Journal of Chemical Ecology

, Volume 36, Issue 4, pp 424–431

Regulation of Reproduction in the Primitively Eusocial Wasp Ropalidia marginata: on the Trail of the Queen Pheromone


  • Anindita Bhadra
    • Centre for Ecological SciencesIndian Institute of Science
  • Aniruddha Mitra
    • Centre for Ecological SciencesIndian Institute of Science
  • Sujata A. Deshpande
    • Centre for Ecological SciencesIndian Institute of Science
  • Kannepalli Chandrasekhar
    • Centre for Ecological SciencesIndian Institute of Science
  • Dattatraya G. Naik
    • Agharkar Research Institute
  • Abraham Hefetz
    • Department of ZoologyTel Aviv University
    • Centre for Ecological SciencesIndian Institute of Science
    • Evolutionary and Organismal Biology UnitJawaharlal Nehru Centre for Advanced Scientific Research

DOI: 10.1007/s10886-010-9770-x

Cite this article as:
Bhadra, A., Mitra, A., Deshpande, S.A. et al. J Chem Ecol (2010) 36: 424. doi:10.1007/s10886-010-9770-x


Queens and workers are not morphologically differentiated in the primitively eusocial wasp, Ropalidia marginata. Upon removal of the queen, one of the workers becomes extremely aggressive, but immediately drops her aggression if the queen is returned. If the queen is not returned, this hyper-aggressive individual, the potential queen (PQ), will develop her ovaries, lose her hyper-aggression, and become the next colony queen. Because of the non-aggressive nature of the queen, and because the PQ loses her aggression by the time she starts laying eggs, we hypothesized that regulation of worker reproduction in R. marginata is mediated by pheromones rather than by physical aggression. Based on the immediate loss of aggression by the PQ upon return of the queen, we developed a bioassay to test whether the queen’s Dufour’s gland is, at least, one of the sources of the queen pheromone. Macerates of the queen’s Dufour’s gland, but not that of the worker’s Dufour’s gland, mimic the queen in making the PQ decrease her aggression. We also correctly distinguished queens and workers of R. marginata nests by a discriminant function analysis based on the chemical composition of their respective Dufour’s glands.

Key Words

Ropalidia marginataQueen pheromonePotential queenDufour’s glandReproductive monopoly


The differentiation of adult colony members into fertile queens and functionally sterile workers is an important feature of insect societies of ants, bees, and wasps. In these highly eusocial insect species, which generally have large colony sizes (>>100 individuals), queens and workers are morphologically differentiated, with queens maintaining their reproductive monopoly with pheromones. On the other hand, in primitively eusocial species, colony sizes are usually small (<100 individuals), and queens and workers are morphologically indistinguishable, with queens believed to suppress worker reproduction by physical aggression (often referred to as dominance behavior) (Wilson 1971). Ropalidia marginata, the subject of this study, is classified as a primitively eusocial species, because of the absence of morphological differentiation between queens and workers. However, the queen in R. marginata is a strikingly non-aggressive and non-interacting individual that, nevertheless, maintains complete reproductive monopoly. Since the queen does not use physical aggression to suppress worker reproduction (Gadagkar 2001), we hypothesized that R. marginata queens may use pheromones for this purpose.

If the queen is lost or removed from a colony, one of the workers becomes highly aggressive within a few minutes; if the queen is not returned, this individual develops ovaries and becomes the new queen of the colony within a few days. We have designated this individual as the potential queen (PQ; Premnath et al. 1996). However, if the queen is returned to the colony within a day of removal, the PQ decreases her aggression and swiftly returns to being a typical worker (Premnath et al. 1996; Kardile and Gadagkar 2002; Sumana and Gadagkar 2003; Deshpande et al. 2006). This suggests that the PQ perceives the absence of the queen within a few minutes. When workers are separated from their queen by a wire mesh screen, one of the workers establishes itself as the PQ and will lay eggs if the wire mesh is not removed (Sumana et al. 2007/2008). How then do workers perceive the presence or absence of their queen? One possibility is through physical interactions with the queen. However, we have shown that the rates of interaction of the PQ with the queen (both direct and indirect, via interactions with other workers) are not frequent enough to explain the rapidity with which the queen’s absence is perceived (Bhadra et al. 2007). Another possibility is that the queen applies her pheromone to the nest surface. This possibility is supported by the observation that the queen, but not the workers, frequently rubs the ventral side of her abdomen on the nest surface (Bhadra et al. 2007). Because the Dufour’s gland, which opens into the tip of the abdomen, is believed to be a source of the queen signal in honeybees (Katzav-Gozansky et al. 1997; Katzav-Gozansky et al. 2002; Dor et al. 2005), we examined the possibility that the queen pheromone of R. marginata is produced in the Dufour’s gland.

Here, we demonstrated that a crude macerate of the queen’s Dufour’s gland mimics her presence in a bioassay, and that queens and workers can be classified correctly by a discriminant function analysis that uses the hydrocarbon profiles of their respective Dufour’s glands.

Methods and Material

We collected post-emergence nests of R. marginata from various localities in Bangalore (13° 00’ N and 77° 32’ E), India, and transplanted them to the Vespiary at the Centre for Ecological Sciences, Indian Institute of Science, Bangalore. The nests were maintained in closed cages made of wood and fine mesh, and provided with food, water, and building material, ad libitum. All adults were uniquely color-coded with small spots of Testors® enamel paints (Gadagkar 2001). The queen was identified by egg-laying behavior prior to beginning the experiment.


We developed a bioassay for the queen pheromone based on the observation that the PQ immediately reduces her aggression when the queen is returned to the colony. Thus, we tested whether applying crude macerate of the Dufour’s gland of the queen but not of the workers mimics the return of the queen by a similar reduction in the aggression of the PQ. The bioassay consisted of three observation sessions lasting 36 min. each. Each session consisted of six observation periods of 5 min. each, interspersed with a 1 min. break. After observing the normal queen-right colony in the first session (Queen-right Session), the queen and a randomly chosen worker were removed, and the queen-less colony was observed in the second session (Queen-less Session). Before beginning observations in the third session (Treatment Session), the contents of the Dufour’s gland of the removed queen or a worker (each crushed in 30 µl of Ringer’s solution) or 30 µl of Ringer’s solution, were applied to the nest with a micropipette. The choice of treatment was decided by drawing random numbers and was unknown to the observer. A total of 25 nests were used in the experiment (8 queen extract; 9 worker extract, and 8 Ringer’s solution). All statistical analyses were performed using the software package STATISTICA 7.

We used Ringer’s solution because we observed that organic solvents, such as pentane or acetone, had an adverse effect when applied to the nest, with the wasps becoming agitated or even dying after contact with these solvents. Later, we validated the use of Ringer’s solution by chemical analysis of crude macerates of glands crushed in Ringer’s solution and found a similar chemical profile to that obtained by extracting glands in pentane.

Chemical Analysis

Preparation of Sample

Seven additional colonies were used for chemical analysis. The queens of each colony and a total of 18 workers (2 to 5 per colony, depending on colony size) had their Dufour’s glands dissected in distilled water under a stereomicroscope, after gently pulling out the sting. The gland was placed in a vial (chilled with dry ice and acetone), containing 5 µl pentane, and crushed with a needle.

Gas Chromatographic-Mass Spectrometric Analysis

After adding another 5 μl pentane to the vial containing the crushed gland, 2 μl extract was injected into an Agilent 6890 gas chromatograph coupled with an Agilent 5973 mass selective detector. A fused-silica capillary column, coated with 100% dimethyl polysiloxane (Agilent HP-1, 60 m × 250 µm × 0.25 µm), was used for the gas chromatography-mass spectrometry (GC-MS) analyses. The injector port and transfer line were set at 250°C, and helium was the carrier gas. The column oven was programmed from 100–275°C at 7°C min−1, held for 5 min, and then to 280°C at 5°C min−1 held for 33 min. Analyses were performed in split mode (split ratio 10:1). Straight-chain and methyl-branched hydrocarbons were identified from their characteristic mass spectral fragmentation patterns, produced by electron impact ionization at 70 eV. The compounds identified were either straight-chain or branched hyrdrocarbons, the spectra of which are very clear and can be interpreted unambiguously. For example, branching at carbon 11 (11-methylHC) gives a distinct peak fragment at m/z 168 with a complementary fragment depending on the chain length of the compound. Likewise, 14-methylHC has a pronounced peak fragment at m/z 196, etc. For dimethyl compounds, we have two branching points and therefore 4 complementary peak fragments, the m/z of which are characteristic to the compound. The spectra of most of these compounds have been previously published with a list of the ions characteristic to the branching points. A blank run with 2 µl pentane confirmed that none of the detected compounds was present as impurities in the solvent.

Statistical Analyses

Both multivariate and univariate analyses were carried out. To reduce the number of peaks used in the multivariate analyses, only those present in at least 70% of all individuals were considered. Before applying any multivariate test, the areas under each peak were transformed by √(X = + 0.5) to eliminate zero values (Zar 1999). Before applying the transformation, we confirmed that group variances were proportional to the means. The transformed data were subjected to the further transformation:
$$ {Z_{p.j}} = \ln \left[ {\frac{{{A_{p.j}}}}{{g\left( {{A_j}} \right)}}} \right] $$
where, Ap.j is the area of the peak p for individual j, g(Aj) is the geometric mean of all peaks considered for analysis in individual j, and Zp.j is the transformed area of peak p for individual j (Reyment 1989). These transformed areas then were subjected to principal components (PC) analysis, followed by stepwise discriminant function analysis. All peaks had communality <0.8 on PC1 and were considered subsequently for discriminant analysis. The significance of Wilk’s λ (for canonical discriminant function), and the percentage of correct assignments (for classification discriminant functions) were used to evaluate the validity of the discriminant functions. For univariate analysis, the areas under each peak were transformed into percentages of total area under all peaks for each individual, and Mann-Whitney U tests were carried out to determine whether queens and workers differed for each compound.
An index of chemical diversity was calculated for each individual, using the Shannon Weiner Index of species diversity (Shannon 1948; Krebs 1989), by the following formula:
$$ {H^\prime} = - \left( {\sum\limits_{i = 1}^S {p_i \ln p_i } } \right) $$
where H’ is the index of chemical diversity, S is the total number of peaks in the respective individual, pi is the relative abundance of each peak (i.e., the area under the respective peak divided by total area under all peaks in that individual). A Mann-Whitney U test was carried out to test whether queens and workers differed with respect to indices of chemical diversity.

Squared Euclidean distances were calculated to estimate the chemical distances between all possible pairs of individuals, using the standardized percentages of area under each peak (percentages calculated out of total area under all peaks as mentioned earlier). Z scores for the matrices were used to standardize the percentages, and the differences between groups were analyzed by Mann-Whitney U tests. Statistical analyses used the software packages StatistiXL, version 1.7 and Mystat 12.

Validation of Bioassay

Because we carried out the bioassay using Ringer’s solution, and the chemicals we identified from the Dufour’s gland are insoluble in Ringer’s, we determined whether the Ringer’s solution macerates used in the bioassays actually contained the Dufour’s gland compounds or not. Each gland was crushed in 30 µl Ringer’s solution, as in the bioassay, and the contents transferred to a clean glass vial. The Ringer’s solution was evaporated from the vial in an oven at 30°C, the dry vial chilled, and 10 µl pentane added to it. Two microliters of this extract were analyzed by GC using an Agilent 6890 gas chromatograph equipped with a flame ionization detector and a fused silica capillary column coated with 5% phenyl methyl siloxane (Agilent HP-5, 30 m × 320 μm × 0.25 μm). General GC conditions were the same as those in the GC-MS analyses. The Ringer’s solution was analyzed similarly to rule out any organic compounds that might be present as impurities. Peaks were identified by analyzing a sample of 21 glands in 20 µl pentane by both GC-MS and GC.

Further, to rule out the possible involvement of other compounds in the Dufour’s gland-Ringer’s solution that are insoluble in pentane, we analyzed glands crushed in acetone. GC conditions were identical to those used in the pentane extracts, and blank runs were also carried out to exclude solvent-based impurities.


The Bioassay

In all nests, a PQ was obvious after queen removal, due to a significant increase in its aggression from the Queen-right Session to the Queen-less Session (Wilcoxon matched pairs signed-ranks test: P = 0.012 for queen-gland and Ringer’s-solution treatments, and P = 0.007 for worker-gland treatment). There was a significant reduction in the PQ’s aggression from the Queen-less Session to the Treatment Session (Fig. 1) in nests treated with the queen’s Dufour’s gland macerate (P = 0.012). However, in nests treated with worker-gland macerate or Ringer’s solution, the PQ did not show any change in aggression from the Queen-less Session to the Treatment Session (P = 0.593 for worker and P = 0.401 for Ringer’s).
Fig. 1

Mean and standard deviation of the frequency per hour of dominance behavior exhibited by the potential queen from Ropalidia marginata nests in the three sessions of the bioassay (Queen-right, Queen-less and Treatment; N = 8, 9, and 8, respectively) when the nest was exposed to queen Dufour’s gland macerate, worker Dufour’s gland macerate or Ringer’s solution (control). Comparisons are by Wilcoxon matched pairs signed-ranks test among the three sessions within each treatment. Different letters denote significant difference among bars (P < 0.05)

Chemical Analysis

Of 18 workers, two were eliminated from statistical analyses because they had poorly developed glands, and most compounds in their glands were below the detectable limit of the GC-MS analysis. All queens had well developed glands. The Dufour’s gland contained a series of linear, monomethyl and dimethyl branched alkanes with 21 to 33 carbon atoms in the main chains. We did not find any unsaturated compounds. Considering all individuals, we found about 30 different compounds (Fig. 2 and Table 1). We did not find any compound or set of compounds present exclusively in queens or in workers.
Fig. 2

Total ion mass chromatograms of a Ropalidia marginata worker and a queen from the same colony. Asterisk signifies contaminant present in blank run. Peak numbers correspond to compounds identified in Table 1

Table 1

Peak numbers, retention times, identity and average percent areas and standard deviations of dufour’s gland compounds of queens and workers in Ropalidia marginata

Peak #

Retention Time (min)

Identity of compound

% area of each compound mean ± standard deviation






0.16 ± 0.46

0.0008 ± 0.002




0.6 ± 0.99

0.04 ± 0.08




0.54 ± 1.51

0.03 ± 0.07




0.37 ± 0.774

0.007 ± 0.012




1.85 ± 3.06

1.12 ± 1.96



Mixture of 11- and 13-methylpentacosane

3.99 ± 3.58

3.28 ± 4.40




0.63 ± 1.14

1.63 ± 2.78



Mixture of 11- and 13-methylheptacosane

1.77 ± 3.08

1.90 ± 1.99




2.19 ± 2.754

1.062 ± 2.069



Mixture of 12-, 14-, 16- and 18-methyloctacosane

0.341 ± 0.59

0.25 ± 0.33




0.56 ± 1.16

0.27 ± 0.53




1.63 ± 3.10

6.74 ± 10.40



Mixture of 11-, 13- and 15-methylnonacosane

41.87 ± 30.06

16.95 ± 14.97




3.24 ± 4.81

7.30 ± 9.29



11-, 15-Dimethylnonacosane

7.814 ± 9.753

4.54 ± 4.09




2.13 ± 2.24

16.43 ± 8.95




0.25 ± 0.91

0.10 ± 0.20




2.70 ± 4.23

0.73 ± 1.20



14-, 16-Dimethyltriacontane

1.92 ± 2.26

1.02 ± 1.95




0.02 ± 0.06

0.05 ± 0.09



Mixture of 11-, 13- and 15-methylhentriacontane

12.03 ± 17.53

14.98 ± 17.40



Mixture of 7- and 9-methylhentriacontane

0.60 ± 1.68

1.58 ± 2.15



Mixture of 11-, 17- and 13-, 17-dimethylhentriacontane

11.10 ± 11.09

9.99 ± 8.40




0.02 ± 0.07

1.39 ± 2.37



Mixture of 5-, 21-, 5-,19- and 5-, 17-dimethyldotriacontane

0.02 ± 0.07

1.11 ± 2.04




0.17 ± 0.32

0.66 ± 0.76




1.84 ± 3.52

0.38 ± 0.7



14-, 18-Dimethyldotriacontane

0.04 ± 0.12

0.63 ± 1.45



Mixture of 13-, 15- and 17-methyltritriacontane

0.43 ± 1.42

5.31 ± 7.22



13-, 19-Dimethyltritriacontane

0.08 ± 0.15

0.55 ± 1.03

We found that: 1) queens had a significantly greater percentage peak area than workers for three compounds (peaks #16, #24, and #29; Mann-Whitney U test, U = 96, P = 0.003 for peak #16; U = 91, P = 0.009 for peak #24; and U = 96.5, P = 0.002 for peak #29); 2) a significantly lower percentage peak area than workers for one compound (peak #13; U = 83, P = 0.038; and 3) they were indistinguishable from workers in percentage peak areas for the remaining compounds (P > 0.05). When the queens and workers of each colony were examined separately, the differences based on these compounds were not consistent. There was no compound that was consistently greater or lower in any queen compared to all the workers tested from her colony. The differences between queens and workers became significant only when all queens were pooled together and compared with all workers pooled together. However, when we added up the percentage areas under peaks 16, 24, and 29, for which queens had higher percentage areas than workers, queens of each colony were greater than workers from their respective colonies, except for one colony, which was a small colony with only three workers, out of which two were newly eclosed. Queens and workers did not differ with respect to total area under all peaks (U = 59, P = 0.871) or total number of peaks present (U = 64.5, P = 0.579).

We differentiated all queens and workers by a stepwise discriminant analysis (Wilk’s λ = 0.214, P < 0.001, classification analysis: 100% correct classification; Fig. 3). Peak numbers 8, 13, 14, 15, 16, 21, and 22 were selected in the discriminant analysis. Analysis of index of chemical diversity showed that queens had higher indices of diversity than workers (Mann Whitney U = 82, P = 0.044), but the difference was not universal. There were two colonies in which the queen had a lower diversity index than a worker from the respective colony. Both of these colonies were small, one had three workers, out of which two were newly eclosed (see above), and the other with four workers out of which three were newly eclosed. Analysis of chemical distances showed that queen-worker distances were significantly greater than worker-worker distances (Mann-Whitney U = 8237, P = 0.001). Queen-queen distances were, however, not different from queen-worker distances (U = 1266, P = 0.583), but were greater than worker-worker distances (U = 1660, P = 0.01).
Fig. 3

Scores on discriminant function analysis for gas chromatographic analyses of Dufour’s glands of Ropalidia marginata queens and workers (nqueens = 7, nworkers = 16, Wilk’s λ = 0.214, P < 0.001, classification analysis: 100% correct classification)

Validation of Bioassay

Glands crushed in Ringer’s solution gave similar GC-MS profiles to those obtained using pentane extracts of glands (Supplementary material S1 and S2). GC analysis of glands crushed in acetone also gave profiles similar to those obtained using pentane (Supplementary material S3).


We have shown that applying an extract of the queen’s Dufour’s gland onto the nest surface results in the PQ behaving as if the queen has returned to the nest. We also have shown that queens and workers can be classified correctly based on the contents of their Dufour’s glands. These results are consistent with the hypothesis that, in R. marginata, the workers (at least the PQ) perceive the presence of their queen through contents of the Dufour’s gland that she applies to the nest surface. It should be noted that our ability to classify queens and workers correctly required that we consider the effect of several compounds simultaneously with no single compound being adequate for this purpose. We do not rule out the role of other compounds or indeed of other mechanisms contributing to queen recognition by colony members. We do not have direct evidence that the compounds tested here are responsible for the lack of ovarian development of the workers in the presence of the queen. However, if workers use an honest signal of their queen’s fertility to refrain from reproduction, as suggested by Keller and Nonacs (1993), the compounds we identified also may be involved in reproductive regulation in R. marginata, thus permitting the queen to maintain reproductive monopoly in spite of her docility.

Analysis of chemical diversity showed that queens, in general, had higher indices than workers, implying that queens have either greater numbers of compounds, or a more even distribution (i.e., less variable ratios, within individuals) of compounds. Since we did not find any difference between queens and workers with respect to numbers of compounds present, it is likely that the distribution of relative proportions of compounds is more even in the case of queens and less even in the case of workers. The two colonies in which the queen did not have the highest index value were small colonies with only one mature worker, with the rest of the workers being newly eclosed. Chemical profiles of old, well established queens could be different from those of newly established queens, as queens are known to be behaviorally dominant during the nest initiation period or pre-emergence phase (Gadagkar 2001). Also, in this species, the old queen sometimes gets overthrown with a worker becoming the PQ and ultimately taking over as the new queen of the colony, a phenomenon known as queen turnover in natural colonies (Gadagkar 2001). It is possible that queens that are on the verge of being overthrown have a declining queen signal and are chemically different from normal queens. Since we do not have any information on the history of each colony, and also did not measure the ovarian development of the queens used in our GC-MS analyses, we cannot make any conclusive statement as to why these colonies were different from the others. Chemical distances show that queen-worker distances are higher than worker-worker distances, indicating that queens and workers are more different from each other than workers are from other workers. However, queen-queen distances are different from worker-worker, but not queen-worker distances, implying that more variability in peak area (relative amount) occurs among queens (between individuals).

Suppression of the PQ’s dominance is not complete after applying the queen’s Dufour’s gland extract to the nest, although the level of dominance is significantly higher than that observed in the queen-right condition. It should be noted that in earlier studies (Sumana and Gadagkar 2003) we saw that, after keeping the queen away from her colony for a day followed by queen reintroduction on the next day, the dominance of the PQ remained significantly higher than in the normal queen-right condition, suggesting that the PQ takes more than a day to lower her dominance to the normal level. Also, in the present study, we observed that the effect of the queen’s Dufour’s gland macerate wanes with time and the PQ again starts to increase her aggression. Overall, we conclude that the queen’s Dufour’s gland chemicals are at least one of the components of the queen signal and can act as a proxy for the queen herself.

Our results add to the growing evidence for chemical communication between queens and workers in primitively eusocial wasps. For example, Downing (1991) demonstrated that the contents of the Dufour’s gland in Polistes fuscatus act as an egg-marking pheromone. Sledge et al. (2001) showed that queens and workers in Polistes dominulus differ in their cuticular hydrocarbon signatures. Dapporto et al. (2007) suggested that the contents of the Van der Vecht’s gland may function as a queen pheromone in Polistes gallicus. Unlike Polistes queens, that are also aggressive in addition to producing a chemical cue signaling their presence, R. marginata queens appear to rely more heavily on pheromones for communication with workers, as they are non-aggressive and interact directly with other members of their colonies relatively infrequently. Our results are similar to the situation in honeybees in which queens are non-aggressive and where there is good evidence for the contents of Dufour’s glands as signals of fertility (Katzav-Gozansky et al. 2002).


We thank the Department of Science and Technology, the Department of Biotechnology, the Council for Scientific and Industrial Research and the Ministry of Environment and Forests, Government of India for financial assistance, and Robin Crewe for helpful comments and encouragement. We also thank Anjali Rajasekharan, Pooja Muralidharan and Gautam Pramanik, Centre for Ecological Sciences, Indian Institute of Science, Bangalore, for technical assistance with the GC-MS analyses and, Prasanta Das, Department of Inorganic and Physical Chemistry, Indian Institute of Science, Bangalore, for providing useful suggestions regarding interpretation of mass spectra. AB carried out the behavioral observations. SAD, KC, and AM did the dissections and prepared the extracts. AM carried out the chemical analysis, with help and guidance from DGN and AH. The paper was co-written by AB, AM, and RG, and RG supervised the overall work. All experiments reported here comply with the current laws of the country in which they were performed.

Supplementary material

10886_2010_9770_MOESM1_ESM.pdf (34 kb)
Supplementary material S1Flame ionization detection gas chromatogram of a Ropalidia marginata Dufour’s gland crushed in Ringer’s solution, evaporated to dryness and resuspended in pentane. Asterisk signifies contaminant from Ringer’s solution. Identities of peak numbers are given in Table 1. Compare the chromatogram of this extraction method with that obtained in S2 (PDF 34 kb)
10886_2010_9770_MOESM2_ESM.pdf (189 kb)
Supplementary material S2Comparison of gas chromatograms of the same sample of a Ropalidia marginata Dufour’s gland extracted in pentane and analyzed by flame ionization detection (FID) and mass spectrometry (total ion chromatogram; TIC). Identities of peak numbers are given in Table 1 (PDF 188 kb)
10886_2010_9770_MOESM3_ESM.pdf (19 kb)
Supplementary material S3Flame ionization detection gas chromatogram of a Ropalidia marginata Dufour’s gland crushed and extracted in acetone. Identities of peak numbers are given in Table 1. Compare the chromatogram of this extraction method with that obtained in S2 (PDF 19 kb)

Copyright information

© Springer Science+Business Media, LLC 2010