Group decision making in nest-site selection by honey bees

HAL Id: hal-00891879https://hal.archives-ouvertes.fr/hal-00891879

Submitted on 1 Jan 2004

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Group decision making in nest-site selection by honeybees

Thomas Seeley, P. Kirk Visscher

To cite this version:Thomas Seeley, P. Kirk Visscher. Group decision making in nest-site selection by honey bees. Api-dologie, Springer Verlag, 2004, 35 (2), pp.101-116. <10.1051/apido:2004004>. <hal-00891879>

https://hal.archives-ouvertes.fr/hal-00891879

https://hal.archives-ouvertes.fr

101Apidologie 35 (2004) 101–116© INRA/DIB-AGIB/ EDP Sciences, 2004DOI: 10.1051/apido:2004004

Review article

Group decision making in nest-site selection by honey bees

Thomas D. SEELEYa*, P. Kirk VISSCHERb

a Department of Neurobiology and Behavior, Cornell University, Ithaca, NY 14853, USAb Department of Entomology, University of California, Riverside, CA 92521, USA

(Received 8 August 2003; revised 5 November 2003; accepted 30 November 2003)

Abstract – In recent years, renewed attention has been paid to the mechanisms of group decision makingthat underlie the nest-site selection process in honey bees. We review the results of these new investigationsby discussing how the recent work builds on the earlier descriptive studies of this decision-making process,how the decision-making abilities of swarms have been tested, and how the mechanisms of this decision-making process have been experimentally analyzed. We conclude by discussing how the scouts in a swarmsense when their group decision making is coming to an end and so should begin stimulating their quiescentswarm-mates to prepare for the flight to their new home.

Apis mellifera / group decision making / honey bee / nest-site selection / swarming / waggle dance

1. INTRODUCTION

For centuries, beekeepers have known thatafter a swarm leaves its hive and coalesces intoa cluster hanging from a tree branch, itsearches for a home and, if left alone, eventu-ally departs for a new abode. Because humansusually have hived the bivouacked swarmsthey have found, and so have cut short thebees’ process of nest-site selection, it is notsurprising that this process long remained adeep mystery. This situation changed in the1950’s when Martin Lindauer (1951, 1953,1955) published his classic studies of househunting by honey bees. Lindauer discoveredthat the bees seen performing waggle danceson the surface of a swarm cluster are scoutsadvertising potential nest sites, not foragersannouncing rich food sources. He then tookadvantage of Karl von Frisch’s recent successin deciphering the waggle dance communica-tion system (von Frisch, 1946) to eavesdrop onthe “deliberations” conducted by a swarm’sscouts as they consider a dozen or more possi-

ble dwelling places and ultimately reachagreement on one of them.

The next phase in the analysis of the nest-site selection process of honey bees came inthe 1970’s when various investigators turnedtheir attention to the actions of scout bees atpossible nest sites rather than the activities ofthese bees at the swarm cluster. Numerousstudies were conducted to address the questionof how scouts judge the quality of a potentialnest site (reviewed by Seeley, 1985; Witherell,1985). For the races of Apis mellifera L. nativeto Europe, it was learned that an attractive nestsite has a cavity volume greater than 10 litersand an entrance hole smaller than 30 cm2,perched several meters off the ground, facingsouth, and located at the floor of the cavity.Neither cavity shape nor entrance shape isimportant. When inspecting a site, a scoutinvests much time in walking the inner sur-faces of the cavity and her perception of cavityvolume is linked somehow to how much walk-ing she must do to circumnavigate a cavity(Seeley, 1977; Franks and Dornhaus, 2003).

* Corresponding author: [email protected]

102 T.D. Seeley, P.K. Visscher

This research on the nest-site preferences ofbees set the stage for the development of effec-tive bait hives (Witherell, 1985; Schmidt andThoenes, 1987, 1990), which have proven val-uable in various settings, including the moni-toring and control of swarms of Africanizedhoney bees in the southwestern US (Schmidt,1990).

Over the last few years, renewed attentionhas been paid to the mechanisms of groupdecision making in honey bee swarms,essentially taking up where Lindauer left off inthe 1950’s. Ever since Lindauer published hisfindings, there has been much speculationabout the mechanisms of this decision-makingprocess (Wilson, 1971; Griffin, 1981, 2001;Markl, 1985) and this has made clear the needfor further analytic work. Moreover, sinceLindauer’s time, there has been a tremendousgrowth in interest in how animal groupsfunction as adaptive units in general (Bourkeand Franks, 1995; Seeley, 1995; Sober andWilson, 1998) and as decision makers inparticular (Franks et al., 2002; Conradt andRoper, 2003). A swarm of honey beeschoosing its future home is one of the mostimpressive examples known of an animalgroup functioning as an adaptive decisionmaker. In this paper, we review what has beenlearned lately about the mechanisms of groupdecision making by honey bee swarms.

Before delving into these mechanisms, wewish to draw attention to three requirementsthat a swarm of bees must fulfill to succeed inthe demanding task of choosing a home. First,it must achieve an accurate decision. A col-ony’s success depends critically on its occupy-ing a cavity that is sufficiently roomy to holdthe combs the colony will need for rearing itsbrood and storing its honey. At the same time,the cavity must be sufficiently tight to providegood protection from predators, robbers, andharsh weather. Second, it must achieve aspeedy decision. Every additional hour that aswarm spends as an exposed cluster, hangingfrom a tree branch, lowers its energy reservesand raises its chances of being soaked by rain.And third, it must achieve a unified decision.A split decision would lead to swarm fragmen-tation, which would be disastrous in mostcases given that a swarm usually has just onequeen and so can establish just one fully func-tioning colony. Thus the central question that

we face is this: how do the bees in a honey beeswarm work together to produce an accurate,speedy, and unified choice of a home?

2. RENEWING THE ANALYSIS

Since Lindauer’s work in the 1950’s, it hasbeen clear that only a small minority of thebees in a swarm cluster – the scout bees – areactively involved in the decision-making proc-ess. The vast majority of a swarm’s membersremain quiescent until a decision has beenmade and it is time to fly to the chosen site. Ithas also been clear that the scout bees searchthe surrounding environment for suitable nestsites, advertise their finds by performing wag-gle dances, and eventually come to an agree-ment on one of these sites. What was not clearuntil recently is how exactly the scout beesbehave to produce this agreement. As a firststep toward solving this puzzle, we repeatedLindauer’s observations of the scout bees’dances and thus their deliberations, but usingvideo equipment to get a more complete pic-ture of the scouts’ dancing than was possiblein the 1950’s (Camazine et al., 1999; Seeleyand Buhrman, 1999). Having only pencil andpaper for recording the scout bees’ behavior,Lindauer was greatly limited in the amount ofdance information he could record. He settledon recording only the first dance performed byeach dancing bee; a paint mark applied to eachdancer as she performed her first dance ena-bled him to distinguish novice dancers(unmarked) from repeat dancers (marked).Working nearly 50 years later and blessed withvideo technology, we were able to record,playback, and analyze every dance performedon a swarm. Moreover, by working with smallswarms of about 4 000 bees, and labeling indi-viduals so that they could be identified, wewere able to attribute each dance to a particu-lar scout bee and so were able to determine thedance history, and other behaviors, of everybee that contributed to a swarm’s decisionmaking.

Figure 1 shows one of the complete dancerecords of a swarm’s decision making. Thepattern resembles closely what Lindauerreported based on his first-dance-only records.The entire decision-making process requiredabout 16 hours of dance activity by the scouts,

Decision making by honey bee swarms 103

spread over 3 days (a period of rain startingaround noon on day 2 extended the processinto day 3). We see that during the first half ofthe decision-making process, the scout beesreported all 11 of the potential nest sites thatthey would consider. We can also see that dur-ing the first half of the process no one sitedominated the dancing; during the second half,however, one of the sites gradually began to beadvertised much more than the others. Indeed,during the last few hours of the decisionmaking, the site that had emerged as the front-runner (site G) became the object of allthe dances – more than 3000 waggle runs –

performed on the swarm. By the end, therewas unanimity among the dancing bees.

This initial work confirmed many of thefeatures of the swarm bees’ decision-makingprocess that had been previously reported byLindauer based on his partial records of thedancing on swarms. These features include:(1) the scout bees locate potential nest sites inall directions and at distances of up to severalkilometers from the swarm; (2) initially, thescout bees advertise a dozen or more potentialnest sites but eventually they advertise just onesite; (3) within an hour or so of the appearanceof unanimity among the dancers, the swarm

Figure 1. History of a swarm’s decision-making process from the time that the first potential nest site wasadvertised on the swarm (shortly after 11:00 hours on July 20, 1997) to when it lifted off to fly to its newhome (at 11:58 hours on July 22, 1997). The circle within each of the panels represents the location of theswarm; each arrow pointing out from the circle indicates the distance and direction of a potential nest site;the width of each arrow denotes the number of different bees that danced for the site in the time periodshown. The set of numbers at the tip of each arrow denotes three things: top the number of bees that dancedfor the site, middle the number of waggle runs performed for the site, bottom the mean number of waggleruns per dance for the site. The numbers after “bees”, “dances”, and “waggle runs” within each panel denotethe total number of each (summed over all the potential nest sites) for the time period shown. (From Behav.Ecol. Sociobiol. 45, Seeley and Buhrman, 1999, © Springer-Verlag.)


lifts off; (4) there is a crescendo of dancing justbefore liftoff; and (5) the chosen site is notnecessarily the one that is first advertised onthe swarm. Furthermore, the analysis of thedancing records of individual scout bees con-firmed something else that Lindauer had sus-pected, which is that there is much turnover inthe dancing bees over the course of a decision-making process. Most bees that dance for a sitecease doing so after a few hours, letting thenext “generation” of dancers carry on thedeliberations. Thus it became clear that aswarm’s choice of a future home is broadlydistributed among the scout bees, and that thisleaderless process of group decision-makingconsists of a friendly competition among thedifferent groups of dancers representing thedifferent potential nest sites. The groups com-pete for additional dancers. Sooner or later,one group of dancers grows numerous andultimately excludes its competitors. The sitewhose dancers prevail in this winners-take-allcontest becomes the swarm’s new home.

3. TESTING A SWARM’SDECISION-MAKING ABILITY

Watching the dancing bees on a swarmbuild a consensus for a nest site naturallyraises the following question: is the site thatwins in the competition for dancers the best ofthe sites that the scout bees have found? Theresults of a recent study (Seeley and Buhrman,2001) indicate that swarms are indeed gener-ally successful in employing the best-of-Ndecision rule: sample some number (N) ofalternatives and then select the best one. In thisstudy, five swarms of bees were transportedone at a time to a windswept, brush-coveredisland along the coast of Maine, which had fewnatural nest sites. There each swarm waspresented with an array of five nest boxes;four provided a mediocre but acceptable homesite – a 15-liter cavity – and one offered anexcellent home site – a 40-liter cavity. Therecord of each swarm’s search for and choiceamong these five alternative nest sites isshown in Figure 2. We see that in each case thescout bees discovered one or more of themediocre sites well before they located theexcellent site, sometimes several hours inadvance, and that they recruited other bees to

the mediocre site. However, we also see ineach case, except trial 4, that as soon as theexcellent site was discovered the beesrecruited very strongly to this site and thatwithin a few hours the number of scout bees atthe excellent site greatly exceeded the numberat any of the mediocre sites. Ultimately, eachswarm, except the fourth, lifted off and flewtoward the excellent site. Thus, in four out offive trials, swarms selected the best site, anoutcome that is extremely unlikely (P =0.0064) to have occurred simply by chance.

4. ANALYZING THE DECISION-MAKING PROCESS

As mentioned already, a key requirement ofthe decision-making process of honey beeswarms is that it produces a unified decision,for ultimately a swarm will occupy just onenest site. It is, therefore, not surprising thatbefore a swarm lifts off to fly to its chosen site,almost always the dancers on the swarm havereached an agreement such that all of theirdances indicate a single location. (Below wewill discuss what happens when a swarm mis-takenly lifts off before the dancers havereached an agreement, and what these mis-takes reveal about how the bees control theirpreparations for liftoff.) Only recently, how-ever, have we gained a clear picture of how thescouts behave so that eventually all of theirdances are performed for one of the severalsites under consideration, usually the best one.

How do the scouts build a consensus for thebest site? The descriptive studies discussedabove indicate that the essence of the nest-siteselection process is a friendly competitionamong the scouts “committed” to the variouspotential nest sites. Each coalition of scoutscommitted to a particular site competes withthe other coalitions for additional membersdrawn from the pool of uncommitted scouts.For the basic situation of a choice between twosites, the swarm bees’ decision-making proc-ess can be conceptualized with the simple pairof equations that were put forth by Winsor(1934) (cited by Hutchinson, 1978) early inthe history of ecological competition theory:

dN1/dt = r1N1U – a1N1 (1)

dN2/dt = r2N2U – a2N2 (2)


Figure 2. Results of five trials of a test of the ability of swarms to select the best of available nest sites. Fivenest boxes were arranged in a fan-shaped array with each box 230–250 m from the swarm. One nest boxwas an excellent nest site (40-L cavity) while the other four were only mediocre sites (15-L cavities). Acount of the number of scout bees visible at each nest box was made every 30 min. In each trial, one or moreof the mediocre sites was discovered 1–14 hours before the excellent site. Nevertheless, by the end of eachtrial, except the fourth, the swarm chose the 40-L site. Thus we see that most of these swarms were accuratedecision makers. (From Behav. Ecol. Sociobiol. 49, Seeley and Buhrman, 2001, © Springer-Verlag.)


Ni is the number of scouts committed to site i,U is the number of uncommitted scouts (theresource being competed for), ri is therecruitment (“birth”) rate per scout committedto site i, and ai is the abandonment (“death”)rate per scout committed to site i. Integratingand eliminating U from the two equationsyields

. (3)

Thus if the scouts behave with a higher percapita rate of recruitment to site 1 than site 2,and a higher per capita rate of abandonment ofsite 2 than site 1, then the quantity (r1a2 – r2a1)has a positive value and the ratio on the leftside of equation (3) is increasing. This impliesthe gradual elimination of the coalition ofscouts committed to site 2. In ecological the-ory, these equations are useful as perhaps thesimplest expression of the principle of com-petitive exclusion. They are useful in thepresent context by showing how differencesin per capita rates of recruitment to and aban-donment of different potential nest sites aresufficient to produce a clear winner in thecompetition among coalitions of scout bees. Inthis model, the competition is for a singleresource – uncommitted scout bees – andoccurs purely by exploitation. In principle, thecompetition among coalitions could alsoinvolve interference, with scouts of a givencoalition directly inhibiting the growth of theother coalitions, and this possibility will beconsidered below.

There is now solid evidence that scouts doadjust their behavior in relation to site qualityso that the best site has both the highest percapita rate of recruitment (ri) and the lowestper capita rate of abandonment (ai). Lindauer(1955) stated that scouts reporting better sitesperform longer and livelier dances (whichshould result in a higher per capita recruitmentrate), but his evidence was only anecdotal.Seeley and Buhrman (2001) obtained experi-mental evidence in support of Lindauer’sclaim. Their approach was to present a swarmwith both an excellent (40 L) nest box and amediocre (15 L) nest box, and then to videorecord the scout bees’ dances for the two nestboxes as they were performed side-by-side onthe swarm. This experiment was performedwith two swarms and in both cases they foundthat the dances for the excellent nest box were

stronger than those for the mediocre nest box,there being on average 36 and 14 waggle runsper dance for the excellent and mediocre nestboxes, respectively. They also found thatLindauer was indeed correct when he statedthat better sites elicit dances that are bothlonger and livelier than those elicited bypoorer sites. Specifically, Seeley and Buhrmanfound that a scout bee tunes her dance strengthby adjusting the number of waggle runs/dance,and that she adjusts the number of waggleruns/dance (W) by changing both the duration(D) and the rate (R) of her waggle-run produc-tion (note that W = D × R). Further analysis ofthe video recorded dances revealed that adancing bee adjusts the rate of her waggle-runproduction by changing the mean duration ofthe return-phase portion of her dance circuits.Evidently, it was these differences in return-phase duration that gave Lindauer (1955) theimpression that dances differ in liveliness.

The rate of recruitment by a dance-produc-ing scout bee depends not only on howstrongly she dances (number of waggle runsper dance), but also on how the dance-follow-ing scouts sample and respond to the variousdances being performed on the surface of aswarm. The simplest possibility is that when ascout decides to follow a dance, she chooses adancer at random and heeds her dance. If so,then the greater the per capita production ofwaggle runs for a site, the higher will bethe per capita rate of recruitment to the site.Visscher and Camazine (1999) report evi-dence that scouts do indeed follow dances cho-sen at random. In a desert area, they provideda swarm with two nest boxes – one east andone west of the swarm – and they labeled eachdancer for individual identification when shefirst danced for one of the nest boxes. Theyalso videotaped all dancing and eventualdance following by the labeled bees through-out the swarm’s decision-making process.When they examined how much time thelabeled bees spent following dances for theeast and west nest boxes, they found that theyfollowed dances for each site in proportion tothe total amount of dancing by other bees forthe site, as shown in Figure 3. Evidently, whenthe labeled bees switched from dance produc-ing to dance following (discussed furtherbelow), they showed no preference for eitherthe dances advertising the site they had already

N1r2 N2

r1⁄ Ce r1a2 r2a1–( )t=


visited or the dances for the other site, andinstead simply followed dances chosen atrandom.

So far we have considered only how scoutbees committed to a site recruit other scouts totheir site (i.e., what determines ri). Let us nowconsider instead how the scout bees commit-ted to a site abandon this site, that is, how theystop dancing for and stop making visits to thesite, and so terminate their commitment to thesite (i.e., what determines ai). This is one of themost curious features of the decision-makingprocess of honey bee swarms. To conceptual-ize the matter, it is useful to recognize twoclasses of hypotheses for why a scout aban-dons a site: H1, an internal stimulus causes herto abandon a site; and H2, an external stimuluscauses her to abandon a site. (Note: we knowthat scouts do not abandon their sites simplybecause they are dying. Seeley and Buhrman(1999) measured the mortality rate of scoutbees in 3 swarms and found it to be low, withonly 6% of the scouts dying over the course ofthe decision-making process.)

Hypothesis 1 is based upon the observation,made repeatedly in the descriptive studies(Lindauer, 1955; Camazine et al., 1999;Seeley and Buhrman, 1999) that scouts visit-ing nest sites rather quickly lose interest intheir sites. With respect to dancing, forexample, most scouts show a statistically sig-nificant tapering off over several hours in the

number of waggle runs performed per dance.(Lindauer aptly called this a decline in Tanz-lust.) Even scouts that start out visiting theultimately chosen site, which presumablyremains a high-quality site throughout thedecision-making process, eventually ceaseperforming dances for and stop making visitsto this site. The fact that even scouts commit-ted to the chosen site eventually abandon thissite suggests strongly that the scouts’ loss ofinterest is not the result of an external influ-ence associated with being committed to aninferior site – for example, experiencing at thesite a weak buildup of scouts or encounteringat the swarm a vigorous dance for an alterna-tive site – but is instead the result of an inter-nally programmed process. One can argue thatit may be highly adaptive for scouts to be pro-grammed to eventually abandon their sites,because it could help prevent the decision-making process from becoming deadlockeddue to unyielding scouts dancing for two ormore sites.

Hypothesis 2 is based upon the observa-tions made by Lindauer (1955) of 3 scout beesthat stopped visiting one site at about the sametime they started following dances for andbegan making visits to a second site. Lindauer(1971, 1975), Griffin (1992) and Gould andGould (1994) have pointed out that theseobservations suggest that scouts will abandonone site only after they have followed dancesfor, inspected, and so gained interest in a supe-rior site. “Bees that had visited a cavity ofmediocre quality sometimes became followersof more enthusiastic dances than their own.Then some of them visited the better cavitythey had learned about as followers of vigor-ous dances, returned, and danced appropri-ately with respect to the superior cavity theyhad now visited” (Griffin, 1992, p. 193). Ifbees decide to abandon a site only after theyhave learned about a superior site, it would bestrong support for hypothesis 2, because itwould appear that information about a supe-rior site (an external stimulus) is what causes ascout to abandon a site. However, the temporalrelationship between abandonment of one siteand recruitment to a second site was, untilrecently, unclear. It may be that scouts gener-ally lose interest in one site even before theylearn about a second site. If so, then this wouldbe evidence against hypothesis 2, because it

Figure 3. Scout bees that had originally danced forone of two available nest boxes (east or west) couldlater follow dances for each site. The expectedamounts of dance following by each class of beesfor each site (Exp) were calculated by parceling outtotal time each labeled dancer followed dancesaccording to the distribution of dances available forher to follow. The actual distribution of dancefollowing (Obs) closely matched this expectation,showing that scouts were not selective either fornovel or familiar sites. (From Nature 397, Visscherand Camazine, 1999, © Nature Publishing Group.)


would appear that the external stimulus ofinformation about a superior site is not neededfor a scout to abandon a site. It should be notedthat the “before” scenario is supported byLindauer’s (1955) original observations. Eachof the 3 scouts that he reported on stopped pro-ducing dances for (hence evidently was losinginterest in) her first site well before she beganfollowing dances for her second site.

A recent study (Seeley, 2003) has add-ressed the critical issue of whether scoutsthat abandon a site do so only after they havefollowed dances for and made visits to a supe-rior site (as predicted by hypothesis 2), orwhether they will abandon a site even beforethey have experienced these external stimuli(as predicted by hypothesis 1). The behaviorsof the first few (4–8) bees that performeddances in each of 6 swarms were monitoredfrom the start of each bee’s dancing to the endof her swarm’s decision making. Because justa few scouts were monitored in each swarm, itwas possible to record each time a focal scoutleft the swarm, returned to the swarm, pro-duced a waggle dance, or followed a waggledance. A total of 37 scout bees were observed,and 33 of them abandoned (ceased dancingfor) the sites that they originally advertisedwith their dances. The critical finding was thatonly one of these 33 abandoners ceased danc-ing for her initial site after she began follow-ing dances for other sites. Most (32 out of 33)of these scouts lost interest in their sites with-out being influenced by the dances of otherscouts, a fact which tends to contradicthypothesis 2. Instead, it looks like these beeslost on their own (without external influence)their motivation to dance for their initial sites.A further piece of evidence that supports theinternal stimulus hypothesis is the finding thatthe number of waggle runs produced by eachbee for her initial site declined noticeably overher consecutive returns to the swarm and, as isshown in Figure 4, the pattern of decline indancing per return to the swarm is strikinglylinear. This linearity suggests that the patternarises from an internal, neurophysiologicalprocess that automatically drives down ascout’s motivation to dance for a site. Thisdecline of dancing over repeated visits to anest site appears to be a special feature ofdancing in the context of house hunting; beesdancing in the context of nectar foraging do

not show a decline in dancing over repeatedvisits to a food source (Seeley, 1994, Visscher,2003).

Another important finding in the Seeley(2003) study is that the scout bees showedsigns of adjusting their per capita rates ofabandonment in relation to the quality of thesite that each bee was advertising. Bees thatperformed dances for their swarm’s chosensite (presumably a superior site) producedstrong initial dances (mean number of waggleruns: 86.0) and continued producing dancesover many (5–6) returns to the swarm. In com-parison, bees that performed dances for one oftheir swarm’s nonchosen sites (presumablysites inferior to the chosen one) producedweaker initial dances (mean number of waggleruns: 34.8) and danced over fewer (1–5)

Figure 4. The number of waggle runs danced for agiven site over consecutive returns to the swarm.Each of 37 scouts made 1–6 consecutive returns tothe swarm with dancing and a final return withoutdancing (= 0 waggle runs/return). Some producedsuch a series for more than one site, so 51 series areincluded. The results show that a scout’s tendencyto dance for a site declines quickly and linearlyover the course of her successive trips to the site.(From Behav. Ecol. Sociobiol. 53, Seeley, 2003,© Springer-Verlag.)


returns to the swarm. On average, therefore,nest-site scouts from a chosen site startedstronger and lasted longer in advertising theirsites than did scouts from a nonchosen site.

It appears that scout bees have a behavioral“rule” regarding the advertising of nest sites.The rule seems to be as follows: uponreturning to the swarm for the first time aftervisiting a possible nest site, produce a dancewhose strength (number of waggle runs)reflects the quality of the site, and uponmaking subsequent returns to the swarm,produce dances whose strength declines byabout 15 waggle runs per return (see Fig. 4).By following this rule, nest-site scouts fromsuperior sites will advertise their sites morestrongly and will abandon their sites moreslowly than will scouts from inferior sites. Asa result, the better the site, the higher the percapita rate of recruitment of additional scouts(ri) and the lower the per capita rate ofabandonment of existing scouts (ai). Asmentioned above, such tuning of the per capitarates of recruitment and abandonment inrelation to site quality is sufficient to explainhow a consensus among the dancing bees isproduced.

We end this section with a bit of specula-tion. Winsor’s competition equations (1 and 2)refer only to competition by exploitation. Dothe scouts of different coalitions competeentirely by mutual exploitation of the pool ofuncommitted scout bees, or do they also com-pete by interference, that is by directly disrupt-ing the growth of competing coalitions? Theevidence presented above (the linear decay indancing) suggests that scout bees are not stop-ping their dancing in response to external stim-uli, but this may not be the whole story.Worker bees can produce a mechanical signal,the “stop signal”, that is directed at waggledancers and that causes them to stop dancing(Esch, 1964; Nieh, 1993; Kirchner, 1993). Sofar, the only context in which stop signals havebeen studied is in the hive, when a colony’snectar influx is excessive. In this situation, thenectar foragers have difficulty finding receiverbees to take their nectar, and these foragersproduce stop signals (usually while trembledancing) to inhibit waggle dancing and therecruitment of additional foragers (reviewed inSeeley, 1995, pp. 162–173). However, in thecourse of our studies, we have heard bees

producing stop signals in swarms. (Stop sig-nals are easily recognized because they areextremely brief, lasting just 0.1–0.2 s, in con-trast to all the other acoustical signals ofworker bees which last much longer.) Thisraises the possibility that the scouts in one coa-lition could reduce the rate of recruitment ofscouts in competing coalitions. And if scoutsin different coalitions do interfere with oneanother, it may happen when a scout hasswitched her commitment from an inferior siteto a superior site. A scout in this situationcould contribute to the decision-making proc-ess by inhibiting dances representing herformer, inferior site. It is known that somescouts do cross over from inferior to superiorsites (Lindauer, 1955; Seeley and Buhrman,1999; Camazine et al., 1999; Visscher andCamazine, 1999). However, no one has lookedto see if crossover scouts produce stop signalsdirected at dancers advertising inferior sites.Visscher and Camazine (1999) have testedwhether the removal of crossover scouts slowsthe decision-making process, as one wouldexpect if interference by crossover scouts wereimportant. They found no slowing. However,in performing their removal experiment,Visscher and Camazine used two identicalnest boxes. Hence their crossover scouts didnot experience a marked difference in qualitybetween the two sites. It may be that directexperience with two sites that differ markedlyin quality is necessary to induce a scout thathas switched sites to inhibit dances for theinferior site. Clearly, the question of whetherthe decision-making process of honey beeswarms involves mechanisms of direct inhibi-tion among the coalitions of dancers remains asubject for future investigation.

5. EXPLORING HOW SWARMS PREPARE FOR LIFTOFF

While the scouts are busy choosing a suita-ble nesting cavity, the other 95% of the bees ina swarm remain quiescent and conserve theswarm’s energy reserve: the 30–40 mg ofconcentrated sugar solution carried insideeach bee (Combs, 1972). As was shown byHeinrich (1981), the bees in the core of aswarm cluster maintain a 30–40 °C microcli-mate, by trapping the metabolic heat produced


by the resting bees and by adjusting the clus-ter’s porosity to control its rate of heat loss.Meanwhile, the outermost bees in a clustermaintain themselves above a relatively lowset-point of 15 °C, thus minimizing theirenergy expenditure for heat production but atthe same time keeping their flight muscleswarm enough to generate heat by shivering(Esch, 1988). In the final 10–30 min beforeliftoff, however, the temperature gradient in aswarm cluster becomes abolished such that thebees in the cluster’s mantle become as warmas those at its core (Heinrich, 1981; Seeley andTautz, 2001). Given that a worker bee needs aflight muscle temperature of at least 33–35 °Cfor rapid flight (Esch, 1976; Heinrich, 1979), itis not surprising that such a striking warm-upphenomenon is observed. Using an infraredcamera to measure the thoracic (flight muscle)temperatures of individual bees on the surfaceof a swarm cluster, Seeley et al. (2003) foundthat liftoff begins just a few seconds after allthe surface-layer bees in a swarm cluster havetheir flight muscles warmed to at least 35 °C.Swarm liftoff is a dazzling display of coordi-nated group behavior. All the bees in theswarm cluster launch into flight in about 60 s,form a cloud of swirling bees, and begin mov-ing off together, with the scouts somehowguiding all the others to their new dwellingplace (Seeley et al., 1979, reviewed by Dyer,2000).

We now understand some of the mecha-nisms whereby the bees in a swarm achievesuch a beautifully coordinated liftoff. One ofthese is the buzz running (Schwirrlaufen) sig-nal. During the final 10 or so minutes beforeliftoff, excited bees force their way throughthe quiet bees in the cluster, running about in azig-zag pattern, butting into the other bees,meanwhile buzzing their wings (Lindauer,1955; Esch, 1967). Many, perhaps most, ofthese buzz runners are scout bees and theiractions appear to loosen up the cluster (Seeleyet al., 1979). Martin (1963) demonstrated,with a split-hive experiment, that only beescontacted by buzz runners will join the massexodus when a swarm initially leaves theparental hive. A second signal that helps acti-vate the quiescent bees in a swarm cluster isthe shaking signal (also called the vibrationsignal). To produce this signal, one bee graspsanother and shakes this bee’s body for 1–2 s at

16–18 Hz (see Fig. 1 in Seeley et al., 1998).There is strong evidence that the shaking sig-nal acts as a modulatory signal that produces ageneral activation of worker bees in swarms(Schneider et al., 1998; Visscher et al., 1999;Lewis and Schneider, 2000; Donahoe et al.,2003; reviewed in Schneider and Lewis, 2004)and in hives (Schneider et al., 1986; Nieh,1998; Seeley et al., 1998). However becausethe shaking signal does not occur solely oreven principally in the last hour before liftoff,it seems that the shaking signal is not the sig-nal that informs the relatively cool bees in aswarm that it is time to warm up for liftoff.

The principal warm-up signal is evidentlythe high-pitched piping signal that is producedduring the last hour or so before departure(Lindauer, 1955; Seeley et al., 1979;Camazine et al., 1999). Seeley and Tautz(2001) investigated the origins and effects ofthis signal. They found that each piper is anexcited bee that scrambles through the swarmcluster, pausing every second or so to press herthorax against another bee, pull her wingstogether over her abdomen, and activate herwing muscles to produce an audible vibration(though probably most of the vibrationalenergy is loaded into the contacted bee). Eachpipe is a pulse of vibration which lasts 0.82 ±0.43 s and rises in fundamental frequencyfrom 100–200 Hz to 200–250 Hz. Many, ifnot all, of the pipers are scouts. The time-course of worker piping matches that of swarmwarming – both start at a low level about anhour before liftoff and both build to a climax atliftoff – which suggests that the scouts use thissignal to stimulate their swarm mates to warmup. The results of an exclusion experimentalso support this view of the function ofworker piping. When pipers were excludedfrom a subset of bees hanging in the cool, out-ermost layer of a swarm cluster, it was foundthat these bees did not warm up, unlike adja-cent bees that were contacted by pipers. Itshould be noted that the form of worker pipingthat occurs in swarms (“wings-together pip-ing”) is distinct from the form of worker pip-ing that has been observed in hives (“wings-apart piping”) (Ohtani and Kamada, 1980;Pratt et al., 1996). The function of the wings-apart piping signal remains a challengingmystery.


How do the scout bees sense when it isappropriate for them to begin producing thepiping signals and so start stimulating theirswarm-mates to prepare for the flight to theirnew home? Seeley and Visscher (2003) testedtwo hypotheses: “consensus sensing”, thescouts noting when all the bees performingwaggle dances are advertising just one site;and “quorum sensing”, the scouts noting whenone site is being visited by a sufficiently largenumber of scouts. To test these hypotheses,they monitored four swarms as they discov-

ered, recruited to, and chose between two nestboxes and as their scouts started producingpiping signals. They found that a consensusamong the dancers was neither necessary norsufficient for the start of worker piping, whichindicates that the consensus-sensing hypothe-sis is false. They also found that a buildup of10–15 or more bees at one of the nest boxeswas consistently associated with the start ofworker piping, which indicates that the quo-rum-sensing hypothesis may be true (Fig. 5).Recently, they tested experimentally the

Figure 5. Results from monitoring a swarm as its scout bees chose between two potential nest sites (one anest box and the other a site in an old house) and then began preparing for liftoff, which is indicated by thestart of worker piping. At the swarm, we recorded the intensity of worker piping every 5 min and the numberof waggle runs that were produced for each site in 20 min intervals before liftoff. At the potential nest sites,we recorded the number of scout bees outside and inside each site every 5 min, except when one of the siteswas an unidentified location in an old house. Note how every time the number of scouts inside the nest boxrose well above 10 bees, piping began to be heard at the swarm, and how when the count fell to 10 or fewerbees, the piping decreased, usually stopping altogether. Note too that there was a consensus among thedancers throughout. (From Behav. Ecol. Sociobiol. 54, Seeley and Visscher, 2003, © Springer-Verlag.)


quorum-sensing hypothesis by seeing if theycould delay the start of piping by slowing thebuildup of scouts (i.e., quorum formation) atthe chosen site but without disturbing anythingelse about the nest-site selection process(Seeley and Visscher, unpublished results). Todo this, they presented swarms, one at a time,with five nest boxes together at one site, andthey monitored the buildup of scouts at thenest boxes and the start of piping at the swarm.For comparison, they performed another trialwith each swarm but with just one nest box ata site (different sites were used for the 1-nest-box and the 5-nest-box trials; also, the order ofthe 1- and 5-nest-box trials was reversedbetween consecutive swarms). They observedthat the buildup of scout bees at a nest box wasslower in the 5-nest-box trials (the recruitsspread themselves among the nest boxes) thanin the 1-nest-box trials (the recruits assembledat the single nest box), and that the onset ofpiping was much delayed in the 5-nest-boxtrials relative to the 1-nest-box trials. Theseresults strongly support the quorum-sensinghypothesis, but leave open the question of howthe scouts are sensing a quorum at the chosensite.

An intriguing question is why the scoutsdon’t use consensus sensing and instead usequorum sensing. This question is particularlypiquant given that a consensus, or at least anear consensus, among the dancers is requiredfor a swarm to execute a successful move to anew home site. In three reported instances of aswarm lifting off when its dancers werestrongly split between two sites (the Balconyand Moosach swarms of Lindauer, 1955, andswarm 3 of Seeley and Visscher, 2003), theairborne swarm divided, stalled in its move,and resettled. (Note: the occasional occurrenceof liftoffs without consensus is further supportfor the quorum-sensing hypothesis.) Two ofthese swarms went on to achieve a danceconsensus and a successful move, but one lostits queen when it split itself in midair and soexperienced a complete failure. One possiblereason for why the bees use quorum sensingis that it would be exceedingly difficult orcostly for the scouts to sense a consensusamong themselves as they perform dances ona swarm, for to do so would presumablyrequire that each scout poll the dances bytraveling over the swarm cluster, reading some

sample of the dances, and keeping a tally ofher readings. Another possible reason forusing quorum sensing is that it provides anadvantage in the speed of the decision making,by enabling the bees to begin preparations forliftoff as soon as enough of the scout bees, butnot all of them, have approved of one of thepotential nest sites. Presumably the reason thatmost liftoffs occur when there is a consensusamong the dancers, even though a consensusis not the trigger for the start of liftoffpreparations, is because there is the strongpositive feedback process of vigorous dancingfor the chosen site, which attracts more andmore dancing for this superior site. It may betoo that the quorum size has been tuned bynatural selection to be sufficiently high toensure that almost always a dance consensus isproduced for a site shortly after a quorum ofscouts is reached at this site.

6. SWARM SMARTS

In closing, we draw attention to two fea-tures of the group decision-making processthat we have just discussed, features that con-tribute importantly to a swarm’s success inchoosing a new home. Let us not forget thateach of these successes is a remarkableachievement. For while it is true that aswarm’s decision making is simplified by hav-ing a clear and stable problem (i.e., find a sin-gle, suitable nest site), it is also complicated bythe realities that the information used in thisdecision making is incomplete (at least earlyon), sometimes inaccurate (as when a scoutdances weakly for a high-quality site; seeSwarm 4 in Fig. 2), and steadily changing.Such informational messiness makes it diffi-cult to apply the classical and powerfulapproach to decision making – identify alter-natives, evaluate these options, and choose theone with the highest value – but swarms man-age to do so. What are their secrets of success?

One is the use of dozens, if not hundreds, ofscout bees that independently, widely, andsimultaneously explore for potential nest sites.These scouts bring back to the swarm clusterheterogeneous information – knowledge ofsuperb, mediocre, and even lousy sites – whichis then shared with the other scouts by meansof waggle dances. The important thing is that


all discoveries of acceptable nest sites arefreely reported; no scout is stifled. Thus aswarm takes advantage of its collective natureto assemble rather quickly, often in just a fewhours (see Fig. 1), a large set of alternativesfrom which to choose. The larger this set, themore likely it includes a first-rate site.

A second important feature of the bee’sdecision-making process becomes apparentwhen we consider how the scouts solve theproblem of finding the best of this large set ofalternatives. At most, any individual has directexperience with only a few of these sites, butthe decision-making process must compare allthe sites. We have described already how thesolution to this problem arises through a proc-ess of friendly competition among the scouts,with the various coalitions of scouts commit-ted to different sites vying to attract uncom-mitted scouts. We have explained how themembers of each coalition attract additionalmembers by performing waggle dances thatare graded in strength in relation to site qual-ity, so that the higher the site quality, thestronger the dance, and the greater the streamof newcomers to it. Furthermore, the appar-ently endogenously programmed extinction ofTanzlust helps sharpen the differences inrecruitment. But what we have not yet pointedout is that the bees have an important check onthe positive feedback that will occur as danc-ers beget more dancers: a dance follower doesnot blindly imitate the dancer that she has fol-lowed. Instead, she leaves the swarm cluster,examines the advertised site, and only if shetoo judges that it is a worthy site does she pro-duce a dance for it. In this way the populationof scouts avoids possible runaway feedback indancing for a poor site that happens to be dis-covered quickly. The procedure of inspectinga site before advertising it also provides thebasis for the scouts deciding when the deci-sion-making process has progressed suffi-ciently far for them to shift to stimulatingswarm liftoff.

These considerations illustrate how thestory of house hunting by honey bees containsvaluable lessons about effective decision mak-ing by groups. There is no need for any indi-vidual to possess a global view of the alterna-tives, nor for any mechanism to tally andcompare “votes” for them. The “smarts” of aswarm derives from a combination of many

individuals working in parallel, each onemaking sophisticated assessments of nest-siteproperties, and a group process of feedbackin recruitment, modulated by these qualityassessments and amplified by the house-hunt-ing-specific decay of Tanzlust. Together, theselead a swarm to an accurate, speedy, and uni-fied decision.

ACKNOWLEDGMENTS

We are grateful to the National Science Founda-tion, the Alexander von Humboldt Foundation, theNational Geographic Society, and the University ofCalifornia-Riverside Academic Senate. Their finan-cial support made possible many of the studies dis-cussed in this article.

Résumé – Prise collective de décision dans lechoix du site de nidification chez l’Abeilledomestique. Une attention renouvelée a récem-ment été portée aux mécanismes de prise de déci-sion collective chez les essaims d’abeilles domesti-ques (Apis mellifera L.) en se basant sur les étudesfaites par Lindauer dans les années 50. Nous pas-sons en revue les connaissances récentes sur lafonction d’un essaim dans la prise de décision.Une première étape dans l’étude renouvelée du pro-cessus de choix du site de nidification a été de faireune description complète, à l’aide d’enregistre-ments vidéo, des danses des éclaireuses, et donc deleurs délibérations. Ceci a confirmé les traits fonda-mentaux du processus de prise de décision : leséclaireuses cherchent partout des sites potentiels denidification, elles recrutent d’abord par les dansesfrétillantes pour de nombreux sites, puis ne recru-tent éventuellement que pour un seul site et, peuaprès l’apparition d’un consensus parmi les danseu-ses, l’essaim s’envole vers le site choisi. Ce travaila aussi montré le taux élevé de renouvellement desdanseuses au cours de la prise de décision del’essaim. Les nouvelles études descriptives ontgénéralement confirmé que le choix d’un domicilepar l’essaim consiste en une compétition amicalepour gagner de nouvelles danseuses entre groupesde danseuses représentant les divers sites potentiels.Un test a été réalisé sur la précision de la prise dedécision par les essaims. Il a montré que les essaimsréussissent généralement à présenter un certainnombre N d’alternatives et choisissent ensuite lameilleure (i.e. ils utilisent la règle de décision « lemeilleur de N »).De nombreux travaux ont cherché à comprendrecomment les éclaireuses se comportent pour pro-duire finalement une danse consensuelle pour lemeilleur site de nidification. En principe cela seproduit si les éclaireuses agissent de façon à ce quele meilleur site ait le plus fort taux de recrutement et


le plus faible taux d’abandon par tête. On sait main-tenant que les éclaireuses qui produisent la danserèglent la force de leur danse frétillante ( = nombrede trajets frétillants par danse) selon la qualité dusite et que celles qui suivent la danse choisissent auhasard la danse à suivre. Le taux de recrutement partête est une fonction positive de la qualité du site denidification. On a aussi maintenant la preuve que leséclaireuses ajustent le taux d’abandon par tête enfonction de la qualité du site. Bien que les éclaireu-ses puissent éventuellement abandonner un site(cesser de danser pour lui), meilleur est le site, pluslongtemps l’éclaireuse dansera pour lui.Peu avant qu’un essaim s’envole pour son nouveaudomicile, les abeilles échauffent leurs muscles devol jusqu’à atteindre au moins 35 °C. Les éclaireu-ses produisent un chant en rassemblant leurs ailespour inciter leurs consoeurs au repos à s’échaufferpour l’envol. Les éclaireuses décident curieusementdu moment où elles se mettent à chanter lorsqu’ellessentent la formation d’un quorum d’éclaireusespour le site choisi et non pas un consensus dans ladanse sur la grappe de l’essaim. Ce sentiment dequorum peut servir à accélérer la prise de décisionde l’essaim.

Apis mellifera / décision collective / choix / site denidification / essaimage / danse frétillante

Zusammenfassung – Gruppenentscheidung beider Nestsuche der Honigbienen. In den letztenJahren wurde die Aufmerksamkeit erneut aufden Mechanismus der Gruppenentscheidung beiSchwärmen der Honigbienen gerichtet, die auf denUntersuchungen von Lindauer in den 1950er Jahrenbasieren. Wir geben eine Übersicht über neuereErkenntnisse über die Funktion eines Schwarms beider Fassung von Entscheidungen.Ein erster Schritt bei der erneuten Untersuchungüber den Entscheidungsprozess während der Suchenach einer geeigneten Nisthöhle bestand aus VideoAufnahmen, die eine vollständige Beschreibung derTänze aller Kundschafterbienen und damit vonihren Beurteilungen ermöglichte. Hierbei bestätig-ten sich die wichtigsten Grundzüge des Ent-scheidungsprozesses: Kundschafter suchen überallnach Nistmöglichkeiten, anfangs werben sie fürviele Nistgelegenheiten mit Schwänzeltänzen, aberallmählich werben sie nur noch für einen Nistplatzund kurz nach der Einigung der Tänzer fliegt derSchwarm zur erwählten Nisthöhle. Diese Filmedokumentieren auch eine hohe Zahl von Umgrup-pierungen der Tänzer während des Entscheidungs-prozesses des Schwarms. Insgesamt haben dieneuen Beobachtungen bestätigt, dass die Wahl desSchwarmes für eine neue Nisthöhle aus einer fried-lichen Konkurrenz bei der Anwerbung von Tänzernzwischen den verschiedenen Gruppen ist, dieNistmöglichkeiten vorstellen.Es wurde ein Test über die Genauigkeit der Ent-scheidungsfindung im Schwarm durchgeführt.

Dabei zeigte sich, dass Schwärme im allgemeinenerfolgreich eine Anzahl (N) von alternativenNistmöglichkeiten aufzeigen und dann den bestenauswählen (i.e., Anwendung der “besten–von-N”Entscheidungsregel).Viele Arbeiten konzentrierten sich auf eineAufklärung des Verhaltens der Kundschafter, daszum Schluss zu einem einheitlichen Tanz für diebeste Nistgelegenheit führt. Im Prinzip geschiehtdas, wenn es den Kundschaftern gelingt, die meis-ten Einzelbienen (pro Kopf) zu rekrutieren undgleichzeitig den geringsten Verlust zu haben.Inzwischen ist bekannt, dass die tanzenden Kund-schafterinnen die Stärke ihres Schwänzeltanzes mitder Qualität des Nistplatzes abstimmen (=Anzahlvon Schwänzelläufen pro Tanz), und dass dieder Tänzerin nachfolgenden Kundschafterinnenzufällig einem Tanz folgen. Eine pro Kopf Rate vonNeulingen hat eine einigende Wirkung und zeigteine positive Reaktion auf der Qualität des Nistplat-zes an. Auch gibt es jetzt Befunde, dass Kundschaf-terinnen die pro Kopf Rate von den Tanz verlassen-den Bienen in Bezug zur Qualität setzen. Obwohlalle Kundschafterinnen nach und nach aufhören(nicht mehr für den Platz tanzen), tanzen sie dochlänger, je besser der Platz ist.Kurz bevor der Schwarm startet und zu seinemneuen Heim fliegt, wärmen die Bienen ihre Flug-muskeln auf mindestens 35 °C auf. Die Kundschaf-terinnen erzeugen ein Piping Signal durch Zusam-menlegung ihrer Flügel, das zur Stimulierungihrer ruhig gebliebenen Schwarmgenossinnen zurAufwärmung für den Abflug führt. Seltsamerweiseentscheiden die Kundschafterinnen über den Beginndes Piping Signals. Sie erkennen die Entscheidungeiner ausreichenden Zahl an Kundschafterinnen füreinen Platz, sie messen nicht die Einigung im Tanzauf der Traube. Dieses Gefühl für ein Quorum magdazu dienen, die Entscheidungsfindung zu be-schleunigen.

Apis mellifera / Findung einer Gruppenent-scheidung / Wahl des Nistplatzes / Schwärmen /Schwänzeltanz

REFERENCES

Bourke A., Franks N.R. (1995) Social evolution inants, Princeton University Press, Princeton, NewJersey.

Camazine S., Visscher P.K., Finley J., Vetter R.S.(1999) House-hunting by honey bee swarms:collective decisions and individual behaviors,Insectes Soc. 46, 348–360.

Combs G.F. (1972) The engorgement of swarmingworker honeybees, J. Apic. Res. 11, 121–128.

Conradt L., Roper T.J. (2003) Group decision-makingin animals, Nature 421, 155–158.

Donahoe K., Lewis L.A., Schneider S.S. (2003) Therole of the vibration signal in the house-hunting


process of honey bee (Apis mellifera) swarms,Behav. Ecol. Sociobiol. 54, 593–600.

Dyer F.C. (2000) Group movement and individualcognition: lessons from social insects, in: BoinskiS., Garber P.A. (Eds.), On the move: how andwhy animals travel in groups, University ofChicago Press, Chicago, pp. 127–164.

Esch H. (1964) Beiträge zum Problem der Entfer-nungsweisung in den Schwänzeltänzen derHonigbiene, Z. Vergl. Physiol. 48, 534–546.

Esch H. (1967) The sounds produced by swarminghoney bees, Z. Vergl. Physiol. 56, 408–411.

Esch H. (1976) Body temperature and flight perform-ance of honey bees in a servomechanically con-trolled wind tunnel, J. Comp. Physiol. 109, 264–277.

Esch H. (1988) The effects of temperature on flightmuscle potentials in honeybees and cuculiinidwinter moths, J. Exp. Biol. 135, 109–117.

Franks N.R., Dornhaus A. (2003) How might individ-ual honeybees measure massive volumes, Proc.R. Soc. London B (Suppl.) 270, S181–S182.

Franks N.R., Pratt S.C., Mallon E.A.B., Britton N.F.,Sumpter D.J.T. (2002) Information flow, opinionpolling and collective intelligence in house-hunting social insects, Philos. Trans. R. Soc.London B 337, 1567–1583.

Frisch K. von (1946) Die Tänze der Bienen, Österr.Zool. Z. 1, 1–48.

Gould J.L., Gould C.G. (1994) The animal mind,Scientific American Library, New York.

Griffin D.R. (1981) The question of animal aware-ness, Rockefeller University Press, New York.

Griffin D.R. (1992) Animal minds, University ofChicago Press, Chicago.

Griffin D.R. (2001) Animal minds: beyond cognitionto consciousness, University of Chicago Press,Chicago.

Heinrich B. (1979) Thermoregulation of African andEuropean honeybees during foraging, attack, andhive exits and returns, J. Exp. Biol. 80, 217–229.

Heinrich B. (1981) The mechanisms and energetics ofhoneybee swarm temperature regulation, J. Exp.Biol. 91, 25–55.

Hutchinson G.E. (1978) An introduction to popula-tion ecology, Yale University Press, New Haven.

Kirchner W.H. (1993) Vibrational signals in thetremble dance of the honeybee, Apis mellifera,Behav. Ecol. Sociobiol. 33, 169–172.

Lewis L.A., Schneider S.S. (2000) The modulation ofworker behavior by the vibration signal duringhouse hunting in swarms of the honeybee, Apismellifera, Behav. Ecol. Sociobiol. 48, 154–164.

Lindauer M. (1951) Bienentänze in der Schwarm-traube, Naturwissenschaften 38, 509–513.

Lindauer M. (1953) Bienentänze in der Schwarm-traube. II, Naturwissenschaften 40, 379–385.

Lindauer M. (1955) Schwarmbienen auf Wohnungs-suche, Z. Vergl. Physiol. 37, 263–324.

Lindauer M. (1971) Communication among socialbees, 2nd ed., Harvard University Press, Cam-bridge, Massachusetts.

Lindauer M. (1975) Verständigung im Bienenstaat,Fischer Verlag, Stuttgart.

Markl H. (1985) Manipulation, modulation, informa-tion, cognition: some riddles of communication,in: Hölldobler B., Lindauer M. (Eds.), Experi-mental behavioral ecology and sociobiology,Fischer Verlag, Stuttgart, pp. 163–194.

Martin P. (1963) Die Steuerung der Volksteilungbeim Schwärmen der Bienen. Zugleich einBeitrag zum Problem der Wanderschwärme,Insectes Soc. 10, 13–42.

Nieh J.C. (1993) The stop signal of honey bees:reconsidering its message, Behav. Ecol. Socio-biol. 33, 51–56.

Nieh J.C. (1998) The honey bee shaking signal: func-tion and design of a modulatory communicationsignal, Behav. Ecol. Sociobiol. 42, 23–36.

Ohtani T., Kamada T. (1980) ‘Worker piping’: thepiping sounds produced by laying and guardingworker honeybees, J. Apic. Res. 19, 154–163.

Pratt S.C., Kühnholz S., Seeley T.D., WeidenmullerA. (1996) Worker piping associated with foragingin undisturbed queenright colonies of honey bees,Apidologie 27, 13–20.

Schmidt J.O. (1990) Swarm traps: an example ofresearch and technology transfer, Am. Bee J. 130,333–334.

Schmidt J.O., Thoenes S.C. (1987) Swarm traps forsurvey and control of Africanized honey bees,Bull. Entomol. Soc. Am. 33, 155–158.

Schmidt J.O., Thoenes S.C. (1990) Honey bee(Hymenoptera: Apidae) preferences among artifi-cial nest cavities, Ann. Entomol. Soc. Am. 83,271–274.

Schneider S.S., Lewis L.A. (2004) The vibrational sig-nal, modulatory communication and the organiza-tion of labor in honey bees, Apis mellifera, Apid-ologie 35, 117–131.

Schneider S.S., Stamps J.A., Gary N.E. (1986) Thevibration dance of the honey bee. I. Communica-tion regulating foraging on two time scales,Anim. Behav. 34, 377–385.

Schneider S.S., Visscher P.K., Camazine S. (1998)Vibration signal behavior of waggle-dancers inswarms of the honey bee, Apis mellifera(Hymenoptera: Apidae), Ethology 104, 963–972.

Seeley T.D. (1977) Measurement of nest cavityvolume by the honeybee (Apis mellifera), Behav.Ecol. Sociobiol. 2, 201–227.

Seeley T.D. (1985) Honeybee ecology, PrincetonUniversity Press, Princeton, New Jersey.

Seeley T.D. (1994) Honey bee foragers as sensoryunits of their colonies, Behav. Ecol. Sociobiol.34, 51–62.

Seeley T.D. (1995) The wisdom of the hive, HarvardUniversity Press, Cambridge, Massachusetts.

Seeley T.D. (2003) Consensus building during nest-site selection in honey bee swarms: the expiration


of dissent, Behav. Ecol. Sociobiol. 53, 417–424.

Seeley T.D., Buhrman S.C. (1999) Group decisionmaking in swarms of honey bees, Behav. Ecol.Sociobiol. 45, 19–31.

Seeley T.D., Buhrman S.C. (2001) Nest-site selectionin honey bees: how well do swarms implementthe ‘best-of-N’ decision rule? Behav. Ecol.Sociobiol. 49, 416–427.

Seeley T.D., Tautz J. (2001) Worker piping in honeybee swarms and its role in preparing for Liftoff, J.Comp. Physiol. A 187, 667–676.

Seeley T.D., Visscher P.K. (2003) Choosing a home:how the scouts in a honey bee swarm perceive thecompletion of their group decision making,Behav. Ecol. Sociobiol. 54, 511–520.

Seeley T.D., Morse R.A., Visscher P.K. (1979) Thenatural history of the flight of honey bee Swarms,Psyche 86, 103–113.

Seeley T.D., Weidenmüller A., Kühnholz S. (1998)The shaking signal of the honey bee informsworkers to prepare for greater activity, Ethology104, 10–26.

Seeley T.D., Kleinhenz M., Bujok B., Tautz J. (2003)Thorough warm-up before take-off in honey beeswarms, Naturwissenschaften 90, 256–260.

Sober E., Wilson D.S. (1998) Unto others, HarvardUniversity Press, Cambridge, Massachusetts.

Visscher P.K. (2003) How self-organization evolves,Nature 421, 799–800.

Visscher P.K., Camazine S. (1999) Collectivedecisions and cognition in bees, Nature 397,400.

Visscher P.K., Shepardson J., Camazine S. (1999)Vibration signal modulates the behavior ofhouse-hunting honey bees (Apis mellifera),Ethology 105, 759–769.

Wilson E.O. (1971) The insect societies, HarvardUniversity Press, Cambridge, Massachusetts.

Winsor C.P. (1934) Mathematical analysis of growthof mixed populations, Cold Spring Harbor Symp.Quant. Biol. 2, 181–187.

Witherell P.C. (1985) A review of the scientificliterature relating to honey bee bait hivesand swarm attractants, Am. Bee J. 125, 823–829.

To access this journal online:www.edpsciences.org

Group decision making in nest-site selection by honey bees

Documents

Transcript of Group decision making in nest-site selection by honey bees