Evolution of Caste in Neotropical Swarm-Founding Wasps(Hymenoptera: Vespidae; Epiponini)

2004·0.86 MB·English

by  Fernando B. Noll

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY
CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024
Number 3467, 24 pp., 8 figures, 10 tables December 30, 2004
Evolution of Caste in Neotropical
Swarm-Founding Wasps
(Hymenoptera: Vespidae; Epiponini)
FERNANDO B. NOLL,1 JOHN W. WENZEL,2 AND RONALDO ZUCCHI3
ABSTRACT
Reproductive castes are compared in species of swarming wasps representing all currently
recognized genera of Epiponini (Polistinae). New morphometric data for nine measures of
bodypartsandovariandataarepresentedfor13species.Theseareintegratedwithallsimilarly
conducted available studies, giving a total of 30 species. Analysis reveals several syndromes
relatingreproductiveandnonreproductiveindividuals:nomeaningfuldistinction,physiological
differencesonly,reproductiveslargerthannonreproductiveswithintermediateindividualspre-
sent, reproductives different in shape from nonreproductives with no intermediates, and re-
productives smaller in some aspects than nonreproductives. Distribution of these syndromes
among species is consistent with phylogenetic relationshipsderivedfromotherdata.Optimiz-
ing these syndromes on the cladogram indicates that the basal condition of Epiponini is a
castelesssocietythatisnotcomparabletotheprimitivelysocialgenusPolisteswheredominant
queens control reproduction.CastesoriginateseveraltimesinEpiponini,withdifferentresults
in different lineages. The best documented evolutionary sequence passes from casteless so-
cieties, to those with reproductives larger, to those with reproductives differing in shape from
nonreproductives, to those with reproductives smaller in some measures. This sequence is
consistent with Wheeler’s theory of the origin of caste through developmental switches, and
represents the most thorough test of that theory to date.
INTRODUCTION In some species, reproductive ‘‘queens’’ and
nonreproductive ‘‘workers’’ differ only in
Separation into castes is one of the cor- behavior, but in other species such behaviors
nerstones of the evolution of social insects. are accompanied by physiological and mor-
1DepartamentodeZoologiaeBotaˆnica;InstitutodeBiocieˆncias,LetraseCieˆnciasExatas,UNESP;RuaCristo´va˜o
Colombo,2265; 15054-000,Sa˜oJose´ do RioPreto,SP;Brazil([email protected]).
2Department of Entomology, The Ohio State University, 1315 Kinnear Road, Columbus, OH 43212 (wenzel.
[email protected]).
3Departamento de Biologia, Faculdade de Filosofia Cieˆncias e Letras de Ribeira˜o Preto, Universidade de Sa˜o
Paulo,Brazil([email protected]).
Copyright(cid:113)AmericanMuseumofNaturalHistory2004 ISSN0003-0082
2 AMERICAN MUSEUM NOVITATES NO. 3467
phological specializations and body-sizevar- diversity that can be arranged along a spec-
iation (Oster and Wilson, 1978). The degree trum from taxa in which queens and workers
to which castes are distinct often serves as are externally similar, to others with fairly
part of the definition of the society itself, distinct caste attributes (reviewed in
with greater caste differentiation indicating O’Donnell, 1998; Shima et al., 1998). The
‘‘higher’’ sociality (Bourke, 1999). In some Paleotropical tribe Ropalidini (perhaps the
groups, such as swarm-founding neotropical sister to the Neotropical tribe Epiponini ex-
wasps,thespeciesaredescribedashighlyso- amined here) shows diversity that is hard to
cial, although the distinction betweenqueens interpret without more study. For instance,
and workers is considered slight and often some species of Ropalidia present distinct
confusing (Richards and Richards, 1951; morphological castes (Yamane et al., 1983),
Richards, 1978, cited as Polybiini, junior includingthoseofsmallcolonysize(Wenzel,
synonym of Epiponini, see Carpenter, 1993, 1992), but others show morphologicallysim-
1997). ilar castes despite larger colony size (Gad-
Wasps (Hymenoptera: Vespidae) have fre- agkar and Joshi, 1982, 1983; Yamane and
quently been used to test evolutionary mod- Itoˆ, 1987; Gadagkar, 1987). Parapolybia has
els for the origin of social behavior because castes (Yamane and Maeta, 1985) but their
of their different levels of sociality—from behavior is poorly known. Belonogaster is
solitary to eusocial (West-Eberhard, 1978, also poorly known (Pardi and Piccioli,1981)
1996; Wilson, 1985; Itoˆ, 1986; Spradbery, but is thought to show great morphological
1991). General theory holds that primitive variation (Keeping, 2000, 2002). In Poly-
systems of dominance rely on physical ag- bioides, morphological castes are known in
gression, with the winner becoming the eg- P. tabidus (Turillazzi et al., 1994), which
glayer. More derived systems require some builds large colonies and reproduces by
kind of caste determination that may occur swarms. However, despite the repeated evo-
during theimmaturestagesofanindividual’s lution of castes in social wasps as a whole,
development (preimaginal determination), the evolution of castes is largely interpreted
with an unambiguous physiological or mor- through our understanding of Epiponini.
phologicalresultthatcannotbereversed.Ab- Epiponini is a tribe of Polistinae, with
sence of morphological castes is a plesiom- about20well-markedgeneraandatleast229
orphic (‘‘primitive’’) condition in the three species, plus another 27 subspecies, all of
subfamilies of social wasps, so that morpho- which are Neotropical. All Epiponini are po-
logical castes were attained independently in lygynic (meaning that many queens com-
different lineages (Carpenter, 1991). Most monlyreproducesimultaneouslyonthesame
distant to the question at hand, females of nest), and all reproduce by swarms, which is
Stenogastrinae are barely distinguished as taken as a sign of sophisticated, obligatory,
queens or as workers (Turillazzi, 1991). high sociality. Initially, Richards(1978)doc-
Closer, and sister to the subfamily examined umented three forms of caste discrimination,
here, queens of Vespinae are larger than and recent research builds upon these: (1)
workers (Spradbery, 1991), but such a trait Queens larger than workers. In some species
may be an adaptation for a queen’s solitary caste differences are so pronouncedthatthey
winter survival in most species (West-Eber- mustbedeterminedbynutritionaldifferences
hard, 1978). In the Polistinae (studied here), during larval development (Evans and West-
the most basal genera (Polistes and Mischo- Eberhard, 1970; Jeanne and Fagen, 1974;
cyttarus), initiate their nests solitarily, with Sakagamietal.,1996;Nolletal.,1997),with
egglayers only slightly larger than non-eg- important consequences for fertility and re-
glayers. However, polistines present a large productive asymmetries (West-Eberhard,
2004 NOLL ET AL.: CASTES IN NEOTROPICAL SWARM-FOUNDING WASPS 3
1978; O’Donnell, 1998). The classical land- TABLE1
mark is a case of different, nonallometric Morphometric Differences Reported from
Literature
castes of Agelaia areata, where queens and
workers have different shapes as well as dif-
ferent sizes (Jeanne and Fagen, 1974). It
seems that these differences must represent
preimaginal bias in nutrition. (2) Allometric
differences, with queens smaller than work-
ers in some body parts, but larger in others,
which may result from ontogenetic repro-
gramming in growth parameters (Jeanne et
al.,1995),accordingtotheideathatthebody
plans represent fundamentally different end
products, not different points along a contin-
uum. Sometimes such castes are barely de-
tectable (Jeanne, 1996). (3) Slight or indis-
tinct morphological differences, with castes
absent, or manifested weakly, based on only
a few traits (Richards, 1978; Jeanne, 1980).
One great problem is that integration of the
empirical studies is not straightforward be-
cause few studies are exactly comparable in
measurement methodology or phase of col-
ony cycle studied, so generalization is diffi-
cult. A rough summary of species known to
fall into these three conditions can be found
in table 1. It has been suggested that prei-
maginal determination may be basal for Epi-
ponini (Jeanne et al., 1995; Jeanne, 1996;
Hunt et al., 1996; O’Donnell, 1998), espe-
cially because such a trait is present in Apoi-
ca (Shima et al., 1994), the most basalgenus
in Epiponini. However, Apoica is peculiar
and highly apomorphic in many ways (gen-
eral morphology, nest architecture, and noc-
turnal habit, see Richards, 1978; Wenzel,
1992; and Pickett, 2003) and so it may not
be a good model of basal synapomorphiesof
Epiponini. It has not been demonstrated that
preimaginal determination is the basal con-
dition for the whole clade.
Integrating studies of caste in Epiponini is
difficult because studies to date have been
limited in scope,with mostpublicationscov-
ering one species at a time. Measurements
are not necessarily comparable across stud-
ies. There are sometimes doubts about the
4 AMERICAN MUSEUM NOVITATES NO. 3467
phase of colony cycle that was measured, Sa˜o Paulo, Brazil, and the AMNH. They
particularly when the colony was collected were taken by being put into a plastic bag
providedwithether-moistenedcottonballsor
opportunistically by visitors on brief field-
with open cyanide tubes. Populations were
trips. In addition, when reproductive status
fixed in Dietrich’s solution and then kept in
does not relate closely to morphological var-
70% ethanol for dissection.
iation,thereisatendencytorescueaconcept
From each of these colonies 100 workers
of caste by categorizing problematicfemales and all queens were measured and dissected.
as ‘‘laying workers’’ or ‘‘replacement When castes could be discriminated by size
queens’’ on an ad hoc basis. Such axiomatic or color, all possible queens were selected,
descriptions may serve only to obscure the and from the remaining individuals 100
original question, protecting an inadequate workers were arbitrarily selected. In small
colonies where castes were not obvious, all
conceptofcasteitselffromrefutationbyem-
individuals were measured and dissected. In
pirical data. We address these issues by pre-
large colonies where castes were hard to
senting here the most complete set of com-
identify, 100 individuals were chosen ran-
parablevaluesregardingcastedetermination.
domly.Afteridentifyingsomemorphological
A total of more than 13,500 measures en- characterstodiscriminatequeens,suchasab-
compassing nine dimensions of 13 species dominal size, all individuals having such a
are offered for the first time. These are in- pattern were chosen and usedintheanalysis.
tegrated with comparable data from other The following characters were measured un-
studies to include 68 species. We establish der a binocular microscope (smallest unit (cid:53)
0.01 mm): head width (HW), minimum in-
the basal condition for Epiponini, providing
terorbital distance (IDm), gena width (GW),
evidence for both preimaginal and postima-
width of mesoscutum (MSW), alitrunk
ginalcastedetermination.Inadditiontomor-
length (AL), length of gastral tergite I (T L),
phometricvariationincaste,westresstheex- 1
basal height of T (T BH), basal widths of
istence of various types of social regulation. tergite II (T BW),1 and1 partial length of the
2
forewing (WL)(fig.1, formoredetailsabout
MATERIALS AND METHODS
the characters see Shima et al.,1994).Ovary
Workers and queens were taken from ma- condition was determined under a stereomi-
turecolonies(i.e.,postworkeremergence)of croscope. In order to analyze insemination,
13 different species of Epiponini collectedin the spermatheca was removed and put on a
three different localities. Angiopolybia pal- slideina1:1solutionofglycerinandethanol
lens (Lepeletier), Asteloeca ujhelyii (Ducke), (70%). The presence ofspermcellswascon-
Charterginus fulvus Fox, Clypearia sulcata firmed by microscope.
(de Saussure), and Leipomeles dorsata (Fa- Before statistical analysis, data were con-
bricius) were collected in Iquitos, Peru. Nec- verted to millimeters and later converted by
tarinella championi (Dover) was collectedin log transformation in order to avoid prob-
Costa Rica, near Atenas. For these two lo- lems of variance. Means and standard devi-
calities, the samples were collected by John ations were calculated from the nine mor-
W.WenzelandJamesCarpenter,killedincy- phological measurements. Statistical charac-
anide and preserved promptly in 70% etha- terization of castewasexploredintwoways.
nol, and deposited in the AMNH. Polybia One-wayANOVAwasusedforcomparisons
(Polybia) liliacea (Fabricius), Polybia (For- of means to demonstrate whether castes (as
micicola)rejecta(Fabricius),Chartergusme- determined by dissection) differ in some
tanotalis Richards, Epipona tatua (Cuvier), bodydimensions.Becausemultiplemeasures
Metapolybia aztecoides Richards, Polybia that relate to shape are not always captured
(Pedothoeca) spinifex Richards,andSynoeca wellinsinglecomparisons,stepwisediscrim-
surinama (Linnaeus) were collected in Nova inant function analysis was used to see if
Xavantina, Mato Grosso, BrazilbyFernando combinations of variables could be useful in
B. Noll and Sidnei Mateus. Vouchers arede- predictingcaste.Inthismethod,variablesare
posited at the Museu de Zoologia (MZUSP), successivelyaddedtothemodelbasedonthe
2004 NOLL ET AL.: CASTES IN NEOTROPICAL SWARM-FOUNDING WASPS 5
Fig. 1. Representative measures for morphometric analyses of this paper: head width (HW), mini-
muminterorbitaldistance(IDm),genawidth(GW),widthofmesoscutum(MSW),alitrunklength(AL),
lengthofgastraltergiteI(TL),basalheightofT (TBH),basalwidthsoftergiteII(TBW),andpartial
1 1 1 2
length of the forewing (WL).
higherFtoentervalues,addingnomorevar- based on the effectiveness in previous pub-
iables when the F-ratio is no longer signifi- lications (Jeanne et al., 1995; Jeanne, 1996;
cant. However,variablesthatenterthemodel Hunt et al., 1996).
early may be completely subsumed by sep-
arate variables that enter later. The signifi- RESULTS
cance of Wilks’ lambda (an estimateofcom-
OVARIAN DEVELOPMENT
mon variance between sets of variates) was
used to determine the degree to which sep- The ovariole number was always three in
aratemeasurescontributedtothefinalmodel. each ovary, and the following types of ovar-
ThisisanalternativetousinganFtoremove ian development were distinguished (table
ateachstep.Variablesthatappearinthefinal 2); type A: with filamentous ovarioles that
model but do not have significant F-ratios had no visible oocytes or with some very
represent variance components that are ex- small oocytes; type B: bearing some young
plained by some combination of the other oocytes; type C: with one or more mature
variables also in the model, and therefore no oocytes in each ovariole; type D: with well-
longer contribute to the discrimination itself. developed and very long ovarioles coiled in-
Wilks’ lambda varies from zero to one: the side the gaster, with at least one mature egg.
lower the value, the greater the significance. Only females with type D ovaries were in-
Analysis of covariance (ANCOVA) was seminated. All four types were found in the
used, with alitrunk length (AL) as a covari- following species: Angiopolybia pallens,
ate,inordertoidentifydifferentgrowthrates Charterginus fulvus, Leipomeles dorsata,
in the body parts measured. AL was chosen and Nectarinella championi (table 2). Type
6 AMERICAN MUSEUM NOVITATES NO. 3467
TABLE2 queens (tables 3–6). Univariate statistics
Ovary Development for All Species Analyzeda (ANOVA) showed that measured characters
could bedifferentornotbetweenqueensand
workers in different species. Also, based on
multivariate statistics, the power of discrim-
ination of each measurement varied in the
different species studied; that is, the mea-
surements may discriminate castes indepen-
dently or only in combination.
In a first group (fig. 2), morphologicaldif-
ferences between castes were totally or prac-
tically absent in the following species: An-
giopolybia pallens, Asteloeca ujhelyii, Cly-
pearia sulcata (table 3), Leipomelesdorsata,
Metapolybia aztecoides, Nectarinella cham-
pioni, and Synoeca surinama (table 4). AN-
OVA showed no significant differences in
any measurementinAn.pallens,As.ujhelyii,
and Cl. sulcata; in S. surinama, only one
measurement was significant, and in L. dor-
C was not found in Asteloeca ujhelyii,Char- sata, M. aztecoides, and N. championi three
tergus metanotalis, Occipitalia sulcata, Epi- measurements were significant. Except for
pona tatua, Metapolybia aztecoides, Polybia M. aztecoides, in which some characters
liliacea, Polybia rejecta, Polybia spinifex, were smaller in queens than in workers, dif-
andSynoecasurinama(table2).Queens’ova- ferences were always based on queens being
ries (type D, inseminated) were much longer larger than workers. ANCOVA, using ali-
than workers’ ovaries in Chartergus metano- trunk length as a dependent variable (table
talis, Epipona tatua, Polybialiliacea,Polybia
7),wasnotsignificantinanycharacterinAn.
rejecta, and Polybia spinifex. No species of
pallens, As. ujhelyii, and Cl. sulcata. In L.
this study lacked both types B and C.
dorsata, M. aztecoides, N. championi, and S.
surinama, significant differences were de-
MORPHOLOGICAL DIFFERENCES tected.
Differences of mean values of nine char- Wilks’ lambda values were used to infer
acters measured were tested in workers and theindependentcontributionofeachvariable
TABLE3
MeansandObservedValuesofANOVATest,forNineCharactersUsedforDiscriminatingCastesof
Angiopolybiapallens,Asteloecaujhelyii,andClypeariasulcata
2004 NOLL ET AL.: CASTES IN NEOTROPICAL SWARM-FOUNDING WASPS 7
TABLE4
MeansandObservedValuesofANOVATest,forNineCharactersUsedforDiscriminatingCastesof
Leipomelesdorsata,Metapolybiaaztecoides,Nectarinellachampioni,andSynoecasurinama
TABLE5
MeansandObservedValuesofANOVATest,forNineCharactersUsedforDiscriminatingCastesof
Charterginusfulvus,Chartergusmetanotalis,andEpiponatatua
8 AMERICAN MUSEUM NOVITATES NO. 3467
TABLE6
MeansandObservedValuesofANOVATest,forNineCharactersUsedforDiscriminatingCastesofPolybia
liliacea,Polybiarejecta,andPolybiaspinifex
to the model that predicts the caste of the surements had alowpowerofdiscrimination
individual wasps. Significance values of the in Cn. fulvus, Cg. metanotalis, and E. tatua
loadings of each morphometric variable in- (tables 8, 9), and thatdiscriminationbetween
dicate whether that variable contributes castescanonlyoccurbasedonacombination
meaningfullytothepredictionofcaste.Anal- of measurements. In Polybia rejecta and Po.
ysis showed that, in the above species, spinifex, differences were much more pro-
Wilks’ lambda values were always high, nounced and all measurementswerestronger
ranging from 0.7 to 1.0 (table 8). Consider- discriminators. Wilks’ lambda ranged from
ing that lambda is defined as 1.0 minus the 0.1 to 0.3 (table 9), and castes were more
squared canonical correlation, the highest clearly discriminated than in the previous
value is 1.0, indicating complete absence of group (table 10). ANCOVA showed signifi-
association.Thesedataindicatethatmorpho- cantdifferencesingrowthratesofbodyparts
logical castes are practically absent in these in at least one character in all species of this
species, based on these measures.Specifical- group (table 7).
ly, even if queens may differ from workers In a third class, Polybia liliacea castes
on average, the ability to identify an individ- were as clearly distinct as in the second
ual as a queen or worker based on morpho- group, but head width was smaller in queens
metric values is very poor. None of the than in workers (table 6). ANOVA showed
queens were correctly classified by discrim- thatallmeasurementsweredifferentbetween
inantscoresinAn.pallensorCl.sulcata,and queens and workers.
relatively few in L. dorsata, M. aztecoides,
N.championi,andS.surinama(seetable10). DISCUSSION
Only in As. ujhelyii arecastesbetterdiscrim-
inated. A distinct physiological separation of fe-
In a second group (fig. 3)—Charterginus males into those that are reproductive and
fulvus, Chartergus metanotalis, Epipona ta- those that are nonreproductive is possible
tua, Polybia rejecta, and P. spinifex—queens based on ovarian condition and insemination
were larger than workers. Mean differences in some of the species reported here. While
using ANOVA (tables 5, 6) were found in complete absence of ovarian development in
five of nine characters in Cn. fulvus and Cg. workers(i.e.,theabsenceofbothtypesBand
metanotalis. In E. tatua, Po. rejecta, and Po. C) was not found in the speciesstudiedhere,
spinifex all body measurements were differ- suchasyndromehasbeenreportedinApoica
ent. flavissima, Agelaia vicina (Sakagami et al.,
Multivariateanalysisshowedthatthemea- 1996), Ag. pallipes, and Ag. multipicta (Noll
2004 NOLL ET AL.: CASTES IN NEOTROPICAL SWARM-FOUNDING WASPS 9
Fig. 2. Slight caste discrimination found in some epiponines based mostly on shape (when appli-
cable). Individuals identified by dissection to be ‘‘queens’’ are represented as squares, while those
identified as ‘‘workers’’ are solid dots. Separate regression linesare drawn forillustrationonly;seetext
for discussion.
et al., 1997). In our present study, queens’ pallipes and Ag. multipicta (Noll et al.,
ovaries (type D) were much longer than 1997). Similarly, though less discrete, inter-
workers’ ovaries in Chartergus metanotalis, mediate type C was not found in Asteloeca
Epipona tatua, Polybia liliacea, Po. rejecta, ujhelyii, Chartergus metanotalis, Clypearia
andPo.spinifex.Thissituationisalsoknown sulcata, Epipona tatua, Metapolybiaaztecoi-
in Apoica flavissima (Shima et al., 1994), des, Polybia liliacea, Po. rejecta, Po. spini-
Protonectarina sylveirae (Shima et al., fex, and Synoeca surinama (table 2). Other
1996a), Polybia dimidiata (Shima et al., species previously reported to have similar
1996b), Po. scutellaris, Po. paulista and Po. distinctionareProtonectarinasylveirae(Shi-
occidentalis (Noll and Zucchi, 2000), Age- ma et al., 1996a), Polybia dimidiata (Shima
laia vicina (Sakagami et al., 1996), and Ag. et al., 1996b), Po. scutellaris and Po.paulis-
weenalues.
betcev
en
erca
wfi
esgni
sesvalupriatesi
ao
cpr
cies hesehap
pe ntwit
S I
ed del.wn,
z oo
y mh
al nes
An ctiosar
eforthe sminantfunmevariable
sCovariat F-Statisticefinaldiscriusingthesa
7a 8dthA
TABLEkLength TABLEmbdaanmeasuretoorANOV
n a f
ngAlitru Wilks’LchseparateF-statistics
Usi eaes.
alyses utionfornofcast
n bo
A ntrinati
ce comi
an ofcri
ari eedis
v grw
Co delo
eg
hn
etati
matdic
stiin
e5,
es0.
valubout
aa
mbdand
La1.0