 (March 8). Each hive had a standard long bottom board that extended 5 cm beyond the hive bodies to facilitate observations of the hive entrance, but before day 67 the entrance width was halved. The entrance was then opened fully to 3 cm high by 35 cm wide to accommodate the increased foraging activity caused by the citrus bloom. To minimize hive disturbance, the hives were only smoked three times during the study period to allow routine bee management. After smoking, hives were allowed to settle for 24 h before further data collection.
Study site
The study was conducted at the Archbold Biological Station, Lake Placid,
Florida, USA, during February and March 1998. We chose this site and time of
year because the floral and climatic conditions were ideal for studying
guarding behavior as environmental conditions change from dearth to abundance
of nectar. A period of great nectar abundance occurs in March when the
extensive citrus groves bordering Archbold bloom. In February, before citrus
bloom, nectar conditions are poor, and hives lose weight as they consume
stored honey (see Figure 1a). The climate is such (mean maximum daily temperature for February = 23°C;
data from Archbold weather station) that honey bees can fly actively before
the citrus bloom, thereby making robbing possible. This is in contrast to more
temperate regions where the temperature before spring bloom is usually too low
for flight.
|
Two apiaries were set up on day 63 (March 4) by moving in hives from
another apiary that was approximately 30 km away: Bee House, and Red Hill. Bee
House had four hives, with 40-60 m between hives. Red Hill had 12 hives
(although only 7 were used for data collection) separated by 4-10 m. We used
two apiaries to create two levels of potential robbing intensity, with Red
Hill having both more hives and less separation between hives. In addition,
the nearby orange groves had apiaries belonging to commercial beekeepers, from
which potential robbing bees could come. Bee House was approximately 1 km from
the nearest citrus grove, and Red Hill was 0.5 km from the nearest grove.
Because honey bees will fly as far as 12 km to flowers
(Seeley, 1995;
von Frisch, 1967) and usually
fly more than 1 km (Seeley,
1995), both apiaries were well placed for the bees to forage on
citrus flowers.
Data collection
Nectar availability
Two standard methods were used to quantify nectar availability: hive
weights and sucrose feeders. Most hive weight change occurring over periods of
less than a few days can be attributed to honey consumption (weight loss) or
nectar collection (weight gain) (Seeley,
1995). We weighed hives using a portable tripod, 4:1 pulley block,
and 0-50 kg dial balance (Salter 233, sensitivity 50 g). Four Red Hill hives
were selected and weighed in the morning before the bees had begun to fly. We
weighed hives every 2 days, except during the final 8 days of the study when
they were weighed daily.
Honey bee foragers will readily collect sucrose solution from artificial
feeders away from the hive (Lindauer,
1961; Seeley,
1995; von Frisch,
1967). However, the minimum concentration that they will collect
depends on nectar availability (Lindauer,
1961). For example, Lindauer
(1961) found that the minimum
concentration collected during the main nectar flow was 2 M, whereas in the
period of dearth after the nectar flow the minimum was less than 0.2 M and as
low as 0.05 M. We set up four feeders
(Seeley, 1995;
von Frisch, 1967) on a table
30 m from the Red Hill apiary. Every second day at 0900 h the feeders were
replenished with syrup (0.1 M, 0.5 M, 1 M, 2 M). At 1200 h we observed the
feeders to determine those from which bees were collecting syrup.
Quantifying guarding and intruder intensities
Guarding and robbing intensities were recorded by observing the entrances
of eight hives, four at Red Hill and four at Bee House. We observed each hive
nine times during the day between 1030 and 1500 h. This was done in three sets
of three observations to reduce the amount of time lost traveling between the
two apiaries. Observations were made on every second day of the study, except
for the final 8 days when they were made daily. A 15-min interval was allowed
between successive observations at a hive. This ensured that fights observed
were independent of those at a previous observation. (Of 10 fights measured,
the longest lasted 152 s; mean = 58.8 s, range = 13-152 s, SD = 48.1 s.) When
making observations the observer stood quietly to the side of the hive to
avoid disturbing the bees. The behaviors at the hive entrance could easily be
seen from this distance. This method does not attempt to quantify absolute
guarding and fighting levels because some guards and fights were probably out
of sight inside the entrance but is designed to detect relative changes.
On each observation we recorded the number of bees seen guarding and the
number of fights. Bees were classified as guards if they showed typical
guarding behaviors (Butler and Free,
1952): patroling the entrance board with their wings held open,
chasing landing bees, inspecting other bees (see below), and fighting (see
below). Inspection involves antenation or licking of the subject's body, or
both. After inspection the guard does one of two things: the subject is either
left to enter the hive or is attacked. The latter involves biting of the
wings, antennae, legs, and sometimes attempted stinging of the abdomen, all of
which we classified as fighting.
Acceptance behavior of guards
We used three hives at Red Hill for the acceptance behavior experiment. A
technique was needed that allowed us to observe the behavior of guards toward
nest mates and non-nest mates which was simple to implement, as natural as
possible, and could give consistent results. Initially we tried to stage
encounters by collecting a guard from the hive entrance and placing her in a
container, to which a second bee was added. This did not work because guards
stopped behaving like guards when removed from the hive entrance. Therefore,
an assay was needed that closely simulated natural colony entrance conditions,
particularly the interactions between guards and nest-mate and non-nest mate
workers. Preliminary experiments showed that direct use of intruders (i.e.,
workers collected from the entrance of another hive) was impractical because
introduction was unreliable. Although workers could easily be caught and
placed at a hive entrance, they frequently flew away. At the other extreme, we
also tried using dead bees. However, workers killed by freezing and used as
intruders immediately after being defrosted were indiscriminately ejected,
similar to the undertaking behavior observed by Visscher
(1983). The technique we
finally adopted was as follows. A hive entrance was blocked for 1 min, and
four returning foragers were collected in individual plastic vials. The vials
were then cooled in a portable ice chest until the bees appeared unable to fly
but were still walking. At this point they could be handled using forceps and
would not fly away when put at a hive entrance. To test the reliability of
this assay we performed 10 nest-mate and 10 non-nest-mate introductions on a
single hive. The results showed a clear discrimination between acceptance
levels of nest mates and non-nest mates, suggesting that the assay was
suitable for the study (Figure
1d). When introducing a chilled bee at the hive entrance, the
experimenter stood to the side of the hive and was careful not to disturb the
hive by making sudden movements or by jolting the hive. To standardize the
returning bees introduced, bees carrying pollen were not used.
The protocol for gathering data on the behavior of a guard toward introduced bees was as follows. We placed a chilled bee on the entrance board of a hive. The entrance was continuously observed, and the behavior of the first guard contacting the introduced bee was recorded. Guard behavior was classified as follows: accepted (introduced bee inspected by the guard and allowed to enter the hive), rejected (introduced bee either bitten or stung by guard, although stinging was rarely observed). In some cases more than one guard contacted the introduced bee. When this occurred, observations were made on the first guard that contacted the introduced bee. Of each group of four chilled bees obtained from the entrance of an experimental hive, two were introduced back into their own hive (nest mates) and one each into the other two hives (non-nest mates). We repeated the procedure with the other two experimental hives, and this constituted a round of introductions. Different vials were used for each hive to eliminate the possibility of inappropriate colony odors being transferred to the bees via the vials.
Preliminary studies showed that repeated introductions alerted the hive and
increased the guarding intensity. Similarly, Butler and Free
(1952) found that guarding
intensity increased if they introduced more than approximately 50
non-nest-mate bees. To prevent this from compromising our results, we allowed
1 h between two rounds of introductions. During 1 day, five rounds of
introductions were made, generating a total of 10 nest-mate and non-nest-mate
introductions per hive per day. This is small in comparison to the probable
rates of natural intrusion (Figure
1c) and to the experimental levels found to increase guarding
(Butler and Free, 1952). The
data were collected on every second day of the study, except for the final 7
days, when data were collected daily.
Although a blind experimental procedure is recommended in studies of
recognition (Gamboa et al.,
1991), it was impossible to adopt such a procedure because one
person (S.G.D.) carried out this experiment. However, because there was such a
large difference between nest-mate and non-nest-mate acceptance at the start
of the experiment, it is unlikely that the results were compromised by not
having a blind procedure. In addition, the study was not primarily
investigating nest-mate recognition, which has been previously studied (Breed,
1983,
1987; Breed et al.
1985,
1988;
Downs and Ratnieks, 1999), and
so the lack of blind testing with respect to worker affiliation is not a
problem. Nectar conditions could have influenced the data collection. We were,
however, blind both to the nectar availability (because we weighed hives the
day after collecting acceptance data) and the nectar load of an introduced
worker.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Nectar availability
In February (days 40-60) nectar conditions were poor, and hives lost weight (Figure 1a). Conditions improved and colonies began to gain weight slightly (about 1 kg per day) in early March (days 62-72), eventually becoming so good that on days 73-78 colonies gained approximately 2-3 kg per day. The sudden weight increases that occurred on days 50, 54, and 70 were caused by overnight rain soaking the hives rather than by nectar collection. Changes in hive weight are mirrored by the sucrose feeder data (Figure 1b). Initially (days 40-46), bees collected from all solutions provided. However, as conditions improved, bees sequentially ignored the more dilute solutions, until, eventually, by day 73, they would only collect 2 M sucrose. These changes correlate well with casual observation of citrus bloom. Trees were not in bloom in early to mid-February and were in full bloom by early to mid-March.
Quantifying guarding and intruder intensities
Guarding declined at both apiaries (Bee House: r2 =.65,
df = 1,18, p <.001; Red Hill: r2 =.96, df =
1,20, p <.001) during the study
(Figure 1c). Colonies at the
Red Hill apiary had significantly more guards than those at the Bee House
apiary (ANCOVA: F = 12.91, df = 1,39, p <.001), although
they both declined to approximately three guards by day 72. Fighting declined
at both apiaries (Bee House: r2 =.75, df = 1,18,
p <.001; Red Hill: r2 =.75, df = 1,20,
p <.001) until zero fights were observed on day 70
(Figure 1c). Fighting
intensities were higher at Red Hill than at Bee House probably due to the
higher colony density at Red Hill (ANCOVA: F = 6.04, df = 1,39,
p =.018). Initially guarding and fighting levels were much lower at
Bee House and increased during the first 6 days of the study, before gradually
declining. In comparison, Red Hill did not have a period of initial increases
but instead declined from the outset.
Acceptance behavior of guards
Acceptance of nest mates and non-nest mates increased significantly during
the study (nest mates acceptance: r2 =.82, df = 1, 16,
p <.001; non-nest mates: r2 =.96, df = 1, 16,
p <.001; Figure
1d). At the start of the study (day 47), 80% of introduced nest
mates, but only 25% of introduced non-nest mates, were accepted by guards.
Nest-mate and non-nest-mate acceptance then increased and converged so that by
day 75 all introduced nest mates and non-nest mates were accepted. Despite the
convergence of nest-mate and non-nest-mate acceptance, nest-mate acceptance
was significantly higher than non-nest-mate acceptance (ANCOVA: F =
61.6, df = 1, 33, p <.001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
During the study, nectar availability changed from extreme dearth to abundance, as shown by the hive weight and syrup collection data. In the period of nectar dearth (days 40-60), robbing intensity was at its greatest in both apiaries, as shown by the numbers of fights observed at the hive entrance (Figure 1c). From day 60 onward, the citrus trees began to bloom, and the number of fights declined, eventually reaching zero by day 72 (March 13). Zero fighting coincides with high hive weight gains (approximately 2-3 kg per day) and suggests that when nectar is abundant bees do not attempt to rob honey from conspecifics. Presumably this is because robbing is then less profitable than collecting nectar from flowers. More fights were observed at the Red Hill apiary (Figure 1c). This was as predicted given that Red Hill hives were both more numerous and closer together than at the Bee House apiary. The difference was particularly obvious at the start of the experiment, days 40-50 (Feb 9 to, 19), when there was a large difference in the number of fights. This was probably because potential robber bees took several days to detect the relatively isolated Bee House hives after they were moved into the apiary on day 37.
Guarding intensity declined at both apiaries during the study
(Figure 1c). In early February
guarding levels were highest and coincided with the period of most intense
fighting (Figure 1c). Guarding
declined from a peak of approximately 30 guards per entrance on day 46 to the
low level of 3 guards per entrance on day 73. This decline was presumably
because colonies needed fewer guards to defend their entrance as attempted
robbing became less frequent. Although robbing had apparently ceased by day
72, as evidenced by the absence of fights, guarding still persisted at a low
level. Guarding probably did not decline to zero because honey bees suffer
robbing and harm from other species and must remain vigilant to such threats
(Seeley, 1985). The declines
in guarding and fighting were highly correlated (Bee House:
R2 =.69, df = 1, 78, p <.001; Red Hill:
R2 =.80, df = 1, 86, p <.001;
Figure 2) and did not differ
significantly between the two apiaries (ANCOVA: F = 2.02, df = 1,
165, p >.05), although robbing and guarding tended to be higher at
Red Hill (Figure 1c). Combining
the data from both apiaries, the overall regression relationship was number of
guards = 5.1 + 8.9 (number of fights) (i.e., a ratio of approximately 9:1,
guards to fights) (Figure 2).
The good fit of the linear model to these data across a wide range in numbers
of guards and fights and two apiaries suggests that colonies have a
generalized guarding response to robbing. Our data are observational and so
cannot show cause and effect. However, Butler and Free
(1952) found that guard levels
could be elevated by introducing non-nest mates to the hive entrance,
suggesting that colonies regulate guard numbers in response to robbing and
intrusion. Alternatively, if the intrusion rate remained constant but the
number of guards declined, we would still have observed a decline in fighting
simply because more intruders were evading detection by guards. This, however,
was unlikely to be occurring because it is well known that robbing only occurs
during periods of nectar dearth (Butler,
1954; Winston,
1987), and so the level of robbing almost certainly declined
during the study. Further studies are therefore needed to show whether
guarding is regulated by intruder intensity in an adaptive manner.
|
Rather than regulating the number of guards, why don't honey bee colonies
simply have a constant and high number of guards? The answer is probably
because nectar conditions vary during the year, resulting in a continually
fluctuating robbing threat. Typically, a year consists of a few ephemeral
nectar flows, lasting from a few days to 2 weeks, interspersed between periods
of modest nectar abundance and dearth
(Seeley, 1995). A colony's
foraging during these periods of abundance can often dictate whether the hive
survives the winter, for which a reserve of approximately 20 kg of honey is
needed (Seeley, 1995), so
colonies should maximize foraging during nectar flows. Selection will
therefore favor an adaptive response over a fixed guarding level because it
will allow colonies to allocate labor in a way that is more appropriate to
conditions.
In addition to the colony-level changes in guarding and fighting discussed
above, there was also a dramatic change in the acceptance behavior of
individual guards. At the start of the study (day 47) guards were
nonpermissive in their acceptance behavior, rejecting almost all non-nest
mates (about 75%) and a considerable proportion of nest mates (about 20%;
Figure 1d). As nectar
availability increased, guards gradually became more permissive, until, by day
76, all bees, both nest mates and non-nest mates, were accepted—an
accept-all strategy (Reeve,
1989). Figure 3
shows that all three colonies demonstrated broadly similar acceptance
threshold shifts, although there were slight differences. For example, colony
B was generally less permissive on any given day (e.g., on day 49 rejection of
non-nest mates was 20% versus 50% in the other two colonies). It seems that
colony B lagged behind the other colonies in the gradual relaxation of its
acceptance threshold, finally reaching accept-all 4 days later than the other
colonies. Colony C is slightly different from the other two hives because any
given shift in nest-mate acceptance has a greater effect on non-nest-mate
acceptance, as seen by the larger gradient in
Figure 4 (A and B have similar,
smaller gradients). To compare these minor differences, we plotted nest-mate
acceptance against non-nest-mate acceptance for each colony and tested this
statistically (Figure 4). There
was no significant difference between the colonies (ANCOVA: F = 2.52,
df = 1, 50, p >.05). The similarity in pattern can most clearly be
seen by visually comparing days 57, 63, and 57 for hives A, B, and C. On these
days all three colonies had reached a point at which they accepted all nest
mates and accepted at least 60% of non-nest mates. Using the means for
nest-mate and non-nest-mate acceptance gives a slope with the equation:
%non-nest-mate acceptance = 0.69 + 0.35 ·(%nest-mate acceptance). In
other words, a 10% increase in nest-mate acceptance results in a 35% increase
in non-nest-mate acceptance.
|
|
The acceptance data are consistent with a shifting rather than a fixed
acceptance threshold. Initially the threshold is quite nonpermissive when
there is high intensity robbing. This is as predicted
(Reeve, 1989) and is adaptive
in that it ensures high levels of non-nest-mate rejection during periods when
acceptance errors are costly and non-nest-mate intrusion is frequent. The
incidence of intrusion and the cost of accepting non-nest mates then declines,
allowing the threshold to gradually become more permissive, until by the end
it was totally permissive (Figure
5a). Accept-all occurs at a time when there is no fighting and
nectar is very abundant (compare Figure
1c with 1d), and so
the colony does not face exploitation. The absence of robbing suggests that
accept-all has no cost in terms of encouraging robbing. There are also
potential ergonomic benefits of the accept-all strategy. In periods of nectar
shortage, foragers will frequently be unsuccessful in trying to locate nectar
or, if successful, make long foraging trips
(Ribbands, 1953). As a result,
delaying their entry into the nest will not severely effect foraging
efficiency. However, during a nectar flow, absence of guarding may reduce the
time taken for a forager to enter the nest, thereby allowing foragers to make
more trips. This argument is illustrated by the following example. If a
forager, on average, spends 100 min foraging and unloading nectar (a foraging
cycle) during a dearth and 30 min in a flow, then a 0.5 min delay caused by
guard inspection (Butler and Free,
1952) will increase the cycle by only 0.5% during dearth but by
1.5% during a flow. In other words, guarding delays to foragers are more
ergonomically costly during a flow.
Two other studies also support a shifting acceptance threshold. In
laboratory experiments it was found that wasps (Polistes dominulus)
were more aggressive to non-nest mates when cues were present that indicated
they were at their nest (i.e., with a nest fragment or nest mate)
(Starks et al., 1998). Starks
et al. propose that wasp acceptance is context dependent, being less
permissive to non-nest mates when the nest is close by and therefore when the
costs of accepting non-nest mates are high. Inoue et al.
(1999) showed that regularly
disturbing colonies of Melipona panamica housed in observation hives
increases intranidal guard rejection of nest mates and non-nest mates. To
disturb the hives, they opened the nest and removed a portion of the
involucrum around brood cells. This was supposed to simulate a natural enemy
attack. Our study is novel in that it is the first to support Reeve's
(1989) predictions using
guards at actual nest entrances and natural environmental changes that affect
non-nest-mate acceptance costs and intruder frequency. Our study not only
confirms that there is a shift in acceptance but also allows the trajectory of
the acceptance threshold to be plotted
(Figure 5a).
Although this study provides data that strongly support the hypothesis that conspecific acceptance thresholds shift adaptively, an alternative hypothesis cannot be excluded as part of the mechanism for increased acceptance: Increase in acceptance could be due to converging colony odors rather than to a shifting threshold (Figure 5b). Colony odor convergence is a possibility because by the end of the study bees from all colonies were foraging almost exclusively on citrus. (From casual observations of pollen color, it was apparent that the bees were foraging from different flower species before the citrus bloom). If odors acquired from flowers are of overriding importance in colony odor, this will make discrimination of non-nest mates more difficult. Figure 5 illustrates how both the acceptance threshold hypothesis and the colony odor convergence hypothesis could explain our data. During the period of nectar dearth at the start of study, nest mates and non-nest mates had discrete colony odors because they were foraging on a wide range of sources, leading to differences among colonies, making non-nest-mate discrimination relatively easy (Figure 5a, b). From this identical starting point, the mechanisms then diverge. Focusing first on the odor convergence hypothesis, it can be seen that as the nectar flow begins, nest mates become more similar and non-nest-mate and nest-mate odors converge (Figure 5b). This means that non-nest-mate discrimination is now more difficult than at the start of the study. As the citrus flow intensifies, colony odors gradually become less distinct and begin to overlap. Eventually the nest mates and non-nest mates become indistinguishable, and accept-all occurs (Figure 5b).
The odor convergence hypothesis, therefore, can logically account for the
increase in acceptance that we observed without invoking any shift in an
acceptance threshold. Alternatively, the odor distributions could remain
constant, and instead the acceptance threshold becomes more permissive
(Figure 5a), until eventually
the threshold is completely permissive and accept-all occurs (5a iii). Our
study cannot reject either hypothesis. However, both previous research and
data from this study suggest that converging odors cannot completely explain
the increased acceptance of non-nest mates. Breed et al.
(1988,
1995) have shown that beeswax
is used for colony recognition in honey bees, and this cue should be
independent of environmentally acquired floral odors. To test this hypothesis,
Bowden et al. (1998) exposed
workers to floral oils and staged encounters between workers and guards. They
found that guards did not respond differently to bees that had been exposed to
floral oils than unexposed guard bees. This finding has since been confirmed
in a field study by Downs et al. (in
press), although they found that floral oil treatment did delay
guard rejection of non-nest mates, but it did not affect the overall
probability of guard acceptance. Therefore, if floral oils do have a role in
nest-mate recognition, it appears to be a minor one and is secondary to cues
acquired from comb wax. This reasoning makes intuitive sense because, if
floral scents were important in nest-mate recognition, foragers that had
discovered a new food source would frequently be rejected.
We also observed small but rapid changes in nest-mate and non-nest-mate acceptance (e.g., days 53-57 and 69-72 in colony B, 57-61 in colony C; Figure 3), and it seems unlikely that these sudden changes could have been caused solely by changes in colony odor, particularly as some occurred during the nectar flow when colonies were mainly foraging on citrus and should have had similar floral odors. In addition, the odor convergence hypothesis does not explain why nest-mate acceptance started at an intermediary level and then rose during the study. If a colony had foragers visiting a variety of flower (odor) sources before the study, the colony (gestalt) odor should have been the same for all the colony members, and acceptance of nest mates should have been close to 100%. The change in nest-mate acceptance that we observed is predicted by the acceptance threshold model, but not by the odor convergence hypothesis. Furthermore, a direct test of the odor convergence hypothesis in the honey bee showed that guard bees did not have a higher probability of accepting non-nest mates from colonies that had been fed syrup with the same odor versus a different odor (Downs et al., unpublished data).
Our data, therefore, strongly support the hypothesis that an adaptive shift in guard acceptance thresholds occurred. However, to show whether such an adaptive shift was the exclusive cause of increased acceptance would require an experiment that can reject either hypotheses. For example, colony odors could be artificially changed during a period of relatively constant nectar conditions, or non-nest-mate intrusion could be increased experimentally to determine whether this causes rapid changes in acceptance. Sudden changes in acceptance in the latter experiment would reject the odor convergence hypothesis because it would be impossible for colony odors to change so rapidly (Downs and Ratnieks, unpublished data).
Our study shows that honey bee colonies regulate guarding in an apparently
adaptive way in response to changing environmental conditions. Previous
studies have shown that honey bees have sophisticated mechanisms for
regulating important colony activities such as comb building, pollen
collection, water collection, and nectar collection
(Seeley, 1995). Guarding is
probably another aspect of social life subject to careful regulation. Our data
show that changes in guarding affect both individuals (i.e., acceptance
behavior) and the colony as a whole (i.e., number of guards at the hive
entrance). Guarding is also an aspect of colony life amenable to experimental
investigation, making the regulation of guarding in honey bees an ideal system
for further studies of the regulation of colony level activities in insect
societies.
|
ACKNOWLEDGEMENTS |
|---|
S.G.D. was funded by a Biotechnology and Biological Sciences Research Council PhD studentship and travel grant. We also thank the Archbold Biological Station for facilities and support, Roger Morse for his expert advice, Paul Cappy for the loan of bee hives, and the referees for their useful comments.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bourke AFG, Franks NR, 1995. Social evolution in ants. Princeton, New Jersey: Princeton University Press.
Bowden RM, Williamson S, Breed MD, 1998. Floral oils: their effects on nestmate recognition in the honey bee, Apis mellifera. Insectes Soc 45: 209-214.
Breed MD, 1983. Nestmate recognition in honey bees. Anim Behav 31: 86-91.
Breed MD, 1987. Multiple inputs in the nestmate discrimination system of the honey bee. In: Chemistry and biology of social insects (Proceedings of the Tenth International Congress of the International Union for the Study of Social Insects, Munich, 1986) (Eder J, Rembold H, eds). Perperny, Munich: Verlag; 461-462.
Breed MD, Butler L, Stiller TM, 1985. Kin
discrimination by workers in genetically mixed groups. Proc Natl Acad
Sci USA 82:
3058-3061.
Breed MD, Garry MF, Pearce AN, Hibbard BE, Bjostad LB, Page Jr RE, 1995. The role of comb wax in honey bee nestmate recognition. Anim Behav 50: 489-496.
Breed MD, Williams KR, Fewell JH, 1988. Comb wax
mediates the acquisition of nest-mate recognition cues in honey bees.
Proc Natl Acad Sci USA 85:
8766-8769.
Butler CG, 1954. The world of the honeybee. London: Collins.
Butler CG, Free JB, 1952. The behaviour of worker honeybees at the hive entrance. Behaviour 4: 263-291.
Crozier RH, Pamilo P, 1996. Evolution of social insect colonies: sex allocation and kin recognition. Oxford: Oxford University Press.
Downs SG, Ratnieks FLW, 1999. Recognition of conspecifics by honeybee guards uses non-heritable cues acquired in the adult stage. Anim Behav 58: 643-648.[Web of Science][Medline]
Downs SG, Ratnieks FLW, Jefferies SL, Rigby HE, in press. The role of floral oils in the nestmate recognition system of honey bees (Apis mellifera L). Apidologie.
Driessen GJJ, Van Raalte AT, De Bruyn GJ, 1984. Cannibalism in the red wood ant, Formica polyctena (Hymenoptera: Formicidae). Oecologia 63: 13-22.
Free JB, 1954. The behaviour of robber honeybees. Behaviour 7: 233-240.
Free JB, 1977. The social organization of honey bees. London: Edward Arnold.
Gamboa GJ, Reeve HK, Holmes WG, 1991. Conceptual issues and methodology in kin recognition research—a critical discussion. Ethology 88: 109-217.
Getz WM, 1981. Genetically based kin recognition system. J Theor Biol 92: 209-226.
Getz WM, 1982. An analysis of learned kin recognition in Hymenoptera. J Theor Biol 99: 585-597.
Getz WM, 1991. The honey bee as a model kin recognition system. In: Kin recognition (Hepper PG eds). Cambridge: Cambridge University Press.
Hölldobler B, 1981. Foraging and spatiotemporal territories in the honey ant Myrmecocystus mimicus Wheeler (Hymenoptera: Formicidae). Behav Ecol Sociobiol 9: 301-314.
Hölldobler B, Wilson EO, 1978. The multiple recruitment system of the African weaver ant Oecophylla longinoda (Latrielle) (Hymenoptera: Formicidae). Behav Ecol Sociobiol 3: 19-60.
Hölldobler B, Wilson EO, 1990. The ants. Cambridge, Massachusetts: Harvard University Press.
Inoue T, Roubik DW, Suka T, 1999. Nestmate recognition in the stingless bee Melipona panamica (Apidae, Meliponini). Insectes Soc 46: 208-218.
Kalmus H, Ribbands CR, 1952. The origin of the odours by which honeybees distinguish their companions. Proc R Soc Lond B 140: 50-59.[Medline]
Kasuya E, Hibino Y, Itô Y, 1980. On `intercolonial' cannibalism in Japanese paper wasps Polistes chinensis antennalis Pérez and P. jadwigae Dalla Torre (Hymenoptera: Vespidae). Res Popul Ecol 22: 255-262.
Lacy RC, Sherman PW, 1983. Kin recognition by phenotype matching. Am Nat 116: 489-512.
Lindauer M, 1961. Communication among social bees. Cambridge, Massachusetts: Harvard University Press.
Ratnieks FLW, 1991. The evolution of genetic cue diversity in social hymenoptera. Am Nat 137: 202-226.
Ratnieks FLW, Reeve HK, 1992. Conflict in single queen Hymenopteran societies: the structure of conflict and processes that reduce conflict in advanced eusocial species. J Theor Biol 158: 33-65.
Reeve HK, 1989. The evolution of conspecific acceptance thresholds. Am Nat 133: 407-435.
Ribbands CR, 1953. The behaviour and social life of bees. London: Bee Research Association.
Saleh-Mghir, E. 1992. Rôle de la nourriture dans la reconnaissance des congénères chez l'abeille, Apis mellifera L. Compt Rend Acad Sci Paris 315: 343-346.
Seeley TD, 1985. Honeybee ecology. Princeton, New Jersey: Princeton University Press.
Seeley TD, 1995. The wisdom of the hive. Cambridge, Massachusetts: Harvard University Press.
Sherman PW, Reeve HK, Pfennig DW, 1997. Recognition systems. In: Behavioural ecology (Krebs JR, Davies NB, eds). Oxford: Blackwell Science; 69-96.
Starks PT, Fischer DJ, Watson RE, Melikian L, Nath SD, 1998. Context-dependent nestmate discrimination in the paper wasp, Polistes dominulus: a critical test of the optimal acceptance threshold model. Anim Behav 56: 449-458.[Web of Science][Medline]
Visscher PK, 1983. The honey bee way of death: necrophoric behaviour in Apis mellifera colonies. Anim Behav 31: 1070-1076.
von Frisch K, 1967. The dance language and orientation of bees. Cambridge, Massachusetts: Harvard University Press.
Wilson EO, 1971. The insect societies. Cambridge, Massachusetts: Harvard University Press.
Winston ML, 1987. The biology of the honey bee. Cambridge, Massachusetts: Harvard University Press.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Segoli, T. Keasar, A. R. Harari, and A. Bouskila Limited kin discrimination abilities mediate tolerance toward relatives in polyembryonic parasitoid wasps Behav. Ecol., November 1, 2009; 20(6): 1262 - 1267. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J Couvillon, J. P Caple, S. L Endsor, M. Karcher, T. E Russell, D. E Storey, and F. L.W Ratnieks Nest-mate recognition template of guard honeybees (Apis mellifera) is modified by wax comb transfer Biol Lett, June 22, 2007; 3(3): 228 - 230. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Cooney Colony defense in Damaraland mole-rats, Cryptomys damarensis Behav. Ecol., March 1, 2002; 13(2): 160 - 162. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. G. Downs, F. L. W. Ratnieks, N. S. Badcock, and A. Mynott Honeybee guards do not use food-derived odors to recognize non-nest mates: a test of the Odor Convergence hypothesis Behav. Ecol., January 1, 2001; 12(1): 47 - 50. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||