Morphological caste differences in the neotropicalswarm-founding Polistinae wasps:Agelaia m. multipicta and A. p. pallipes(Hymenoptera Vespidae)

F.B. NOLL 1, D. SIMÕES 2 and R. ZUCCHI 1

1 Departamento de Biologia, Faculdade de Filosofia Ciências e Letras de RibeirãoPreto, Universidade de São Paulo, 14040-901 Ribeirão Preto (SP), Brasil(E-mail: fernnoll@usp.br)2 Departamento de Zoologia e Centro de Estudos de Insetos Sociais, Instituto deBiociências, UNESP, 13506-900 Rio Claro (SP), Brasil

Received 31 October 1996, accepted 22 September 1997

Distinct size dimorphism was found between queens (inseminated egg-layers) and workers (uninseminated non egg-layers) sampled from mature colo-nies of Agelaia pallipes and A. multipicta. In both species, among 22 measuredbody parts queens were larger than workers in 17 characters. The Mahalanobis’distance (D2) between castes was 122.74 in A. pallipes and 110.99 in A. multipic-ta, giving clear-cut evidence of caste separation, a conclusion which is strength-ened by the absence of intermediate females. Comparisons with A. vicina and A.flavipennis demonstrated that morphological castes attributes are less developedin A. pallipes and A. multipicta.

KEY WORDS: Vespidae, Polistinae, Agelaia multipicta, Agelaia pallipes, caste differ-ences, multivariate analysis.

Introduction . . . . . . . . . . . . . . . . . 362Material and methods . . . . . . . . . . . . . . 362Results . . . . . . . . . . . . . . . . . . 363

Nest composition . . . . . . . . . . . . . . 363Ovary development and insemination . . . . . . . . . 365Queen-worker differences in relation to morphometry, hamulus number

and morphology . . . . . . . . . . . . . 366Discussion . . . . . . . . . . . . . . . . . 370Acknowledgments . . . . . . . . . . . . . . . 371References . . . . . . . . . . . . . . . . . 371

Ethology Ecology & Evolution 9: 361-372, 1997

362 F.B. Noll, D. Simões and R. Zucchi

INTRODUCTION

Caste differentiation in social insects, whether behavioural, physiological ormorphological, is a means of evaluating the level of sociality in a given taxon(RICHARDS 1978, JEANNE 1991). In Vespidae, castes are more conspicuous in theVespinae (BLACKITH 1958). Although caste differences in the Polistinae are muchless conspicuous, such traits are probably more complex here than in Vespinae.Morphological differences between queens and workers have been detected inBelonogaster griseus (PARDI & MARINO PICCIOLI 1981), Ropalidia montana (YAMANE etal. 1983), Ropalidia ignobilis (WENZEL 1992). The tribe Epiponini of the polistinewasps (CARPENTER 1993) is remarkable for their polygyny and reputed minor differ-entiation between castes. Preliminary overall studies (RICHARDS & RICHARDS 1951,RICHARDS 1978 and references therein) demonstrated a wide range of queen-workerdimorphism among different taxa. In previous papers, including those of ourseries, some phenomena demonstrating the diversity of morphological caste dis-tinction were: distinct queen dwarfism in Apoica flavissima and Polybia dimidiata(SHIMA et al. 1994, 1996a, respectively), queens larger than workers in Protonectari-na sylveirae (SHIMA et al. 1966b), and queens and workers morphologically verysimilar in Pseudopolybia vespiceps (S.N. SHIMA et al. in prep.) and Protopolybia exi-gua (NOLL et al. 1997). But, as already stated by RICHARDS (1978) the most conspic-uous caste differences were found in some Agelaia (= Stelopolybia): A. flavipennis(EVANS & WEST-EBERHARD 1970), A. areata (JEANNE & FAGEN 1974), A. vicina (VON

IHERING 1903, SAKAGAMI et al. 1996). This paper deals with caste differences andrelated aspects in Agelaia pallipes pallipes (Olivier) and Agelaia multipicta multipicta(Haliday).

MATERIAL AND METHODS

One mature colony for each species was used (see Nest composition). In Agelaia pallipespallipes (Olivier) the analysed sample of workers (n = 100) and queens (n = 100) was collectedin Ribeirão Preto (South eastern Brazil) by introducing CO2 into the nest. Using the same col-lecting procedure 100 workers and 88 queens (total population) of Agelaia multipicta multipic-ta were randomly chosen for examination.

In order to detect morphological differences between the castes, lengths of the follow-ing 21 external body parts (Fig. 1) were measured and the number of hamuli was counted foreach specimen under a binocular microscope: head width (HW), head length (HL), maximumand minimum interorbital distances (IDx and IDm, respectively), gena width (GW), eye width(EW), pronotum width (PW), length and width of mesoscutum (MSL and MSW, respectively),scutellum length (MTL), metanotum length (MNL), mesosomum height (MSH), alitrunklength (AL), propodeum length (PL), length of gastral tergite I (T1L), basal and apical heightsof T1 (T1BH and T1AH), basal and apical widths of tergite II (T2BW and T2AW), length of T2

(T2L) and partial length of the forewing (WL). The numerical data were statistically analyzedin relation to the ovarian and spermathecal states. The statistical analysis including canonicaldiscriminant analysis (CDA: RAO 1973) were performed with the SAS Program Package for PCcomputers.

363Caste differences in polistine wasps

RESULTS

Nest composition

A. pallipes

A. pallipes pallipes occurs from Costa Rica to Argentina (Misiones) and Para-guay (RICHARDS 1978). The nest is constructed in several kinds of cavities, such ashollows in the ground (which include the abandoned chambers of Atta nests), inarmadillo tunnels, among tree roots, hollow logs and several man-made construc-tions. In commercial apiaries the wasps often invade abandoned hives. The nest isalways devoid of an envelope and if the available cavity is narrow the verticalcombs composed of horizontal cells form a single row of concentric hemispheres,as described by RICHARDS (1978). A similar pattern of nest construction (concentricarcuate combs) is also observed in A. angulata (WENZEL 1991). Larger cavities, how-ever, may produce the kind of nest shown in Fig. 2A which was found (February1975) hanging from the roof of an abandoned honeybee hive. There were more

Fig. 1. — Locations of the measured characters. HW: head width, HL: head length, IDx and IDm:maximum and minimum interorbital distances, respectively, GW: gena width, EW: eye width, PW:pronotum width, MSL and MSW: length and width of mesoscutum, respectively, MTL: scutellumlength, MNL: metanotum length, MSH: mesosomum height, AL: alitrunk length, PL: propodeumlength, T1L: length of gastral tergite I, T1BH and T1AH: basal and apical heights of T1, T2BW andT2AW: basal and apical widths of tergite II, T2L: length of T2 and WL: partial length of the forewing.

364 F.B. Noll, D. Simões and R. Zucchi

than 20 combs arranged in various directions. Each comb hung from the woodenlid by means of thin paper pillars. New combs were started attached to a singlesupporting pillar (Fig. 2B). Additional similar structures are made as the combgrows. Interspaces between combs varied considerably and this probably happensbecause of the comb’s irregular growth. Horizontal or tilted pillars generally inter-connected combs when interspaces were 1 cm or less. The brood was concentratedat the core of the nest. Thus, the outer sterile combs, often composed of incompletecells, may function as a kind of envelope as suggested by WENZEL (1991) and ZUC-CHI et al. (1995). On account of its generally populous nests, A. pallipes is reputedto be aggressive. A nest collected in Ribeirão Preto (July 8, 1976) had 20,400 work-ers, about 200 queens and 13,990 males.

A. multipicta

A mature colony of Agelaia multipicta multipicta was collected (August, 1975)in Ribeirão Preto (SE Brazil). The colony occupied a space among tree roots. Inthe course of nest collection, the combs were damaged and part of the adult popu-

Fig. 2. — Nest of Agelaia pallipes pallipes hanging from the roof of an abandoned honeybee hive. A,details of the upper part of the same nest B, showing details of the primordium of a new comb.Mature nest of Agelaia multipicta multipicta found inside an abandoned honeybee hive. C, the nestarchitecture seen from above, D, part of the same nest seen laterally.

365Caste differences in polistine wasps

lation escaped. However, colony maturity could be ascertained by the presence ofabundant brood ranging from eggs to pupae; 88 queens and 15,522 adults werecounted.

Details of nest structure (Fig. 2C-D) refer to a colony that had occupied (Jan-uary, 1975) an abandoned honeybee hive. There were nine complete curved verticalcombs that, as in A. pallipes, were composed of horizontal cells. Two young combswere also present. This colony was transferred to the laboratory, but escaped in thenext few days. In the course of migration the nest population issued out as twowaves of migrating wasps. These established two congregation sites from which therespective final migrations took place separately, suggesting that a single migratingswarm can eventually produce more than one daughter colony.

Ovary development and insemination

In both species two kinds of ovary development were recognized (Fig. 3): typeA (Fig. 3A) with long very well developed ovaries bearing some mature eggs andmany oocytes, and B (Fig. 3B) with shorter and thread-like ovaries. Since insemi-nation was detected exclusively in the females bearing type A ovaries, these wereclearly queens and the remaining B type uninseminated females were workers.Moreover, the absence of intermediate females (uninseminated individuals withdeveloped ovaries, RICHARDS & RICHARDS 1951) is remarkable and contributes tocharacterizing the clear-cut distinction between castes.

Fig. 3. — Types of ovary development found among the females of Agelaia pallipes and Agelaia mul-tipicta. A, developed ovaries found in inseminated females (mature queens, left; younger queens,right); B, undeveloped ovaries found in uninseminated females (workers).

366 F.B. Noll, D. Simões and R. Zucchi

Queen-worker differences in relation to morphometry,hamulus number and morphology

A. pallipes

Among the mean relationships of the 22 analysed characters (Table 1, Fig. 4),17 showed queen/worker ratios higher than 1.00. In contrast, only five characters,mostly of the head, showed queen/worker ratios lower than 1.00. This indicatesthat queens are larger than workers except in five characters (HW, HL, IDx, EW,NH). Among the analysed characters, only hamuli number did not differ signifi-cantly between castes (t-test, P < 0.05).

The result of the canonical discriminant analysis based on 22 morphologicalcharacters supports the clear dimorphism between egg-layers (queens) and unin-seminated sterile females (workers). Queens showed higher values of the firstcanonical variate (CAN1) than workers (3.0 to 7.0 vs –8 to –3) (Fig. 5). To calculatethe CAN1 the following equation was used:

Table 1.

Means, queen/worker ratios, CAN1 values and observed values of t-test for 22 characters used fordiscriminating the castes of Agelaia pallipes.

Means (mm) t-test Queen/ CAN1

Characters Queens Worker observed Workers Stand- Rawvalues ratios ardized

HeadHW 2.59 ± 0.03 2.70 ± 0.05 – 19.45* 0.96 – 1.36 – 18.21HL 2.31 ± 0.06 2.34 ± 0.07 – 3.13* 0.99 – 0.18 – 2.77IDx 2.26 ± 0.03 2.34 ± 0.05 – 13.30* 0.97 – 0.43 – 7.36IDm 1.19 ± 0.02 1.18 ± 0.02 5.54* 1.01 – 0.004 – 0.18GW 0.68 ± 0.03 0.61 ± 0.04 16.45* 1.11 0.37 7.47EW 0.60 ± 0.03 0.66 ± 0.04 – 12.75* 0.91 0.11 2.45

MesosomaMSL 1.83 ± 0.06 1.73 ± 0.07 11.42* 1.06 – 0.007 – 0.08MSW 1.94 ± 0.04 1.76 ± 0.06 26.38* 1.10 0.28 2.87PW 2.33 ± 0.05 2.13 ± 0.06 26.62* 1.09 0.41 3.60MTL 0.69 ± 0.02 0.65 ± 0.03 11.13* 1.06 – 0.13 – 4.10MNL 0.49 ± 0.03 0.41 ± 0.02 20.67* 1.20 0.37 7.61MSH 2.81 ± 0.03 2.67 ± 0.07 17.67* 1.05 0.36 4.00AL 4.43 ± 0.09 4.10 ± 0.10 21.91* 1.08 0.14 0.70

MetasomaPL 1.34 ± 0.03 1.22 ± 0.03 26.81* 1.10 0.86 12.23T1L 2.37 ± 0.05 2.17 ± 0.08 20.11* 1.09 0.14 1.14T1BH 0.54 ± 0.03 0.44 ± 0.03 21.98* 1.23 0.56 9.34T1AH 0.83 ± 0.04 0.71 ± 0.03 25.24* 1.17 0.29 4.06T2L 2.30 ± 0.07 2.09 ± 0.11 16.39* 1.10 0.05 0.39T2BW 1.18 ± 0.04 1.03 ± 0.03 28.31* 1.15 0.35 4.47T2AW 2.73 ± 0.07 2.38 ± 0.08 30.96* 1.15 0.69 3.64

WingWL 6.13 ± 0.12 5.80 ± 0.21 13.69* 1.06 – 0.08 0.45NH 0.35 ± 0.03 0.36 ± 0.03 – 1.79 0.97 0.14 – 2.48

Symbols are explained in the text; * significant difference at 5% level.

367Caste differences in polistine wasps

Fig. 4. — Mean queen/workers ratios for 21 morphometric characters and hamuli number in Age-laia pallipes, A. multipicta compared with A. vicina and A. flavipennis.

Fig. 5. — Discrimination between queens and workers of Agelaia pallipes pallipes based on thecanonical discriminant analysis using 21 metric characters and the number of hamuli. The equationto calculate the first canonical variate (CAN1) is shown in the text.

368 F.B. Noll, D. Simões and R. Zucchi

CAN1 = -18.21(HW-2.65)-2.77(HL-2.33)-7.36(IDx-2.30)-0.18(IDm-1.19)+7.47( G W- 0 . 6 5 ) + 2 . 4 5 ( E W- 0 . 6 3 ) - 0 . 0 8 ( M S L - 1 . 7 8 ) + 2 . 8 7 ( M S W-1.85)+3.60(PW-2.23)-4.10(MTL-0.67)+7.61(MNL-0.45)+4.00(MSH-2.74)+0.70(AL-4.27)+12.23(PL-1.28)+1.14(T1L-2.27)+9.34(T1BH-0.49)+4.06(T1AH-0.77)+0.39(T2L-2.27)+4.47(T2BW-1.11)+3.64(T2AW-2.55)+0.45(WL-5.97)-2.48(NH-0.35).

For determining CAN1 (Table 1), basal height of T1 (T1BH), basal width of ter-gite II (T2BW), propodeum length (PL) and, especially, head width (HW) were themost important among the characters examined. The Mahalanobis distance (D2,ANDERSON 1958) calculated through CDA was 122.74 between queens and workers.

As observed by RICHARDS (1978) another queen/worker difference is: “queenlacks the brown or black stripe on the humeri which seems to be universally pre-sent in the workers and her first gastral tergite is little longer and wider”.

Table 2.

Means, queen/worker ratios, CAN1 values and observed values of t-test for 22 characters used fordiscriminating the castes of Agelaia multipicta multipicta.

Means (mm) t-test Queen/ CAN1

Characters Queens Worker observed Worker Stand- Rawvalues ratios ardized

HeadHW 2.58 ± 0.04 2.66 ± 0.05 – 11.90* 0.97 – 1.14 – 20.19HL 2.35 ± 0.06 2.28 ± 0.06 8.15* 1.03 – 0.04 – 0.57IDx 2.25 ± 0.03 2.25 ± 0.04 – 0.47 1.00 – 0.31 – 8.37IDm 1.23 ± 0.02 1.17 ± 0.03 20.14* 1.05 0.54 13.51GW 0.65 ± 0.03 0.60 ± 0.03 12.91* 1.08 0.06 1.41EW 0.60 ± 0.02 0.65 ± 0.03 – 11.83* 0.92 0.05 1.41

MesosomaMSL 1.89 ± 0.05 1.73 ± 0.06 19.42* 1.09 0.16 1.75MSW 1.98 ± 0.05 1.78 ± 0.05 28.76* 1.11 0.42 3.79PW 2.40 ± 0.05 2.16 ± 0.04 33.56* 1.11 1.16 9.08MTL 0.70 ± 0.02 0.63 ± 0.02 19.48* 1.11 0.24 5.66MNL 0.53 ± 0.03 0.54 ± 0.09 – 1.22 0.98 – 0.06 – 0.82MSH 2.84 ± 0.06 2.68 ± 0.06 17.59* 1.06 – 0.27 – 2.71AL 4.43 ± 0.11 4.10 ± 0.12 19.84* 1.08 0.21 1.09

MetasomaPL 1.34 ± 0.04 1.23 ± 0.04 19.44* 1.09 0.20 2.93 T1L 2.43 ± 0.07 2.21 ± 0.09 18.46* 1.10 0.27 1.98T1BH 0.55 ± 0.04 0.43 ± 0.03 24.41* 1.28 0.20 3.04T1AH 0.91 ± 0.05 0.75 ± 0.04 24.94* 1.21 0.56 6.12T2L 2.30 ± 0.14 2.03 ± 0.14 12.67* 1.13 0.15 0.79T2BW 1.31 ± 0.05 1.09 ± 0.04 35.39* 1.20 0.91 7.61T2AW 2.79 ± 0.09 2.37 ± 0.08 33.84* 1.18 0.50 2.20

WingWL 6.08 ± 0.16 5.78 ± 0.21 11.18* 1.05 – 0.22 – 0.93NH 0.36 ± 0.03 0.38 ± 0.04 – 3.43* 0.95 – 0.02 – 0.62

Symbols are explained in the text; * significant difference at 5% level.

369Caste differences in polistine wasps

A. multipicta

Among the mean relationships of the 22 analysed characters (Table 2, Fig. 4),17 of them show queen/worker ratios higher than 1.00. In contrast, only four char-acters, primarily of the head plus hamuli number (NH) and metanotum length(MNL), showed queen/worker ratios lower than 1.00. Only the maximum interorbi-tal distance (IDx) was equal to 1.00. Among the measured characters, metanotumlength (MNL) and the maximum interorbital distance (IDx) did not differ signifi-cantly between castes (t-test, P < 0.05). This indicates that queens are larger thanworkers except in five characters (HW, IDx, EW, NH, MNL).

The result of the canonical discriminant analysis based on 22 charactersstrengthens the clear dimorphism between inseminated egg-layers (queens) anduninseminated sterile females (workers). Queens showed higher values of the firstcanonical variate (CAN1) than workers (1.0 to 9.0 vs –7 to –1) (Fig. 6). To calculatethe CAN1 the following equation was used:

CAN1 = -20.19(HW-2.63)-0.57(HL-2.31)-8.37(IDx-2.25)+13.51(IDm-1.20)+1.41(GW-0.62)+1.41(EW-0.62)+1.75(MSL-1.80)+3.79(MSW-1.87)+9.08(PW-2.28)+5.66(MTL-0.66)-0.82(MNL-0.54)-2.71(MSH-2.76)+1.09(AL-4.25)+2.93(PL-1.28)+1.98(T1L-2.31)+3.04(T1BH-0.48)+6.12(T1AH-0.83)+0.79(T2L-2.16)+7.61(T2BW-1.19)+2.20(T2AW-2.57)-0.93(WL-5.92)-0.62(NH-0.36).

For determining CAN1 (Table 1), basal width of tergite II (T2BW), head width(HW) and, especially, pronotum width (PW) were the most important among theexamined characters. The Mahalanobis distance (D2, ANDERSON 1958) calculated

Fig. 6. — Discrimination between queens and workers of Agelaia multipicta multipicta based on thecanonical discriminant analysis using 21 metric characters and the number of hamuli. The equationto calculate the first canonical variate (CAN1) is shown in the text.

370 F.B. Noll, D. Simões and R. Zucchi

through the CDA was 110.99 between queens and workers. In addition some mor-phological queen-worker differences were shown: (1) petiole larger with more hairsin queens, (2) queen’s gaster larger and paler than workers, (3) worker’s middletibia brown but yellow in queens, (4) interspace between antennal insertion brown-ish in workers but yellow in queens.

DISCUSSION

Three patterns of caste differentiation have been recognized for epiponinewasps according to RICHARDS (1978): (1) conspicuous size and allometric differenc-es present, with queens larger than workers in the absence of intermediates (A.areata, JEANNE & FAGEN 1974; A. vicina, SAKAGAMI et al. 1996; Protonectarina sylvei-rae, SHIMA et al. 1996b); (2) conspicuous dimorphism present, with queens smallerthan workers and no intermediates present (Apoica flavissima, SHIMA et al. 1994;Polybia dimidiata, SHIMA et al. 1996a); (3) morphological differences slight or indis-tinct, and intermediates present (Pseudopolybia vespiceps, S.N. SHIMA et al. in prep.;Protopolybia exigua, NOLL et al. 1997). However, in some groups, queens are signifi-cantly smaller than workers in some characters and larger in others. According toJEANNE et al. (1995), in Apoica pallens this is considered as non-size-based allome-try probably due to a reprogramming in growth parameters (WHEELER 1991). Sucha pattern was detected also in Epipona guerini (HUNT et al. 1996), Pseudopolybia dif-ficilis (JEANNE 1996), Apoica flavissima (SHIMA et al. 1994), and Polybia dimidiata(SHIMA et al. 1996a).

Queen-worker dimorphism in A. multipicta and A. pallipes is similar to otherAgelaia previously analysed, that is, queens are larger than workers with no inter-mediates present. In addition, colour and morphological differences were alsodetected. In A. multipicta, CAN1 values of queens and workers displayed a widerrange than those of A. pallipes. Although these two species are morphologically verysimilar (RICHARDS 1978), among the four most discriminant characters observed,only head width (HW) applies to both species, although most of the other charac-ters are gaster related. The finding that queen-worker mean ratios are not signifi-cantly different in two characters (IDx and MNL), for A. multipicta and for onlyone (NH), for A. pallipes, indicates that queens are larger than workers in practical-ly all the analysed characters.

The Mahalanobis distances obtained for A. pallipes and A. multipicta (124.67and 110.99, respectively) are considered high if compared with A. vicina (176.7,F.B. NOLL et al. in prep.). Species from other genera show lower values of Mahalan-obis distances: Pseudopolybia vespiceps (0.99, S.N. SHIMA et al. in prep.), Polybiadimidiata (6.76, SHIMA et al. 1996a), Protonectarina sylveirae (8.64, SHIMA et al.1996b), Protopolybia exigua (4.95, NOLL et al. 1997), Parachartergus smithii (10.31,S. MATEUS et al. in prep.) and Apoica flavissima (11.27, SHIMA et al. 1994).

As previously expected, A. pallipes and A. multipicta demonstrated the largestcaste dimorphism among the species studied in our serial work. These two speciesdiffer from Apoica flavissima (SHIMA et al. 1994) and Polybia dimidiata (SHIMA et al.1996a), in that queens are larger than workers, which is comparable to Protonecta-rina sylveirae but much more pronounced. Except for Agelaia lobipleura (RICHARDS

1978), marked queen/worker dimorphism has been reported in other Agelaia spp.(A. flavipennis, EVANS & WEST-EBERHARD 1970; A. areata, JEANNE & FAGEN 1974; A.

371Caste differences in polistine wasps

vicina, VON IHERING 1903, SAKAGAMI et al. 1996), indicating that well-developed castedifferentiation is widely distributed in this taxon as suggested by RICHARDS (1978).Although the same pattern of caste differentiation is observed, A. vicina and A. fla-vipennis have larger queens than A. pallipes and A. multipicta, especially the gasterrelated characters (Fig. 4). This conclusion is reinforced by the queen/workerdimorphism based on the alitrunk length: 1.08 for A. pallipes and A. multipicta,1.12 for A. vicina (F.B. NOLL et al. in prep.), 1.12-1.16 for A. areata (JEANNE & FAGEN

1974) and 1.10 for A. flavipennis (F.B. NOLL unpub.). These values can be consid-ered because in vespines queen/worker dimorphism based on the thorax width wasestimated in 1.37 and 1.34 for Vespula germanica and V. rufa, respectively (BLACKITH

1958). In relation to ovary development, A. pallipes and A. multipicta are closely

related to Apoica flavissima (SHIMA et al. 1994) and Protonectarina sylveirae (SHIMA

et al. 1996b), in which only the queens were inseminated and had well-developedovaries, unlike workers which have the opposite conditions and, in addition, inter-mediates were absent.

ACKNOWLEDGMENTS

This study was supported by grants from Fundação de Amparo à Pesquisa do Estado deSão Paulo (FAPESP). The identification of Agelaia multipicta multipicta and Agelaia pallipespallipes was confirmed by the late Professor O.W. Richards. Special thanks to Professor Wil-liam D. Hamilton for offering the specimens (one worker and one queen) of Agelaia flavipen-nis analysed. This paper was improved due to the comments of three anonymous refereesand, especially those of Dr John Wenzel.

REFERENCES

ANDERSON T.W. 1958. Introduction to multivariate statistical analysis. New York: John Wileyand Sons Inc., 374 pp.

BLACKITH R.E. 1958. An analysis of polymorphism in social wasps. Insectes Sociaux 5: 263-272.

CARPENTER J.M. 1993. Biogeographics patterns in the Vespidae (Hymenoptera): two views ofAfrica and South America, pp. 139-155. In: Goldbatt P., Edit. Biological relationshipsbetween Africa and South America. New Haven and London: Yale University Press.

EVANS H.E. & WEST-EBERHARD M.J. 1970. The wasps. Ann Arbor University, Michigan, 265 pp.HUNT J.H., SCHMIDT D.K., MULKEY S.S. & WILLIAMS M.A. 1996. Caste dimorphism in Epipona

guerini (Hymenoptera: Vespidae): further evidence for larval determination. Journal ofKansas Entomological Society 69 (4): 362-369.

IHERING R. VON 1903. Des Vespides de l’Amérique du Sud. Annales de la Société Entomologiquede France 72: 144-155.

JEANNE R.L. 1991. The swarm-founding Polistinae, pp. 191-231. In: Ross K.G. & MatthewsR.W., Edits. The social biology of wasps. Ithaca: Cornell University Press.

JEANNE R.L. 1996. Non-allometric queen-worker dimorphism in Pseudopolybia difficilis(Hymenoptera: Vespidae). Journal of Kansas Entomological Society 69 (4): 370-374.

JEANNE R.L. & FAGEN R. 1974. Polymorphism in Stelopolybia areata (Hymenoptera, Vespidae).Psyche 81: 155-166.

JEANNE R.L., GRAF C.A. & YANDELL B.S. 1995. Non-size-based morphological castes in a socialinsect. Naturwissenschaften 82: 296-298.

372 F.B. Noll, D. Simões and R. Zucchi

NOLL F.B., MATEUS S. & ZUCCHI R. 1997. Morphological caste differences in neotropicalswarm-founding Polistinae wasps. V- Protopolybia exigua exigua (Hymenoptera: Vespi-dae). Journal of New York Entomological Society 104 (1): 61-68.

PARDI L. & MARINO PICCIOLI M.T. 1981. Studies on the biology of Belonogaster (HymenopteraVespidae). 4. On caste differences in Belonogaster griseus (Fab.) and the position of thisgenus among social wasps. Monitore Zoologico Italiano (Nuova Serie) Supplemento XIV(9): 131-146.

RAO C.R. 1973. Linear statistical inference. New York: John Wiley and Sons.RICHARDS O.W. 1978. The social wasps of the Americas excluding the Vespinae. London: Brit-

ish Museum (Natural History), 580 pp.RICHARDS O.W. & RICHARDS M.J. 1951. Observations on the social wasps of South America

(Hymenoptera, Vespidae). Transactions of the Royal Entomological Society of London102: 1-170.

SAKAGAMI S.F., ZUCCHI R., YAMANE S., NOLL F.B. & CAMARGO J.M.F. 1996. Morphological castedifferences in Agelaia vicina, the Neotropical swarm-founding polistine wasp with thelargest colony size among social wasps (Hymenoptera: Vespidae). Sociobiology 28 (2):207-223.

SHIMA S.N., YAMANE S. & ZUCCHI R. 1994. Morphological caste differences in some Neotropi-cal swarm-founding Polistinae wasps I. Apoica flavissima (Hymenoptera, Vespidae). Jap-anese Journal of Entomology 62 (4): 811-822.

SHIMA S.N., YAMANE S. & ZUCCHI R. 1996a. Morphological caste differences in some Neotropi-cal swarm-founding polistine wasps II. Polybia dimidiata (Hymenoptera, Vespidae). Jap-anese Journal of Entomology 64 (1): 131-144.

SHIMA S.N., YAMANE S. & ZUCCHI R. 1996b. Morphological caste differences in some Neotropi-cal swarm-founding polistine wasps III. Protonectarina sylveirae (Hymenoptera, Vespi-dae). Bulletin of the Faculty of Educaction, Ibaraki University 45: 57-67.

WENZEL J.W. 1991. Evolution of the nest architecture, pp. 480-509. In: Ross K.G. & MatthewsR.W., Edits. The social biology of wasps. Ithaca: Cornell University Press.

WENZEL J.W. 1992. Extreme queen-worker dimorphism in Ropalidia ignobilis, a small-colonywasp (Hymenoptera, Vespidae). Insectes Sociaux 39: 31-43.

WHEELER D.E. 1991. The developmental basis of worker castes polymorphism in ants. Ameri-can Naturalist 138: 1218-1238.

YAMANE S., KOJIMA J. & YAMANE S. 1983. Queen/worker size dimorphism in an Oriental polis-tine wasp, Ropalidia montana Carl (Hymenoptera, Vespidae). Insectes Sociaux 30: 416-422.

ZUCCHI R., SAKAGAMI S.F., NOLL F.B., MECHI M.R., BAIO M.V., MATEUS S. & SHIMA S.N. 1995.Agelaia vicina, a swarm-founding Polistinae with the largest colony size among waspsand bees. Journal of New York Entomological Society 103 (2): 129-137.