13 December 1988

PROC. ENTOMOL. SOC. W ASH.91(1), 1989, pp. 81-87

ANTS ASSOCIATED WITH A COLEOPTEROUS LEAF-BUDGALL ON XYLOPIA AROMATICA (ANNONACEAE)

O. WILSON FERNANDES, WILLIAM J. BOECKLEN, RoGÉRIO P. MARTINS,AND ALEXANDRE G- CASTRO

(GWF, WJB) Box 5640, Department of Biological Sciences, Northem Arizona Uni-versity, F1agstaff, Arizona 86011 ; RPM, AGC Departamento de Biologia Geral, CaixaPostal 2486, ICB/Universidade Federal de Minas Gerais, 30.000-Belo Hofizonte-MG,Brazil.

Abstract. -We describe the ant fauna (11 species divided into four genera) associatedwith a coleopterous leaf-bud gall on Xylopia aromatica (Annonaceae) in Minas Gerais,Brazil. Ant occupied ga11s are significantly larger than are unoccupied galls, and the vari-ance of gall diameters for occupied galls is significantly smaller than is the variance forall galls. There is a significant positive relationship between ant colony size and galldiameter. The six most common species of ants may partition galls according to gall size,as there are significant differences in the diameters of occupied galls among species. Atthe community level, pattems of co-occurrence of ants are indistinguishable from thoseexpected under a random assortment model. Individual ant species do exhibit non-randompattems of co-occurrence.

Key Words; Annonaceae, ants, Brazil, coleopterous gall, community ecology, habitatselection, insect galls, Minas Gerais, resource partitioning, Xylopia aro-matica

ga11s. Mani (1964) divides the 1ocatari into33 categories according to their eco1ogica1niches. The 1ocatari often represent a vastfauna with the most numerous categoriesconsisting of parasites and predators. Forexamp1e, Stegagno (1904) reported 177species associated with cynipid ga11s onQuercus in Ita1y; ofthese, 138 are parasitesand predators.

Species that inhabit ga11s after emergenceof ga11 formers and inqui1ines are the "suc-cessori" (Mani 1964, Yukawa 1983). Mostof these are p1ant-nesting ants and myr-mecophi1ous insects, such as aphids andcoccids. Others inc1ude mites, spiders,thrips, co11embo1a, bees, and wasps (re-viewed by Mani 1964). The most commonant genera associated with old ga11s are:Camponotus, Cataulacus, Crematogaster,

P1ant ga11s represent an imp rtant re-source for many species other th n the ga11formers. Owing to their 1oca1iz d concen-tration ofnutritive tissues and th .r markedsuccu1ence, ga11s provide favoura e breed-ing sites for a variety of species (Bra orst1962, Mani 1964, Shorthouse 1973, Yu-kawa 1983). Ga11s protect their inhabitantsfrom inc1ement weather (Fe1t 1940, Uh1er1951, Sand1ant 1979; but see Baust et a1.1979) and natural enemies (Askew 1961,1980, but see Price et a1. 1986, 1987). Useof ga11s ranges from species that open ga11ssimp1y to prey upon ga11 formers and in-qui1ines to those that depend exc1usive1y onga11 tissues for food and she1ter.

Beauvisage (1883, cited in Mani 1964)app1ied the term "locatari" to species, otherthan the ga11 formers, associated with insect

82 PROCEEDINGS OF THE ENTOMOLOGICAL SOCIETY OF WASHINGTON

45° 40°15°

?nO

Fig. Location of the Ecological Station of Parapitinga in Três Macias, Minas Gerais, Brasil.

Colobopsis, Lasius, Leptothora.x, and Olop-sis(Walsh 1864, Patton 1879,Whee1er 1910,Ping 1920, Sturtevant 1925, Mani 1964,Torossian 1971a, b). Despite their impor-tance and dominancewithin the succe~sori,1itt1e is known about the eco1ogies of/theseants other than compendiums ofant ~ Cies associated with particu1ar species of 9 1 for-

mers (e.g. Brandhorst 1962, Espada1e and

Nieves 1983).

Here, we describe the ant fauna assocl-

ated with a co1eopterous leaf-bud ga11 on

Xylopia aromatica (Annonaceae). We de-

scribe pattems in distribution and abun-

dance of the ant fauna and document e1e-

ments of habitat se1ection and resource

partitioning.

MATERIAL AND METHODS

One ofus (G.W.F.) collected galls fromthe Ecological Station of Parapitinga, TrêsMarias, Minas Gerais, Brazil during onesample period in mid-May, 1984. The sta-tion is a continental island located in a man-

made lake (Três Marias Lake) between lat-itude 18°-19° south and longitude 45°-46°west (Fig. 1).

The host tree, Xylopia aromatica, oc-curred in a gallery forest along the northwestborder ofthe island. Only three individuaIsof approximately six meters tall bemng gallswere observed in the area surveyed. Thegalls are induced by an unidentified speciesof curculionid (Coleoptera). Gall formers-held as vouchers were kept in the author'scollection. The galls are spherical and gla-brous and occur on leaf-buds (Fig. 2). Theyare green when occupied by the gall former ,but turn brown after its emergence. AlI gallswhich were on trees (n = 114) were collectedand subsequently measured and dissectedin the laboratory .Galls that falI on the forestfloor are also utilized by ants; however theyare not included here because of insufficientsample size.

In this paper, we use the word "colony"as any group of ants, composed of workersand larvae (eggs, queen, and pupae if pres-

VOLUME 91, NUMBER 83

~~I //,

~

I~ -/

-

~

-,.""'

~

"'~

:.: ?/

---

~

~

, ,""'

~

,--f

-

<~~ /

I; -.' ,. .

::'.' ;

~

~

~i

TFig. 2. Co1eopterous 1eaf-bud ga11s found on Xylopia aromatica. Gal1 diameter ranged from 5.0 to

21.5 mm.

ent), occupying a single gall. In addition, wedefined those galls in which we found onlyworkers as "groups ofworkers."

analysis ofvariance. We also compared thevariaIice of occupied galls with the varianceof all galls (including occupied galls) with aChi-square test (Sokal and Rohlf 1969, page175). We assumed that our collection con-sisted of the entire population of galls onthe island and that the set of occupied gallswas a sample from that popu1ation.

STATISTICAL ANALYSES

We compared the mean diameter ofgalls

that were occupied by ants with the mean

diameter ofunoccupied galls using one-way

PROCEEDINGS OF THE ENTOMOLOGICAL SOCIETY OF WASHINGTON84

Table 1 Ant species associated with coleopterous leaf-bud galls on Xylopia aromatica (Annonaceae).

(x) IndividuaisGaJl Diameter

(X)mmAnt Species Number of Colonies Adult Larvae/Pupae

1.028.1

5.445.0

0.81.01.02.0

51.35.74.0

14.516.414.717.016.412.515.012.915.914.910.0

11713

I4117

811

1

-

8.9

4.6

37.0

5.5

12.5

2.2

4.0

Azteca bicolorAzteca sp.Leptothorax wilda

Leptothorax sp.Pseudomyrmex jlavidusPseudomyrmex sp. APseudomyrmex sp. B

Zacryptocerus pallensZacryptocerus pusillusZacryptocerus sp. AZacryptocerus sp. B

We compared the mean diameters of oc-cupied galls among the six most commonspecies of ants by one-way analysis of vari-ance.

We estimated the relationship between thelogarithm of colony size (number of indi-viduals, of all castes, in a gall) and gall di-ameter through linear regression.

Finally, we examined pattems of occur-rence within galls. We compared the num-ber of galls that contained O, 1, 2, and 3species of ants with those expected under ~Poisson distribution. We estimated À, t1iePoisson parameter, from the sample. ~ ecompared the observed distribution to teexpected distribution with a Chi-squa egoodness-of-fit test.

RESULTS

We co11ected 114 ga11s ofwhich 52 (45.6%)were occupied by ants. E1even species ofants were represented in the samp1e; themost common were Azteca sp. which oc-cupied 17 ga11s, Leptothora.x wilda whichoccupied 13 galls, and Zacryptocerus sp. " A"

which occupied 11 galls (Tab1e 1). Fivespecies were represented by a sing1e indi-vidual. "Co1ony sizes" ranged from threeindividua1s to a co1ony of Zacryptoceruspusillus which contained 129 adu1ts and 1ar-vae.

The interna1 stf1,lcture of the galls washigh1y modified in some cases, unmodified

in others (Fig. 3). Modification is defined asany intemal architectural departure, such astunnel and gall wall holes, from the usualspheroid larval chamber. In particular, gallsinhabited by Azteca sp. and Zacryptoceruspusillus exhibited extensive modificationswhen compared to unmodified galls. Gallsinhabited by Pseudomyrmex jlavidus wereunmodified. Modified galls typically con-tained larger colonies than did unmodifiedgalls.

Ants typically occupied large galls (Fig.4). The mean diameter of occupied galls,15.6 mm, was significantly larger than themean diameter of unoccupied galls, 13. 6 mm(F1.112 = 12.8, P < 0.001). The variance ofdiameters of occupied galls was significantly~ler than would be expected if ants wereselecting galls at random (X2113 = 72.1, P <

0.005).There was a significantly positive linear

relationship between the logarithm of col-ony size (number of individuals) and galldiameter (Fig. 5).

The six more common species of ants fur-ther partitioned the subset of occupied gallsaccording to gall size; mean gall diametersof occupied galls were significantly differentamong species (F5,54 = 2.74, P < 0.05). Az-teca sp. typically occupied the largest galls(x = 16.4 mm), Zacryptocerus pallens thesmallest (x = 12.9 mm).

We collected 65 colonies distributed

VOLUME 91, NUMBER 85

BA

Dc

among 114 ga11s. We estimated À, the Pois-son parameter, as 65/114 = 0.57. The ob-served distribution of co1onies among ga11sclose1y resembled the expected under aPoisson distribution (Table 2). The good-ness-of-fit test indicated no significant dif-

ference between the observed and expected~istributions (X23 = 2.51, p > 0.10).

~vidua1 species did exhibit frequenciesof co-occurrence that appeared differentfrom those expected by chance alone (Table3). For example, Zacryptocerus pusillus

~,

2.4.

.. ....... .

~ : ..., .: ..

..

,2.0.24D < 0001U)

~<::>o

>

õz

(.!>o~

~ '

./4

.$

.6

2

0.6

-;-0.0

9:0 12.0 15.0 IB.o 2i.0

GALL DIAMETERFig. 5. Relationship between the 10garithm of col-

ony size (number of individuais) and gall diameter.TherelationshipisY= -0.929 + 0.122X(F,.63=20.1,p < 0_001: R2 = 0_24)-

>-(.)zw;:)d

~LL

W>

~<-JWa:

5 7 .III. 15 17 ,. 21 2.

GALL DIAMETER ( )

Fig. 4. Relative frequency histograms of gall di-

ameter for the entire set of galls (foreground) and for

the subset of galls that were occupied by ants (back-

ground).

Fig. 3. Cross-sectio;; of galls showing intemal modtfications by ants. Gall (A) contained Azteca sp., (B)Leptothorax sp., (C) Zacryptocerus pallens, (D) Zacryplocerus pu.sillu.s, and (E) Pseudomyrmex jlavidu.s.

PROCEEDINGS OF THE ENTOMOLOGICAL SOCIETY OF WASHINGTON86

Table 2. Distribution of colonies among galls. Ex-pected values are based on a Poisson distribution withÀ= 0.57.

Table 3. Pattems of co-occurrences among antspecies.

In;nt """'._n~

Colonies/Gall Observed Expected Species Observed Expected

oI23

62

42

7

3

64.536.810.5

2.0

0.587.155.930.582.130.580.583.543.995.190.58

43O31O5O5O

Azteca bicolorAzteca sp.Leptothorax wilda

Leptothorax sp.Pseudomyrmex jlavidusPseudomyrmex sp. APseudomyrmex sp. B

Zacryptocerus pallensZacryptocerus pusillusZacryptocerus sp. AZacryptocerus sp. B

never jointly occupied a gall, while approx-imately 4 such co-occurrences wou1d be ex-pected at random. Azteca sp. and Lepto-thorax wilda were roughly half as likely toco-inhabit a ga11 as chance alone would pre-dict. On the other hand, Zacryptocerus pal-lens exhibited a s1ight tendency to share ga11s.

DISCUSSION

Ga11s represent an important resource forthis ant community, and ants exp1oit thisresource in a non-random fashion. Ants se-1ect significant1y 1arger ga11s, over a narrow-er range of sizes, than chance a1one wou1dpredict. Both the difference in means andthe difference in variance indicate habitatse1ection. There may be strong se1ectivepressures for this habitat se1ection as indi-cated by the significant positive re1ationshipbetween co1ony size and ga11 size. Se1ectionof1arger ga11s by ants may be due to severa1reasons, among them increase in queen fit-ness. However, more work is ca11ed for toobserve the se1ection and the use of 1argerversus sma11er ga11s.

It is not c1ear whether ga11s are a 1imitingresource to these ants. On the other hand,the most common species of ants exhibitedresource partitioning. This is a necessarycondition for stab1e coexistence of mu1ti-species assemb1ages exp1oiting similar, 1im-iting resources predicted by Lotka- Vo1terrabased ana1yses of community dynamics(May 1973, Schoener 1974). Ofcourse, nei-ther observation by itse1fisconc1usive. Un-occupied ga11s may be the resu1t of interfer-ence competition 9r priority effects (sensuTorres 1984). Significant differences among

species in the diameters of ()ccupied gallsmay not be the result of competitively in-duced resource partitioning, but rather ,species' idiosyncratic responses to their en-vironment (James et al. 1984). Correlativedata/f1rovide notoriously weak inferencesregarding mechanisms (Brady 1979); ex-~riments are required to establish causa-

tlon.\ Patterns of species co-occurrence at the~ ~munitY level are indistinguishable from

th se predicted from a random assortment

mo. However at the constituent level,individual species deviate from expecta-tion. In particular, Zacryptocerus pusillusnever shares a gall, although approximatelyfour joint occurrences are predicted. Fowleret al. (1985) provide evidence that Z. pusil-lus is interspecifica1ly territorial. Whetherdeviations from chance for the other speciesare biologically significant is unclear. Un-fortunately, little is known about the ecol-ogies of these ants.

Gall-inhabiting ants may provide an idealsystem to examine the determinants ofcommunity organization in arboreal ants.Galls are a discrete resource, amenable toexperimental manipu1ation in time andspace.

ACKNOWLEDGMENTS

We thank Hélcio R. Pimenta and Ivo dasChagas for field assistance. We also thank

87VOLUME 91. NUMBER 1

A. Mabelis and w. w. Benson for infor-mation on ants, H. Castanheira and L. M.Araújo for identification of ants, L. Ki-noshita for identification of Xylopia aro-matica, and W. E. Clark for identificationof the gall former insect. In addition, wethank the comments of H. Larew and ananonymous reViewer on an early draft ofthis manuscript. The Departamento deBiologia Geral, ICB, Universidade Federalde Minas Gerais and the staff of Codevasf-Três Marias, MG provided logistical sup-port. This work was supported by a spe-cialization fellowship from SEMAlCNPq (n°10.6193/84) to G. W. Femandes.

May, R. M. 1973. Stability and complexity in modelecosystems. Princeton University Press, Prince-ton.

Patton, W. H. 1879. Agall-inhabitingant. AmericanNaturalist 13: 126-127.

Ping, C. 1920. Some inhabitants of the round gall ofgolden-rod. Journal of Entomological Zoology 7:161-177.

Price, P. W., G. W. Fernandes, and G. L. Waring.1987. Adaptive nature of insect galls. Environ-mental Entomology 16: 15-24.

Price, P. W., G. L. Waring, and G. W. Fernandes.1986. Hypothesis on the adaptive nature of galls.Proceedings of the Entomological Society ofWashington 88: 361-363.

Sandlant, G. R. 1979. Arthropod successori inhab-iting willow galls during autumn in Christchurch,New Zealand. Mauri Ora 7: 83-93.

Schoener, T. W. 1974. Resource partitioning in eco-logical communities. Scienc,e 185: 27-39.

Shorthouse, J. D. 1973. The insect community as-sociated with rose galls of Diplolepis polita (Cy-nipidae, Hymenoptera). Quaestones Entomologi-cae 9: 55-98.

Sokal, R. R. and F. J. Rohlf. 1969. Biometry. Free-man, San Francisco.

Stegagno, G. 1904. I locatari dei cecidozoi sin quinoti in Italia. Marcellia 3: 18-53.

Sturtevant, A. H. 1925. Notes on the ant fauna ofoak galls in Woods Hole region. Psyche 32: 313-314.

Torossian, C. 1971a. Faune secondaire des galles deCynipidae: I. Etude systématique des fourmis etdes principaux arthropods récoltés dans les galles.Insectes Sociaux 18: 135-154.

-.1971 b. Etude biologique des fourmis forest-ieres peuplant les galles de Cynipidae des chênes.Insectes Sociaux 18: 193-202.

Torres, J. A. 1984. Niches and coexistence of antcommunities in Puerto Rico: Repeated patterns.Biotropica 16: 284-295.

Uhler,C.D. 1951. BiologyandEcologyofthegolden-rod gall fly, Eurosta solidaginis (Fitch). CornellUniversity Agricultural Station Memoirs 300.

Walsh, B. D. 1864. On insects, coleopterous, hy-menopterous, and dipterous, inhabiting the gallsof certain species of willows. Proceedings of theEntomological Society ofPhiladelphia 3: 543-641.

Wheeler; W. M. 1910. Ants. Columbia UniversityPress, New York.

Yukawa, J. 1983. Arthropod community centeredupon the neolitsea leaf gall midge, Pseudoasphon-dylia neolitseae Yukawa (Diptera: Cecidomyiidae)and its host plant, Neolitsea sericea (Blume) Koldz.(Lauraceae). Memoirs of the Faculty of Agricul-ture 19: 89-96.

LITERA TURE CITED

Askew,R.R. 1961. Onthebiologyoftheinhabitantsof oak galls of Cynipidae (Hymenoptera) in Brit-ain. Transactions of the Society of British Ento-mology 14: 237-268.

-.1980. The diversity of insect com nitiesin leaf mines and plant galls. Journal of Anima1Ecology 49: 817-829.

Baust, J. G., R. Grandee, G. Condon, and .E. Mor-rissey. 1979. The diversity of ove nteringstrategies utilized by separate populatio of ga11insects. Physiological Zoology 52: 572-58

Brady, R. H. 1979. Natural selection and the crite .aby which a theory is judged. Systematic Zoology28: 600-621.

Brandhorst, C. T. 1962. The microcommunity as-sociated with the gall of Walshia amorphella (Lep-idoptera: Cosmopterygidae) on Amorpha fruti-cosa. Annals of the Entomological Society ofAmerica 55: 476-479.

Espadaler, X. andJ. L. Nieves. 1983. Hormigas(Hy-menoptera, Formicidae) probladoras de aga11asabandonadas de cinipideos (Hymenoptera, Cyni-pidae) sobre Quercus sp. en la peninsu1a Iberica.Boletin de la Estación Central de Ecologia 12: 89-93.

Felt, E. P. 1940. Plant Ga11s and Gall Makers. Com-stock, Ithaca. New York.

Fowler, H. G., M. M. Costa, and J. Justi. 1985. Re-lação e reconhecimento individua1 em Paracryp-tocerus pusillus (K1ug). XII Congresso Brasileirode Zoologia (abstract), Campinas, Brasil, Page 130.

James, F. C., R. F. Johnston, N. O. Wamer, J. Niemi,and W. J. Boeck1en. 1984. The Grinne1liannicheofthe wood thrush. American Natura1ist 124: 17-30.

Mani, M. S. 1964, Ecology ofPlant Ga1ls. W. Junk,The Hague.