JournalofResearchontheLepidoptera 31(1-2):1-11, 1992 Biology and behavior ofthe neotropical butterfly Eunica hechina (Nymphalidae) with special reference to larval defence against ant predation.,^ Andre V. L. Freitas and Paulo S. Oliveira DepartamentodeZoologia, InstitutedeBiologia, UniversidadeEstadualdeCampinas, C.P. 6109, 13081-970 Campinas, Sao Paulo, BRAZIL ‘ = Abstract. This paper describes the biology and behavior ofEunica bechina (Nymphalidae). Eggs are laid singly on Caryocar hrasiliense (Caryocaraceae), a plantthatbears extrafloral nectaries. Mostofthe eggsarelaidonyoungleaves,onwhichthecaterpillarspreferablyfeed. Pupationoccursoffthehostplant.Thefifthinstarlarvaeandpupaeare like those ofNica flavilla and Temenis laothoCy suggesting that the genusis amongthe Callicorini. Firsttofourthinstarlarvae construct frass chains, where they rest immune from attacks by foraging ants that climb on Caryocar for its nectary secretions. While feeding on leaves, however, caterpillars maybe killed and removed by foraging ants. When attacked by ants, Eunica caterpillars may repel their aggressors by regurgitating and/or bleeding. Ants receiving these fluids exhibited strong disturbance and conspicuously cleaned their mandiblesandhead. Droppingofftheplantandhangingontheendof a drag line was also observed inEunica larvae after successive bites from the ants. We suggest that frass chains are probably related to defence against “walking” predators, especially ants, who have diffi- cultyinattackingthe caterpillars atthese refuges. KeyWords:Eunicahechina,Eurytelinae,Caryocarhrasiliense,cerrado vegetation, ant predation, extrafloral nectaries, herbivores, defensive behavior, frass chains ThegermsEunicaHtibner(1819),includes45speciesand24additional subspecies distributed throughout the Neotropical region, the majority intheAndeanRegionandtheAmazonBasin(Jenkins, 1990). Thegenus has an uncertain systematic position v^ithin the Nymphalidae (Otero, 1990)andfewlarvaeandhostplants areknown(Barcant, 1970; DeVries 1986, 1987; Ackery, 1988; Jenkins, 1990, Oliveira & Freitas, 1991). Eunica bechina magnipunctata Talbot 1928 occurs in the cerrados (savanna-like vegetation) of Central and Southeast Brazil (Jenkins, 1990; Oliveira & Freitas, 1991). Larvae ofE. bechina feed on leaves of Caryocarhrasiliense Camb. (Caryocaraceae), aplantbearingextrafloral nectariesandfrequentlyvisitedbyants(Oliveira&Oliveira-Filho, 1991; Oliveira & Brandao, 1991). Early instar larvae construct frass chains & (Oliveira Freitas,1991),abehavioralsoobservedinotherNymphalidae, especially amongthe Eurytelinae and Charaxinae (Muyshondt, 1973a, b, c 1974, 1976; Muyshondt & Muyshondt, 1976; Casagrande & Mielke, 2 J. Res. Lepid. 1985;DeVries, 1987;Aiello, 1991).Immaturestagesarestillundescribed formanygeneraandspeciesofNymphalidae;studiesoftheirmorphology andbehaviorcouldhelptounderstandtherelationshipamongmembers ofthisfamilyofbutterflies. Thepresentstudydescribestheearlystages ofE. bechina. We also provide dataonthenaturalhistoryofimmatures andadults, aswellasonlarvalbehavioranditsrelationtoantpredation on the host plants. Study sites and methods The study was carried out in a cerrado area in Itirapina (21°15'S, 47°49'W), Sao Paulo, SE Brazil during 1987, 1991 and 1992. The vegetation consists ofa scrub ofshrubs and trees, which is the cerrado sensu stricto ofGoodland (1971). Average annual rainfall and tempera- mm ture are ca. 1400 and 21°C respectively (Setzer, 1949). A total of 27 shrubs of Caryocar brasiliense (35-150 cm tall) were censused to determine the preference for oviposition sites by Eunica bechina. The eggs were collected and the larvae were reared in plastic boxes containingleaves ofCaryocar. Boxeswere cleaned andthe leaves replaceddaily.Eggsizeisgivenasheightanddiameter;theheadcapsule sizeisthedistancebetweenthetwogroupsofocelli;sizeofcephalichorns was also measured. The behavior ofEunica caterpillars andvisiting ants, as well as their responses to one another, were investigated through natural and pro- voked encounters on Caryocar shrubs. Encounters were provoked by removingthecaterpillarsfromtheirfrasschainsandplacingtheminthe proximityofdifferentantspecies.Larvaeofdiferentsizeswereplacedon leavesorbudsofant-occupiedshrubs.Aftertheantshadencounteredthe caterpillar,thebehavioralinteractionsbetweenthemwereregisteredin observation sessions lasting 15 - 30 min. A detailed account ofthe ant faunaassociatedwithCaryocarbrasilienseisgiveninOliveira&Brandao (1991). Results Descriptions ofearly stages Egg(Fig. lA):yellowish, conical, andflattenedatthetop,with 12to 14 longitudinal ridges and 10 to 12 transverse ridges. Average height 0.76 mm (a=0.03 mm, n=15); average diameter 0.72 mm (a=0.06 mm, n=15). Larvae hatch 5 days after oviposition (n=5). Firstinstarlarva(Fig. IB): headtranslucentbrown, bodytranslucent yellow changing to pale red after feeding (due to visible intestinal contents), legs and prolegs translucentyellow; maximum length 3 mm; average width of head capsule 0.42 mm (a=0.02 mm, n=16), average duration 2.6 days (a=0.54 days, n=47). The distribution ofsetae in the first instar larva is given in Fig 2A. Secondinstarlarva(Fig. 1C): headblackwithtwo shortstubbyhorns; bodypalebrownwithshortconicalscoli;maximumlength6mm;average ventral). (lateral, Pupa H, G, larva. instar Fifth F, 4 J. Res. Lepid. mm widthofheadcapsule0.67 (a=0.06,n=30);averagelengthofthehorn 0.33 mm (a=0.05 mm, n=30), average duration 1.5 days (a=0.59 days, n=46). Thirdinstarlarva(Fig. ID):Headblackwithwhitewartsandwithtwo long diverging horns armed with accessory spines in the middle and ending distally in a knob crowned with short spines; body dark brown with several scoli; maximum length 12 mm; average width of head capsule 1.26mm(a=0.06mm,n=43);averagelengthofthehorn2.36mm (a=0.20 mm, n=43); average duration 2.4 days (a=0.75 days, n=44). Fourthinstarlarva(Fig. IE): Headasinthirdinstar;bodydarkbrown withapalebrownlateralstripe;maximumlength20mm;averagewidth ofhead capsule 1.95 mm (a=0.06 mm, n=36); averagelengthofthehorn 4.56 mm (a=0.09 mm, n=36); average duration 3.7 days (a=1.47 days, n=33). Fifthinstarlarva(Fig. IF):Headasinfourthinstar;bodybrown,dorsal region dark brown with white lines and stripes, ventral region varies fromyellowtoorangeorred, sublateralstripeorangeorpaleyellow,legs dark brown and prolegs red exceptthe anal prolegs, (shinyblack), scoli blackwithyellow andreddots. The placementofthe scoliinthe bodyis shown in Figure 2B. Maximum length 40 mm; average width ofhead capsule 3.28 mm (a=0.10 mm, n=9); averagelengthofthehorn 6.62 mm ((j=0.30 mm, n=9); average duration 6.33 days (a=1.77 days, n=12). Prepupaassumesa“J”position,fixedonthesubstratebytheanalprolegs and abundant silk. There is no great change in color. Pupa (Fig. 1G,H): Green, purple or yellowish, changing to brown and grayorgreenafteroneortwodays;spiraculainconspicuouslightbrown; a dorsal indentation separates abdomen from thorax. Abdominal seg- ments are mobile; average size 2.2 cm (a=0.15 cm, n=ll), average duration 8.7 days (a=1.35 days, n=18). The sex ratio ofthe adults obtained in the laboratory (13 males and 9 females),canbeconsidered1:1(chisquaretest; =0.72,p>0.20;D.F.=1). Natural history Females ofEunica hechina lay their eggs singly on small shrubs of Caryocarbrasiliense between 10.00 and 13.00hours. Atotal of141 eggs werecensusedonCaryocar.Mosteggswerefoundonyoungleaves(87%), andlessfrequentlyonshoottips(10%),petioles(1%),andstems(1%).The verticaldistributionoftheeggsonthehostplantvariedfrom3to 150cm above the ground (x=60.5 cm, a=44.8 cm, n=141). The caterpillars eat part of the egg shell after hatching and feed preferentially on young leaves of C. brasiliense. Although E. bechina larvae were seen on CaryocarfromSeptembertoJanuary(rainyseason),thehighestinfesta- tionleveloccurredbetweenSeptemberandOctoberwhenthemajorityof the leaves are still young, soft, and red in color. FirsttofourthinstarlarvaeofE. bechinaconstructfrasschains, onthe tip ofwhichtheyrest (Figure 3A). When disturbed the caterpillars may 31(1-2):1-11, 1992 in O present LU scolum may 5-furcate larvae Many subdorsal absent. The generally larva. is instar A2 fifth a segment in scoli the of in scolum Distribution dorsal B, 5-furcate. simple bechina. The as A8 Eunica T2. of and of that A7 larva than segments instar larger bit first a the of is in T3 scoli Chaetotaxy segment dorsal A, the the 2. Figure 6 J. Res. Lepid. C, larva. instar third a retrieving nestmates. of blandus ait recruitment Camponotus after and Worker before B, Eunica, chain. of frass a larva on instar resting third a bechina attacking Eunica sp. of Azteca larva of instar Workers Third A, D, 3. Figure 1 31(1-2):1-11, 1992 7 jump off the leaf suspending themselves from silk threads. Pupation usually occurs offthe host plant, on neighbouring shrubs. Adults ofE. m hechina areeasilyseeninthefieldthroughouttheyearfl3dngabout3 high. Agonistic behavior and chases between males ofE. hechina are frequentlyobserved, suggestingakindofterritoriality. The males were seenfeedingonsapoozingfromtreewounds,andtheyprobablyalsofeed on decayingfruits and mud puddles (K. S. BrownJr., personal commu- nication), like other “fruit feedingnymphalids” (DeVries, 1988). Interactions between ants and caterpillars When not on their frass chains, E. hechina caterpillars may interact aggressively with the ants that climb on Caryocar attracted to its extrafloral nectary secretions. Behavioral interactions betweenEunica caterpillars and ants are summarized in Table 1. In all, 47 ant X caterpillarsencounterswereprovokedonCaryocarshrubs;in36ofthese the larvae were attacked byforaging ants. Such attacks resulted in the deathandremovalofthe caterpillarfromtheplantin20instances. Two Camponotus species (C. crassus and C. aff. hlandus) and one species of Azteca were most aggressive towards caterpillars. Unlike Camponotus whicharelargeenoughtosubdueandcarrythecaterpillaralonetotheir A nests(Fig. 3B),thesmall 2^eca antsrecruitedtensofnestmatestohelp with these tasks (Fig. 3C, D). Besidesthe7speciestestedagainstE. hechinalarvae(Table 1),wealso observed 6 other species attacking the caterpillars in the field (but withoutquantification),including2speciesofCrematogaster,2ofPheidole, one of Pseudomyrmex {pallidus group) and Ectatomma tuherculatum (Olivier). Table 1. Behavioral interactions between larvae of Eunica hechinaand ants on shrubs of Caryocarbrasiiiense. Results are based on 15-30 min of observation afterprovoked encounters between Eunica larvae and ants. Ant species N° of N° of N° of N° of ant Xlarvae larvae larvae larvae encounters attacked removed jumping of the leaf Camponotus crassus Mayr 15 10 6 3 C. aff. blandus (Fr. Smith) 10 10 7 1 C. rufipes (Fabricius) 1 1 1 0 C. renggeriEmery 2 2 0 1 C. aff. cingulatus Mayr 3 0 0 0 Azteca sp. 5 5 5 0 Zacryptoceruspusillus (Klug) 1 8 1 2 Total 47 36 20 7 8 J. Res. Lepid. When bitten by the ants, Eunica caterpillars frequently regurgitate and/orbleed, abehaviorshownto effectivelyrepeltheir aggressorswho mayendup abandoningthelarvae. Aftersuccessivebites onthelarvae, attackingantsfrequentlyexhibitstrongdisturbancebehaviorandvigor- ously clean their mandibles, antennae and head. Sevenlarvaedroppedofftheplantaftersuccessivebitesfromtheants (Table 1). In three instances the caterpillars suspended themselves on theendofasilkenlineforapproximately20minbeforeclimbingbackto theleaf. Fourotherlarvaedroppeddirectlytothegroundandhidamong the leaflitter. Discussion General biology This is the first detailed description of the biology and behavior of Eunica immatures (see also Jenkins, 1990). The distributionofsetaeinthefirstinstarlarvaeisverysimilartothe “primitive”patternofNymphalidae(Nakanishi, 1988). Thedistribution patternofthescoliofthefifthinstarlarvaeandpupaeislikethatoiNica flavilla and Temenis laothoe (Muyshondt, 1973b, c); however, the spine distribution pattern in N. flavilla is quite distinctive. The pupae of Eunica suggest that the genus is among the Callicorini, as stated by Otero(1990),whoplacesEunicainthemostadvancedbranchofCallicorini, paraphyleticwithTemenis andA^ica. However,Otero’sresultsarebased on only eight characters of adult morphology (apparently with high consistency). Onthe otherhand, Harvey(1991)proposes thditEunica is related with Myscelia, Catonephele, Nessaea, Cyhdelis and Lihythina (this one very close to Eunica) and the paleotropical genus Sallya (see alsoJenkins, 1990andOtero, 1990).Furtherstudyofimmaturesofother Eurytelinae genera may help solve some systematic problems in this group, as has been done for the Ithomiinae (K. S. Brown & A. V. L. Freitas, in preparation). Defence against ants Ants are the most frequent visitors to the extrafloral nectaries of Caryocar hrasiliense in the cerrado (Oliveira & Oliveira-Filho, 1991; Oliveira & Brandao, 1991). Foragingants mayencounterEunica cater- pillars on leaves and occasionally remove them from Caryocar. All ant genera observed attacking E. hechina caterpillars are known to tend Lycaenidae and Riodinidae larvae, Homoptera, and to visit extrafloral nectaries (see DeVries, 1991; Oliveira & Brandao, 1991). Azteca ants, however, were also observedkillingThisbe irenea (Riodinidae) caterpil- lars (DeVries, 1991). We noted that some caterpillars can overcome predation orinjurybyants through an arrayofbehavioral mechanisms (see also Heads & Lawton, 1985; Costa et al. 1992). Thebehaviorofsuspendingthemselvesbysilkenthreadsiscommonin E. hechina caterpillars and appears to be widespread among the Lepi- 31(1-2):1-11, 1992 9 doptera(see alsoDeVries, 1987 andseveralcitationstherein). Dropping and suspending on the end of a drag line is a technique known to be employed by arthropods who live in close proximity to aggressive ants (Robinson & Valerio, 1977; Oliveira & Sazima, 1984, 1985; Heads & Lawton, 1985). Regurgitation is also common in butterflies (see Brower, 1984). Just afterattackingthecaterpillars,antsreceivingthisfluidexhibitedstrong disturbance (walking erratically and shaking the body) and conspicu- ously cleaned their mandibles and head. Ant deterrence can also occur from bleeding by the injured caterpillars, as also noted by Heads & Lawton 1985)forsomeherbivoresofbrackenfern(Pteridiumaquilinum). ( Rearingup the body, curling andwrigglingvigorously (beat reflex) can intimidateortemporarilyexpelsomepredatorsfromtheplant. Forsome ants,however,thebeatreflexmaystimulateadditionalattacks(Malicky, 1970). Thesebehaviors areverycommon amongbutterflylarvae, except forsome Lycaenoideae(Malicky, 1970), andseemtobemoreeffectivein lateinstarcaterpillarsduetotheirlargersizeinrelationtotheants(see also Heads & Lawton, 1985). The frass chains constructed bythelarvae may diminishtheirpreda- tion/removal by ants, since the latter were never observed climbing on thisstructure.Thebehaviorofrestingortakingrefugeonfrasschainsis analogoustothatexhibitedbysomeHeliconiinilarvae,whichrestatthe end of tendrils or on “island-like” leaf segments (Benson et al., 1976; Bentley & Benson, 1988). Frass chains are observed in several other larvaeofNymphalidaebutterfliesfeedingonvariousplantfamilies.This trait is especially common among the Charaxinae and Limenitidinae (sensu Harvey, 1991), a fact supporting the idea that this structure permitstheutilizationbyEunica ofaplantoftenoccupiedbyants. This behavior needs to be studied in other genera ofNymphalidae such as Hamadryas sndAnaea, whose larvae commonlyfeed on plants bearing extrafloral nectaries. Although the primaryrole offrass chains has not been tested so far, we suggest that it is related to defence against “walking”predators, speciallyants,thatwouldhavedifficultyinattack- ingthe caterpillars on their chains. Acknowledgements. We thank Dr. K. S. Brown Jr., C. F. Klitzke, K. Fiedler and P. J. DeVriesfordiscussions andhelpful comments onthe manuscript. R. B. Francinihelpedwiththephotographs. S. GeraldiandE. Z. Borghimadethe linedrawings. Financialsupportto P. S. Oliveirawas providedbygrants from the CNPq (no.300101/90-2 and 400692/92-9), FAPESP (no. 90/2775-6) and FAEP/UNICAMP(no. 634/91). A. V. L. Freitas also acknowledges afellowship from CAPES. Literature cited Ackery, P. R. 1988. Hostplants and classification: a review of nymphalid butterflies. Biological Journal ofthe Linnean Society, 33: 95-203. 10 J. Res. Lepid. Aiello, A. 1991. Adelpha ixia leucas immature stages and position within Adelpha(Nymphalidae).JournaloftheLepidopterists’Society,45: 181-187. Barcant, M. 1970. Butterflies ofTrinidad and Tobago. Collins, London. Benson,W.W.,K. S. Brownjr. &L. E. Gilbert. 1976. Coevolutionofplantsand herbivores: passion flowerbutterflies. Evolution, 29: 659-680. Bentley, B. L. &W.W. Benson. 1988.Theinfluenceofantforagingpatternson thebehaviorofherbivores.In:Trager,J.C.(ed.):AdvancesinMyrmecology. E. J. Brill, NewYork. 297-306. Brower, L. P. 1984. Chemical defence in butterflies, in R. I. Vane-Wright and P. R.Ackery(eds). TheBiologyofButterflies. Academicpress, pp. 109-134. Casagrande, M. M. & O. H. H. Mielke. 1985. Estagios imaturos de Agrias claudinaclaudianusStaudinger(Lepidoptera,Nymphalidae,Charaxinae). Revistabrasileira de Entomologia, 29: 139-142. Costa, F. M. C. B., A. T. Oliveira-Filho & P. S. Oliveira. 1992. The role of extrafloral nectaries in Qualea grandiflora (Vochysiaceae) in limiting herbivory:anexperimentofantprotectionincerradovegetation. Ecological Entomology, 17: 363-365. DeVries,P.J. 1986.HostplantrecordsandnaturalhistorynotesonCostaRican butterflies(Papilionidae,Pieridae&Nymphalidae).JournalofResearchon the Lepidoptera, 24: 290-333. 1987. The butterflies ofCosta Rica and their natural history. Princeton University Press, Princeton, New Jersey. 1988.Stratificationoffruit-feedingnymphalidbutterfliesinaCostaRican rainforest. Journal ofResearch on the Lepidoptera, 26: 98-108. 1991. MutualismbetweenThisheireneabutterfliesandants, andtherole ofantecologyintheevolutionoflarval-antassociations. BiologicalJournal ofthe Linnean Society, 43: 179-195. Goodland,R. 1971.Aphysiognomicanalysisofthecerradovegetationofcentral Brazil. Journal ofEcology, 59: 411-419. Harvey, D. J. 1991. Higherclassification oftheNymphalidae (AppendixB). in H. F. Nijhout, The development and evolution ofbutterfly wing patterns, Smithsonian Press, pp. 255-273. Heads, P. A. &J. H. Lawton. 1985. Braken, ants and extrafloral nectaries. III. Howinsecthebivoresavoidantpredation.EcologicalEntomology, 10:29-42. Jenkins, D. W. 1990. Neotropical Nymphalidae VIII. Revision of Eunica. Bulletin ofthe Allyn Museum, 131: 1-177. Malicky, H. 1970. New aspects of the association between lycaenid larvae (Lycaenidae) and ants (Formicidae, Hymenoptera). Journal of the Lepidopterists’ society, 24: 109-202. Muyshondt, a. 1973a. Notes onthelife cycle and naturalhistoryofbutterflies of El Salvador. II A.-Epiphile adrasta adrasta (Nymphalidae- Catonephelinae). Journal oftheNewYorkEntomological Society, 81: 214- 223. 1973b. Notes on the life cycle and natural history of butterflies of El Salvador. Ill A.-Temenis laothoe liheria (Nymphalidae-Catonephelinae). Journal ofthe NewYork Entomological Society, 81: 224-233. 1973c. Notes on the life cycle and natural history of butterflies of El Salvador.IVA.-PseudonicafLavillacanthara(Nymphalidae-Catonephelinae). Journal ofthe NewYork Entomological Society, 81: 234-242.