Behavioral Ecology Advance Access originally published online on January 19, 2005
Behavioral Ecology 2005 16(2):488-496; doi:10.1093/beheco/ari020
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Quorum sensing by encounter rates in the ant Temnothorax albipennis
Department of Ecology and Evolutionary Biology, Princeton University, Princeton, NJ 08544, USA
Address correspondence to S.C. Pratt. E-mail: spratt{at}princeton.edu.
Received 21 June 2004; revised 24 October 2004; accepted 25 November 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Emigrating colonies of the ant Temnothorax (formerly Leptothorax) albipennis can choose the best of several nest sites, even when the active ants organizing the move do not compare sites. This collective ability depends on a quorum rule used by ants assessing a candidate site. Only when the site's population has surpassed a threshold do they switch from slow recruitment of fellow active ants by tandem runs to rapid transport of the majority of the colony. Here, I show that ants perceive the achievement of a quorum through their rate of direct encounters with nest mates at the site. When ants in a crowded site were prevented from tactile contact with nest mates, they recruited by tandem runs, as though to an empty nest. Furthermore, when the encounter rate was raised independent of population, by reducing the size of the candidate nest, ants started to transport at a significantly lower population. The switch occurred at the same encounter rate regardless of nest size, whether the rate was measured as the mean over the entire visit or as the inverse of the latency until the first encounter. Because encounter rate reflects the density of nest mates and thus varies with nest size as well as population, the ants' collective decision-making algorithm may be robust to the exact population at which the switch to transport occurs. Ants cease monitoring quorum presence after switching to transport, coincident with an abrupt shortening of visit duration by approximately 2 min, which may be interpreted as the time required for quorum detection.
Key words: ants, collective decision making, nest-site selection, quorum sensing, Temothorax albipemis.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Insect societies perform complex collective tasks in the absence of any form of central control, partly through the reliance by individual workers on simple but informative local cues. One class of such cues—the local number or density of nest mates—may commonly play an important role in coordinating a worker's actions with those of her nest mates. The allocation of labor across different tasks, for example, requires colony members to respond to the numbers already working at a given job (Gordon and Mehdiabadi, 1999
; Pacala et al., 1996). Similar responsiveness can help during conflicts with other societies, when the best strategy depends on the disparity in size between the colony and its opponent (Lumsden and Hölldobler, 1983), or for tasks whose successful completion requires a minimum number of workers, such as comb construction by honey bees (Darchen, 1968). Nest-mate numbers may be especially relevant for decisions requiring a colony-wide consensus, as when an emigrating colony must ensure that all of its members move to the same new home (Franks et al., 2002; Seeley, 2003).
Understanding the emergence of collective behavior requires knowledge of how individual workers perceive crucial proximate cues like nest-mate numbers. Here, I examine this problem in the context of colony emigration by the ant Temnothorax (formerly Leptothorax) albipennis, where response to nest-mate numbers plays a key role. These ants typically inhabit rock crevices, hollow nuts, or other preformed cavities whose fragility makes necessary frequent emigrations. Laboratory studies have shown that emigrating colonies can reliably choose the best of several potential nest sites, even when most of the individual workers responsible for organizing the move do not directly compare sites (Mallon et al., 2001). A collective decision instead emerges from a competition between independent groups of ants recruiting nest mates to alternative sites. Two crucial behavioral rules increase the chance that the better site wins this competition. First, an active ant initiates recruitment to a promising site only after a delay that varies inversely with site quality (Mallon et al., 2001). This ensures that better candidates enjoy stronger positive feedback on population growth. Second, the ants amplify this difference by using two distinct forms of recruitment: slow tandem runs, in which fellow active ants are painstakingly led to the new site, and speedier transports, in which the passive majority of the colony is simply carried to the site. Ants choose between recruitment types on the basis of a quorum rule, leading tandem runs early on but switching to transports as the population of the new site surpasses a threshold size (Pratt et al., 2002). This rule sharpens the precision of collective decision-making, by making full commitment to a site contingent on the "votes" cast by nest mates that continue to visit and recruit to the site (Pratt et al., 2002).
How do ants actually implement the quorum rule? That is, what proximate cues cause them to change their recruitment behavior? When we say that this change relies on attainment of a quorum, this reflects both the experimental observation that the number of nest mates at a site determines subsequent recruitment behavior as well as the likelihood that nest-mate number is the functionally important feature for decision-making performance at the colony level. The ants themselves, however, are not likely to assess nest-mate numbers directly. Although evidence exists for a limited number sense in many animals, including honey bees (Chittka and Geiger, 1995; Dehaene et al., 1998), insects and simpler organisms likely rely instead on simple cues correlated with either the absolute numbers or the density of conspecifics.
Chemical cues are one likely kind of population indicator. For example, quorum-sensing bacteria monitor a diffusible signal emitted by all individuals, with suprathreshold concentrations triggering specific group behaviors that require a minimum population density (Bassler, 2002). Pheromones might play a similar role in ants, with the choice of recruitment method depending on the concentration of a volatile pheromone diffusing from nest mates in the closed cavity or a contact pheromone deposited by them on the cavity surface. Consistent with this mechanism, scouts of T. albipennis are known to chemically mark candidate nest sites (Mallon and Franks, 2000).
Alternatively, ants may respond to their rate of encounter with nest mates. During each visit to her candidate site, the scout wanders through it, approaching nest mates and touching them with her antennae. The rate of such encounters is likely to increase with the local density of ants (Gordon et al., 1993). One possibility is that an ant simply notes the time elapsed between her entry at the site and her first encounter. If this interval declines with increasing site population, it could serve as a simple cue of nest-mate density. Similar time intervals play an important role in allocating labor in partitioned tasks, efficiently balancing the numbers of workers foraging for material with the numbers using this material at the nest (Jeanne, 1986; Seeley, 1995). A second possibility is that each ant integrates information from many encounters over the duration of her visit. Use of multiple encounters could improve the precision of her estimate, as others have suggested for partitioned tasks (Ratnieks and Anderson, 1999). Similar cues have already been implicated in labor allocation within harvester ant colonies, where an ant's probability of switching to a given task is positively related to her rate of encounter with nest mates already performing that task (Greene and Gordon, 2003; Gordon and Mehdiabadi, 1999).
In this study, I examined the potential roles of encounter rate and indirect pheromonal signals in detection of a nest-mate quorum. I began by determining when, during an emigration, scouts pay attention to quorum attainment, and by estimating the time they devote to its measurement. I then explicitly tested the importance of each cue by manipulating its availability to scouts and determining how this affected their subsequent recruitment decisions.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Reversibility of switch to transport
I first determined whether ants assess nest population throughout the emigration or only until they begin to transport. I induced emigrations and followed recruiters until they switched from tandem runs to transports. I then emptied the new nest of ants and noted whether the recruiters continued to transport or reverted to tandem runs. Two colonies of T. albipennis were housed in observation nests constructed of thin cardboard or wood partitions sandwiched between glass microscope slides (50 x 76 mm). These colonies and all others used in this study were collected at Portland Bill, Dorset, U.K. Ants were individually marked with four drops of paint on the head, thorax, and gaster. Emigrations were induced by removing the roof slide from the old nest, forcing the ants to find a new home.
To aid manipulation of the new nest's population, emigrations were observed in a special arena consisting of two petri dishes joined by a short tunnel (Figure 1). The colony in its original nest was placed in one dish and the empty new nest in the other. The walls of both dishes were coated with Fluon to prevent the ants' escape. A rack-and-pinion mechanism allowed one dish to be smoothly shifted into or out of contact with the other, controlling the ants' passage between them. In each emigration, free passage was allowed until a focal ant had switched from tandem runs to transports. When this ant returned to the old nest to fetch a nest mate, I separated the dishes, removed the roof of the new nest, aspirated all ants and brood items within the nest and in the surrounding dish, and replaced the roof. No further crossings between the dishes were allowed, except by the focal transporter. After she had deposited her recruit in the new nest and returned to the old dish, I noted her subsequent recruitment behavior—tandem run, transport, or no recruitment. During her absence from the new nest, any newly recruited ant or brood item was removed. This cycle was repeated for at least five recruitment acts. Then, while the focal ant was again in the old dish, the new nest was replaced with an identical fresh nest. I noted the behavior of the focal ant on encountering this nest, as well as her recruitment behavior, if any, on her next return to the old nest. Three emigrations by each of two colonies (A5 and A8) were observed.
|
Timing and duration of quorum detection
The time required by ants to decide whether a quorum is present may help to distinguish among potential mechanisms. Detection of volatile pheromones is potentially very rapid, while measurement of lag to the first encounter may be more time consuming, and integration of a series of encounters should take longer still. To determine whether time devoted to quorum detection changed as ants switched from tandem runs to transports, I measured the duration of each visit to the new site by individually marked recruiters. Emigrations were observed in a large tray with Fluon-coated walls. An inhabited nest was placed against one wall of the tray and a single empty nest against the opposite wall, 60–65 cm away. Digital video cameras recorded the ants' activity at each nest throughout the emigration. Visit durations and subsequent recruitment behavior were measured from the videotapes. Only data from ants that led tandem runs were analyzed. I excluded the visit before each ant's first recruitment to avoid confounding time devoted to quorum detection with time needed to assess the nest. Also excluded were visits that followed searches in the arena, to avoid confounding quorum assessment with prolonged stays in the nest that may follow these searches. Six emigrations were observed: three by colony A6, two by colony 6, and one by colony 1.
Quorum size
To determine the site population at which ants switch from tandem runs to transports, I induced an emigration by colony A6 and noted, for each independent recruitment decision, the mean population of the new nest during the recruiter's immediately preceding stay. Independent decisions included only those preceding tandem runs from the old nest or each ant's first transport because the results of the reversibility experiment described above showed that ants cease monitoring nest population once they have switched to transport. All ants were individually marked, and the new nest was videotaped throughout the emigration to allow later measurement of nest population. Recruitment decisions were scored as either 0 (tandem run) or 1 (transport) and their dependence on population fitted with a Hill function, a mathematically simple and general way to represent switch-like responses:
 |
Necessity of direct contact with nest mates
To determine whether quorum detection requires physical contact between the detector and her nest mates, I allowed a focal ant access to pheromonal cues in a nest that had reached the quorum but denied her direct contact with nest mates. I then noted whether she subsequently recruited by transport, indicating that she had decided a quorum was present, or by tandem run, indicating that she had decided that it was not. Two individually marked colonies (A5 and A8) were induced to emigrate several times to a specially designed two-chamber nest (Figure 2). Ants could enter the lower chamber through a 2-mm-wide entrance hole. The inaccessible upper chamber was separated from the lower one by a double layer of very fine mesh, impassable by the ants. Because the glass roof and both chambers were not fixed to one another, each chamber could be opened during the experiment and ants quickly shuttled in or out. To control the population of ants in the new nest, emigrations were carried out in the two-dish apparatus described above (Figure 1).
|
At the start of each emigration, motivated searchers were selected as candidate focal ants by taking the first six untested ants that crossed the tunnel to the new nest dish. These ants were sequestered in a closed petri dish, while the rest of the colony was allowed to launch an emigration. The new nest was judged to have reached the quorum once a total of five transports had been completed. This value was a rough compromise between ensuring that a quorum had been attained and maintaining many naïve active ants as potential tandem run followers, should the focal ant choose to lead a tandem run. Traffic between the dishes was then cut off, and the colony was subjected to one of three treatments. (1) No ants: the new nest, its occupants, and all the ants in the surrounding dish were removed from the apparatus and replaced with an empty nest that had never been visited by any ant. (2) Ants/no contact: the ants in the lower chamber were moved to the upper chamber, and the remaining ants in the new dish were removed from the apparatus. (3) Ants/full contact: all ants were removed from and then immediately returned to the lower chamber and the new nest dish.
The six candidate focal ants were then returned to the old nest dish. After a 3-min recovery period, these ants, but no others, were allowed to pass freely between the dishes. After one of them entered the new nest, she alone was allowed subsequent access and was followed until she began to recruit. The first and third treatments served as controls, with the focal ant expected to choose tandem runs when recruiting to a completely empty nest (no ants) and transports when allowed full contact with a quorum of nest mates (ants/full contact). Behavior in the second treatment (ants/no contact) provided the critical test, with the focal ant expected to choose tandem runs if physical contact is required for her to detect nest mates and transports if pheromonal cues suffice, whether deposited by her nest mates before their removal or passing through the permeable membrane separating the chambers.
To confirm that the focal ants experienced similar nest populations in the ants/full contact and ants/no contact treatments, I counted the ants inside the nest at the end of each experiment. To test for effects of decision latency on recruitment behavior, I also noted the time elapsed between the manipulation and the focal ant's decision, that is, the point at which she left the new nest to recruit.
Role of encounter rate
To determine whether ants detect the presence of a quorum by monitoring their rate of encounter with nest mates, I manipulated this rate independent of nest population and tested whether rate or population better predicted recruitment behavior. A colony of individually marked ants (A6) was induced to emigrate first to a nest with a small floor area (3.9 cm2) and then to one with a large floor area (35.8 cm2). At a given nest population, the greater dispersal of ants over the large cavity yields a lower encounter rate than in the small cavity. Thus, if encounter rate is the important cue, ants should switch to transport at a higher absolute population—and identical encounter rate—in the large nest than in the small nest. If rate is not important, the ants should switch at the same population for both cavities, corresponding to a higher encounter rate in the small nest.
Emigrations were induced in a large plastic tray (75 x 60 x 7 cm), with the old nest placed against one wall and the new one near the opposite wall, 65 cm away. In order to reduce the discovery rate and thus ensure that many recruitment decisions were made at low nest population, the new nest was placed inside a petri dish (14-cm diam; 2.5-cm height), the sides of which were coated on both surfaces with Fluon. Ants could enter the dish only through a single small hole drilled in its side. Digital video cameras above the old and new nests recorded all entries and departures; I independently recorded the timing and participants of all tandem runs. From the videotapes, I examined visits to the new nest preceding independent recruitment decisions and measured the mean nest population experienced by the recruiter, as well as her rate of encounter with nest mates. An encounter was scored whenever the recruiter touched a nest mate with her antennae. Total encounter rate was estimated as the number of encounters during a visit, divided by visit duration. I then fitted the probability of transporting to a Hill function of either nest population or total encounter rate, as described above. To account for the possibility that ants do not make use of multiple encounters but simply respond to the latency between the start of a visit and the first encounter, I also fitted transport probability as a Hill function of the inverse of this latency.
As a second test of the role of encounter rate, I observed the decisions made by recruiters to either a large or a small nest held at a target population of approximately 15 ants. Because this population is near the estimated threshold for an intermediate sized nest, I expected that the large nest would yield a subthreshold encounter rate and the small nest a suprathreshold rate. Thus, if encounter rate is the important cue, an ant at the large nest should be more likely to lead a tandem run at her first recruitment than an ant at the small nest. To control nest population, emigrations were carried out in the two-dish apparatus described above (Figure 1). Before the start of the emigration, approximately 20 ants were moved directly from the old nest to the new dish and allowed 30 min to find their way into the nest. Because these ants moved in and out of the nest during the experiment, I occasionally needed to adjust the population by removing excess ants from the new dish or adding more from the old nest. Up to three focal ants were allowed to find their way from the old dish to the new nest and followed until their first recruitment. A total of 23 ants from two colonies (A6 and A14) were observed.
To confirm the control of nest population, I measured the mean population experienced by each ant during her last visit to the new nest before recruiting. To account for possible effects of decision latency on recruitment behavior, I also noted the time elapsed between the start of each emigration and the focal ant's decision, that is, the time at which she left the new nest to recruit.
To test the possibility that any observed effect of nest size actually resulted from differences in nest quality, I noted the preferences of 17 colonies induced to choose between the small and large nests. Each colony, housed in an intact nest, was placed against one wall of a 20 x 20–cm clear plastic assay dish, with empty small and large nests placed at opposite corners. Emigrations were induced by removing the roof of the old nest. Three test sessions were run, using approximately one third of the colonies each time. Within each session, tests were performed simultaneously, the dishes stacked on top of one another and separated by a paper towel to avoid potential visual cues offered by the behavior of neighboring colonies. The stack was centered beneath a fluorescent ceiling light and its sides were shielded with white fiberglass trays to prevent external light gradients or other visual cues from biasing the ants. The left and right positions of the nest types were alternated through the stack to control for any remaining side bias. The nest to which each colony had moved was noted the day after the old nests had been opened and the choices tested for a significant preponderance of one design over the other.
Statistical analysis
Analyses were performed with the statistical package R (Venables and Ripley, 2002). Summary data are reported as mean ± standard deviation. In box plots, the ends of each box mark the upper and lower quartiles, the filled circle inside the box gives the median, the brackets extend to the furthest data point no more than 1.5 times the interquartile range from the box, and the open circles show outliers.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Reversibility of switch to transport
In each of six cases, an ant that had already started to transport continued to do so energetically even after the new nest population was experimentally reduced to zero (Figure 3). Ants showed little sign of disturbance, although their first visits to the new nest after it was emptied sometimes appeared prolonged. Once the new nest was replaced with an identical but unfamiliar nest, however, five ants ceased recruiting altogether and one waited 30 min before transporting again. This manipulation seemed clearly to disturb the ants, with each one hesitating before entering the nest and frequently leaving again before putting down her transportee. In one case, the ant refused to deposit her nest mate, instead returning to the old nest.
|
Timing and duration of quorum detection
After discovering or being recruited to the new nest, an active ant typically made a series of round trips to the old nest, separated by brief but variable stays in the new nest. The durations of these stays shortened abruptly once an ant began to transport (Figure 4). Visits fell into two distinct groups: early long stays preceding tandem runs and the first transport and later brief stays preceding all subsequent transports. (ANOVA:
post hoc comparisons by Tukey's honest significant difference method). As described in Methods, this analysis excluded certain visits to avoid confounding effects of time devoted to site assessment or rest periods after search trips. Dropping any or all of the exclusion criteria, however, did not substantially alter the results.
|
The sequence of each ant's recruitment decisions was consistent with an irreversible switch to recruitment. Many ants led one or a few tandem runs before switching to transport (24 cases), but most transported exclusively (212 cases), and a few quit recruiting after a few tandem runs (5 cases). The exclusive transporters typically started to recruit later in the emigration, once the nest population was already large. In no case did an ant lead a tandem run from the old nest to the new once she had begun transporting. Many transporters, however, led occasional tandem runs from the new nest back to the old (34 cases).
Quorum size
According to a Hill function fit to data from 72 ants, the probability of transporting reached 50% at a mean population, T, of 20.2 ants, which may be taken as the size of the transport quorum (Figure 5). The increase in probability with population was roughly step-like, as reflected in the relatively high Hill coefficient (k = 6.3).
|
Necessity of direct contact with nest mates
When encountering a cavity with nest mates isolated behind a permeable membrane, ants consistently responded by leading tandem runs (Figure 6). This behavior was identical to that of ants encountering a truly empty cavity and unlike that of ants allowed full contact with nest mates in the cavity, who always performed transports. The opportunity for contact with nest mates strongly predicted recruitment decision (analysis of deviance:
), while the population of the nest did not (analysis of deviance: 
). Thus, in the absence of direct contact with nest mates, ants responded as though the quorum had not been met, despite the availability of pheromonal or other indirect cues. Decision latency did not vary with treatment (ANOVA: 
) nor did it influence recruitment decisions (analysis of deviance: 
).
|
Role of encounter rate
When moving to a large nest, ants switched to transport at a significantly higher mean nest population than they did when moving to a small nest (ANOVA:
) (small nest: T = 8.1; large nest: T = 23.9, k = 4.4) (Figure 7). In contrast, no effect of size could be detected when the probability of switching was instead fitted as a function of total encounter rate (ANOVA: 
) (small nest: T = 6.2 encounters per minute; large nest: T = 7.7 encounters per minute, k = 2.4). Size similarly showed no effect when recruitment choice was fitted to the inverse of the lag between the start of a visit and the first encounter (ANOVA: 
) (small nest: 1/T = 11.8 s lag; large nest: 1/T = 2.4 s lag, k = 1.8).
|
In a second experiment, ants that encountered a small nest proved significantly more likely to transport than ants that encountered a large nest, when the two nest types were fixed at roughly equal populations (analysis of deviance:
) (Figure 8) Although the exact populations of the nests varied across replicates, mean populations did not differ significantly between treatments (ANOVA: 
), and population had no effect on recruitment decision (analysis of deviance: 
). Similarly, treatments did not differ in the latency of the recruitment decision (ANOVA: 
) nor did this latency influence the ants' decisions (analysis of deviance: 
).
|
Colonies showed no consistent preference for one nest design over the other. Most colonies split between the two nests, with each type predominating in roughly half of the colonies (Figure 9).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The collective choice of a new home by emigrating colonies of the ant T. albipennis depends critically on the detection of a nest-mate quorum by the ants evaluating each candidate site. The quorum allows ants to condition their full commitment to a site—signaled by the switch from slow tandem runs to rapid transports—on the parallel decisions of their nest mates (Pratt et al., 2002
). In this study, I show that quorum detection depends on direct contact between an ant and her nest mates at the site, probably acting through density-dependent encounter rates. I further show that ants cease monitoring the presence of a quorum once they have made the switch to transport, failing to revert to tandem runs even if the site's population falls well below the quorum. Finally, my observations suggest that quorum detection is not instantaneous but requires ants to invest 1–2 min patrolling the site to gather information.
The finding that direct contact is required for quorum detection is crucial to interpreting the other experimental results. For example, the observation that ants carry on transporting to an experimentally emptied nest might otherwise be explained by their continuing to respond to lingering pheromonal cues. Likewise, the higher quorum seen in larger nests could be attributed to lower concentration of a chemical cue generated by a given population of ants in a larger cavity. These alternative explanations, however, are rejected by the necessity of direct contact to trigger the switch from tandem runs to transports. Chemical cues are not, however, irrelevant to the recruitment decisions of ants. If their candidate nest is replaced by an identical but unfamiliar one, recruiters are clearly disturbed and generally cease transporting to the site. This is consistent with previous findings that ants deposit individual-specific chemical markers in candidate sites that they have found (Mallon and Franks, 2000).
The effects of cavity area on quorum size probably do not reflect differences in perceived nest quality. Although no evidence has been reported for an effect of nest quality on quorum size, this question has not been thoroughly examined and the possibility must be considered credible in the light of recent work, suggesting that ants adjust their quorum size according to the degree of crisis faced at the old nest and hence the urgency with which they must complete the move to a new one (Franks et al., 2003a). Even if such effects exist, however, they cannot explain differences in behavior at the large and small nests used here because preference tests show that the ants found them equally attractive. This may seem surprising, given previous studies showing a clear preference for larger nests over cavities similar to the small ones used here (Franks et al., 2003b; Mallon and Franks, 2000). The largest nests used in those tests, however, were only about one quarter the size of the large nest used here. My results therefore suggest that colonies favor an intermediate size range, although firm conclusions on this point will require simultaneous comparison of all three designs in the same experiment.
Exactly how encounter rate indicates quorum attainment remains unclear. In particular, how many of her encounters does an ant actually use? At one extreme, she might integrate all of them to assay a mean encounter rate over her entire visit. At the other extreme, she might respond only to the interval between entering the nest and her first encounter. The current results are consistent with both these hypotheses, as well as intermediate possibilities. Favoring the single-encounter hypothesis is its obvious cognitive simplicity and the evidence for similar interval measurements in the decision-making of other insects, including collective decisions by bees and wasps (Jeanne, 1986; Seeley, 1995). On the other hand, integration of multiple encounters might improve precision by reducing the impact of stochastic effects. A similar suggestion has been made about honey bee nectar foragers, who apparently use the time interval until their first unloading by a receiver bee to decide whether their recruitment effort should be directed at foragers or receivers (Seeley, 1995). Because foragers often unload to many receivers, the possibility remains open that they actually integrate several search intervals (Hart and Ratnieks, 2001; Ratnieks and Anderson, 1999; but see Huang and Seeley, 2003). Moreover, other insects clearly detect local crowding via elevated rates of tactile contact. In the locust Schistocerca gregaria, the shift from solitary to gregarious behavior in crowded conditions is sparked by a prolonged high rate of touching of the hind legs (Simpson et al., 2001). In the social aphid Tuberaphis styraci, tactile contact similarly triggers production of a specialized soldier morph in crowded conditions (Shibao et al., 2004).
A role for multiple encounters is supported by evidence for a significant time cost of quorum detection. The switch to transport, when ants stop attending to nest population, coincides with an abrupt shortening of stays at the new nest, from about 3 min to less than 1 min. The roughly 2-min difference may represent the time needed for quorum detection. In contrast, the longest latency until the first encounter was less than 1 min and most were less than 20 s, even in relatively empty nests. If the extra length of early visits truly reflects investment in quorum detection, this implies that information is also gathered from later encounters. This interpretation must be considered tentative, however, as it also remains possible that the extra length reflects time devoted to assessing other aspects of the site that are ignored once the switch to transport is made.
Implicit in this discussion is an assumption about what constitutes an encounter. Although this study defined encounters as direct antennal contact, more distant interactions are possible. Workers of Lasius niger, for example, can detect one another at a distance of about 12 mm, probably by sight, and then decide whether to approach more closely for antennal contact (Gordon et al., 1993). T. albipennis conceivably have similar abilities, but there are several reasons to suppose that distant interactions would not be adequate for quorum detection. First, ants can identify the distinctive cuticular hydrocarbon profiles of nest mates only on close inspection with antenna-born chemoreceptors (Hölldobler and Wilson, 1990). Second, T. albipennis appear to have trouble spotting one another even at very short distances. A tandem follower who loses contact with her leader enters a distinctive looping search pattern that she continues even if it brings her within a few millimeters of the waiting leader, apparently recognizing her partner only on making direct antennal contact. Finally, visual perception of distant nest mates may be of little use in the darkness of natural nest cavities.
Another assumption of the current study is that all encounters have the same effect on quorum detection. Future work may show that ants discern various sorts of encounter, such as vigorous mutual antennations versus glancing touches of a nest mate's abdomen. A particularly interesting question is whether ants can distinguish encounters with different types of ant—fellow recruiters, uncommitted scouts, passive ants, etc.—on the basis of hydrocarbon signatures perceived during antennation. Harvester ants appear to use similar cues to detect the task specialization of nest mates (Greene and Gordon, 2003). Temnothorax scouts conceivably learn more from some encounters than others and might take this into account when deciding whether to switch to transport. For example, the presence of passive transportees or brood items would signal clearly that at least one fellow scout has begun to transport.
This study's findings have implications for the collective decision-making algorithm used by colonies to choose the best of several potential nests. The quorum rule contributes to this decision by amplifying differences in recruitment strength to nests of different quality and by acting as a check on erroneous recruitment initiations by any single ant. The delay imposed by waiting for a quorum can be seen as a cost paid for better decision-making. The discovery here that the switch to transport is irreversible, even if the population again falls below the quorum, may imply a limit placed on this cost. The maintenance of fast recruitment prioritizes speed of abandonment of the old nest over avoidance of movement to an inferior site. In nature, this situation may arise rather rarely because transports, once started, will tend to bolster the nest's population. A possible exception may occur when a mediocre site is simultaneously the target of transports by ants carrying nest mates from the old nest and the source of transports by others who have found a better site.
If quorum attainment is signaled by ant density, rather than absolute population, the effective quorum size will depend on nest area. Ants might be expected to control for this by adjusting the encounter rate at which they switch according to the area of the nest to which they are recruiting. The current results show that they do not do so. When recruiting to nests of different size, ants switch to transport at very different nest populations but similar encounter rates. Combined with the considerable variation among individuals in the absolute population at which the switch is made, these results suggest that ants do not closely regulate quorum size. This in turn suggests that the collective decision-making algorithm may be robust to variation in this parameter. The effectiveness of the quorum may derive not from its having a specific, optimal value but from its contribution to the structure of the decision algorithm—a step change in recruitment strength conditioned on information from other ants. This possibility merits further investigation, as the emergence of robust performance directly from the architecture of complex biological networks may be a widespread phenomenon (Yi et al., 2000).
The results of this study illuminate the proximate cues guiding the individual decisions from which collective choices emerge. This kind of knowledge clarifies the links between cognitive mechanisms of individual insects and the "cognition" of the colony as a whole. Ultimately, this should allow a better understanding of how selection has shaped individual behavior to produce functional colony-level phenotypes, through a network of interactions among colony members and their environment.
|
ACKNOWLEDGEMENTS |
|---|
This work was supported by the Pew Charitable Trusts (award 2000-002558).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Bassler BL, 2002. Small talk: cell-to-cell communication in bacteria. Cell 109:421–424.[CrossRef][ISI][Medline]
Chittka L, Geiger K, 1995. Can honey bees count landmarks? Anim Behav 49:159–164.[CrossRef]
Darchen R, 1968. Le travail de la cire et la construction dans la ruche. In: Traité de biologie de l'abeille (Chauvin R, ed). Paris: Masson; 566.
Dehaene S, Dehaene-Lambertz G, Cohen L, 1998. Abstract representations of numbers in the animal and human brain. Trends Neurosci 21:355–361.[CrossRef][ISI][Medline]
Franks NR, Dornhaus A, Fitzsimmons JP, Stevens M, 2003a. Speed versus accuracy in collective decision making. Proc R Soc Lond B Biol Sci 270:2457–2463.[Medline]
Franks NR, Mallon EB, Bray HE, Hamilton MJ, Mischler TC, 2003b. Strategies for choosing between alternatives with different attributes: exemplified by house-hunting ants. Anim Behav 65:215–223.[CrossRef]
Franks NR, Pratt SC, Mallon EB, Britton NF, Sumpter DJT, 2002. Information flow, opinion polling and collective intelligence in house-hunting social insects. Philos Trans R Soc Lond B Biol Sci 357:1567–1583.[CrossRef][ISI][Medline]
Gordon DM, Mehdiabadi NJ, 1999. Encounter rate and task allocation in harvester ants. Behav Ecol Sociobiol 45:370–377.[CrossRef]
Gordon DM, Paul RE, Thorpe K, 1993. What is the function of encounter patterns in ant colonies? Anim Behav 45:1083–1100.[CrossRef]
Greene MJ, Gordon DM, 2003. Social insects—cuticular hydrocarbons inform task decisions. Nature 423:32.[CrossRef][Medline]
Hart AG, Ratnieks FLW, 2001. Why do honey-bee (Apis mellifera) foragers transfer nectar to several receivers? Information improvement through multiple sampling in a biological system. Behav Ecol Sociobiol 49:244–250.[CrossRef]
Hölldobler B, Wilson EO, 1990. The ants. Cambridge, Massachusetts: Belknap Press of Harvard University Press.
Huang MH, Seeley TD, 2003. Multiple unloadings by nectar foragers in honey bees: a matter of information improvement or crop fullness? Insectes Soc 50:330–339.[CrossRef]
Jeanne RL, 1986. The organization of work in Polybia occidentalis—costs and benefits of specialization in a social wasp. Behav Ecol Sociobiol 19:333–341.[CrossRef][ISI]
Lumsden CJ, Hölldobler B, 1983. Ritualized combat and intercolony communication in ants. J Theor Biol 100:81–98.[CrossRef]
Mallon EB, Franks NR, 2000. Ants estimate area using Buffon's needle. Proc R Soc Lond B 267:765–770.[Medline]
Mallon EB, Pratt SC, Franks NR, 2001. Individual and collective decision-making during nest site selection by the ant Leptothorax albipennis. Behav Ecol Sociobiol 50:352–359.[CrossRef]
Pacala SW, Gordon DM, Godfray HCJ, 1996. Effects of social group size on information transfer and task allocation. Evol Ecol 10:127–165.
Pratt SC, Mallon EB, Sumpter DJT, Franks NR, 2002. Quorum sensing, recruitment, and collective decision-making during colony emigration by the ant Leptothorax albipennis. Behav Ecol Sociobiol 52:117–127.[CrossRef]
Ratnieks FLW, Anderson C, 1999. Task partitioning in insect societies. II. Use of queueing delay information in recruitment. Am Nat 154:536–548.[CrossRef][Medline]
Seeley TD, 1995. The wisdom of the hive. Cambridge, Massachusetts: Belknap Press of Harvard University Press.
Seeley TD, 2003. Consensus building during nest-site selection in honey bee swarms: the expiration of dissent. Behav Ecol Sociobiol 53:417–424.
Shibao H, Kutsukake M, Fukatsu T, 2004. The proximate cue of density-dependent soldier production in a social aphid. J Insect Physiol 50:143–147.[CrossRef][ISI][Medline]
Simpson SJ, Despland E, Hagele BF, Dodgson T, 2001. Gregarious behavior in desert locusts is evoked by touching their back legs. Proc Natl Acad Sci USA 98:3895–3897.
Venables WN, Ripley BD, 2002. Modern applied statistics with S, 4th ed. New York: Springer.
Yi TM, Huang Y, Simon MI, Doyle J, 2000. Robust perfect adaptation in bacterial chemotaxis through integral feedback control. Proc Natl Acad Sci USA 97:4649–4653.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. M. Gordon, S. Holmes, and S. Nacu The short-term regulation of foraging in harvester ants Behav. Ecol., January 1, 2008; 19(1): 217 - 222. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Pratt and D. J. T. Sumpter A tunable algorithm for collective decision-making PNAS, October 24, 2006; 103(43): 15906 - 15910. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||