HYM. RES. J. Vol. 19(1), 2010,pp. 66-76 Among Social Behaviours in Solitary Bees: Interactions Individuals in Xeralictus bicuspidariae Snelling (Hymenoptera: Halictidae: Rophitinae) Miriam H. Richards* and Laurence Packer (MHR) Department of Biological Sciences, Brock University, 500 Glenridge Ave., St. Catharines, ON, Canada L2S 3A1 (LP) Department of Biology, York University, 4700 Keele St., Toronto, ON, Canada M3J 1P3 — Abstract. Understanding behavioural interactions among ancestrally solitary species is key to understanding the evolutionary origins ofgroup living and cooperation. Previously, Packer (2006) showed thatcircletube arenas canbe used toevaluate the social status ofspeciesforwhichnesting data are unavailable. We used circle tube arenas to study thebehaviour among30 female dyads of the solitary halictidbee, Xeralictus bicuspidariaeSnelling, a member ofthe subfamily Rophitinae, all members of which are ancestrally solitary. Overall, 75.2% of frontal encounters resulted in avoidance, 20.7% in aggression, and 4.1% in a successful pass, values which are similar to those previously observed in solitary halictids. Although passing events, which are interpreted as cooperative behaviour, were rare, they were significantly correlated with bees' rates of approach and avoidance, and also with differencesbetween dyad members inrates ofovarian development. Ratesofaggressionwerenotcorrelatedwithphysicaltraitsoffemalesorwithotherbehaviours. We compare the circle tube behaviour of X. bicuspidariae to previously studied solitary and social halictids, and provide statistical support for this method of assessing social status. The origin of eusociality is one of the eusociality stems from our poor knowl- major events in the evolutionary history of edge of solitary species, whose behaviour life (Szathmary and Maynard Smith 1995), is most likely to represent the ancestral yet our understanding of w^hat transpires forms from which sociality evolved. during transitions to sociality remains Sweat bees (Halictidae: Halictini and poor. One reason is the great age at which Augochlorini) are the most socially vari- most solitary to eusocial transitions took able group of animals on earth, including place - over 100 million years ago for species that run the gamut from obligately termites, ants, and vespid w^asps (Wenzel solitary to obligately social, with sociality 1990; Martinez-Delclos and Martinell varying from communal to semisocial and 1995), perhaps somewhat less in the bees eusocial forms (Schwarz et al. 2007). There (Michener and Grimaldi 1988), and around are even examples of intraspecific social 20 million years ago in the three main polymorphism, in which solitary or social lineages of eusocial Halictinae (Brady et al. behaviour is expressed within or among 2006). The great age of these social lineages populations, often in response to variabil- means that to investigate the evolutionary ity in environmental conditions (Schwarz origins of sociality, we must often use et al. 2007). The ecological processes that comparative methods based on detailed shape the social behaviour of modern knowledge of the behaviour of extant halictines are often considered to be analo- species. However, a second reason for our gous to those that shaped the evolution of incomplete understanding of the origins of major social transitions in the subfamily as a whole, including at least three origins of •Author forcorrespondence: [email protected] eusociality and multiple reversions to Volume 19, Number 1, 2010 67 solitary behaviour (Danforth et al. 2003; forced to interact (Breed et al. 1978), is one Danforth et al. 2008). route to obtain sufficient behavioural data Recent behavioural studies provide in- on interactions among individuals of soli- triguing evidence that behavioural transi- tary species (Packer 2006). tions from solitary to socialbehaviour may In this paper we analyse the results of a occur easily and rapidly. Jeanson et al. detailed study of interactions among in- (2005) observed that in forced associations dividuals in a solitary species of the bee of solitary L. (Chilalictus) NDA-1 or of family Halictidae. Xeralictus bicuspidariae communal L. hemichalceum, both domi- Snelling is a member of the halictid nance interactions and division of labour subfamily Rophitinae with several advan- arose as natural outcomes of normal tages as a study organism. First, the solitary behavioural patterns expressed in phylogenetic position of rophitine bees the context of novel, social environments, suggests that their solitary behaviour is theeffectbeingstrongerinthesolitarythan ancestral,i.e. thereisnoevidencethatthere in the communal species. A similar phe- has been any sociality in the evolutionary nomenon occurred in experiments on history ofthe entire subfamily (Danforth et solitary Ceratina carpenter bees (Apidae, al. 2008; Patiny etal. 2008), so anypotential Xylocopinae) when females were forced to for social interactions that might be in- nest in social associations (Sakagami and duced experimentally, is part of its solitary Maeta 1977). This suggests that at the very ground plan. Second, this bee exhibits beginrung of evolutionary transitions to considerable variation in colour of the caste-based sociality, 'emergent' social be- metasoma of females (Snelling and Stage haviour by solitary bees could provide the 1995) such that pairs can easily be chosen behavioural substrate upon which natural to permit individual recognition without selection acts,before the evolution ofcaste- the intervention of artificially marking the based sociality. This fascinating possibility bees (markinghasbeen shown to influence makes understanding the behaviour of interactions among individuals; Packer obligately solitarybees all the more critical. 2005). Third, it is a large bee, facilitating Althoughbehaviouralinteractionsamong observations of behaviours. individuals of social and socially poly- METHODS morphic sweat bees have been studied in detail on numerous occasions (Breed et al. Xeralictus bicuspidariae was studied at 1978; Buckle 1984; McConnell-Garner and Dome RockRoad, La Paz County, Arizona, Kukuk 1997; Wcislo 1997; Soucy 2002), USA, in April 2005. Female bees were solitaryspecieshavereceivedless attention. collected from flowers of Mentzelia (Loasa- Consequently, we know little about the ceae) andretained inmicrocentrifuge tubes potential for social interactions among for no more than 30 minutes before individuals of solitarily nesting species, behaviouralobservationscommenced. This and how naturally occurring variation in duration between capture and observation individual behaviour might impinge upon was maintained to reduce the effect of the development of sociality remains ob- captivity-induced physiological changes scure. Of course, one problem with study- upon behaviour (Pabalan et al. 2000). Two ingthebehaviourofsoUtarybeesisadearth bees were then placed simultaneously in a of opportunities for observing known in- clean, plastic circle tube of internal diam- mm dividuals under natural conditions at suffi- eter 7 and length 20 cm. Simultaneous cient frequency to permit detailed analysis. entry precludes ownership effects (Wcislo Fortunately, a recent comparative study 1997), and this tube diameterwas sufficient suggests that the circle tube arena, a circle for the two individuals to pass one another of clear plastic tubing in which bees are and to turn around (Packer 2005), but 68 Journalof Hymenoptera Research: Festschrift Honoring Roy Snelling narrow enough that onebee could block an Reproductive status was estimatedbased attempted pass by the other. Observations upon dissection of the metasoma. The lasted for fifteen minutes, a time period spermathecawasinspected forthepresence sufficient for differences in behaviour of sperm and ovarian development was between individuals within a pair to be assessedbyestimatingthe size ofoocytesin detected, and took place outdoors in the each ovariole relative to the size of a fully shade. The metasomal colour of females developed oocyte, and summing the result- varied from entirely brick red to entirely ing proportions across all six ovarioles. As dark brown (Snelling and Stage 1995); expected for a solitary bee during nest pairs were set up with one red and one provisioning, all females had mated and dark female that could be easily differen- so matedness was not considered further. — tiated by the observer without being Statistical analysis. In circle tubes, the artificially marked. Several experiments behaviour of each member of a pair is were terminated when discrimination be- affected by the behaviour of the second tween the individuals was found to be member ofthe pair. This creates a problem more difficult than expected. of statistical non-independence between An approachwas takentohave occurred members of each dyad. A second problem when individuals came within a distance is variation in behavioural rates among of one body length of each other (Kukuk pairs-somepairs areveryactive and some 1992; Packer 2005). Both frontal (head to do almost nothing. A common approach head) and front-to-back (head to tail) has been to standardize focal behaviors by encounters were assessed and their out- the encounter rate, which in effect means comes classified into categories: ap- allthebehaviours are analyzed as ratios, so proaches, aggression, avoidance and information related to absolute frequency passes. Aggressive interactions included is lost and the statistical problems of nudges, lunges and C-postures. Interac- analysing ratios are gained. To address tions thatresulted in avoidance arosewhen these issues, we present an approach one individual moved away from a sta- somewhat different than in previous circle tionary individual or they both moved tube studies. First, when behavioural pat- away from each other. A pass was scored terns of individuals are considered, we when the two bees manoeuvred to permit analyse only one individual per dyad (red one to move past the other, or they both bees or dark bees), which avoids inflating moved past one another simultaneously. the number of degrees of freedom in each All behavioural observations were carried measurement. Second, when properties of out by LP, and are therefore directly dyads are considered, we analyse both comparable to the data presented in Packer behavioural frequencies and physical traits (2006). For more detailed descriptions of in terms of differencesbetween eachbee in individual behaviours, see Batra (1966). a pair. Correlations between trait differ- Allbees were measured and assessed for ences can be interpreted in the same way relative age and reproductive condition as as correlations between the traits them- follows. Head width was measured as the selves. For instance, a negative correlation greatest distance across the compound between head width difference and wear eyes; this was the greatest diameter of the difference would indicate that larger bees head in dorsal view. Relative wear was tended to be less worn. All differences assessed from mandibles, scored from between pair members were calculated as (unworn) to 6 (worn to the base of the (value for red bee) - (value for dark bee), subapical mandibular tooth), and from except for head width (HW) difference, HW wings (the total number of nicks along which was calculated as (red - dark the margin of the left forewing). HW) / average HW. Volume 19, Number 1, 2010 69 Table 1. Physical characteristics of adult female Xeralictus bicuspidariae used in circle tube experiments. Signedranktestswereusedtocomparethephysicalcharacteristicsofredanddarkfemalesineachdyad;non- significant (n.s.) results indicate thatoverall, red and darkfemales were equivalent. Variable Mean SD Range Signedranktest Head width (mm) (n=56) 7.4 0.24 7.0-8.0 S = -22.5, n.s. Mandibularwear (n=56) 2.1 1.2 0-5 S = 31.5,n.s. Wingwear (n=48) 8.0 5.1 0-20 S - 4.0, n.s. Total ovarian score (n=55) 1.7 0.4 0.5-2.6 S = 8.0,n.s. Allvariables, includingdifferences,were bees by low levels of cooperation and checked for normality using the array of high levels of both aggression and avoid- tests in SAS 9.1 (PROC UNIVARIATE); ance. We used discriminant functions since several variables were non-normally analysis (DFA) to assess how accurately distributed, we mainly used non-para- X. bicuspidariae and 21 other species (refer- metric statistical methods. Additionally ences in Packer 2006) canbe categorized as we used principal components analysis solitary, communal, or semi and eusocial, (PCA) to further explore and confirm based on the percentages of avoidance, relationships among physical and behav- aggression, and passing behaviours in ioural variables in X. bicuspidariae. Initially, circle tubes. the PCA was based on eight variables RESULTS (entered as untransformed differences be- — tween females in mandibular wear, wing Circle tube assays. Physical traits of the wear, head width, total ovarian score, 60 females used in 30 circle tube trials are approach frequency, aggressive frequency, presented in Table 1. There were no sig- avoidance frequency, and pass frequency). nificant correlations among body size, However, since Kaiser's Measure of Sam- degree of wear, and degree of ovarian pling Adequacy (MSA) with all eight development within individuals used in variables had a value of only 0.467, the the behavioural tests, although degree of variable with the lowest communality mandibular wear was positively correlated measure (head width) was dropped from with degree ofwingwear (Pearson correla- the PCA. With the remaining seven vari- tion coefficient, r=0.55, n=48, p<0.0001). ables, MSA=0.63, which exceeds the 0.6 All females had at least one V4-size, criterion. We present both factor loading developing oocyte, and 25 of 52 (48%) scores (the degree to which each variable dissected females contained a full-size influences the inferred factors) and com- oocyte, ready to lay. The mean difference munality estimates (a reliability score between red and dark females in each which estimates the proportion ofvariance dyad for each of these characteristics, in eachvariable that isjointly explainedby was zero (Table 1), so bee colouration had all three factors). no significance other than providing a Packer (2006) argued that the social convenient identification tool for the ob- status of halictine bees can be accurately server. assessed using circle tube assays of fe- The frequencies of each of the four males, even in the absence of nesting data. classes of behaviour per dyad and per Solitary bees should be characterized by female are given in Table 2. The most high levels of avoidance behaviour, com- frequent behaviours were approaches munal bees by high levels of cooperative (32.5 per dyad) and avoidance (25.1 per behaviour (passing) and low levels of dyad), followed by aggressive behaviours aggression, and semisocial and eusocial (6.7 per dyad) and passing (1.2 per dyad). 70 Journalof Hymenoptera Research: Festschrift Honoring Roy Snelling Table 2. Behavioural frequencies for approach, avoid, pass, and aggressive behaviours. Note that 'aggression' includes C-postures, biting, and pushing. Since the behavioural rates of each member of a dyad are non-independent, only one bee per dyad is used to provide an estimate of behavioural frequencies per indi\idual. Measurements of mutual behaviour refer to simultaneous performance of that behaviour by both members ofa dyad. Sample size N=30 dyads, except where otherw^ise noted. Behaviour Rate Mean SD Range Frontal encounters Dyad total 33.0 8.3 13-46 Approach Dyad total ' 32.5 9.1 11-52 Red female ' 4.6 4.8 0-22 Darkfemale 7.6 7.4 0-35 Mutual 20.3 10.6 0-44 Avoid Dyad total 25.1 7.5 10-37 Red female 9.3 4.3 3-19 Dark female 9.7 4.3 0-20 Mutual 6.1 4.0 0-13 Pass Dyad total 1.2 1.1 0-i Red female 0.3 0.5 0-2 Darkfemale 0.2 0.6 0-3 Mutual 0.7 0.9 0-3 Aggression Dyad total 6.7 5.0 1-19 Red female 3.8 3.8 0-16 Dark female 2.9 3.0 0-11 N=29 Overall, 75.2% of frontal encounters re- vs. approach: r=—0.224,n=29,n.s.; aggress sulted in avoidance, 20.7% in aggression, vs. pass: r=—0.106, n=30, n.s.; aggress vs. and 4.1% in a successful pass. avoid: r=-0.193, n=30, n.s.). Aggressive acts were observed in all 30 Differences betweenbees with respect to pairs, andby 52 ofthe 60 (87%) individuals head width, wing wear, and mandibular assayed. Withdrawals were also observed wear were notsignificantly correlated with in all 30 pairs; only 1 bee of 60 (2%) did not differences in behavioural frequency for display a unilateral withdrawal, but she any of the behaviours. Differences in total did take part in a mutual (bilateral) with- ovarian score did correlate positively with drawal. Passing or cooperative acts were the rates of approach and pass, although rare, being observed in only 22 of 30 (73%) not with either avoidance or aggressive pairs. Of a total of35 passes, 20 {57%) were frequencies (Table 3, Fig. 1). In other bilateral (both bees moved past each other) words, the female with greater ovarian and 15 (43%) were unilateral (1 bee moved development was almost significantly past the other bee). likely to approach and was significantly Based on behavioural frequency differ- more likely to pass than the female with ences (red bee - dark bee), three behav- lesser ovarian development. iours, approach, avoid, and pass were A principal components analysis (PCA) found to be mutually positively correlated further describes behavioural and physical (i.e. the bee that did one behaviour more variation among female interactants in frequently also did the other behaviour circle tubes. As outlined in the Methods, more frequently; Spearman rank correla- the PCA (Table 4) included all variables tions: approach vs. avoid, r=0.798, n=29, except head width, which contributed little p<0.0001, approach vs. pass: r=0.577, to understanding variation among the n=29, p=0.001; avoid vs. pass: r=0.460, dyads. Three factors had eigenvalues > n=30, p=0.010), but none was correlated 1.0 and were retained, explaining 77.9% of with the frequency of aggression (aggress the variation among dyads. Factor 1 was Volume 19, Number 1, 2010 71 Table 3. Influence of female physical status on behaviour. Spearman rank correlations were based on differences in both female traits and differences in behaviour frequencies (red bee - dark bee). Positive correlations indicate thatthebee withthe greater traitvalueexhibited thebehaviour more frequently. Behaviour(N=numberofdvads) Physicaltraitoffemales Approach Avoid Cooperate Aggression Head width 0.197(27) 0.018 (28) -0.054 (28) 0.292 (28) Wingwear 0.020 (20) -0.123 (21) -0.092 (21) 0.077 (21) Mandibularwear -0.007 (27) -0.100 (28) -0.068 (28) 0.215 (28) Total ovarianscore 0.451 (26),p=^0.062 0.250 (27) 0.435 (27), p=^0.023 -0.132 (27) most influenced by non-aggressive behav- bility in wear differences among dyads, iour and ovarian development, reflecting and so does not reflect behavioural varia- the previously noted positive association tion. Factor 3 was influenced mainly by between ovarian development and ap- aggression. PCA based only on the four proach and passing frequencies. Factor 2 behavioural frequency differences, re- was influenced mainly by mandibular and sulted in two factors that together ex- wing wear; thus Factor 2 describes varia- plained 79.7% of the variation among oaggress joavoid I | I 20 ' o 10 o o • o oO oO o °oo5 o a -^ -2 -1.5 -1 1.5 2 8 D -5- • o°.,o. o -15 -20- :approach IAcooperateI 3- 2i1 1 A4 A A A )5 ^ ><l -2 -1.5 -1 -0.5 1 0.5 1 1.5 2 xX-10- -1 A -2 -3 -4 Fig. 1. Influenceofovariandevelopment(OD,horizontalaxes) ondifferentbehaviours (verticalaxes),scored in terms ofthe differencesbetweenindividuals (red-dark). Positive values on the horizontal and vertical axes indicate a greatervaluefortheredbee,whereasnegativevalues indicate a greatervalue forthe darkbee. Top left:ODvs.aggression.Topright:ODvs. avoidance.Bottomleft:ODvs.approachfrequency. Bottomright: OD vs. cooperation (pass). 72 Journalof Hymenoptera Research: Festschrift Honoring Roy Snelling Table 4. Principal components analysis describing variation annong dyads based on differences between interactants in both physical and behavioural traits. Three factors were retained with eigenvalues > 1, explaining a cumulative total of 77.9% of the variation among dyads. Relatively strong factor loading scores (>0.6)areindicated inboldface.Communalityestimatesdescribetheproportionofvarianceineachtraitthatis jointly explained by Factors 1, 2 and 3. Kaiser's overall Measure ofSamplingAdequacy (MSA) was 0.6312. Trait(differencebetweenfemales) Factor1 Factor2 Factor3 Communalityestimate Mandibular wear 0.0124 0.9064 -0.0567 0.4962 Wing wear -0.1391 0.7738 -0.5193 0.4465 CKarian score 0.6217 0.3534 0.2441 0.7520 Approaches 0.9017 0.0752 -0.1593 0.6582 Avoidance 0.8311 -0.0742 -0.2158 0.7008 Pass 0.8116 -0.0438 0.2910 0.7450 Aggression -0.1848 0.5085 0.7388 0.5381 Eigenvalue 2.602 1.817 1.035 Variance explained 37.2% 26.0% 14.8% dyads. Factor 1, which explained 55.0% of caste-based social bees in the proportions the behavioural variation among dyads, of aggressive behaviour (ANOVA, was strongly influenced by approaches F=50.32, df=2,19, p<0.0001) and avoid- (loading score 0.910), avoidance (0.866), ance behaviour (ANOVA, F=15.15, and passes (0.752). Factor 2, which ex- df=2,19, p<0.0001), as well as significantly plained 24.6% of the variation, was more frequent passing behaviour in com- strongly influenced by aggression (loading munal species, as compared to both soli- score 0.959). tary and social species (ANOVA, F=62.55, Comparison ofX. bicuspidariae with other df=2,19, p<0.0001). — halictines. Fig. 2 compares X. bicuspidariae DISCUSSION to 21 other species, in terms of the propor- — tion ofavoidance, aggressive, and coopera- Solitary behaviour ofX. bicuspidariae. In tive (passing) behaviours observed in circle X. bicuspidariae, differences between circle tube assays. It most closely resembles tube interactants in head width, wing Penapis toroi, another solitary rophitine. wear, and mandibular wear were not Discriminant functions analysis (DFA) associated with differences in behaviour, based on four putative categories (solitary, suggesting that neither body size nor wear communal, semisocial, and eusocial) per- (and possibly age) structured interactions fectly assigned solitary and communal among adult females. Differences in ovar- species, but failed to distinguish between ian development (OD) also did not predict the latter two, assigning 1 eusocial species differences in either aggression or avoid- to the semisocial category and 1 semisocial ance, but were associated with rates of species to the eusocial category. DFAbased approach and pass behaviours, thesebeing on three putative categories (solitary, com- exhibited more frequently by the bee with munal, and caste-based social) reassigned greater ovarian development. Why would each species into the category presented in high OD females be more likely to ap- Fig. 2. Moreover, when Caenohalictus pygo- proach and especially, to pass? One possi- sinuatum was categorized as communal bility is that the closer a female is to laying (Michener et al. 1979), then DFA assigned an egg, the more active she is likely to be. it to the solitary group (as suggested by Under natural circumstances, a female Packer 2006). The success of the DFA halictine getting ready to lay an egg should approach is based on significant differ- be spending considerable time readying a ences among solitary, communal, and brood cell and provisioning it. In a circle Volume 19, Number 1, 2010 73 0% 20% 40% 60% 80% 100% SOLITARY X.blcuspidariae Pe.toroi II ^ T.prognathus Ca.pygosinuatum Ca.cuprellus Lplatycephalum L(Ct.)NDA-1 L.ancense L.figueresi Co corinogaster Co.melanocladius SEMISOCIAL Co.chloris p*j avoidance Co.pCaot.ahgeornbisctai m^v^, naggression pass EUSOCIAL H.lanei z^ Lzephymm 1 COMMUNAL Lhemichalceum ^^^^^1 ^^^^^1 R.mutabilis R.proxima Ca.dolator Ca.sp.A ^ Ps.pississi VERTICALTUBES 1 L(Ct.)NDA-1 ' L.hemichalceum Fig. 2. Comparison ofcircletubebehaviour ofX. blcuspidariaewithliteraturevalues (partiallyredrawnfrom Packer 2006, which also contains a complete list ofreferences). Solitarybee species are characterized by high rates of avoidance (withdrawals), communal bees by high rates of cooperation or tolerance (passing), and semisocialandeusocialspeciesbyhighratesofaggression,coupledwithverylowratesofpassing. Differences inbehaviouralprofilesofbeestestedincircletubesversusverticaltubes(Jeansonetal.2005)areevident.Genera represented (toptobottom) areXeralictus (X.),Penapis (Pe.), ThrincoJmlictus (T.),Caenohalictus (Ca.),Lasioglossum (L.), Cojynura (Co.), Halictus (H.), Ruizantheda (R.), and Pseudagapostemon (Ps.). tube, heightened activity rates make it the bee with smaller ovaries was often first more likely that a bee will approach the to withdraw (Wcislo 1997). Even in obli- second bee, and then perhaps continue gately eusocial species like Halictus ligatus, right past her. In other words, bees with dominance, aggression, and defensive be- higher ovarian development may be more haviours are most likely and most severe motivated to remain active, which in a when the two members of a dyad both circle tube would result in higher rates of have relatively high ovarian development, approaching and passing. for example, when a queenis paired with a The lack of correlation between ovarian worker with highly developed ovaries status and aggression might seem surpris- (Pabalan et al., 2000). This suggests that ing, but is consistent with observations in correlations between OD and aggression other species, including the solitary halic- might have more to do with reactions to tine, L. (Ctenonomia) NDA-1, and the the threat of egg replacement than with communalspecies,L. (Chilalictus)'platyceph- dominance behaviour per se. In solitary alum, in which ovarian status was not bees, we should not then expect to see a associated with aggression (McConnell- correlation between ovarian development Gamer and Kukuk 1997). In contrast, in and aggressive behaviour (or withdrawal Lasioglossumfigueresithe femalewithlarger behaviour) except perhaps in species with ovarieswas oftenfirsttobe aggressive,and high rates of intraspecific egg parasitism. 74 Journalof Hymenoptera Research: Festschrift Honoring Roy Snelling Our experiments uncovered consider- horizontal circle tubes versus vertical lin- able variability among individuals. Ran- ear tubes (Fig. 2), implying that major dom pairings of individuals with different differences in behavioural frequencies are behavioural tendencies (personalities) produced by different experimental meth- must then have contributed to variation odologies. Nevertheless, the interspecific among dyads, creatingbehavioural scenar- consistency of behavioural syndromes ob- ios that bear a marked resemblance to served in circle tube assays of solitary, those expressed by social bees. Indeed, this communal, and caste-based social species v^ould seem tobe thebasis ofthe phenom- is striking and statistically supportable, enon of 'emergent' sociality as described suggesting that when circle tube assays by Jeanson et al. (2005). However, the use are used consistently, they uncover funda- of the term 'emergent' to describe forced mental differences in behaviour among social interactions among solitary bees is solitary, communal, and semisocial and somewhat problematic, even in those eusocial species. These differences, if not which like X. bicuspidariae are ancestrally the behavioural frequencies themselves, and monomorphically solitary. This is can be used to infer general behavioural because solitary bees may also experience tendencies inbees of different social levels. social interactions that insect sociobiolo- In halictids, the ancestral trait of intoler- gists do not usually categorize as 'social', ance is suggested by high rates of avoid- such as interactions between foragers on ance in solitarybees such as X. bicuspidariae flowers, between nest residents in dense and another solitary rophitine species, nesting aggregations, between nest resi- Penapis toroi, in which avoidance behav- dents and would-be nest usurpers, or iours comprise about 75% of encounters between residents and egg kleptoparasites. (Fig. 2). Transitions to communal versus Moreover, group living may occur at very caste-based social behaviour may be quite low frequencies in some solitary species different. Circle tube assays imply that without extensive nest observations, as has solitary-communal transitions involve sig- recently been found for several species of nificant decreases in both aggression and the apid genus, Cemtina (Rehan etal. 2009). avoidance, whereas transitions to caste- In other words, many solitary bees, both based eusociality involve a significant ancestrally solitary and ancestrally social, increase in aggression, coupled with a may have considerable scope for intraspe- decrease in avoidance. To the extent that cific socialbehaviour, even if theyrarely or passes represent cooperative interactions, never nest in multifemale groups. The solitary-communal transitions would ap- variability in behavioural syndromes of pear to involve huge increases in coopera- solitary halictines (Fig. 2) suggests that tion whereas transitions to caste-based eventually it may be possible to detect eusociality involve little change or perhaps differences among obligately solitary, so- even a decrease in cooperative behaviour. cially polymorphic, and reversed solitary It will be important in future studies of species, especially based on the frequency both solitary and social halictines, to assess of avoidance and aggession. the degree of behavioural concordance — Behavioural changes in social transitions. between natural versus artificial contexts One caveat to the use ofartificial arenas for whenever possible, so that we can actually observing bee behaviour is that the fre- understand how representative circle tube quencies of circle tube behaviours may or behaviour is for those species for which may not represent the frequencies of same nesting data are unobtainable. or similar behaviours in natural settings. Given that one of the most outstanding Indeed, there are obvious differences in features of the eusocial insects is their behavioural frequencies assessed using frequent and sophisticated cooperative : VOLL-X^Z If X"_^.-^ZR !010 b€ha\'iour, the h}"pothesis that transitions fonvard and reverse social transitions, to caste-based socialit}' should involve especially where these involve the expres- decreases in cooperation coupled with sion of context-dependentbehaviour. increases in aggression seems contradic- ACKXO^\TEDGMI\TS tor'. However, semisocial and eusocial halictines not only interact \s'ith. n\any more indi\iduals than solitary- bees do, =rr.;r;: :f—i.if :: i 5rf;:f5 si', .sasdescnre:: : but they must also cope %\-ith dominance- ^r.e_ir:: is z.iz.:.ii.'iz. :: 'Si ~einor\' ot itlLS subordinance relationships, m.any ofwhich r> Rob Paxtcm fcH" heljrful are regulated bv aggressive beha\"ioui r :r -j-f manuscript and (Kukuk and May f991; Pabalan et al. :zre:f:e:i:r::!.'5 su1g5gefstuinodnesd fboyr 2000). Semisocial and eusocial bees must —rir.gRefearchCouncilof be able to exercise both tolerance and Car.- r.- Lr- tobothauthors. aggression with the same individuals. -\lthough aggressive beha\iours by soli- LITER.^TL'RI CITED tar}* and caste-based social bees in circle tubes may appear to be similar, a m.ajor Batra,S. \N. '^y66. ine Aiecyiuearidsocial£>eria\"iorof difference in natural settings is that ag- the priniiti\ely social bee Lasioglossum zeptofrum gressive beha^-iouT by the latter is likely Hy~erzr^exa: Halictidae). Unwersity ofKansas 5; ;: 5 fr:ti46: 359-423. modulated by nest-mate recognition, such Era- r C 5 Sre? A. Pearson,andB.N.Danforih. that encounters \\*ith non-nestmates ^sill 1. z Re:er: ari simultaneouscaiginsofeusod- likely provoke aggression, whereas en- alir IT ral:;-a zas ~'::£edmgs cf tiie Royal counters ^N-ith nestmates m.av engender S:::-r, 3-3-1^—^ S::^--^i 273:164^-1649. aggression, tolerance, or cooperation (Peso Breed, M. D., J. M. Slverman, and \V. J. BelL 1978. Agonistic behavior, social interactions, and be- and Richards 2010), depending on the havicKal spedaHzatiOTi in a primitively eusocial immediate behavioural context. bee. InsectesSociatu25: 351-364 Transitionstosocialbehaviour,especially Buckle, G. K 1984. A second lock a: aueer-forager to caste-based sociality' are rarer in halic- inteiactions in tiie priniiti\e- easza. haBctid, , tines than reversals to solitary- behav- Lasioglossum zephyrum. Jcu:?iz. .':-: Kansas EntomologiailSociety57: 1-6. iour (DaTLtorth et al. 2003). Recent evidence Daniorth,B.N.,L.Conway,andS.Ji2ij03.rmiogeny suggests that reversals to soHtar}' behav- ofeusocialLasioglossumrevealsmultiplelossesof iour do not necessarily retrace the original eusodalitywitfainaprimiti\'elyeusocialdadeof evolutionary* steps that led to socialitv'. For bees (Hyinenc^>fcera: Halictidae). Systematic Bid- instance, reversed-solitarv* Lasioglossum c^52: 23-36. have retained the social nesting character- , C Eardley, L. Packer K ^.'."alker. A. Pauly, istic of constructingbrood ceUs close to the aofndHaFl.iJc.tRidaanedriwa:r:ra—arr:raar—:irc:-.as1i1s[^on~r:e.n\d:ecicrracy m.ain burrow, facilitating both maternal .AtricanHaHctinae -: .; .r- ?-: 86-101. inspection and care of the cells (Plateaux- JeanscHi, R-, P. F. Kukuk, and J. H. FeweD JH. 20C5. Quenu 2008), and the potential for social EmergOKeofdi\isionoflabourinhaUctinebees: interactions among newly emerged, adult ccHitributi«is of social intraacticxts and behav- brood. Thus reversed soHtarv* bees mav iouralvariance.AnimalBduwkmr70: 1183-1193. Kukuk,P.F. andB.May.1991.CcJonydynamicsina have lost caste-based sodalitv*, but may primitively eusodal haBctine bee Lasioglossum haveretainedthecontext-dependentability' (LHalictus) zephyrum (Hymenc^>taa: Halictidae). to discriminate nestmates from non-nest- InsectesSodaux38: 171-189. mates. Circle tube comparisons of ances- . 1992. Social interactions and familiarity- in a X ccHimimialhalictinebeeLasioglossum (ChUalictus) trally sohtarv' species Like hiaispidariae, henddudceum.Ethology91:291-300. easni,dmraevyerhseeldp-stooliitaQru}m-inspaetceieasnldikdeisLt.ifnigguiiiesrh- Martkinneozw-nDdldeoosM,xiX.ofansodciJ.alMairntseicntesl.LJ1o9u9r5n.alHoifePoalldeeons-t the evolutionarv- sequences involved in rjJooy69: 595-599.