Behavioral Ecology Vol. 11 No. 6: 676-685
© 2000 International Society for Behavioral Ecology
Nest mate recognition in ants with complex colonies: within- and between-population variation
Department of Biology, University of Vermont, Burlington, VT 05405, USA
Address correspondence to J. M. Herbers, who is now at the Department of Biology, Colorado State University, Fort Collins, CO 80523. E-mail: herbers{at}lamar.colostate.edu .
Received 3 November 1999; revised 18 May 2000; accepted 25 May 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The ability to recognize kin is widespread, and especially important in highly social organisms. We studied kin recognition by assessing patterns of aggression within and between nests of the ant Leptothorax longispinosus. Colonies of this species can be fractionated into subunits, a condition called polydomy. The problem of recognizing relatives is therefore more complex when those relatives can live in two or more different places. We hypothesized that spatial subdivision may have resulted in a stronger genetic component to kin recognition than in cases where colonies live in a single location. To test our hypothesis we assessed recognition capabilities for two populations of this ant that differ in the complexity of their colonies. In a New York, USA, population, polydomy is very common, and colonies also can have multiple queens. By contrast, a population in West Virginia, USA, has colonies that typically are monogynous and rarely are polydomous. We conducted introductions of ants between different nests collected in the same neighborhood, with self-introductions and alien introductions as controls. Nests from the two populations showed corresponding differences in their aggression towards intruders. For New York nests, the extent of genetic similarity was the single best predictor of aggression, whereas for West Virginia nests aggression was jointly influenced by genetic similarity and spatial distance. In both populations, we found nest pairs for which aggression was nonreciprocal; these probably reflect recognition errors by one of the nests. After the ants were maintained in the laboratory for 3 months, their aggression scores rose and fewer recognition errors were made. Thus nest-mate and colony-mate recognition in this species are mediated primarily by endogenous cues (genetic similarity); the importance of exogenous cues for nest mate recognition depends on the population's social system.
Key words: nest mate recognition, social insects, ants, colony structure.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
One of the most successful derivatives from Hamilton's landmark work on inclusive fitness theory (1964
) is the field of kin
recognition. Hamilton's theory suggested that selection favoring behaviors
that benefit kin could also produce the ability to discriminate between kin
and non-kin. Indeed, a voluminous literature (reviews in
Fletcher and Michener, 1987;
Hepper, 1991;
Sherman et al., 1997) has
confirmed that kin recognition shapes social behavior in a wide range of
organisms. The field of kin recognition explores three categories of proximate
questions that must be carefully disentangled: signal production and/or
acquisition; signal perception; and behavioral responses to perceived signals
(Gamboa et al., 1991;
Sherman et al., 1997). Here we
investigate the source of recognition signals for a social insect with complex
colony structure. We show that two ant populations with different social
structures have correspondingly different sources of recognition cues which
reflect their ecological backgrounds. Therefore, kin recognition should be
studied with explicit references to the ecological context within which the
recognition behavior is expressed.
The hymenoptera comprise a continuum from solitary life to
tightly-structured family groups (Wilson,
1971). In the highly social forms, colonies are family groups
living in a central nest location. Because the nest represents the focus for
most interactions, students of social insects focus on the ability to
recognize nest mates rather than on broader kin recognition abilities. The
distinction can be illustrated by the behavior of honeybees which reproduce by
binary fission during swarming; the daughter colonies retain close kin
relationships, but they acquire distinct recognition cues within a few days of
swarming (Breed et al., 1998).
Here we introduce another complication by considering insects whose colonies
can be physically subdivided; when a colony is fractionated into multiple
nesting sites, the ability to recognize colonial relatives must transcend the
confines of the nest itself.
Recognition behavior for social insects typically involves active
antennation of individuals seeking entrance to the nest; when a stranger is
recognized, agonistic behavior ensues. Aggression towards non-nest mates is
provoked almost exclusively by chemical cues that are learned shortly after
eclosion from the pupal life stage (Alloway
and Hare, 1989; Gamboa,
1995; Smith and Breed,
1995). Two species have been especially well-studied with respect
to their nest mate discrmination abilities, the honey bee Apis
mellifera (Breed, 1998;
Getz, 1991) and the paper wasp
Polistes fuscatus (Gamboa,
1995). Both have sophisticated recognition abilities based on
olfactory cues emanating from the exoskeletal cuticle. Cuticular hydrocarbons
serving as signals for nest mate recognition are absorbed by the insects from
their nest environment (comb wax and paper, respectively
[Breed et al., 1995;
Gamboa et al., 1996]). Because
nest materials derive from sources outside the colony, subtle differences in
the resource bases used by different colonies result in different blends of
cuticular hydrocarbons. Thus exogenous signals have priority for nest mate
recognition in these species, but there is evidence that endogenous cues
(i.e., genetic differences) can mediate recognition as well
(Breed, 1998;
Gamboa et al., 1996; but see
Downs and Ratnieks, 1999).
The literature on nest mate recognition in ants is scattered, with only a
few studies on any given species (Breed and
Bennett, 1987; Jaisson,
1991). Even so, the ability of ants to recognize nest mates has
been well-established, and olfactory cues mediate that recognition; both
endogenous and exogenous cues have been described
(Beye et al., 1998;
Breed et al., 1992;
Crosland, 1990;
Morel et al., 1990;
Stuart, 1988a). An additional
complication for studying nest mate recognition in some ant species is their
complex colony structure. Colonies can consist of physically discrete local
nesting units, sometimes separated from each other by several meters, a
condition termed polydomy (Alloway et al.,
1982). Spatial substructuring in polydomous colonies requires
workers to move frequently among different nest sites to maintain intra-colony
communication (Herbers, 1986;
Stuart, 1985), and identifying
colony boundaries in these species requires a synthesis of data from
behavioral aggression tests and genetic characterization
(Banschbach and Herbers,
1996).
In some species, polydomy can lead to fission reproduction. Should a colony
containing more than one queen be separated into subunits, any subunit
containing a queen can become independent over time. This process is
functionally equivalent to swarming in honeybees, but with a longer time
scale. In honeybees, new recognition cues to differentiate between nest mates
are acquired within a few days of swarming
(Breed et al., 1998). By
contrast, polydomy that leads to fission reproduction is relatively passive,
with new colonies gradually becoming behaviorally independent over time
(Crozier et al., 1984;
Pedersen and Boomsma,
1999).
We hypothesize that for polydomous species, a genetic component to nest
mate recognition should predominate. Exogenous cues acquired from
environmental sources should be less important in these species, because they
could interfere with recognition between colony members living in different
domiciles. To test this hypothesis, we examined nest mate recognition in two
populations of the tiny forest ant Leptothorax longispinosus, which
exhibit important differences in social structure. Colonies in a New York
population vary in queen number, with some nests having no queens, others
having one, and still others having more than one; this syndrome of variable
queen number is known as facultative polygyny
(Herbers, 1984). By contrast,
colonies in a West Virginia population rarely have multiple queens. In New
York, colonies split into multiple nesting units, and there is an annual cycle
of colony fractionation and re-coalition called seasonal polydomy; this
accounts for a high frequency of queenless nests, which are in fact colony
subunits. In West Virginia, colonies are rarely polydomous, and queenless
nests tend to be orphans (Herbers and
Stuart, 1996a). These differences in social structure cause a
cascade of other effects, notably on reproduction and sex ratios
(Herbers and Stuart, 1996b),
and so we suspected they might differ in their nest mate recognition as well.
In particular, we hypothesized that complex colony structure in the New York
population would be accompanied by an enhanced role for endogenous sources of
nest mate recognition cues there.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Nests of L. longispinosus were collected from two sites in 1987-1989. The New York site (NY) is located in the E.N. Huyck Preserve (Albany County) and the West Virginia population (WV) at Watoga State Park (Pocahontas County). Our study sites and the ant communities living therein have been described elsewhere (Herbers, 1989
; Herbers and Stuart,
1996a,b).
At each site, a series of 49-m2 quadrats was set up and then
completely excavated and mapped; three quadrats excavated in each of the four
seasons provided the ants for our tests. Nests of L. longispinosus
were bagged and returned to the laboratory, where they were settled into 10 mm
x 2 mm glass tubes. These nest tubes were placed in nest boxes and
maintained according to standard conditions
(Herbers, 1984). The occupants
were censused periodically and fed a combination of ant food
(Bhatkar and Whitcomb, 1970)
and frozen fruitflies, with water provided ad libitum. The spatial
structure and genetics of these collections have been explored in detail
elsewhere (Herbers, 1989;
Herbers and Stuart,
1996a).
Within each plot, we identified a series of focal nests, using only nests that had at least ten workers. For each nest, we identified its nearest neighbors in the plot (up to the sixth nearest neighbor). We then conducted trials between each focal nest and its neighbors, with focal nests receiving intruders from test neighbors as described below. Because we tested within plots, many nests served as focal nests and as neighbors to other focal nests. We also had two sets of controls: selfintroductions and introduction to/from "alien" nests that were collected at least 100 m from the focal nest. We therefore had four classes of tests: (1) nests collected from the same plot (pairwise introductions); (2) self-introductions; (3) introduction of alien workers into focal nests; and (4) introduction of ants from focal nests into alien nests. Our experiment, then, conformed to a split-split-split plot design (focal nests within plots within seasons within site), with repeated measures (neighbors) at the lowest level.
Within 3 weeks of collection, the first round of aggression tests was
initiated. We repeated aggression tests 3 months after collection, to
ascertain whether signals used for colony recognition changed over time; these
tests were designated collectively as round two. In both rounds, the protocols
were identical. A single worker from a donor nest was selected; we used
foragers outside the nest if possible, and only used workers with dark
exoskeletons to ensure we had older workers. We chilled the ant to anesthetize
her, and then placed a loop of polyester thread around her alitrunk
(Stuart, 1986). The thread was
glued to a thin stick such that the tethered ant could be gently walked into
the nesting tube of a recipient nest. The donor ant was left in the recipient
nesting tube until either 3 min had elapsed or until a clear aggressive
response had been elicited. A different donor worker was used for each
introduction, and a given nest was not designated as recipient more than twice
on the same day.
During trials, we videotaped the ensuing behavior both of the donor ant and of workers in the recipient nest, characterizing the interaction along three axes of grooming, biting, and stinging. From those raw observations an aggression score was developed on a scale of zero to four. A score of zero meant the introduction was completely amicable, with the donor entering and being groomed by the recipients. A score of one meant very mild aggression, with a short bout of dragging the donor ant or a recipient opening her mandibles. A score of two indicated escalated aggression with open mandibles directed toward the donor, often accompanied by dragging or biting. If there was prolonged dragging, with actual biting and attempts to sting, a score of three was given. Scores higher than three were given for attempts to maim or kill the donor ant. All experiments were done blind to mitigate observer bias: one investigator tracked the identity of donor and recipient nests, while a second scored the ensuing behavior.
Because we tested within plots, many nests served as focal nests and as neighbors to other focal nests. After all data had been collected, we characterized interactions between a pair of nests as friendly (both introductions with scores less than two); hostile (both introductions scoring two or higher); or asymmetric (one introduction scoring below two and the reciprocal test scoring two or above). Clearly, the reciprocal tests were not independent, nor were the repeated observations between focal nests and their neighbors. For the smallest data sets (plots with seven or fewer nests) we were able to use matrix correlation methods to correct for nonindependence. With denser plots, however, our focal nest design produced many missing cells in distance matrices, thereby precluding use of matrix correlations. Rather, in order to achieve independence within our data sets, we used each pair of nests only once in statistical analysis, randomly assigning the status of donor and recipient to members of the pair. Therefore our fundamental unit for data analysis was a pair of nests, and any given pair appeared only once in the data.
Data on allozyme variability at five polymorphic loci were used to infer
genetic structure within and between nests. Because allozyme markers have low
resolution, we combined information from all five loci to compute a genetic
similarity index (for details see Herbers
and Stuart, 1996a). This index varied from zero (genetic profiles
completely incompatible between two nests) to one (genetic profiles identical
between the two nests). We included these values, worker and queen numbers in
both donor and recipient, nest, and spatial location of nests in our
database.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Table 1 shows the size of the database we built from aggression tests. Between sites, New York, had significantly higher nest density than West Virginia (K-W test, G = 8.2, 1 df, p <.01), yielding significant differences in distances to nearest neighbors as well (Figure 1; ANOVA, effect of site p <.01). Nests in WV were typically isolated from their nearest neighbors by more than a meter; in NY only about 60 cms separated the average nest from its nearest neighbor. Because differences in density introduced differences in distance to neighbors, we are careful to distinguish neighbor rank from absolute spatial distance below.
|
|
Logistically, we were able to conduct a second round of aggression tests for only half the sampled plots from NY; thus comparisons between round one and round two are available only for summer and autumn collections. By contrast, we were able to conduct both rounds of tests for all the plots collected in West Virginia (Table 1). Totals of 171 nests from New York and 102 nests from West Virginia were used; respectively, 1662 and 1435 tests were conducted, including self-introductions, pairwise tests, and tests with aliens. To our knowledge, these represent the largest sample sizes for aggression tests available for any ant species.
Overall patterns
Raw data for an uncrowded plot are shown in
Figure 2. For this plot, we
conducted all six possible pairwise introductions between nests, and the
bi-directional arrows in Figure
2 show those aggression scores. We also conducted
self-introductions and tests with aliens for these nests (data not shown),
yielding a total of 20 tests altogether in this plot
(Table 1). The data in
Figure 2 illustrate that in
general introductions were symmetric; if the introduction of a worker from
nest i elicited aggression from workers in nest j, usually
we recorded aggression for the reverse introduction as well. On that
particular plot, we found one friendly pair (BC), four hostile pairs (AB, AD,
BD, CD), and one asymmetric pair (AC).
|
Table 2 gives statistics for the four classes of introductions, with data pooled over all tests. Self-introduction controls, by which a worker was tethered and reintroduced into her own nest, were almost uniformly friendly (Table 2). With rare exceptions, workers were allowed re-entry to their own nest without the slightest display of hostility, and usually with considerable grooming (medians of zero in both sites, and arithmetic means very close to zero). Thus our manipulation of workers by tethering did not appear to induce aggressive behavior as artifacts. Conversely, our controls that included "aliens" typically elicited hostile behavior (Table 2). There were a few friendly introductions involving aliens from each population (minimum scores in Table 2), but these were rare. Median scores were 3.0 in NY and 2.5 in WV for introductions involving aliens, showing that aggression was high between ants living in different parts of the forest. Taken together, all these results support our contention that aggression reflected the degree of familiarity between ants. Our experimental protocol, by which we gently introduced ants from one nest into the entrance of the focal nest, mimicked the natural situation when guard ants encounter returning foragers. Individuals seeking entrance to a nest must pass antennal inspection, and presumably must present the appropriate combination of olfactory cues. We had very low frequencies of apparent errors in our control trials involving self-introductions and introductions into/from alien nests, which strongly suggests that our observations reflected natural behavior. We proceed then to interpret results from the pairwise trials in that light.
|
Introduction of ants from different nests collected from the same plot elicited a range of reactions (pairwise interactions in Table 2). On average, pairwise interactions were much more aggressive than self-introductions (Friedman's tests, p <.0001), yet in both sites perfectly friendly introductions of ants into other nests were observed (minimum scores of 0 for pairwise tests). There was no difference between the sites in average pairwise aggression scores (medians of 2.5 in both NY and WV).
In the NY population, introductions to/from alien nests elicited stronger aggression on average (median of 3.0) than did pairwise introductions (median of 2.5; K-W test, H' = 10.9, 1 df, p <.0001). On average, then, pairwise interactions were friendlier in NY than were alien interactions. Conversely, both groups had identical median scores of 2.5 for the WV population (Table 2); tests involving aliens and pairwise tests provoked the same levels of aggression on average. We interpret the difference between sites to reflect their different social structures. Because polydomy is common in NY but rare in WV, pairwise tests in NY had a higher frequency of zeroes.
Effects of distance and genetic similarity
The aggression scores we collected represented up to six pairwise
introductions for each focal nest. Each focal nest was collected from a
particular plot, which was part of a seasonal collection within each site.
Unfortunately, we were unable to normalize the aggression scores via standard
transformations, and thus could not use the hierarchical parametric approach
that fit our experimental design.
Our principal goal was to sort out the effects of distance and genetic
similarity on aggression scores. We first examined data from six plots small
enough to allow all pairwise interactions. For those plots, we constructed
three matrices, one with the aggression data, a second with spatial distance,
and the third with genetic similarity data; we then tested for correlations
among the matrices via Mantel tests (cf.
Manly, 1995). Those tests
showed a significant positive correlation between distance and aggression for
one plot (G = 2.12 for WV 11, p <.05), and significant
negative correlations between genetic similarity and aggression for two plots
(G = -2.34 for NY 7 and G = -2.18 for WV 11, both p
<.05). Therefore, even for small data sets with low power (since Mantel
tests are conservative), we found evidence for effects of both genetic and
spatial distance on the aggression between ant nests. For larger datasets, we
were unable to perform Mantel tests due to the large number of missing cells.
Rather, we applied nonparametric methods and used each pair of nests only
once. Below we report on patterns for data pooled across plots and seasons,
because preliminary analysis showed such pooling did not obscure important
patterns.
In Figure 3 we show average aggression scores as a function of neighbor rank (nearest neighbor has rank one, second-nearest has rank two, and so on). Aggression changed with neighbor rank in both sites (Figure 3a; K-W tests, H' = 13.26 and 13.75 respectively, each with 5 df; p <.05 for each), due to lower aggression towards the first nearest neighbor. Aggression scores among higher neighbors were no different (nonparametric multiple contrasts; in both sites p <.05 for differences between the first neighbor and all others; p >.05 for differences among neighbors two through six). That is, focal nests were less aggressive towards their closest neighbor than to other nests in the vicinity. We saw the same pattern in round two, which was conducted several months later (Figure 3b): aggression towards the nearest neighbor was significantly less than aggression towards further neighbors, but there were no differences among second through sixth neighbors (K-W tests with multiple contrasts; p <.05 for the first neighbor, and p >.05 among higher-ranking neighbors). Thus reduced aggression was associated with close spatial proximity.
|
Overall, genetic similarity values were higher in NY than those in WV (K-W
test, H' = 47.9, 1 df, p <.0001); this result
reflects colony subdivision in NY. Neighbor rank was not correlated with
genetic similarity in either population, however (Spearman's rank
correlations; in NY
= -0.07 and in WV = -0.06; for both p
>.05). That is, the first nearest neighbor was no more similar or
dissimilar genetically to the focal nest than more distant neighbors. On the
other hand, genetic similarity was inversely related to absolute spatial
distance between nest pairs in New York (Spearman's = -0.248, p
<.0001); no such correlation existed for West Virginia ( = -0.058,
p >.05). We interpret the fact that absolute spatial distance but
not neighbor rank was correlated with genetic similarity in NY to mean that
polydomous colonies intercalated. To be sure, colony subunits were closer to
each other than two randomly chosen nests, but they were not necessarily each
other's closest neighbors. By contrast, the lack of any correlation between
spatial distance and genetic similarity in WV reflected the low frequency of
colony subdivision there.
The above analysis shows that we could examine the effects of spatial
distance and genetic similarity independently for WV aggression data. For NY
data, however, the two were confounded. To tease them apart, we analyzed the
aggression data via nonparametric partial correlation analysis
(Siegal and Castellan Jr.,
1988). The coefficients in
Table 3 reinforce our
conclusion above that distance and genetic similarity were confounded in NY
(partial correlation = -0.225, p <.001) but not in WV (partial
correlation = -0.046, p >.05). In NY, only genetic similarity gave
a significant partial correlation with aggression in round one
(Table 3). By contrast, both
distance and genetic similarity were correlated with round one scores in WV.
The same pattern was observed for round two aggression
(Table 3). These results, which
are based on large sample sizes and thus have high power
(Table 3), show that for the NY
population only genetic similarity influenced aggression scores, whereas in WV
both distance and genetic similarity did.
|
Effects of demography
We wondered whether demographic factors affected our aggression scores.
Because we had multiple tests with each focal nest, we avoided
pseudoreplication by using only aggression towards the first nearest neighbor
for this analysis. We found no effects of queen number or worker number in the
donor nest on aggression shown to an intruder from that nest (Kruskal-Wallis
tests, p >.05). That is, the demographic background of an intruder
worker's home did not affect how she was treated by a recipient nest.
Similarly, we found no effect of worker number of the recipient nest on
aggression: small and large nests alike reacted similarly to intruders.
However, we did find an effect of queen number in the recipient nest for both
populations (Figure 4); workers
in queenless nests were less aggressive towards intruders than were workers in
monogynous and polygynous nests (K-W tests; H' = 7.44, 2 df for
NY; and H' = 6.44, 1 df for WV; both p <.05). In
general, then, nest demography did not influence aggression, with the
important exception of recipient queen number.
|
Reciprocity of interactions
Many nests served as both donor and recipient to each other; that is, we
had a number of pairs for which we could assess reciprocity of aggression.
These reciprocal tests gave the most complete picture of between-nest
aggression possible, and so we examined them in detail. In general the
aggression of ants in nest j towards an introduced ant from nest
i was correlated with the aggression of ants in nest i
towards an introduced ant from nest j (Kendall's 
= 0.24 for NY
and 0.21 WV, p <.0001 for both). Those correlations provide
support for our interpretation that we were in fact measuring some aspect of
nest mate recognition in our assay.
We characterized nest pairs as friendly, hostile, or asymmetric in each round (Table 4). Most pairwise interactions were hostile, and fewer than 10% were friendly. We were surprised to find that about 16% of all interactions were asymmetric; that is, introduction of i into j was peaceful whereas the reverse was not. These asymmetries were equally common in the two sites for round one (Table 4; G-test, p >.05), and we return to them below.
|
Those pairs that were friendly in round one tended to be first neighbors to each other (data not shown; G-tests with five df; for NY, G = 38.9 and for WV, G = 30.3; both p <.001). Indeed, the average distance between friendly pairs was considerably smaller than the average distance between hostile pairs (Figure 5; ANOVA, p <.001 for both sites in round one). We reclassified pairs based on their interactions in round two (Table 4) and found the same patterns: nests that were friendly tended to be near neighbors, and were significantly closer to each other than hostile or asymmetric pairs (Figure 5; ANOVA, p <.001 for both sites). The data in Figure 5 also confirmed our earlier inferences: friendly pairs were significantly more similar genetically than hostile or asymmetric pairs (K-W tests; H' = 15.9 for NY and 14.8 for WV; both with 2 df and p <.001). The same pattern was observed in round two but was significant only in WV (H' = 26.3, 2 df, p <.0001; in NY, H' = 4.93, 2 df, p >.05).
|
We were intrigued by pairs of nests that gave asymmetric aggression scores. In these pairs, one nest was friendly to an intruder but the reciprocal introduction yielded an aggressive response. We first looked to see if particular nests were overrepresented in this category: from the frequency of asymmetries (Table 4) we generated a Poisson distribution, and then compared the number of times a particular nest was classified as part of an asymmetric pair to that distribution. The distribution of asymmetries per nest conformed to Poisson expectation (data not shown), meaning that individual nests were involved in asymmetric interactions according to chance expectation. We examined our data further to see if demographic characteristics could distinguish asymmetric pairs from others, but found no such effects (Gtests on queen numbers and ANOVA on log-transformed worker numbers; in all casesp >.05). Taken together, our data show that asymmetric pairs could not be distinguished in any way from mutually hostile pairs.
To explore the problem of asymmetric pairs more closely, we categorized interactions with alien ants (Table 4). Only one test involving aliens in each site was friendly, and those occurred in round one (Table 4). This result confirms that our tests with aliens were solid controls. Even so, tests involving aliens were asymmetric in 10.2% of the NY trials and in 17.5% of the WV trials, proportions that were not significantly different (G test, p >.05). Asymmetries involving aliens were rarer than those involving pairs (compare results of pairwise interactions with results of tests with aliens, Table 4; G = 20.0 in NY and 7.4 in WV, both with 2 df; p <.0001 and p <.05, respectively). In round two, asymmetries involving aliens were very rare for the WV data, but represented 12.9% of all the NY trials.
Round one versus round two
Finally, we compared interactions of nest pairs tested right after
collection and again after three months of laboratory domestication
(Figure 6). There was an
overall positive correlation between aggression scores for the same pairs of
nests between rounds (Spearman's = 0.24 in NY and 0.42 in WV; p
<.001 for both sites), but scores were higher in round two than in round
one (sign tests, p <.05 for both sites). Furthermore, median
aggression between pairs was higher in round two
(Figure 6; K-W tests; in NY
H' = 10.6 and in WV H' = 49.2 with 1 df,
p <.01 for both). For WV nests, both sets of control introductions
(involving aliens and self-introductions) also yielded higher aggression
scores in round two than they had in round one (p <.05); for the
NY nests, however, aggression in control trials did not change between round
one and round two (p >.05).
|
A pattern of increased aggression after lab domestication was evident from the classification of individual interactions as well (Table 4). The proportion of hostile pairwise interactions rose in round two (Table 4, p <.01 for each site), and there was a higher frequency of hostility for WV pairs than for NY pairs (p <.05). Similarly, tests involving aliens in round two provoked a higher frequency of hostile interactions than had been seen in round one (Table 4). Further analysis showed that only 59% of pairs that were characterized as friendly in round one were still friendly 3 months later (Figure 7); one-fourth became hostile, and the remainder showed asymmetric behavior in round two. Furthermore, 84% of all nests that showed asymmetric aggressive behavior in round one became hostile after a lapse of three months, and only 3% became friendly. Only one pair of nests that had been hostile in round one became friendly in round two, and only 8% made the transition to being asymmetric (Figure 7). The transitions frequencies in Figure 7 deviate strongly from chance expectation (G = 163.4, 2 df, p <.0001), with friendly pairs remaining friendly and hostile pairs remaining hostile at a higher than expected frequency. By contrast, asymmetric pairs remained asymmetric at a rate lower than chance expectation (G = 7.5, 2 df, p =.02), with the majority becoming hostile (Figure 7). The overwhelming pattern then was for pairs of nests to become more hostile to each other after 3 months had elapsed in the laboratory.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The ability to discriminate kin, while widespread, need not imply the operation of kin selection (Barnard, 1991
; Grafen,
1990). Indeed, confirming that kin selection is the best
explanation for kin discrimination requires rigorous testing of alternatives
(Pfennig et al., 1999), and
evidence linking kin discrimination to kin selection is surprisingly thin
(Sherman et al., 1997). Even
so, we can safely look to species with well-developed social systems that have
themselves been shaped by kin selection in order to infer the evolutionary
basis of kin recognition. Not surprisingly, then, highly social species that
live in family groups have been especially well-studied with respect to their
ability to recognize kin (Crozier and Pamilo, 1996). Our work fits squarely
into the large body of research on social insects by demonstrating that
ecological factors affecting colony structure can exert strong influences on
how kin recognition is achieved.
Most authors have characterized recognition systems within a single
population. We have shown here that between-population variation in nest mate
recognition gives important insight to the ecology of recognition systems (see
also Morel et al., 1990). Our
use of two populations that differ in social structure uncovered important
differences in the way they use cues to distinguish between nest mates and
strangers. Our West Virginia and New York populations differ in nest density,
frequency of polydomy, and prevalence of polygyny. These differences produce
higher within-nest relatedness and lower between-nest relatedness in WV
relative to NY (Herbers and Stuart,
1996a). Furthermore, colonies are more tightly packed in NY, which
produces spatially intercalated polydomous colonies. As a consequence, genetic
similarity was of overwhelming importance to predicting between-nest
aggression there. Use of exogenous cues to distinguish nest mates in NY would
lead to high error rates: environmentally derived cues are most similar over
short distances, and the intercalation of polydomous colonies could lead to
odor similarity among non-relatives and dissimilarity among relatives. Thus
the combination of polydomy and tight colony packing should have selected for
endogenous cues having priority, as we found. Conversely, in WV ants used cues
correlated both with genetic similarity and distance between nest pairs. Given
that colonies have simple structure there and are widely spaced, exogenous
cues have greater relevance for nest mate recognition.
Despite the differing priority of endogenous cues between the two populations, exogenous cues must have contributed to aggression both in NY and WV. After 3 months of lab domestication, overall aggression rose and pairs that had been friendly became hostile. We interpret that clear pattern to show that repeated interactions in the field, which we prevented in the lab, are crucial for maintaining recognition among subunits of polydomous colonies. Thus, while endogenous cues are important, they must be reinforced with exogenous cues. Clearly, an understanding of population structure is essential for interpreting data on aggression between non-nest mates.
In general, environmental influences are extremely important for predicting
aggressive interactions between colonies of social insects
(Smith and Breed, 1995).
Nesting material is an important source of recognition cues for wasps and
bees, but rarely plays a role in producing recognition cues for ants
(Gamboa, 1995;
Smith and Breed, 1995; but see
Heinze et al., 1996).
Demographic variables can influence nest mate recognition; for example,
variable queen number has been linked to differential aggression in
Leptothorax lichsteini (Provost,
1989) and Solenopsis invicta
(Morel et al., 1990), but was
excluded as important for Leptothorax ambiguus
(Stuart, 1991) and
Rhytidoponera confusa (Crosland,
1990). Similarly, colony size (i.e., worker number) was correlated
with aggression in L. ambiguus
(Stuart, 1991) but not R.
confusa (Crosland, 1990).
We have shown that in general, nest demography had few effects on nest mate
recognition in Leptothorax longispinosus. The exception was that
queenless nests, presumably subunits of polydomous colonies, were less
aggressive to neighbors than nests containing queens. Again, a thorough
understanding of the social system is necessary to interpret demographic
effects on recognition systems.
Hostile versus friendly behavior among pairs of nests was most strongly
dependent on two factors, distance and genetic similarity. These two were
themselves strongly correlated in our NY population, which makes its nest mate
recognition system similar to that of F. pratensis (Beye et al.,
1997,
1998). Studies on a European
Leptothorax species (Heinze et
al., 1996) also implicated spatial distance between nests; whether
or not genetic similarity was confounded with distance in that study is
unknown. Certainly, our discovery of inter-population differences in the
importance of distance and genetic similarity mandates a thorough exploration
of their priority in future studies.
The pairs of nests for which we recorded asymmetric interactions are of
special interest. While we cannot rule out the possibility that those
asymmetries are artifacts, the data suggest otherwise. We were unable to
identify any demographic, spatial, or genetic variable associated with
asymmetric pairs. This contrasts with paper wasps, for which genetic
relationships predict asymmetric aggressive behavior; for example, nieces
tolerate aunts but aunts do not tolerate nieces in Polistes fuscatus
(Bura and Gamboa, 1994).
Furthermore, the vast majority of pairs classified as asymmetric in round one
were classified as hostile in round two
(Figure 7; p =.02). We
therefore suspect that asymmetries represented mistakes of recognition by the
tolerant nest of the pair. Our interpretation that most asymmetries involved
recognition mistakes is reinforced by their presence among tests that included
aliens (Table 4); any nest not
showing aggression to an alien intruder clearly made a mistake. Whether or not
such errors occur in the field at the rates we found them in the laboratory
awaits future study.
Between-nest aggression in our study species shows many similarities to its
close relatives L. curvispinosus and L. ambiguus (Stuart,
1987a,
b,
c,
1988a,
b,
1991,
1992). Stuart
(1988a) has established that
L. curvispinosus uses a collective odor system to recognize nest
mates, and our data echo many results of his experiments. There are some
important differences, however, among the species. Most striking, Stuart
(1987c) found that aggression
between nests of L. curvispinosus waned over time in the laboratory,
which he ascribed to the loss of environmentally-based colony recognition
odors as nests were kept on a uniform diet. We found, on the other hand, that
aggression increased as nests were held in the lab, which we also ascribe to
the loss of collectively-held colony-specific odors. Nest mate recognition
cues still important shortly after collection (round one) apparently
disappeared by round two (Figure
6). To explain these differences, we postulate that for L.
longispinosus, repeated interactions among nest subunits are needed to
maintain colony integrity in the field. Exchange of recognition cues among
workers from different nest subunits is maintained after colonies break up in
the spring, either by grooming or trophollaxis, both of which have been
observed in laboratory reconstructions of polydomous colonies
(Herbers and Tucker, 1986).
When that exchange of cues is interrupted by physical segregation in the
laboratory, colony recognition between nest subunits can no longer be
effected, and introductions at a later date elicited aggressive behavior.
We have shown here that recognition systems in these little ants reflect their ecological background. Nest mate recognition in L. longispinosus appears to be but one dimension of a complex social system that responds to environmental parameters. Our results show that a full understanding of social ecology can allow accurate predictions about how nest mate recognition is mediated. Indeed, we assert that future studies of nest mate recognition in social insects are incomplete unless they are placed squarely within the context of the species' social organization.
|
ACKNOWLEDGEMENTS |
|---|
We thank Vickie Backus, Laura Snyder, and Linda Prince for assistance with fieldwork and aggression experiments. Chris DeHeer, Susanne Foitzik, Tim Judd, and an anonymous reviewer made many suggestions for improving the manuscript. This work was supported by grants from Vermont EPSCoR and the National Science Foundation.
|
FOOTNOTES |
|---|
R. J. Stuart is now at the University of Florida, Citrus Research and Education Center, 700 Experiment Station Road, Lake Alfred, FL 33850, USA.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Alloway TM, Buschinger A, Talbot M, Stuart R, Thomas C, 1982. Polygyny and polydomy in three North American species of the ant genus Leptothorax Mayr (Hymenoptera: Formicidae). Psyche 89: 249-274.
Alloway TM, Hare JF, 1989. Experience independent attraction to slave-maker ant larvae in host-species ant workers (Leptothorax longispinosus; Hymenoptera: Formicidae). Behaviour 110: 93-105.
Banschbach VS, Herbers JM, 1996. Complex colony structure in social insects: I. Ecological determinants and genetic consequences. Evolution 50: 285-297.
Barnard C, 1991. Kinship and social behaviour: the trouble with relatives. Trends Ecol Evol 6: 310-312.
Beye M, Neumann P, Chapuisat M, Pamilo P, Moritz RFA, 1998. Nest mate recognition and the genetic relatedness of nests in the ant Formica pratensis. Behav Ecol Sociobiol 43: 67-72.
Beye N, Neumann P, Moritz RFA, 1997. Nest mate recognition and the genetic gestalt in the mound-building ant Formica polyctena. Insectes Soc 44: 1-73.
Bhatkar A, Whitcomb WH, 1970. Artificial diet for rearing various species of ants. Florida Entomol 53: 217-232.
Breed MD, 1998. Recognition pheremones in the honey bee. Bioscience 48: 463-470.
Breed MD, Bennett, B, 1987. Kin recognition in highly eusocial insects. In: Kin recognition in animals. (Fletcher DJC, Michener CD, eds). Chichester: John Wiley; 243-285.
Breed MD, Garry MF, Pearce AN, Hibbard BE, Bjostad LB, Page RE Jr, 1995. The role of wax comb in honey bee nest mate recognition. Anim Behav 50: 489-496.
Breed MD, Leger EA, Pearce AN. Wang YJ, 1998. Comb wax effects on the ontogeny of honey bee nest mate recognition. Anim Behav 55: 13-20.
Breed MD, Snyder LE, Lynn TL. Morhart JA, 1992. Acquired chemical camouflage in a tropical ant. Anim Behav 44: 519-523.
Bura EA, Gamboa GJ, 1994. Kin recognition by social wasps: asymmetric tolerance between aunts and nieces. Anim Behav 47: 977-999.
Crosland M, 1990. The influence of the queen, colony size, and worker ovarian development on nest mate recognition in the ant Rhytidoponera confusa. Anim Behav 39: 413-425.
Crozer RH, Pamilo, P, 1996. Evolution of social insect colonies. Oxford: Oxford University Press.
Crozier RH, Pamilo P, Crozier YC, 1984. Relatedness and microgeographic genetic variation in Rhytidoponera mayri, an Australian aridzone ant. Behav Ecol Sociobiol 15: 143-150.[Web of Science]
Downs SG, Ratnieks FLW, 1999. Recognition of conspecifics by honeybee guards uses nonheritable cues acquired in the adult stage. Anim Behav 58: 643-648.[Web of Science][Medline]
Fletcher DJC, Michener CD, 1987. Kin recognition in animals. New York: John Wiley.
Gamboa GJ, 1995. Kin recognition in social wasps. In: Natural history and evolution of paper-wasps. (Turillazzi S, West-Eberhard MJ, eds). Oxford: Oxford University Press; 161-177.
Gamboa GJ, Grudzien TA, Espelie KE, Bura EA, 1996. Kin recognition pheremones in social wasps: combining chemical and behavioural evidence. Anim Behav 51: 625-629.
Gamboa GJ, Reeve HK, Holmes WG, 1991. Conceptual issues and methodology in kin-recognition research: a critical discussion. Ethol 88: 109-127.
Getz W, 1991. The honey bee as a model kin recognition system. In: Kin recognition. (Hepper P, ed). Cambridge: Cambridge University Press; 358-412.
Grafen A, 1990. Do animals really recognize kin? Anim Behav 39: 42-54.
Hamilton W, 1964. The genetical evolution of social behavior. J Theor Biol 7: 1-52.[Web of Science][Medline]
Heinze J, Foitzik S, Hippert A, Hölldobler B, 1996. Apparent dearenemy phenomenon and environment-based recognition cues in the ant Leptothorax nylanderi. Ethology 102: 510-522.
Hepper P, 1991. Kin recognition. Cambridge: Cambridge University Press.
Herbers JM, 1984. Queen-worker conflict and eusocial evolution in a polygynous ant species. Evolution 38: 631-643.[Web of Science]
Herbers JM, 1986. Nest site limitation and facultative polygyny in the ant Leptothorax longispinosus. Behav Ecol Sociobiol 19: 115-122.
Herbers JM, 1989. Community structure in north temperate ants: temporal and spatial variation. Oecologia 81: 201-211.
Herbers JM, Stuart RJ, 1996a,. Multiple queens in ant nests: impact on genetic structure and inclusive fitness. Am Nat 147: 161-187.
Herbers JM, Stuart RJ, 1996b. Patterns of reproduction in southern versus northern populations of Leptothorax ants (Hymenoptera:Formicidae). Ann Entomol Soc Amer 89: 354-360.
Herbers JM, Tucker CW, 1986. Population fluidity in Leptothorax longispinosus (Hymenoptera: Formicidae). Psyche 93: 217-229.
Jaisson P, 1991. Kinship and fellowship in ants and social wasps. In: Kin recognition. (Hepper P, ed). Cambridge: Cambridge University Press; 60-93.
Manly BFJ, 1995. The statistics of natural selection. London: Chapman and Hall.
Morel L, VanderMeer RK, Lofgren CS, 1990. Comparison of nest mate recognition between monogyne and polygyne populations of Solenopsis invicta (Hymenoptera: Formicidae). Ann Ent Soc Amer 83: 642-647.
Pedersen J, Boomsma JJ, 1999. Genetic analysis of colony structure in polydomous and polygynous ant populations. Biol J Linn Soc 66: 115-144.
Pfennig DW, Collins JP, Ziemba RE, 1999. A test of
alternative hypotheses for kin recognition in cannibalistic tiger salamanders.
Behav Ecol 10:
436-443.
Provost E, 1989. Social environmental factors influencing mutual recognition of individuals in the ant Leptothorax lichsteini Bondr. (Hymenoptera: Formicidae). Behav Proc 18: 35-59.
Sherman PW, Reeve HK, Pfennig DW, 1997. Recognition systems. In: Behavioural ecology. (Krebs JR, Davies NB, eds). Oxford: Blackwell Science; 69-96.
Siegal S, Castellan NJ Jr, 1988. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill.
Smith BH, Breed MD, 1995. The chemical basis for nest mate recognition and mate discrimination in social insects. In: Chemical ecology of insects 2. (Carde RT, Bell WJ, eds). New York: Chapman and Hall; 287-317.
Stuart RJ, 1985. Spontaneous polydomy in laboratory colonies of the ant Leptothorax curvispinosus Mayr (Hymenoptera: Formicidae). Psyche 92: 71-81.
Stuart RJ, 1986. Use of polyester fibers to mark small leptothoracine ants (Hymenoptera: Formicidae). J Kansas Entomol Soc 59: 566-568.
Stuart RJ, 1987a. Individual workers produce colony-specific nest mate recognition cues in the ant Leptothorax curvispinosus. Anim Behav 35: 1062-1069.
Stuart RJ, 1987b. Nest mate recognition in leptothoracine ants: testing Fielde's progressive odor hypothesis. Ethol 76: 116-123.
Stuart RJ, 1987c. Transient nest mate recognition cues contribute to a multicolonial population structure in the ant Leptothorax curvispinosus. Behav Ecol Sociobiol 21: 229-235.
Stuart RJ, 1988a. Collective cues as a basis for nest mate recognition in polygynous leptothoracine ants. Proc Nat Acad Sci 85: 4572-4575.
Stuart RJ, 1988b. Development and evolution in the nest mate recognition systems of social insects. In: Evolution of social behavior and integrative levels. (Greenberg G, Tobach E, eds). Hillsdale, New Jersey: Erlbaum; 177-195.
Stuart RJ, 1991. Nest mate recognition in leptothoracine ants: testing for effects of queen number, colony size, and species of intruder. Anim Behav 42: 277-284.
Stuart RJ, 1992. Nest mate recognition and the ontogeny of acceptability in the ant Leptothorax curvispinosus. Behav Ecol Sociobiol 30: 403-408.
Wilson EO, 1971. The insect societies. Cambridge: Belknap.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Sorvari, P. Theodora, S. Turillazzi, H. Hakkarainen, and L. Sundstrom Food resources, chemical signaling, and nest mate recognition in the ant Formica aquilonia Behav. Ecol., March 1, 2008; 19(2): 441 - 447. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Foitzik, B. Fischer, and J. Heinze Arms races between social parasites and their hosts: geographic patterns of manipulation and resistance Behav. Ecol., January 1, 2003; 14(1): 80 - 88. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||