Behaviour 88 (2014) 113e124

Contents lists available at ScienceDirect

Animal Behaviour

journal homepage: www.elsevier.com/locate/anbehav

Nestmate discrimination in the social Ropalidia marginata: chemical cues and chemosensory mechanism

Aniruddha Mitra a,b,*, Arathy Ramachandran a, Raghavendra Gadagkar a,c a Centre for Ecological Sciences, Indian Institute of Science, Bangalore, India b Laboratoire Evolution, Génomes et Spéciation, Centre National de la Recherche Scientifique, Gif sur Yvette, c Evolutionary and Organismal Biology Unit, Jawaharlal Nehru Centre for Advanced Scientific Research, Bangalore, India article info Nestmate discrimination plays an important role in preserving the integrity of social colonies. It is fi Article history: known to occur in the primitively eusocial wasp Ropalidia marginata in which non-nestmate conspeci cs Received 3 April 2013 are not allowed to come near a nest. However, newly eclosed females are accepted in foreign colonies, Initial acceptance 10 June 2013 suggesting that such individuals may not express the cues that permit differentiation between nestmates Final acceptance 10 October 2013 and non-nestmates. As cuticular hydrocarbons (CHCs) have been implicated as chemosensory cues used Available online in nestmate recognition in other species, we investigated, using bioassays and chemical analyses, MS. number: 13-00292R whether CHCs can play a role in nestmate recognition in R. marginata. We found that individuals can be differentiated according to colony membership using their CHC profiles, suggesting a role of CHCs in Keywords: nestmate discrimination. Non-nestmate CHCs of adult females received more aggression than nestmate adult CHCs, thereby showing that CHCs are used as cues for nestmate recognition. Contrarily, and as expected, behavioural sequence analysis CHCs of newly eclosed females were not discriminated against when presented to a foreign colony. bioassay chemosensory cue Behavioural sequence analysis revealed the behavioural mechanism involved in sensing nestmate fi cuticular hydrocarbon recognition cues. We also found that newly eclosed females had a different CHC pro le from that of adult gas chromatography females, thereby providing an explanation for why young females are accepted in foreign colonies. nestmate discrimination Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. newly eclosed individual Ropalidia marginata

Communication is an integral part of all living beings, and can be Individuals recognize and differentiate between nestmates and found at various levels of biological organization, ranging from the non-nestmates and accordingly do or do not allow them to enter molecular level (intra- and intercellular communication) to the their nest, thus preventing social or theft of valuable organismal and colonial or societal levels (communication between resources from the nest. This also helps maintain group adhesion, individuals and between groups of individuals). Interindividual allowing individuals to show altruistic behaviours towards nest- communication plays an important role in the lives of social ani- mates, while avoiding or attacking non-nestmates, thus having an mals, as they have to perform various complicated social behav- implication for colony fitness (Wilson, 1971). Thereby nestmate iours and social decision making. This is best exemplified in social discrimination plays a key role in preserving the integrity of social , in which a large number of individuals (often thousands or insect colonies and is important for the organization and mainte- millions) communicate and coordinate with each other to perform nance of eusociality. the tasks necessary for their existence efficiently. Nestmate The mechanism of nestmate recognition involves one individual discrimination, that is, recognizing colony members (nestmates) perceiving the chemical cues present on the body surface of and differentiating them from nonmembers (non-nestmates) is an another (Dapporto, Fondelli, & Turillazzi, 2006; Howard & example of interindividual communication that is widespread in Blomquist, 2005; Martin, Vitikainen, Helantera, & Drijfout, 2008; social insects (Gadagkar, 1985; Gamboa, Reeve, Ferguson, & Wacker, van Zweden, Dreier, & d’Ettorre, 2009; van Zweden & d’Ettorre, 1986; Gamboa, Reeve, & Pfennig, 1986; Pfennig, Gamboa, Reeve, 2010). It is believed to occur by matching a label (chemical signa- Reeve, & Ferguson, 1983; van Zweden & d’Ettorre, 2010). ture containing nestmate/non-nestmate cues) with a template (representation of colony odour in memory), and, depending on the similarity or difference between the label and template, a conspe- * Correspondence: A. Mitra, Laboratoire Evolution, Génomes et Spéciation, Centre cific is accepted into or rejected from the colony (Gadagkar, 1985; National de la Recherche Scientifique, Number 1, avenue de la Terrasse, Batiment Sturgis & Gordon, 2012; Tsutsui, 2004; van Zweden & d’Ettorre, 13, Gif sur Yvette 91198, France. E-mail address: [email protected] (A. Mitra). 2010). Over the years, cuticular hydrocarbons (CHCs) have come

0003-3472/$38.00 Ó 2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anbehav.2013.11.017 114 A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 to be recognized as chemical cues used in nestmate recognition in aseasonal, that is, nests are founded and abandoned throughout the various social insect species (Akino, Yamamura, Wakamura, & year, and there is no body size variation between individual Yamaoka, 2004; Dapporto et al., 2006; Gamboa, Grudzien, from season to season. These wasps generally do not allow any non- Espelie, & Bura, 1996; Howard & Blomquist, 2005; Lenoir, nestmate conspecifics to come near their nest (Venkataraman & Fresneau, Errard, & Hefetz, 1999; Lorenzi, Sledge, Laiolo, Sturlini, Gadagkar, 1992; Venkataraman, Swarnalatha, Nair, & Gadagkar, & Turillazzi, 2004; Martin et al., 2008; Ozaki et al., 2005; van 1988). Young or newly eclosed individuals, however, are often Zweden et al., 2009; van Zweden & d’Ettorre, 2010). The relative accepted into foreign colonies, suggesting that such individuals do proportion of compounds in CHCs have been found to differ be- not express the cues that permit differentiation between nestmates tween different colonies of a species, thereby giving rise to a and non-nestmates (Arathi et al., 1997; Venkataraman & Gadagkar, colony-specific blend, and there is behavioural evidence for the 1995). As CHCs have been implicated as chemosensory cues involvement of CHCs in nestmate discrimination (Akino et al., involved in nestmate recognition in various species, we investi- 2004; Foitzik, Sturm, Pusch, d’Ettorre, & Heinze, 2007; Lahav, gated whether CHCs can be used for nestmate recognition in Soroker, & Hefetz, 1999; Lorenzi, Bagnères, Clement, & Turillazzi, R. marginata. Since non-nestmates are attacked when they come 1997). Thus CHCs can function as chemical cues perceived when near a foreign colony, we predicted that non-nestmate CHCs should individuals come into contact and can be used by individuals to receive more aggression than nestmate CHCs. Contrarily, since differentiate nestmates from non-nestmates. newly eclosed individuals are not discriminated against when The study of nestmate discrimination can be broadly divided presented to a foreign colony, we predicted that there should not be into two components: (1) chemical cues or chemosensory cues, any difference between the aggression received by nestmate CHCs that is, the chemicals involved in forming the signal based on which and that received by non-nestmate CHCs, when CHCs of newly interindividual chemical communication resulting in nestmate eclosed individuals are used. Hence in addition to revealing the recognition occurs, generally exemplified by CHCs present on the chemical cues used in nestmate recognition in R. marginata, this cuticle of individuals, and (2) chemosensory mechanisms, that is, study also allowed us to show how the age of an individual has an the proximate mechanisms (behavioural and neurobiological) used effect on whether the individual will or will not be discriminated to perceive the chemical cues involved in nestmate recognition, and against in a foreign colony. thereby to make the decision whether or not to allow an individual We also unravelled, using behavioural sequence analysis, the into a colony. The chemosensory cues that form the nestmate signal behavioural mechanism used in perceiving nestmate discrimina- have been considered as the ‘expression component’ of nestmate tion cues. Behavioural sequence analysis was first developed as a recognition, wherein the polymorphic phenotypic cues or labels technique for the social sciences to study events that are sequential (generally CHCs in the case of social insects) are expressed on the in nature or to analyse any context in a temporal manner (Abbott, external body surface of individuals, and such cues can be both 1995). It has been applied in animal behaviour, often to observe inherent and derived from the environment (Tsutsui, 2004). The interactive social behaviour, and has mostly been used to look at chemosensory mechanisms used in perceiving nestmate cues fall behaviour occurring in successive time units or at which behaviour under the purview of the ‘perception component’ of nestmate follows which behaviour (Bakeman & Gottman, 1997; Fagen & recognition (Mateo, 2004). Lastly, following perception of the Young, 1978; Maubourguet, Lesne, Changeux, Maskous, & Faure, ‘nestmate’ cues, depending on the similarity to or difference be- 2008; Slooten, 1994). It has not yet been applied to look at chem- tween the perceived cue and template, a decision is made whether ical communication in social insects. We used behavioural or not to allow an individual into a colony. This falls under the sequence analysis to look at which behaviour follows which ‘action component’ of nestmate recognition (Liebert & Starks, behaviour and thereby to find out the behavioural chemosensory 2004). Studies that have investigated nestmate discrimination in mechanism that is used for perceiving the chemical cues used in social insects have primarily focused on the chemosensory cues nestmate recognition. Aggression is the key behaviour for catego- (the expression component) by looking at chemical differences in rizing the response of a wasp towards a stimulus containing CHC profiles of colonies, sometimes followed by bioassays ‘nestmate’ cues as acceptance or rejection. High aggression is ex- (involving the action component) to show that CHCs are indeed pected with non-nestmate CHCs implying rejection, and lower the chemical cues involved in nestmate discrimination (Akino et al., aggression or no aggression is expected with nestmate CHCs 2004; Dapporto et al., 2006; Foitzik et al., 2007; Gamboa et al., implying acceptance. Hence by looking at the behaviours that 1996; Howard & Blomquist, 2005; Lahav et al., 1999; Lenoir precede aggression, in a stimulus-specific manner, and based on et al., 1999; Lorenzi et al., 1997; Lorenzi et al., 2004; Martin et al., the relative frequency with which aggression follows a particular 2008; van Zweden et al., 2009; van Zweden & d’Ettorre, 2010). given behaviour in response to a nestmate recognition stimulus, we Studies investigating the chemosensory mechanisms that are could find out the behaviour that is involved in perceiving nestmate used to perceive nestmate discrimination cues (the perception cues. component) are relatively rare (Brandstaetter & Kleineidam, 2011; Brandstaetter, Rössler, & Kleineidam, 2011; Leonhardt, METHODS Brandstaetter, & Kleineidam, 2007; Ozaki et al., 2005). Hence there is a need to incorporate investigation into the chemosensory Chemical Analysis mechanisms in addition to looking at the chemosensory cues in order to have a more comprehensive understanding of chemical Colony differences in CHC profiles communication. To investigate whether CHCs can convey information on nest Ropalidia marginata is a primitively eusocial found membership, we analysed the CHCs of adult females (>5 days old) in peninsular India. These wasps build nests out of cellulose ma- from six different colonies. Postemergence nests of R. marginata terial collected from plants. Each colony has one queen that mo- were collected from various localities in Bangalore (13000N, nopolizes reproduction, and the other individuals function as 77320E), India, and transplanted to the vespiary at the Centre for sterile workers that perform foraging and other tasks needed to Ecological Sciences, Indian Institute of Science, Bangalore. The nests maintain a colony. Males stay on the nest for a brief period (usually were maintained in closed cages made of wood and fine mesh, and up to 7 days), after which they leave the nest for a solitary life, and provided with food, water and building material ad libitum. All thus nest members are primarily females. The colony cycle is adults were uniquely colour-coded with small spots of Testors A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 115 enamel paints on their thorax (Gadagkar, 2001). The queen of each randomly divided into six arbitrary categories, wherein individuals colony was identified by observing egg-laying behaviour before the from different real colonies could be clumped together into the experiment. The queen, along with six randomly chosen workers, same arbitrary category, and then we tested whether individuals was collected from each of the six colonies resulting in 42 in- from these arbitrary categories could be differentiated from each dividuals (7 6) in total. The wasps were killed by anaesthesia on other using discriminant analysis. The process was iterated 10 000 ice followed by freezing at 20 C, and were stored at 20 C until times, and we counted the times both the significance of Wilks’s l further analysis. Extraction of CHCs was done following Mitra and was <0.050 and the percentage of correct classifications was Gadagkar (in press) by rubbing a clean cotton wrapped toothpick greater than or equal to that in discriminant analysis applied to real on the dorsal part of the second gastral segment of each wasp data. Thereby, the percentage of permutations for which discrimi- applying uniform pressure for 30 s, followed by washing the cotton nant analysis on arbitrary data gave better or equally good in 200 ml of pentane, evaporation of pentane under a 40 W table discrimination as present in real data was calculated (Mitra & lamp and finally resuspension of CHCs in 10 ml of chilled pentane Gadagkar, 2012). We expected that statistical analysis on such from which 2 ml was used in gas chromatography. Gas chroma- arbitrary data should be significant in only a very small proportion tography conditions were identical to those described in Mitra and of permutations, if the pattern observed in the real data was robust. Gadagkar (in press). Blank runs were done to exclude any con- As mentioned earlier, discriminant analysis used for analysing taminants that may be present in cotton, toothpick and solvent. CHC data can potentially lead to overfitting as the number of pre- After gas chromatography one worker was found to have very low dictor variables often exceeds the sample size in the smallest group CHC levels as most peaks were either absent or below the detection (a problem that can be overcome by checking the results of the limit of the gas chromatography machine. Hence data from this discriminant analysis using simulations, see above). Thus we also individual were discarded and data from the remaining 41 in- used another technique, random forest for classification, using the dividuals were used for statistical analysis. package randomForest 4.6-6 in R 2.13.2 (Mitra & Gadagkar, 2012). Identification of CHCs (34 peaks) had been done earlier by gas Random forest is a robust technique that can be used to differen- chromatographyemass spectrometry (Mitra & Gadagkar, in press), tiate individuals using classification trees (Breiman, 2001). It does and for discriminant analysis, we used peaks that were present in at not have any assumptions about the distribution of data or the least 70% of all samples (resulting in 32 peaks). The area under each minimum required sample size, and hence should be adequate for peak was converted to percentage area. Since some minor peaks analysing small sample sizes with a large number of predictor were occasionally below the detection limit of the gas chroma- variables (Cutler et al., 2007). Further details on the methods and tography machine, occasionally resulting in a zero percentage area results of analysis by random forest are given in the Appendix. under a peak, we added 0.001 to these zero values to eliminate zeros (to avoid computational problems in the next step), and CHC profiles of newly eclosed and older females subjected the resulting data to log ratio transformation using the As it is known that R. marginata wasps cannot discriminate following formula between newly eclosed nestmates and non-nestmates (Arathi, " # Shakarad, & Gadagkar, 1997; Venkataraman & Gadagkar, 1995), A ¼ p$j which is in accordance with the results of our bioassays (see Zp$j ln g Aj Results), we did further analyses with CHCs of newly eclosed fe- males (posteclosion age < 5 days) to see whether their CHC profiles where Ap$j is the area of the peak p for individual j,g(Aj) is the are different from those of older adult females (posteclosion geometric mean of all peaks considered for analysis in individual j, age > 5 days). We compared 10 newly eclosed individuals (easily and Zp$j is the transformed peak area of peak p for individual j identifiable by their black eyes; older adults have brown eyes) with (Aitchison, 1986; Reyment, 1989). Following log ratio trans- 11 adults. The wasps were collected from seven colonies at the formation, Levene’s test (Brown Forsythe version) was performed same time. Thus all newly eclosed individuals and adults came from on each variable to check for homogeneity of variance, and only the same phase of the colonies, and were expected to be of the those peaks that had homogeneous variances were used in same sizes. Also since R. marginata has an aseasonal colony cycle, discriminant analysis (29 peaks). Discriminant analysis was per- there is not much difference in body size of individuals eclosing in formed to see whether individuals from different colonies can be different seasons, and newly eclosed individuals are similar in body differentiated from each other using their CHC profiles. During size parameters to adults (personal observation). The wasps were discriminant analysis, we opted to omit from the analysis variables killed and stored at 20 C until further analysis as described having very low tolerance (<0.001) to avoid multicollinearity. The above. Since rubbing cotton on the gaster did not extract sufficient significance of Wilks’s l and the percentage of correct classifica- quantities of CHCs from newly eclosed females, CHC extraction was tions were used to interpret the results. Peaks 17, 25 and 28a (2- done by washing each individual in 200 ml of chilled pentane for methyl nonacosane, mixture of 5-, 21-, 5-, 19- and 5-, 17- 5 min on an orbital shaker (for both newly eclosed and adult dimethyldotriacontane and 6-, 14-dimethyldotriacontane) were wasps), followed by evaporation of pentane under a 40 W incan- eliminated from discriminant analysis because they had hetero- descent table lamp. Finally, extracted CHCs were resuspended in geneous variances (Levene’s test: P < Pcritical for Bonferroni 10 ml of chilled pentane to which 3.745 pg of dodecane was added correction, i.e. 0.002), and peaks 4 and 19a (11-methyl tricosane as an internal standard, and 2 ml of extract drawn out immediately and 6-methyl triacontane) were also eliminated (present in <70% for gas chromatography (done as described above). Unmarked in- samples). Peaks 4 and 19a were also removed from analysis by dividuals were chosen for this analysis to exclude any pentane- random forest (see below). soluble contaminants that may be present in the paint used for Since discriminant analysis done on CHC data can potentially marking the wasps. Blank runs were done to exclude any lead to overfitting, we performed simulations to check the contaminant that may be present in the solvent. Multivariate sta- robustness of the results of the discriminant analysis. Individuals tistical analyses using discriminant analysis (including simulations) were randomly assigned to any one of the six colonies, keeping the and random forest were carried out as described above. Peaks 17, sample size of each arbitrary category the same as that in the real 19a, 25 and 28a were removed from the discriminant analysis and data, and the significance of Wilks’s l and the percentage of correct random forest (present in <70% samples). Peaks 4, 6, 11, 20 and 22 classifications were calculated. Thus all 41 individuals were were also removed from the discriminant analysis as they had 116 A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 heterogeneous variances (Levene’s test: P < Pcritical for Bonferroni expected proportion of each behavioural dyad was calculated by correction, i.e. 0.002), and peaks 1e3, 7e9 and 13 were removed multiplying the proportions of each of the two behaviours (pro- from the discriminant analysis because of low tolerance (<0.001). portion of each behaviour ¼ acts of that particular behaviour/total Thus discriminant analysis used 18 peaks in total, and random acts of all behaviours). The expected proportion was compared to forest used 30 peaks in total. In addition, the area under each peak the observed proportion by a chi-square test. This was done was converted to picograms by calibrating with an internal stan- separately for each kind of dyad, for each kind of stimulus (blank, dard, and the difference between adults and newly eclosed in- nestmate CHCs and non-nestmate CHCs). We used a Fisher’s exact dividuals checked for each individual peak (for 30 peaks, i.e. peaks test of proportions to compare the proportions of each kind of present in at least 70% of all individuals), as well as for the total dyad among responses shown towards the blank, nestmate CHCs quantity of all CHCs, by ManneWhitney U tests. and non-nestmate CHCs. Thus behavioural sequence analysis was used to look at which behaviour follows which behaviour. Since Behavioural Assays aggression or dominance behaviour shown by the wasps of a colony towards a non-nestmate decides whether the individual Nestmate discrimination from CHCs of adult females will or will not be accepted into the colony, by finding out the We used 12 individuals from four colonies (collected and behaviours that precede aggression in a statistically significant maintained in the vespiary in the same way as described above) in a manner, we could find out the behaviours used to perceive nest- bioassay to test whether CHCs of adult females are involved in mate recognition cues. If the occurrence of a particular behaviour nestmate discrimination. For each round of stimulus presentation, X following another particular behaviour Y is more than that ex- CHCs were extracted from a pair of unmarked workers taken from pected from the relative frequencies of each behaviour alone, it two different colonies, by washing each individual in 200 mlof can be concluded that Y inherently follows X and thereby they pentane for 5 min in chilled conditions on an orbital shaker. Prior to together form an integral unit of a behavioural mechanism. CHC extraction, each wasp was anaesthetized on ice and killed as Comparison of proportions of different dyads between different described above. Each extract was then added drop by drop onto a kinds of stimuli further helped to reveal the behavioural chemo- small piece of filter paper, approximately 0.5 0.5 cm (previously sensory mechanisms, as particular dyads can be expected to vary cleaned by washing in pentane), and then dried under a 40 W in- consistently according to the kind of stimulus (nestmate versus candescent table lamp. We added 200 ml of pentane on a third filter non-nestmate CHCs). paper to be used as a blank (prior to beginning the behavioural assays, gas-chromatographic analyses of pieces of filter paper with Nestmate discrimination from CHCs of newly eclosed females CHCs added were done to confirm that CHCs were transferred We used 10 newly eclosed individuals (posteclosion age < 5 successfully to the filter paper by this process and that the same set days) from five colonies (including three of the four colonies used of peaks was obtained from the CHC-added filter paper as was earlier in the bioassay on CHCs of adult females) to test whether obtained from the cuticle of wasps). Thus three kinds of stimuli nestmate discrimination is possible from CHCs of newly eclosed were available for presenting to each colony: nestmate CHCs, non- individuals. As it was difficult to obtain simultaneously a pair of nestmate CHCs and a blank. Each stimulus was presented sepa- newly eclosed individuals from a pair of nests, CHCs were extracted rately to each of the two colonies from which the two individuals from an unmarked newly eclosed female taken from a single colony had been removed, and held for 1 min. The stimuli were held very using the same procedure as described above. After preparation of close to each nest (within 1e2 cm from the nest surface) ensuring CHC extract, each stimulus (nestmate or non-nestmate CHCs and that the colony members had ample opportunity to perceive and blank) was presented to the colony from which the individual had respond to the stimuli. The order of presentation of stimuli was been removed, as well as to another colony, in the same way as random and was unknown to the observers. A gap of 2e5 min was described above, and we recorded each act of interaction of the maintained between two successive presentations. We recorded wasps of a colony with the stimulus presented for 1 min (video each act of interaction of the wasps of a colony with the stimulus recording not done). Thus each round of presentations yielded one presented and also videorecorded them to calculate the duration set of responses to nestmate CHCs and the blank from one colony, spent in interacting with the stimuli. Thus each round of pre- and another set of responses to non-nestmate CHCs and the blank sentations yielded two sets of responses in total from two colonies from another colony. As earlier, the experiment was done blind. The (one set of responses per colony) to nestmate CHCs, non-nestmate procedure followed to perform one round of presentations was CHCs and the blank. The procedure followed to perform one round repeated nine more times to obtain 10 sets of responses in total to of presentations was repeated five more times to obtain 12 sets of each of the three kinds of stimuli: nestmate CHCs, non-nestmate responses in total. The relative frequency of each behaviour shown CHCs and blank. For each behaviour, the response shown towards by each colony in response to each stimulus was calculated by either nestmate or non-nestmate CHCs was converted to relative dividing the number of occurrences of the particular behaviour by frequency as described above, and then corrected by subtracting the total number of occurrences of all behaviours shown towards the response shown towards the corresponding blank, and finally that particular stimulus. For each behaviour, the responses shown compared by ManneWhitney U test. towards nestmate and non-nestmate CHCs (relative frequencies) For statistical analyses we used StatistiXL 1.8 (Microsoft Corp., were corrected by subtracting the value for the response shown Redmond, WA, U.S.A.) and R 2.13.2 (R Foundation for Statistical towards the corresponding blank, and compared by a Wilcoxon Computing, Vienna, ). paired-sample test. RESULTS Behavioural sequence analysis Behavioural sequence analysis was done for dyads, that is, two Chemical Analysis consecutive behaviours, to check whether a particular behaviour following another particular behaviour occurred more than that Colony differences in CHC profiles expected from the relative frequencies of each behaviour alone, The CHC profiles of adult females were the same as has been and transition diagrams were drawn to visualize the behavioural reported before, that is, 34 peaks in total, identified as linear, transitions (Bakeman & Gottman, 1997; Fagen & Young, 1978). The monomethyl and dimethyl branched alkanes (Mitra & Gadagkar, in A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 117 press). Individuals could be differentiated and segregated into Behavioural Assays nonoverlapping clusters based on their colony membership using discriminant analysis (Fig. 1). All five discriminant functions were Nestmate discrimination from CHCs of adult females significant (function 1: Wilks’s l < 0.0001, P < 0.0001; function 2: Three kinds of behaviours were observed in the bioassays (in Wilks’s l < 0.0001, P < 0.0001; function 3: Wilks’s l ¼ 0.001, response to both adult female CHCs as well as newly eclosed female P < 0.0001; function 4: Wilks’s l ¼ 0.022, P ¼ 0.002; function 5: CHCs): antennation (wasps touched and explored the stimulus Wilks’s l ¼ 0.176, P ¼ 0.036; N ¼ 41 individuals from six colonies), with their antennae), mouthing (touched and explored the stim- and classification analysis gave 100% correct classification. ulus with mouthparts), and dominance behaviour or aggression. Randomly assigning individuals to categories and running Peck was the predominantly observed aggressive behaviour (wasp discriminant analysis showed that discrimination as good as or rapidly pecks the stimulus with mandibles); attack (wasp grapples better than that on real data was obtained in only 1.59% of iterations with the stimulus, forming a characteristic ‘C’ shape over it, and (out of 10 000), suggesting that the results of discriminant analysis attempts to bite and sometimes also to sting it) and nibble (wasp carried out on real data are likely to be robust. Analysis by random aggressively nibbles the stimulus with mandibles) were also oc- forest also segregated individuals into clusters based on colony casionally seen (the Supplementary videos show the most membership. The degree of overlap between colonies on the commonly seen behaviours: antennation, mouthing and peck). random forest appeared to be higher than with discriminant In the bioassay on CHCs of adult females, there was no difference analysis (see Appendix Fig. A1a). between nestmate and non-nestmate CHCs with respect to antennation, but nestmate CHCs received significantly more mouthing than non-nestmate CHCs, whereas non-nestmate CHCs fi CHC pro les of newly eclosed and older females received significantly more dominance behaviour than nestmate fi The CHC pro les of newly eclosed females were distinctly CHCs (Fig. 2; Wilcoxon paired-sample tests: antennation: T ¼ 22, different from those of older adult females (see Appendix Fig. A2). P ¼ 0.204; mouthing: T ¼ 15, P ¼ 0.032; dominance behaviour: Newly eclosed individuals could be differentiated from adults on a T ¼ 2, P ¼ 0.001; N ¼ 24). ’ l ¼ discriminant function (Appendix Fig. A3; Wilks s 0.005, There was no difference between nestmate CHCs, non-nestmate < fi fi P 0.0001; classi cation analysis: 100% correct classi cation). CHCs and the blank with respect to the time the wasps spent Randomly assigning individuals to categories and running interacting with each kind of stimulus (bioassay on CHCs of adult discriminant analysis showed that discrimination as good as or females; Wilcoxon paired-sample test: nestmate CHCs versus better than that on real data was obtained in only 4.45% of itera- blank: T ¼ 28, P ¼ 0.700; non-nestmate CHCs versus blank: T ¼ 31, tions (out of 10 000), suggesting that the results of discriminant P ¼ 0.898; nestmate CHCs versus non-nestmate CHCs: T ¼ 27, analysis carried out on real data are likely to be robust. Newly P ¼ 0.638; N ¼ 24). Also, the response to a stimulus that had already eclosed and adult females were well separated on the random been presented to a colony did not differ from the response when it forest (see Appendix Fig. A1b). was presented again to another colony (ManneWhitney U tests: Newly eclosed individuals had lower amounts of CHCs than P > 0.050 for all three behaviours, antennation, mouthing and adults: 156.248 58.414 pg/individual (mean SD) for newly dominance behaviour, for all stimuli, blank, nestmate CHCs and ¼ eclosed individuals (N 10) and 369.632 204.157 pg/individual non-nestmate CHCs). Since each stimulus was used twice, this ¼ e ¼ < for adults (N 11; Mann Whitney U test: U 99, P 0.001). confirmed that the first use did not affect the response on the sec- When the quantities of each peak were compared between newly ond use, implying that the chemical signature present on each eclosed individuals and adults, after Bonferroni correction stimulus remains unchanged after wasps of a colony interact with it. (Pcritical ¼ 0.002), a significant difference was found for 20 of the 30 peaks tested (see Appendix Table A1). Behavioural sequence analysis Behavioural transition diagrams showed that the relative fre- quency of antennation following antennation was roughly the same for all three stimuli whereas dominance behaviour following 8 antennation and antennation following dominance behaviour were lower with nestmate CHCs than with non-nestmate CHCs or the blank (Fig. 3). Mouthing following antennation and antennation following mouthing were higher for nestmate CHCs, while domi- 3 nance following mouthing was similar for all three stimuli. Mouthing following mouthing and dominance following domi- nance (absent for nestmate CHCs) were also roughly similar for all kinds of stimuli. − 2 Behavioural sequence analysis revealed the behavioural dyads V1006 that were significantly different from what can be expected from the V1028

Function 2 (27.9%) relative frequencies of each behaviour alone (Table 1). For nestmate V1029 −7 CHCs, mouthing followed by antennation and antennation followed V1069 by mouthing were more than expected, and antennation followed V1082 by dominance behaviour and dominance followed by antennation V1109 were less than expected. Dominance followed by mouthing was also −12 less than expected. For non-nestmate CHCs, antennation followed − − 10 5 0 5 10 15 by dominance behaviour was more than expected. Function 1 (49.3%) Fisher’s exact test of proportions showed that following anten- nation, nestmate CHCs received more mouthing than the blank or Figure 1. Adult females from six colonies of R. marginata (N ¼ 41; colony identities, i.e. non-nestmate CHCs. Non-nestmate CHCs and blank stimuli nest numbers given in the key) plotted on discriminant functions 1 and 2 that differentiate them using their CHC profiles. Figures in parenthesis beside each function received more dominance following antennation than nestmate denote the percentage variance explained by that function. CHCs (Table 2). 118 A. Mitra et al. / Animal Behaviour 88 (2014) 113e124

0.45 0.3 (a) a AN 0.2 a 0.14 0.08 0.12 0.1 0.11

0 M DB 0.02 0.02 −0.1 Blank 0.07 Antennation

−0.2 0.41 −0.3

AN −0.4

0.05 0.4 0.22 0.03 a (b) 0.26 0.3 M DB

0.02 0.02 0.2 b Nestmate CHCs 0.1

0.33 0 Mouthing

−0.1 AN

0.13 −0.2 0.15 0.16 0.16 − 0.3 0.01 M DB 0.4 0.01 0.01 (c) Non-nestmate CHCs 0.04 0.3 b

0.2 Figure 3. Transition diagrams showing transition frequencies of all behavioural dyads shown in response to a blank stimulus, nestmate cuticular hydrocarbons (CHCs) and 0.1 non-nestmate CHCs of adult R. marginata females. AN: antennation; M: mouthing; DB: dominance behaviour. Each transition frequency represents the relative frequency of a a particular behaviour following another behaviour out of the total number of behav- 0 ioural dyad sequences shown in response to each kind of stimulus. Only those behavioural dyads that were observed have been depicted. −0.1 Nestmate discrimination from CHCs of newly eclosed females Dominance behaviour −0.2 Behaviours observed in the bioassay of CHCs of newly eclosed individuals were the same as what was observed in the bioassay of −0.3 CHCs of adults. There was no difference between responses shown towards nestmate CHCs and non-nestmate CHCs for any of the −0.4 three behaviours, antennation, mouthing and dominance behav- Nestmate Non-nestmate iour (Fig. 4; ManneWhitney U tests: antennation: U ¼ 54, CHCs CHCs P ¼ 0.796; mouthing: U ¼ 71, P ¼ 0.123; dominance behaviour: ¼ ¼ Figure 2. Relative frequency of (a) antennation, (b) mouthing, and (c) dominance U 65, P 0.28). behaviour (corrected by subtracting the values for the corresponding blanks) shown towards stimuli containing nestmate and non-nestmate cuticular hydrocarbons (CHCs) DISCUSSION of adult females of R. marginata. Different letters signify significant difference between distributions (Wilcoxon paired-sample test: P < 0.05, N ¼ 12 for each distribution). Box plots show median, interquartile range and range. Adult females could be segregated according to their colony membership using the relative proportions of their CHCs, sug- gesting that CHCs can be involved in nestmate discrimination in A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 119

Table 1 Statistically significant behavioural dyads 0.3 (a) Dyad Stimulus Relative frequency c2 P Result

Expected Observed 0.2

AN/M Nestmate CHCs 71/521 46/130 19.934 <0.0001 >Expected M/AN Nestmate CHCs 71/521 38/138 9.487 0.002 >Expected 0.1 AN/DB Nestmate CHCs 61/531 5/171 8.693 0.003 Expected CHCs 0 DB/AN Nestmate CHCs 61/531 8/168 4.820 0.028 DB DB Blank 14/578 12/167 6.665 0.010 Expected −0.2 Results of chi-square analysis (df ¼ 1 for all comparisons) of behavioural sequences showing behavioural dyads that are significantly different (P < 0.05) from what can be expected from the relative frequencies of each behaviour alone, for responses −0.3 shown by R. marginata wasps towards blank stimuli, nestmate CHCs and non- nestmate CHCs. P value that remains significant after Bonferroni correction − (Pcritical ¼ 0.001852) is highlighted in bold. AN: antennation; M: mouthing; DB: 0.4 dominance behaviour. 0.2 (b) R. marginata. This was corroborated by the bioassay results of adult female CHCs in which non-nestmate CHCs received more domi- 0.1 nance behaviour than nestmate CHCs, showing that CHCs contain cues that are used to differentiate members of one’s own nest from those of another nest. It is known that in R. marginata individuals 0 receive aggression from members of a foreign colony, which will generally not allow any non-nestmate to come near their nest (Venkataraman et al., 1988; Venkataraman & Gadagkar, 1992, 1993). −0.1 Hence our bioassay results are in accordance with these observations. Mouthing Behavioural sequence analysis showed that there were several −0.2 dyads whose relative frequencies were significantly different from what should be expected from the relative frequencies of each component of the dyad, showing that behaviours can co-occur −0.3

Table 2 −0.4 Behavioural dyads that differ significantly from each other

Dyad Proportion Result P 0.4 (c) Blank Nestmate Non-nestmate CHCs CHCs 0.3 AN/AN 80/99 66/136 Non-nestmate 0.0201 CHCsblank 0.0002 160 46/130 32/170 Nestmate CHCs>non- 0.0155 0.1 nestmate CHCs / < AN DB 22/ 5/171 Nestmate CHCs blank 0.0010 0 157 5/171 32/170 Nestmate CHCsblank 0.0003

165 Dominance behaviour − 14/ 30/172 Non-nestmate 0.0367 0.2 165 CHCs>blank DB/AN 25/ 8/168 Nestmate CHCsnestmate CHCs −0.4 DB/DB 12/ 0/176 Nestmate CHCsnestmate CHCs Figure 4. Relative frequency of (a) antennation, (b) mouthing and (c) dominance behaviour (corrected by subtracting the values for the corresponding blanks) shown ’ Results of Fisher s exact test comparing proportions of behavioural dyads shown by towards stimuli containing nestmate and non-nestmate cuticular hydrocarbons (CHCs) R. marginata wasps towards different stimuli (blank stimuli, nestmate CHCs and of newly eclosed females of R. marginata. Comparison between distributions: Manne fi < non-nestmate CHCs). Only comparisons that are signi cantly different (P 0.05) are Whitney U test: P > 0.05, N ¼ 10 for each distribution. Box plots show median, inter- fi ¼ shown. P values that remain signi cant after Bonferroni correction (Pcritical 0.002) quartile range and range. are highlighted in bold. AN: antennation; M: mouthing; DB: dominance behaviour. 120 A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 sequentially in the form of discrete units. Aggression or dominance amounts of CHCs of newly eclosed females could be why they stay behaviour is the key behaviour for categorizing as acceptance or on the nest, in a relatively protected environment, and do not go rejection the response of a wasp towards a stimulus containing nest out to forage, as they could be physiologically predisposed against membership cues. When a non-nestmate tries to enter a colony, she leaving the nest. Hence the probability of a newly eclosed indi- is immediately shown aggression by its members, and hence she vidual attempting to enter a foreign colony is low. For this reason, leaves (rejection), whereas a nestmate entering a colony is not R. marginata may not have evolved any mechanism to discriminate usually shown aggression (acceptance). Dominance behaviour newly eclosed non-nestmates. following antennation and antennation following dominance Age-based acceptance/rejection of non-nestmate conspecifics behaviour were higher with non-nestmate CHCs than nestmate has been reported in other social hymenopterans (Morel, Vander CHCs, suggesting that antennation can elicit aggression when a Meer, Lavine, & Blum, 1988; Panek, Gamboa, & Espelie, 2001; stimulus contains non-nestmate CHCs. Honeybees, Apis mellifera, Wilson, 1971). It has also been found that young individuals have are known to use their antennae for both olfaction and contact lower amounts of CHCs on their body surface, similar to what we chemosensation (gustation), and chemosensory sensilla that detect found in R. marginata, and gradually with time they show an in- CHCs have also been found on the antennae of ants, suggesting that crease in the amount of CHCs and also acquire or develop the hymenopterans use their antennae for detecting both volatile colony-specific CHC profile (Lenoir, d’Ettorre, Errard, & Hefetz, odours and surface-borne chemicals (Ozaki et al., 2005; Sanchez, 2001; Neves, Andrade, Súarez, Lima, & Antonialli-Junior, 2012). 2012). Hence R. marginata wasps can also be expected to use The lack of CHCs on the cuticle of newly eclosed individuals has their antennae for contact chemosensation (for sensing nest been termed ‘cuticular chemical insignificance’ (Lenoir et al., 1999). membership cues), which is supported by the findings of the This has been considered the reason why such young individuals behavioural sequence analysis. can be accepted into foreign colonies. As they get older, they may Mouthing the stimulus is very similar to ‘allogrooming’, which is start synthesizing more CHCs and may also acquire CHCs from their a behaviour commonly shown towards nestmates, in which one nestmates or from the nest material, thereby developing the wasp grooms another with its mouthparts, mouthing the body colony-specific blend of CHCs (Lenoir et al., 2001; Lorenzi et al., surface of the wasp being groomed. We found that nestmate CHCs 2004). Our results on R. marginata are in agreement with these receive more mouthing than non-nestmate CHCs, and also that previous studies. Based on behavioural experiments, it has been mouthing follows antennation more frequently with nestmate suggested that young R. marginata individuals may acquire colony- CHCs than non-nestmate CHCs or a blank stimulus, suggesting that specific chemical cues from their nest surface (Arathi et al., 1997; a stimulus containing nestmate cues may inherently elicit allog- Venkataraman et al., 1988; Venkataraman & Gadagkar, 1992, rooming behaviour. Allogrooming or mouthing can help spread 1995). Since we found that young R. marginata individuals have chemical cues across colony members giving rise to a colony- low amounts of CHCs, and since it has been shown in another specific blend of chemicals, sometimes referred to as the ‘gestalt’ polistine wasp that young individuals having low CHCs can acquire (Crozier & Dix, 1979). Since some mouthparts in insects are known hydrocarbons from the environment (Lorenzi et al., 2004), it is to contain chemosensory sensilla, this can have further implica- possible that young R. marginata wasps develop their colony- tions for sensing nest membership cues (de Bruyne & Warr, 2006). specific CHC signature by acquiring it from the nest surface and When newly eclosed females were used in the bioassay, there from their nestmates. was no difference in aggression received by non-nestmate and Our study using bioassays and chemical analyses shows the nestmate CHCs, suggesting that these wasps are unable to involvement of CHCs in nestmate discrimination in R. marginata, discriminate nestmate CHCs from CHCs of newly eclosed females. and also that antennation is the behavioural mechanism used to In contrast to adults, newly eclosed individuals can be accepted into perceive nestmate and non-nestmate cues. We have shown that foreign colonies (Arathi et al., 1997; Venkataraman & Gadagkar, newly eclosed females have a different CHC profile from that of 1995). This suggests that young individuals do not express the adult females, thereby providing an explanation for why young cues needed to differentiate nestmates and non-nestmates. Our females are accepted into foreign colonies. We also showed how results showing that R. marginata wasps do not discriminate nest- the age of an individual plays a role in deciding whether it will or mate CHCs of newly eclosed individuals are in accordance with will not be discriminated against in a foreign colony. Our work these observations. throws light on how chemical communication is linked with the The results of the chemical analysis of CHCs of newly eclosed evolution of colony integrity in social insects, and highlights the females showed that they have lower quantities of CHCs than adult importance of focusing on both chemical cues and the chemo- females and their overall CHC profiles are also largely different. This sensory mechanism used in perceiving the cues to get a more may be why they are not discriminated against when presented to comprehensive understanding of chemical communication. foreign colonies. Newly eclosed individuals probably require a few days to acquire the CHC profile found in adults, and hence can be Acknowledgments accepted into foreign colonies at this early stage. Acceptance of unrelated newly eclosed non-nestmates into a colony can poten- We thank the Department of Science and Technology, the tially have a negative impact on the colony’s fitness as resources Department of Biotechnology, the Council of Scientific and Indus- and altruistic behaviours are shared with an individual having low trial Research, and the Ministry of Environment and Forests, Gov- relatedness with other colony members. However, since newly ernment of India for financial assistance, Souvik Mandal, Centre for eclosed individuals are known to have poorly developed ovaries, Ecological Sciences, IISc, Bangalore, for help with recording videos, they pose no reproductive or social parasitism threat to a colony, Paromita Saha, CES, IISc, Bangalore, for help with collecting nests, and may instead serve to augment its workforce (Venkataraman & and Dr Sharoni Shafir (Editor) and three anonymous referees for Gadagkar, 1993). Also, since newly eclosed individuals mostly stay their helpful comments on the manuscript. The study was designed on the nest and generally do not go out to forage (Gadagkar, 2001), by R.G., A.R. and A.M., chemical and statistical analyses were done the probability of a newly eclosed individual attempting to enter a by A.M., and bioassays were done by A.M. and A.R. The paper was foreign colony should also be low. The primary function of CHCs is cowritten by A.M. and R.G., and R.G. supervised the overall work. All to prevent desiccation and provide protection from environmental experiments reported here comply with the current laws of the stress (Schal, Sevala, Young, & Bachmann, 1998), and the lower country in which they were performed. A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 121

Supplementary Material Mateo, J. M. (2004). Recognition systems and biological organization: the percep- tion component of social recognition. Annales Zoologici Fennici, 41,729e745. Maubourguet, N., Lesne, A., Changeux, J.-P., Maskous, U., & Faure, P. (2008). Supplementary material associated with this article is available, Behavioral sequence analysis reveals a novel role for b2* nicotinic receptors in in the online version, at http://dx.doi.org/10.1016/j.anbehav.2013. exploration. PLoS Computational Biology, 4, e1000229. ’ 11.017. Mitra, A., & Gadagkar, R. (in press). Dufour s gland and cuticle in the social wasp Ropalidia marginata contain the same hydrocarbons in similar proportions. Journal of Insect Science. References Mitra, A., & Gadagkar, R. (2012). Road to RoyaltyeTransition of potential queen to queen in the primitively eusocial wasp Ropalidia marginata. Ethology, 118, e Abbott, A. (1995). Sequence analysis: new methods for old ideas. Annual Review of 694 702. Sociology, 21,93e113. Morel, L., Vander Meer, R. K., Lavine, B. K., & Blum, M. S. (1988). Ontogeny of fl Aitchison, J. (1986). The statistical analysis of compositional data. London: Chapman nestmate recognition cues in the red carpenter ant (Camponotus oridanus): & Hall. behavioral and chemical evidence for the role of age and social experience. e Akino, T., Yamamura, K., Wakamura, S., & Yamaoka, R. (2004). Direct behavioral Behavioural Ecology and Sociobiology, 22,175 183. evidence for hydrocarbons as nestmate recognition cues in Formica japonica Neves, E. F., Andrade, L. H. C., Súarez, Y. R., Lima, S. M., & Antonialli-Junior, W. F. (: Formicidae). Applied Entomology and Zoology, 39, 381e387. (2012). Age related changes in the surface pheromones of the wasp Mis- Arathi, H. S., Shakarad, M., & Gadagkar, R. (1997). Factors affecting the acceptance of chocyttarus consimilis (Hymenoptera: ). Genetics and Molecular e alien conspecifics on nests of the primitively eusocial wasp, Ropalidia marginata Research, 11, 1891 1898. (Hymenoptera: Vespidae). Journal of Insect Behavior, 10, 343e353. Ozaki, M., Wada-Katsumata, A., Fujikawa, K., Iwasaki, M., Yokohari, F., Satoji, Y., et al. Bakeman, R., & Gottman, J. M. (1997). Observing interaction: An introduction to (2005). Ant nestmate and non-nestmate discrimination by a chemosensory e sequential analysis. New York: Cambridge University Press. sensillum. Science, 309,311 314. ’ Brandstaetter, A., & Kleineidam, C. J. (2011). Distributed representation of social Panek, L. M., Gamboa, G. J., & Espelie, K. E. (2001). The effect of a wasp s age on its fi odors indicates parallel processing in the antennal lobe of ants. Journal of cuticular hydrocarbon pro le and its tolerance by nestmate and non-nestmate fi e Neurophysiology, 106, 2437e2449. conspeci cs ( fuscatus, Hymenoptera: Vespidae). Ethology, 107,55 63. Brandstaetter, A., Rössler, W., & Kleineidam, C. J. (2011). Friends and foes from an Pfennig, D. W., Gamboa, G. J., Reeve, H. K., Reeve, J. S., & Ferguson, I. D. (1983). The ant brain’s point of view e neuronal correlates of colony odors in a social insect. mechanism of nestmate discrimination in social wasps (Polistes, Hymenoptera: e PLoS One, 6, e21383. Vespidae). Behavioral Ecology and Sociobiology, 13, 299 305. e Breiman, L. (2001). Random forests. Machine Learning, 45,15e32. Reyment, R. A. (1989). Compositional data analysis. Terra Nova, 1,29 34. de Bruyne, M., & Warr, C. G. (2006). Molecular and cellular organization of insect Sanchez, Maria Gabriela de Brito (2012). Taste perception in honey bees. In chemosensory neurons. Bioessays, 28,23e34. C. G. Galizia, D. Eisenhardt, & M. Giurfa (Eds.), Honeybee neurobiology and e Crozier, R. H., & Dix, M. W. (1979). Analysis of two genetic models for the innate behavior: a tribute to Randolf Menzel (pp. 253 267). Dordrecht: Springer. components of colony odor in social Hymenoptera. Behavioral Ecology and So- Schal, C., Sevala, V. L., Young, H. P., & Bachmann, J. A. S. (1998). Sites of synthesis and ciobiology, 4,217e224. transport pathways of insect hydrocarbons: cuticle and ovary as target tissues. e Cutler, D. R., Edwards, T. C., Beard, K. H., Cutler, A., Hess, K. T., Gibson, J., et al. (2007). American Zoologist, 38, 382 393. ’ Random forests for classification in ecology. Ecology, 88, 2783e2792. Slooten, E. (1994). Behavior of Hector s dolphin: classifying behavior by sequence e Dapporto, L., Fondelli, L., & Turillazzi, S. (2006). Nestmate recognition and identi- analysis. Journal of Mammalogy, 75, 956 964. fication of cuticular hydrocarbons composition in the swarm founding paper Sturgis, S. J., & Gordon, D. M. (2012). Nestmate recognition in ants (Hymenoptera: e wasp Ropalidia opifex. Biochemical Systematics and Ecology, 34,617e625. Formicidae): a review. Myrmecological News, 16,101 110. Fagen, R. M., & Young, D. Y. (1978). Temporal patterns of behaviour: durations, in- Tsutsui, N. D. (2004). Scents of self: the expression component of self/non-self e tervals, latencies, and sequences. In P. W. Colgan (Ed.), Quantitative ethology (pp. recognition systems. Annales Zoologici Fennici, 41,713 727. 79e114). New York: J. Wiley. Venkataraman, A. B., & Gadagkar, R. (1992). Kin recognition in a semi-natural fi Foitzik, S., Sturm, H., Pusch, K., d’Ettorre, P., & Heinze, J. (2007). Nestmate recog- context: behaviour towards foreign conspeci cs in the social wasp Ropalidia e nition and intraspecific chemical and genetic variation in Temnothorax ants. marginata (Lep.) (Hymenoptera: Vespidae). Insectes Sociaux, 39, 285 299. Animal Behaviour, 73, 999e1007. Venkataraman, A. B., & Gadagkar, R. (1993). Differential aggression towards alien fi e Gadagkar, R. (1985). Kin recognition in social insects and other e areviewof conspeci cs in a primitively eusocial wasp. Current Science, 64,601 603. fi recent findings and a consideration of their relevance for the theory of kin selec- Venkataraman, A. B., & Gadagkar, R. (1995). Age speci c acceptance of unrelated fi tion. Proceedings of the Indian Academy of Sciences (Animal Sciences), 94,587e621. conspeci cs on nests of the primitively eusocial wasp, Ropalidia marginata. e Gadagkar, R. (2001). The social biology of Ropalidia marginata: Toward understanding Proceedings of the Indian National Science Academy, B61, 299 314. the evolution of eusociality. Cambridge, MA: Harvard University Press. Venkataraman, A. B., Swarnalatha, V. B., Nair, P., & Gadagkar, R. (1988). The mech- Gamboa, G. J., Grudzien, T. A., Espelie, K. E., & Bura, E. A. (1996). Kin recognition anism of nestmate discrimination in the tropical social wasp Ropalidia mar- pheromones in social wasps: combining chemical and behavioural evidence. ginata and its implications for the evolution of eusociality. Behavioral Ecology e Animal Behaviour, 51, 625e629. and Sociobiology, 23, 271 279. Gamboa, G. J., Reeve, H. K., Ferguson, I. D., & Wacker, T. L. (1986). Nestmate Wilson, E. O. (1971). The insect societies. Cambridge, MA: Harvard University Press. ’ recognition in social wasps: the origin and acquisition of recognition odours. van Zweden, J. S., & d Ettorre, P. (2010). Nestmate recognition in social insects and Animal Behaviour, 34, 685e695. the role of hydrocarbons. In G. J. Blomquist, & A.-G. Bagneres (Eds.), Insect hy- e Gamboa, G. J., Reeve, H. K., & Pfennig, D. W. (1986). The evolution and ontogeny of drocarbons biology, biochemistry and chemical ecology (pp. 222 243). New York: nestmate recognition in social wasps. Annual Review of Entomology, 31,431e454. Cambridge University Press. ’ Howard, R. W., & Blomquist, G. J. (2005). Ecological, behavioral, and biochemical van Zweden, J. S., Dreier, S., & d Ettorre, P. (2009). Disentangling environmental and aspects of insect hydrocarbons. Annual Review of Entomology, 50,371e393. heritable nestmate recognition cues in a carpenter ant. Journal of Insect Physi- e Lahav, S., Soroker, V., & Hefetz, A. (1999). Direct behavioral evidence for hydrocar- ology, 55,159 164. bons as ant recognition discriminator. Naturwissenschaften, 86, 246e249. Lenoir, A., d’Ettorre, P., Errard, C., & Hefetz, A. (2001). Chemical ecology and social APPENDIX parasitism in ants. Annual Review of Entomology, 46,573e599. Lenoir, A., Fresneau, D., Errard, C., & Hefetz, A. (1999). Individuality and colonial identity in ants: the emergence of the social representation concept. In For analysis by random forest, peaks present in at least 70% of all C. Detrain, J.-L. Deneubourg, & J. M. Pasteels (Eds.), Information processing in samples were considered. The areas under each peak in an indi- e social insects (pp. 219 237). Basel: Birkhauser Verlag. vidual were divided by the total area under all peaks in that indi- Leonhardt, S. D., Brandstaetter, A. S., & Kleineidam, C. J. (2007). Reformation process of the neuronal template for nestmate-recognition cues in the carpenter ant vidual and thereby converted to percentage areas. The resulting Camponotus floridanus. Journal of Comparative Physiology A, 193, 993e1000. data were analysed by random forest for classification by the Liebert, A. E., & Starks, P. T. (2004). The action component of recognition systems: a package randomForest 4.6-6 in R 2.13.1 (Mitra & Gadagkar, 2012). focus on the response. Annales Zoologici Fennici, 41,747e764. Lorenzi, M. C., Bagnères, A. G., Clement, J.-L., & Turillazzi, S. (1997). Polistes biglumis The number of trees generated was 100 000 and the number of bimaculatus epicuticular hydrocarbons and nestmate recognition (Hymenop- variables to be used at each decision branch was set to six (rec- tera, Vespidae). Insectes Sociaux, 44,123e138. ommended to be close to the square root of the number of predictor Lorenzi, M. C., Sledge, M. F., Laiolo, P., Sturlini, E., & Turillazzi, S. (2004). Cuticular hydrocarbon dynamics in young adult Polistes dominulus (Hymenoptera: Ves- variables). The proximities among individuals on the random forest pidae) and the role of linear hydrocarbons in nestmate recognition systems. were used to get a measure of Euclidean distance (one minus Journal of Insect Physiology, 50, 935e941. proximity) between them on the random forest, and the distance Martin, S. J., Vitikainen, E., Helantera, H., & Drijfout, F. P. (2008). Chemical basis of nest-mate discrimination in the ant Formica exsecta. Proceedings of the Royal matrix was scaled on two dimensions to visualize the relative po- Society B, 275,1271e1278. sitions of individuals on the random forest. 122

Table A1 Peak numbers, identities, quantities (mean SD), ManneWhitney U statistics and P values for comparisons between adults and newly eclosed females for compounds found in gas-chromatographic analysis of cuticular hydrocarbons of R. marginata

Peak Identity of compound Quantity (pg)

Adult Newly eclosed UP

1 Heneicosane 0.3820.312 4.9781.812 110 <0.0001* 2 11-Methylheneicosane 4.6622.143 24.98014.166 110 <0.0001* 3 Tricosane 0.8280.846 21.9378.854 110 <0.0001*

4 11-Methyltricosane 0.3200.316 1.7430.861 104 <0.0001* 113 (2014) 88 Behaviour Animal / al. et Mitra A. 5 Pentacosane 1.5730.988 10.8055.014 110 <0.0001* 6 Mixture of 11- and 13-methylpentacosane 1.3461.548 0.4830.215 77 0.132 7 Heptacosane 3.7522.411 9.7905.941 89 0.008 8 Mixture of 11- and 13-methylheptacosane 2.7821.919 0.3200.254 110 <0.0001* 9 3-Methylheptacosane 4.5473.472 0.6980.313 102 <0.001 9a Octacosane 1.1560.619 0.6100.284 82 0.031 10 Mixture of 12-, 14-, 16- and 18-methyloctacosane 2.4671.607 0.4600.288 107 <0.0001* 11 4-Methyloctacosane 2.856 3.002 0.092 0.118 110 <0.0001* 0.022 1213 Nonacosane Mixture of 11-, 13- and 15-methylnonacosane 20.921 15.764 74.91455.972 10.514 5.764 13.2627.154 84 100 <0.001 14 5-Methylnonacosane 17.83535.179 1.0310.407 110 <0.0001* 15 11-, 15-Dimethylnonacosane 7.0764.557 0.7860.707 107 <0.0001* 16 3-Methylnonacosane 21.96018.340 2.4551.041 101 <0.001 18 8-Methyltriacontane 7.3824.215 0.7520.611 109 <0.0001* 19 14-, 16-Dimethyltriacontane 7.8064.686 1.0380.638 110 <0.0001*

20 2-Methyltriacontane 1.4490.917 0.8691.181 84 0.022 e 20a Hentriacontane 1.7200.897 3.1391.787 86 0.015 124 21 Mixture of 11-, 13- and 15-methylhentriacontane 105.05458.566 19.6657.468 110 <0.0001* 22 Mixture of 7- and 9-methylhentriacontane 5.1452.502 3.3623.468 83 0.026 23 Mixture of 11-, 17- and 13-, 17-dimethylhentriacontane 21.37012.013 6.0144.434 101 <0.001 24 3-Methylhentriacontane 4.0723.071 0.4720.212 110 <0.0001* 26 14-Methyldotriacontane 3.6002.022 0.4090.237 110 <0.0001* 27 8-Methyldotriacontane 10.8936.898 1.2100.604 110 <0.0001* 28 14-, 18-Dimethyldotraicontane 3.5862.281 1.6431.083 92 0.004 6.067 6.0931.890 86 0.015 29 Mixture of 13-, 15- and 17-methyltritriacontane 11.266 30 13-, 19-Dimethyltritriacontane 9.2695.330 6.4504.686 74 0.197

Compounds are those identified previously by gas chromatographyemass spectrometry (Mitra & Gadagkar, in press). Significant P values are highlighted in bold. Asterisk signifies P values that remain significant after Bonferroni correctioncritical (P ¼ 0.002). A. Mitra et al. / Animal Behaviour 88 (2014) 113e124 123

(a)

0.2

0

−0.2

Dimension 2 V1006 V1028 −0.4 V1029 V1069 V1082 V1109 −0.6 −0.4 −0.2 0 0.2 0.4 Dimension 1

0.1 Adult (b) Newly eclosed 0.08

0.06

0.04

Dimension 2 0.02

0

−0.02

−0.4 −0.2 0 0.2 0.4 Dimension 1

Figure A1. (a) Colony differences in CHC profiles: proximities among R. marginata adult females (N ¼ 41) from six colonies on random forest (100 000 trees, number of randomly selected variables used at each decision branch ¼ 6) that differentiates individuals using relative abundances of 32 hydrocarbon peaks from the cuticle. Nest numbers are given in the key. (b) Comparison of CHC profiles of newly eclosed and older females: proximities among newly eclosed (N ¼ 10) and older adult females (N ¼ 11) of R. marginata on random forest (100 000 trees, number of randomly selected variables used at each decision branch ¼ 6) that differentiates individuals using relative abundances of 30 hydrocarbon peaks from the cuticle. In both (a) and (b) proximity distances are scaled on two dimensions. 124 A. Mitra et al. / Animal Behaviour 88 (2014) 113e124

13 (a) 21 400 14

350 15 300 16 22 250

200 12 26

150 9a 10 20a 27 100 3 9 19 28a 4 7 18 20 23 29 2 6 8 11 28 30 50 5 24

2022.5 25 27.5 30 32.5 35 pA (b) 3

200 2 175

150 12 21 5 125 18 100 22 24 75 1 9a 13 14 7 4 19 20a 6 9 23 27 50 16 29 28 30 25 20 22.525 27.5 30 32.5 35 Retention time (min)

Figure A2. Flame ionization detection gas chromatograms of cuticular hydrocarbons of (a) an adult female and (b) a newly eclosed female of R. marginata.

20

15

10

5

0

−5

−10 Scores on discriminant function

−15

−20 Adult Newly eclosed

Figure A3. Adult (N ¼ 11) and newly eclosed females (N ¼ 10) of R. marginata plotted on a discriminant function that differentiates them using their cuticular hydrocarbon profiles.