Insects as Selective Agents on Plant Vegetative Morphology: Egg Mimicry Reduces Egg Laying by Butterflies Author(s): Kathy S. Williams and Lawrence E. Gilbert Source: Science, New Series, Vol. 212, No. 4493 (Apr. 24, 1981), pp. 467-469 Published by: American Association for the Advancement of Science Stable URL: http://www.jstor.org/stable/1686077 Accessed: 07/04/2010 14:48

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aaas.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

American Association for the Advancement of Science is collaborating with JSTOR to digitize, preserve and extend access to Science.

http://www.jstor.org in the teleosts; living rep- 15. C. J. Platt, T. H. Bullock, G. Cieh, N. Kovace- 20. T. H. Bullock, in (II), p. 1. independently vic, D. Konjevic, M. Gojkovic, ibid. 95, 323 21. J. B. Johnston,J. Comp. Neurol. 12, 1 (1902). resentatives of the holostean fishes, the (1974);T. H. Bullock,J. Physiol. (Paris)75, 397 22. R. W. Murray,in (11), p. 125;J. H. Teeter, R. teleosts lack (1970). B. Szamier, M. V. L. Bennett, J. Comp. Phy- groupfrom-which evolved, 16. R. Schultz, E. C. Berkowitz, D. C. Pease, J. siol. 138, 213 (1980). a dorsalnucleus (1) and are not electro- Morphol.98, 251 (1956);J. K. Moore and T. H. 23. T. Szabo and A. Fessard, in (11), p. 60. Bullock, personalcommunication. 24. We thankT. H. Bullockfor supportthroughout receptive (19), and only 3 of 35 teleost 17. H. Kleerekoperand K. Sibakin,J. Fish, Res. the project and C. C. Bell for valuable com- orders are.electroreceptive(20). Board Can. 14, 145 (1957); E. A. Stensio, in ments on the manuscript.We also thank B. Traite de Zoologie, P. Grass6, Ed. (Masson, Monk and the National Marine Fisheries Ser- Although only a single type of lateral Paris, 1958), vol. 13, part 1, p. 173; K. S. vice for animalcollection. This study was sup- line has been for lam- Thomson,in Problemsin VertebrateEvolution, portedby an NIH fellowshipto D.B., and NIH receptor reported S. M. Andrews,R. S. Miles, A. D. Walker,Eds. and NSF researchgrants to R.G.N. and T. H. preys (2I), we have histologicallyidenti- (AcademicPress, New York, 1977),p. 247. Bullock. R.G.N.'s participationwas made pos- fied at that differ 18. A. J. Kalmijn,in (3), p. 507. sible by a GuggenheimFoundation fellowship. least.two populations 19. R. G. Northcutt, D. Bodznick, T. H. Bullock, * Present address: Departmentof Biology, Wes- with respect to their depth from the paper presented at the 28th InternationalCon- leyan University, Middletown,Conn. 06457. gress of PhysiologicalSciences, Budapest, 13 to epidermal surface and the diameter of 19 July 1980. 16 July 1980;revised 25 November 1980 their canals. We have not yet positively identifiedthe electroreceptors. The ho- mologies suggest, however, that the re- ceptors,-likethose of chondrichthianand chondrosteanfishes (22), will be ciliated Insects as Selective on Plant and excited by cathodal electric stimu- Agents Vegetative Morphology: li-that is,-negative electrical potential at Egg Mimicry Reduces Egg Laying by Butterflies the outer face of the receptor relative to the rest of the fish. In contrast, the Abstract. Experiments show that Heliconius butterflies are less likely to oviposit electroreceptorsof teleosts bear micro- on host plants that possess eggs or egglike plant structures. These egg mimics are an villi on their apical surfaces and are unambiguous example of a plant trait evolved in response to a host-restricted group excited by anodal stimuli (23). of insect herbivores. If electrosensory systems in lampreys and primitive jawed fishes are indeed The idea of coevolution between in- have helped to drive the evolution of the homologous, by definition they have sects and plants is attractiveto biologists presumeddefensive chemistry. Thus, in- been inheritedfrom the common ances- attempting to account for patterns of stead of a gene-for-gene coevolution, tor of agnathans and gnathostomes. plant chemistry and the use of plants by there may be a more diffuse process (3) Thus, the earliest vertebrateswere prob- insects (1). However, it is difficult to that is intractableto experimentalanaly- ably electroreceptive, and vertebrate demonstrate a causal connection be- sis. electroreceptors may be as ancient as tween a plant characteristicand a partic- One approachis to study plant groups lateralline mechanoreceptors. ular selective agent (2) because most that supportonly one or a few important DAVID BODZNICK* plants have been exposed through time herbivore taxa and that may therefore Departmentof Neurosciences, to a multitude of pathogens and herbi- have traits attributable to coevolution Universityof Californiaat San Diego, vores, any complement of which may with such taxa. For neotropicalvines of La Jolla 92095 the genus Passiflora, heliconiine butter- R. GLENN NORTHCUTT flies are likely selective agents on several Division of Biological Sciences, features of morphology(4). Plants such Universityof Michigan, as Passiflora may have effectively "fil- Ann Arbor48109 tered out" most potential herbivores at an early stage of the evolution of their Referencesand Notes defensive chemistry, so that the insects 1. C. A. McCormick,thesis, University of Michi- as herbivores of gan (1978). remaining significant 2. A. A. Pearson,J. Comp.Neurol. 64, 235 (1936); these plants are those that circumvent W. J. A. J. Smeets and R. Nieuwenhuys, ibid. 165, 333 (1976);R. G. Northcutt,in (3), p. 117. the chemical defenses of these plants. 3. E. S. Hodgson and R. F. Mathewson, Eds., Morphologicalrather than chemical in- Sensory Biology of Sharks, Skates, and Rays (Office of Naval Research, Departmentof the novation is the effective evolutionary Navy, Arlington,Va., 1978). response to such herbivores (5), and 4. R. L. Boord and C. B. G. Campbell,Am. Zool. 17, 431 (1977). C herbivorebehavior rather than counter- 5. D. M. Koester and R. L. Boord, ibid. 18, 587 defensive the course of (1978). chemistryshapes 6. R./G. Northcutt,Brain Res. 167, 163 (1979);in evolution (4). ComparativeStudies of Hearingin Vertebrates, A. N. Popper and R. R. Fay, Eds. (Springer- Structuresresembling the yellow eggs Verlag,New York, 1980). of Heliconius butterflies have arisen in- 7. P. J. McCreadyand R. L. Boord, J. Morphol. 150, 527 (1976). dependently on a number of Passiflora 8. D. Bodznick and R. G. Northcutt, Brain Res. species in several subgenera and are 195, 313 (1980). A if 9. Voltagepulses were suppliedby a square-pulse derived from several distinct structures stimulatorand isolation unit, and a 105to 106 (6). That these structures have evolved ohm series resistance was used to minimize 1. Passiflora used in electrode polarizationeffects. Electric field in- Fig. cuttings experiment: specifically to mimic Heliconius eggs is tensities were monitoredwith a pair of elec- (A) Passifloracyanea, showingdisplay of egg trodesin the water.The fieldswere only approx- mimics on stipule tips. (B) Passiflora oerste- indicatedby the facts that (i) heliconiines imatelyhomogeneous in the central area of the are of Passi- tank where the fish were held. dii, showing yellow egg (open circle) placed importantdefoliating agents 10. J. D. Green, Nature (London)182, 962 (1958). near green egg (closed circle) on tendril. (C) flora (7); (ii) larvae of many Heliconius 11. A. Fessard, Ed., Handbookof SensoryPhysiol- Enlargedview of P. cyanea stipules showing feed on congeneric eggs and larvae (6); ogy (Springer,New York, 1974),vol. 3, part 3. (top) unalteredstipule, (middle) stipule with 12. and females exhibit care in A. J. Kalmijn,in (II), p. 147. egg mimic removed, and (bottom) stipule cut (iii) great 13. , Nature (London)212, 1232(1966). sites In this 14. T. H. Bullock and J. T. Corwin, J. Comp. but retainingegg mimicfor control. Passiflora inspectingoviposition (6, 8). Physiol. 129A, 223 (1979). cyanea stipules are 3 to 4 cm in length. reportwe provideexperimental evidence SCIENCE,VOL. 212, 24 APRIL 1981 0036-8075/81/0424-0467$00.50/0Copyright ? 1981AAAS 467 which shows that Heliconius females green eggs, which were eggs that had eggs depositedto numberof inspections) discriminateagainst plants with eggs or been tinted with food coloringand rinsed on plants with no eggs was significantly egg mimics and are thereforelikely to be with distilled water to blend with the higher than on plants that had either a selective agents in the evolution of these plants' coloring; and washed yellow natural or washed yellow egg present structures. eggs, which were yellow eggs washed (Fig. 2, A and B) (13), indicatingthat the We studiedthe ovipositionbehavior of with distilled water and which served as presenceof a yellow Heliconius egg does Heliconius cydno in greenhouses, using controls. indeed reduce oviposition on plants. Passiflora oerstedii, a naturalhost plant In each test of oviposition preference, When eggs were laid on plants already without mimetic structures, and P. the butterflieswere offered four P. oer- bearinga yellow egg, they were usually cyanea, which has egg mimics on stip- stedii cuttings; two had single eggs of placed several centimeters away on an- ules (9). The golden color of the mimics one type and two had no eggs or had a other part of the cutting. closely resemblesthat of Heliconius eggs single egg of a different type. The cut- Egg-freecuttings and those with green just priorto hatching,and mimicstend to tings were arrangedat random with re- eggs had equal oviposition frequencies be clustered at meristems where many spect to one another and the butterflies (Fig. 2, C and D), showing that green Heliconius prefer to oviposit (Fig. 1A). were allowed to oviposit until they lost eggs were not recognized as something Althoughthe ranges of P. cyanea and H. interestin the plants. Most trials lasted 1 to be avoided. In fact, several eggs were cydno do not naturallyoverlap, H. cydno to 2 hours and the butterflieslaid eight to laid within 2 cm of green eggs (Fig. lB). larvae eat and develop successfully on ten eggs per trial. Washedyellow eggs reducedoviposition the plant (10). Both P. cyanea and P. The oviposition behavior of H. cydno as effectively as untreated yellow eggs oerstediiare membersof the series loba- was consistent. The butterflies,probably (Fig. 2, A and B), demonstratingthat no tae. respondingto a combinationof olfactory repellant chemical was washed off the In the first set of experiments, we and visual cues (11), usually noticed the eggs in the coloring treatment. examinedthe response of the butterflies host plants as soon as the plants were In experiments in which P. cyanea to the presence of real eggs on P. oerste- broughtinto the greenhouse. While flut- was used to test responses to egg mim- dii, the host withoutmimics. Host plants tering around a plant, they repeatedly ics, the H. cydno were kept in a green- were available to the butterflies only tappedit with their antennae, then land- house with Passifora hosts and with duringexperiments, when females were ed on the leaves to drumthe cuticle with other species of Heliconius. On days of presented with combinations of plant their forelegs, presumablyusing chemo- experimentsall Passiflora were removed cuttingswith and without eggs. The cut- receptorsto "taste" and furtheridentify from the greenhouse so that butterflies tings were of similarmorphology, and H. the plant (12). They would then fly would oviposit only on the test plants. cydno eggs were placed on tendrils near aroundthe plant, tapping and searching For each experiment two P. cyanea of meristemswhere eggs are naturallylaid. for a satisfactory oviposition site, or similar morphology were brought into Eggs laid in the course of each trial were reject the plant by flying away. Often, the greenhouse, and all butterflieswere immediately removed from the test when a butterfly noticed an egg or egg allowed to oviposit. plants. mimics, it would stop searchingthe plant Three types of plants were paired in Three types of H. cydno eggs were and fly to some other part of the green- these tests: one with naturalegg mimics, placed on the cuttings: bright yellow house. one with all egg mimics removed, and eggs, just as they appearedin the field; Percentoviposition (ratio of numberof one with the tip of each mimicclipped off so that the mimic's general appearance was not affected (control plants). The mimictips were clippedoff to controlfor chemical effects of removingstipule tips when mimicswere removedfrom experi-

40 40 -N=113 eggs mentalplants (Fig. 1C). II.:.:?.< Is.:: ~ Mean times: No egg 9.7 seconds Plants without egg mimics seemed to 20.?sl~~~ Yellowe I egg 21,8 seconds .: 20 be more satisfactoryfor ovipositionthan [ __ -] No egg plants with mimics 2E). The :::...... : egg (Fig. 0O?????~?[W gN S Yellow egg , Fig. 2. Graphs which percentage of oviposition on control No Yellow No Washed show re- eggs eggs eggs yellow 20 oviposition was not different from that on sponse to various plants to D natural plants (Fig. 2F), so again, the > treatments.(A 60 - p 5-_ p<.001P<- c and G) Experiments negativeoviposition response appearsto p were done with P. be due to visual ratherthan to chemical FieiN mimic oerstedii.(E to G) Ex- cues. i s c i s2-n - p c o 3 1 imi10c M perimentswere done .el with P. In The mean oviposition times for eggs 2020 - cyanea. (G) a. and the x-axis laid on P. oerstedii with real yellow eggs 40 - N=180 (H) M ? eggs No Na a aM e a n time s : indicates seconds was longer than for P. oerstedii with no 20 >ur4::] No Green Green Washed0 No mimic 15.4 seconds betweenrec- No Green Green Washed elapsed eggs present (Fig. 2, G and H) (14). A eggs eggs eggs yellow Mimic 21.9 seconds ognitionand oviposi- E ~~~~E F ~eggs , female spends more time selecting a sat- - - El No mimic tion. See text for de- N=690 inspections N=265 inspections - I to on with 60p< .001 _ P >.5 Mimic I tails. isfactoryplace oviposit plants or mimicsthan on 60 - 1 - 1 yellow eggs egg plants g g qi20O iiiii devoid of eggs or structures mimicking eggs. Indeed, in crowded greenhouse when all tendrils have 20 :U' cultures, eggs, females switch to secondary sites such as older leaves, dead tendrils, or nearby No Natural Control Natural 1-5 6-10 t1-15 16-20 21-25 >25 mimic mimic mimic mimic structures not normally used. The in- 468 SCIENCE, VOL. 212 crease in oviposition time appears to host plants less apparent or less recog- 12. Tarsalchemoreceptors are presentin manyoth- er insects [W. H. Calvert,Ann. Entomol. Soc. arise from the search for safe sites that nizable to the butterflies (21), egg mimic- Am. 67, 853 (1974);C. J. C. Rees, Entomol.Exp. are close to new without ry deters after the host has Appl. 12, 565 (1969)]. growth expos- oviposition 13. Datawere analyzedwith an arc sine transforma- ing eggs to cannibalism. Although egg- been discovered (22). Our study sup- tion for testing the equality of two percentages like structures do not the and is [R. R. Sokol and J. Rohlf, Biometry(Freeman, completely prevent ports egg mimicry hypothesis San Francisco, 1969),pp. 607-610]. oviposition, they do reduce oviposition an instance of a plant structural trait 14. Time was measuredfrom the butterfly'sinitial inspectionuntil her abdomencurled and ovipo- frequency. resulting from coevolution with an insect sition began. A t-test of the differencebetween Heliconius can have a severe impact herbivore. two means was used (ibid., pp. 220-223). 15. This'statementis based on 10 years' experience on its host plants. Females of several KATHYS. WILLIAMS* by L.E.G. in growing Heliconius with Passi- on LAWRENCEE. GILBERT flora. species frequently oviposit seedlings 16. D. Wiens, in Evolutionary Biology, M. K. and small vines in their forest understory Department of Zoology, Hecht, W. C. Steere, B. Wallace,Eds. (Plenum, habitats larva can de- Austin 78712 New York, 1978),vol. I1, p. 365. (10); single totally University of Texas, 17. Passifora ariculatahas well-developedegg mi- foliate individual plants. We routinely mics in Trinidad (modified petiolar nectar Referencesand Notes glands) but P. ariculata in Costa Rica is not observe suppression of flowering and mimetic. increased incidence of root disease in 1. P. R. Ehrlichand P. H. Raven,Evolution 18, 586 18. Selection of oviposition hosts by Heliconius (1965). may reduce the numberof potentially suitable greenhouse plants exposed to heliconiine 2. D. A. Jones, in Taxonomyand Ecology, V. H. Passiflora species used as larval hosts [J. T. Ed. New defoliation (15). Heywood, (Academic Press, York, Smiley, Science 201, 745 (1978)]. 1973),p. 213. 19. Heliconiusethilla in Trinidad,for example. Although we have stressed coevolu- 3. L. E. Gilbert, in Coevolutionof Animals and 20. L. E. Gilbertand J. T. Smiley, in Diversity of L. E. Gilbert and P. H. Eds. tion, the of this Plants, Raven, Insect Faunas, L. A. Mound and N. Waloff, plant mimicry example (Univ. of Texas Press, Austin, 1980),p. 247. Eds. (Blackwell,Oxford, 1978),pp. 89-104. has also received attention (16). The 4. , in ibid., p. 210. 21. L. E. Gilbert, in Analysis of Ecological Sys- 5. , Science 172, 585 (1971). tems, D. Horn et al.. Eds. (Ohio State existence of nonmimetic populations of 6. W. W. Benson, K. Brown, L. E. Gilbert,Evolu- Univ. Press, Columbus, 1979),p. 117. Passifora within some widespread spe- tion 29, 659 (1975). 22. In our trials, butterflieswere forced to choose 7. L. E. Gilbert, Colloq. Int. C.N.R.S. 265, 399 between two potential oviposition hosts, since cies suggests that the evolution of egg (1977). Flea beetles apparently are the only Heliconius butterfliesrarely encounterpatches other serious herbivoreof is an (17). Passiflora. of Passiflorain nature.M. Rothschildand L. M. mimicry ongoing process 8. The visual acuity of Heliconius is documented Schoonvoven[Nature (London) 266, 352 (1977)] However, every population in which we by C. A. Swihartand S. L. Swihart[Zoologica reported that Pieris brassicae chose between 155 Anim. Behav. 60 observed was for 48, (1963); 18, (1970);see two hosts with discriminatorybehavior similar mimicry monomorphic also (6)]. to that of Heliconius.With more potentialhosts the trait and was used by at least one 9. OriginalH. cydno stocks were collected by L. available, Pieris butterfliesmight spread their E. Gilbertat the Organizationfor TropicalStud- eggs less discriminantlythroughout the plant Heliconius species. Under conditions in ies, La Selva station, Costa Rica. Passiflora population,spreading the risksof overcrowding, which a nonmimetic population contain- oerstedii and P. cyanea stocks originatedfrom cannibalism,predation, and parasitismas P. M. populationsat La Selva and ArimaPass, Trini- Ives [Aust. J. Ecol. 3, 261 (1978)]proposes. ing a few mutant individuals is used by dad, respectively. Heliconius and Passflora 23. Facilitiesnecessary for this workwere provided one or more cannibalistic Heliconius species are maintained in temperature-con- by grants from the University Research Insti- trolled Lord and Burnhamglass houses (5 by 8 tute, University of Texas, and NSF grant GB species, the egg-mimic genotypes might m) at the Universityof Texas. 4074-P to L.E.G. We thank many friends and 10. J. of all nonmimetic indi- Smiley, thesis, University Texas, Austin colleaguesfor discussion of this work. eventually replace (1978). * Present address:Department of Biological Sci- viduals on the basis of relative (rather 11. M. D. Rausher has recently described search ences, Stanford University, Stanford, Calif. imageformation for leaf shape by the pipe vine 94305. than absolute) protection. swallowtail,Battus philenor [Science 200, 1071 In a complex community containing (1978)]. 10 July 1980;revised 26 November 1980 ten or more species each of Passifora and Heliconius, the microevolution of egg mimics by one Passiflora could all but eliminate that species from the host Internal Fertilization in an Oviparous Frog range of several coexisting Heliconius (18). At the same time, a new niche is Abstract. Eleutherodactylus coqui, an oviparous frog, undergoes internalfertiliza- then available to specialists able to uti- tion. If this mode offertilization occurs in other species of anurans, interpretations of lize the mimetic species (19). Passifiora anuran reproductive strategies based on the assumption of externalfertilization must with egg mimics should also excel at be reviewed. expanding into new habitats already in- fested with Heliconius since fake eggs Mode of fertilization is a fundamental truei, a unique species in which adult would provide some protection during factor in the evolution of reproductive males have a tail used as an intromittent the vulnerable establishment phase. strategies, in part because it constrains organ. Internal fertilization has been in- Thus it appears that egg mimicry repre- the nature of the parental investment by ferred for the few known live-bearing sents one way that a coevolutionary step each sex (1). In species with internal frog species (4, 5) but has never been might promote local diversity within a fertilization, the lower certainty of genet- demonstrated for an oviparous frog. We food web (20). ic relatedness of the male to the offspring present evidence that the oviparous frog We have demonstrated that (i) Heli- (compared to that of the female) has Eleutherodactylus coqui Thomas has in- conius females respond to the presence often resulted in the evolution of mater- ternal fertilization. of eggs; (ii) this response has a strong nal care systems, along with various Eleutherodactylus coqui is a noctur- visual basis (8), although chemical cues mechanisms to ensure paternity. Paren- nal, terrestrial-breeding frog of Puerto are not altogether excluded; and (iii) the tal care by males alone has been reported Rico (6). Males call from elevated perch- response to egglike structures of Passi- primarily in species with external fertil- es and exhibit parental care by attending flora and to real eggs both reduces the ization (2). eggs from oviposition to hatching. Eggs probability that eggs will be laid after Biologists have long assumed that undergo direct development and hatch as host discovery and increases the time most frogs employ external fertilization, tiny froglets. Males may remain with the required to oviposit. Whereas other visu- and anuran reproductive strategies have froglets for up to 5 days after hatching. al aspects of Passiflora that may be been interpreted in light of this assump- Courtship may occur at any time of related to the defense of Heliconius are tion (3). The only frog demonstrated to night. A gravid female approaches a call- directed toward making the potential have internal fertilization is Ascaphus ing male and makes contact. The male SCIENCE,VOL. 212. 24 APRIL 1981 0036-8075/81/0424-0469$00.50/0Copyright ? 1981AAAS 469