BIOLOGY Saliva or Regurgitated Nectar? What Butterflies (: ) Use for Pollen Feeding

1,2 1 3 1 STEFAN H. EBERHARD, ANNA L. HIKL, CAROL L. BOGGS, AND HARALD W. KRENN

Ann. Entomol. Soc. Am. 102(6): 1105Ð1108 (2009) ABSTRACT The Neotropical Heliconius butterßies are well known to supplement their nectar diet by active pollen collecting. They extract proteins and free amino acids from pollen grains, exhibiting a particular behavior that involves the use of a ßuid of uncertain origin. It has been assumed that this ßuid is either regurgitated nectar or saliva, because for anatomical reasons a butterßy is able to release only these two ßuids through its proboscis. In an experimental approach, 27 individuals of Heliconius melpomene (L.) were given red-dyed sugar solution and subsequently we observed whether the ßuid used in pollen feeding was dyed or not dyed. Because regurgitated nectar should contain sugar, ßuid samples were taken from the proboscis of butterßies from natural populations in Costa Rica. Samples of 44 individuals from seven species were tested for the presence of fructose and glucose with the aid of aniline phthalate. This study is the Þrst detailed investigation of the origin of the ßuid used by Heliconius butterßies in pollen feeding. The results are discussed in terms of already existing hints in literature concerning the true nature of that ßuid.

KEY WORDS heliconiines, Laparus, sugar assay, nutritional ecology

Butterßies of the genera Heliconius and Laparus ac- speciÞcally are able to discharge saliva from the tip of tively collect pollen with their proboscises, in addition the proboscis to dissolve solid or viscous substances to nectar feeding. This unique behavior supplies these (Knopp and Krenn 2003). Although the salivary glands butterßies with proteins and free amino acids and is a of Heliconius do not differ in terms of anatomy and key innovation for the sophisticated biology of pollen histology from those of other nonpollen feeding feeding heliconiines (Gilbert 1972, Gilbert 1991). Es- nymphalids (Eberhard and Krenn 2003), pollen feed- sential amino acids derived from pollen feeding have ers have disproportionately larger salivary glands been shown to be transferred directly to eggs (OÕBrien (Eberhard et al. 2009). Combined with the fact that a et al. 2003). greater amount of ßuid is used for pollen feeding than During pollen feeding, the pollen remains on the for nectar feeding (Penz and Krenn 2000), this hints outer proboscis surface. A clear liquid is discharged, that saliva is important in pollen feeding. saliva wherein pollen is processed for up to several hours by is an aqueous solution of different enzymes, depend- partial uncoiling and recoiling of the proboscis. Dur- ing on the mode of nutrition (Ribeiro 1995). Eberhard ing pollen processing the pollen grains become me- et al. (2007) detected protease activity in the saliva of chanically damaged, which aids nutrient acquisition Heliconius melpomene (L.), which may improve pollen for Heliconius butterßies (Krenn et al. 2009). For an- digestion. In contrast, nectar is a solution of varying atomical reasons, only saliva or regurgitated nectar concentrations of different sugars, primarily sucrose, can be released from the food canal of a butterßyÕs glucose, and fructose (Galetto and Bernardello 2005). proboscis (Eberhard and Krenn 2003), so the process- In addition, germination of pollen grains in differently ing liquid must be one or the other, or a mixture of concentrated sugar solutions could not be observed both. Gilbert (1972) had good reasons to suppose that (H.W.K., unpublished data). the liquid used for pollen processing is regurgitated In the current study, we use an experimental ap- nectar. He referred to studies that indicated that ger- proach to determine whether pollen processing in- minating pollen placed in a sucrose medium released volves regurgitated nectar or salivary ßuid or a mixture free amino acids and proteins (Stanley and Linskens of both. 1965, Linskens and Schrauwen 1969). However, in general use saliva to facilitate Materials and Methods the uptake of food (Ribeiro 1995), and butterßies Experiment 1: Dyed Nectar. To check if the ßuid 1 Department of Evolutionary Biology, University of Vienna, Al- used in pollen feeding is regurgitated nectar, 27 indi- thanstrasse 14, A-1090 Vienna, Austria. viduals from a H. melpomene greenhouse population 2 Corresponding author, e-mail: [email protected]. 3 Department of Biology, Stanford University, Stanford, CA 94305- were allowed to feed on nectar, which had been dyed 5020. with a red food coloring. Subsequently, pollen of ei-

0013-8746/09/1105Ð1108$04.00/0 ᭧ 2009 Entomological Society of America 1106 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 6

proboscis with a needle. Most butterßies started pol- len processing behavior and released a clear liquid. This mixture of liquid and glass beads was rinsed off the proboscis into a vessel with 0.05 ml of distilled water. To test for sugar, a small amount of the liquid was applied with a micropipette onto a precoated sheet for thin layer chromatography (TLC). The TLC sheet was put on a hot-plate (45ЊC) until the applied liquid spots were dried. This procedure was repeated until the whole sample from an individual was applied on one spot on the TLC sheet. Three control samples (distilled water and 3 and 0.3% sugar solution) were applied in the same way. The TLC sheet was then sprayed with aniline phthalate and heated at 125ЊC. Aniline phthalate reacts only with reducing sugars, such as glucose and fructose (not with sucrose) during heat exposure. Presence of any kind of reducing sugar H melpomene Fig. 1. Pollen-feeding . (Nymphalidae). in the sample results in a color reaction. In preliminary The pollen on the proboscis (arrow) is processed by using tests, even a 0.05% glucose solution resulted in a color salivary ßuid. reaction. Test 2 was performed similarly, but without Þrst ther pumpkin (Cucurbita pepo L.) or sunßower (He- isolating and putting the individuals on sugar solution. lianthus annuus L.) was placed on the proboscis by The proboscis was only rinsed with distilled water if it using a needle. These speciesÕ pollen was chosen be- was contaminated with pollen. In total, 12 individuals cause the size is similar to pollen used in the wild. In (eight females and four males) were tested; these most cases, the butterßy immediately released a drop- individuals also had been used in test 1. let of ßuid from the proboscis and started pollen- Experiment 3: Field Samples. In total, 44 Heliconius processing behavior (Fig. 1). The color of this ßuid and Laparus butterßies were netted at the Tropical was recorded. A few individuals did not start process- Research Station La Gamba, Puntarenas, Costa Rica ing pollen; thus, no ßuid release was observable. Some (8Њ 45Ј N, 83Њ 10Ј W; 81 m above sea level) in February butterßies were tested with pollen from both plant 2007. The butterßies were put in a net cage, and ßuid species (Table 1). samples were extracted by putting glass beads on the Experiment 2: Sugar Content of Fluid. We per- proboscis. When a butterßy performed pollen-pro- formed two different tests for sugar in the regurgitated cessing behavior, the mixture of released ßuid and ßuid, by using only those individuals that exhibited glass beads was rinsed off with distilled water into a processing behavior and ßuid release. small vessel. Afterward, the butterßies were released In test 1, 15 individuals of H. melpomene were tested where they had been caught. The ßuid sample from in total (10 females, four males, and one undetermined each individual was applied onto a Þlter paper (seven gender). The individuals were isolated overnight and droplets of Ϸ3.3 ␮l each) by using a micropipette. provided with water next morning. One hour later, These Þlter papers were dried at 50ЊC for 12 h and then they were put on cotton soaked with a 30% (wt:wt) stored in an airtight container with silica gel. Those sugar solution (sucrose:glucose:fructose, 1:1:1) and Þlter papers were tested for reducing sugars using were allowed to feed ad libitum. After a butterßy had aniline phthalate as described above. The threshold ceased feeding, its proboscis was rinsed twice with for detecting reducing sugars with this method is be- distilled water to remove any remaining sugar. Then, tween 0.1 and 1% reducing sugar solution. glass beads (Ϸ106 ␮m in diameter) were placed on the Results from a laboratory population Experiment 1: Dyed Nectar. In total, 33 trials were (27 ؍ Table 1. H. melpomene (N were tested for the color of the released fluid during pollen pro- performed with 27 H. melpomene individuals, which cessing after intake of dyed sugar solution (experiment 1) had been fed red-dyed nectar. In 18 trials with Tested with pumpkin pollen, the ßuid droplets released on the Pumpkin Sunßower Total proboscis during pollen processing were crystal pollen pollen clear and thus recorded as undyed (Table 1). No droplets were released in four further trials with Individuals tested 22 11 27 (6 twice) Undyed ßuid released 18 5 23 pumpkin pollen. In Þve tests using sunßower pollen, Dyed ßuid released 0 0 0 the ßuid droplets also were undyed and in six tests No ßuid released 4 6 10 no ßuid was released. Red dyed ßuid was never Total no. of tests 22 11 33 visible on the proboscis (Table 1). Experiment 2: Sugar Content of Fluid. In test 1, For these tests, pumpkin pollen (C. pepo) and sunßower pollen (H. annuus) were used. Six individuals were tested with both speciesÕ butterßies were isolated overnight and then fed a pollen. sugar solution before the test. Three ßuid samples November 2009 EBERHARD ET AL.: POLLEN FEEDING IN Heliconius 1107

Table 2. Evidence for sugar in the fluid used in pollen pro- without feeding the butterßies before ßuid sampling, cessing by H. melpomene no sugar was detected. This indicates that neither lab populations of H. melpomene nor natural populations Evidence for sugar Test Total of various Heliconius species and Laparus doris (L.) Positive Negative use regurgitated nectar for pollen processing. How- 1 3 12 15 ever, sugar was present in three ßuid samples from 2 0 12 12 butterßies that were fed on cotton soaked with a sugar solution before sampling. Because the butterßiesÕ tarsi Test 1 was performed with isolated individuals that had been fed with a 30% sugar solution directly before the test. For test 2, the came into contact with sugar, we cannot exclude butterßies were not specially isolated or fed. Postive, color reaction the possibility that rapidly moving legs might have indicating presence of reducing sugars. Negative, no color reaction. touched the proboscis. This may have caused contam- ination with sugar before the ßuid sample was re- moved. Furthermore, it cannot be entirely excluded from 15 tested H. melpomene individuals displayed a that the butterßies may have released small amounts positive color reaction with aniline phthalate, indicat- of nectar through the proboscis during experimental ing that sugar was present. In 12 ßuid samples no handling. So, the positive sugar samples are not proof reducing sugar could be found (Table 2). For com- for regurgitated nectar. parison, all control sugar solutions of 3 and 0.3% ex- The food canal of the butterßiesÕ proboscis is con- hibited the brownish staining for a positive reaction of nected with the sucking pump in the head and leads reducing sugars with aniline phthalate after heating. into the esophagus. Beneath the sucking pump, the In test 2, butterßies were not isolated before the salivary gland opens via a salivary pump into the food test. All ßuid samples from the 12 H. melpomene in- canal (Eberhard and Krenn 2003). Therefore, if re- dividuals displayed a negative color reaction (Table gurgitated nectar is excluded as the pollen processing 2). No reducing sugar could be detected in the ßuid, ßuid, the only ßuid that may enter the food canal is whereas the control sugar solutions (3 and 0.3%) ex- saliva. Moreover, the oral valve, which is situated in hibited a positive reaction. butterßies between the food canal and the sucking Experiment 3: Field Samples. Fluid samples from pump, should prevent the reßux of nectar into the the proboscis of 44 butterßies from seven species were proboscis (Eberhard and Krenn 2005). tested for reducing sugars (Table 3). No color reaction The ability of butterßies to discharge a ßuid from was detectable with aniline phthalate, indicating that the proboscis (Knopp and Krenn 2003) and the fact no reducing sugars were present in the ßuid released that Heliconius needs a comparatively high amount of to process the glass beads (Table 3). ßuid for pollen processing (Penz and Krenn 2000) suggested the crucial role of saliva in pollen feeding. Discussion In previous studies, protease activity was detected in the pollen processing ßuid (Eberhard et al. 2007), and All three experiments clearly indicate that pollen it is very unlikely that those proteases originate from feeding butterßies do not use regurgitated nectar for nectar. A biometric study of the salivary glands pollen processing. None of the tested H. melpomene showed that pollen-feeding heliconiines have dispro- individuals released a red dyed ßuid during pollen portionately larger glands than comparable nymphalid feeding, although they had been fed with red-dyed species (Eberhard et al. 2009). The larger salivary nectar before (Table 1). The possibility that the food glands of pollen feeding species were interpreted as an coloring was chemically destroyed in the butterßyÕs adaptation to pollen feeding (Eberhard et al. 2009). digestive tract and that therefore no dye could be Taken together with the results presented in this detected seems unlikely, because the droppings of the study, we conclude that the ßuid used in pollen feed- butterßies were red colored for hours after the trials ing probably is saliva. were performed. In addition, the tests for the presence of sugar in the released ßuid gave unambiguous results. In all tests Acknowledgments We thank S. Schmid, S. Suette, and D. Senkpiel for help in Table 3. Species, number of individuals, and results of fluid performing the experiments and the staff of the Tropical samples tested for reducing sugars from butterflies of the Tropical Research Station La Gamba, Costa Rica, for providing ex- Research Station La Gamba, Costa Rica cellent facilities. The Costa Rican Ministerio del Ambients y Energia kindly granted research permits. We also thank the No. Evidence for sugar Department of Ecophysiology and Functional Anatomy of Species individuals Positive Negative Plants (University of Vienna) for granting us permission to a greenhouse. The study was funded by the Austrian Science Heliconius charithonia (L.) 1 0 1 Fund (FWF-Project 18425-B03). Heliconius hecale (F.) 4 0 4 Heliconius hewitsoni Hewitson 11 0 11 H. melpomene 404 References Cited Heliconius pachinus Salvin 12 0 12 Heliconius sara (F.) 8 0 8 L. doris Eberhard, S. H., and H. W. Krenn. 2003. Salivary glands and 404 salivary pumps in adult Nymphalidae (Lepidoptera). Total 44 0 44 Zoomorphology 122: 161Ð167. 1108 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 102, no. 6

Eberhard, S. H., and H. W. Krenn. 2005. Anatomy of the Knopp, M.C.N., and H. W. Krenn. 2003. EfÞciency of fruit oral valve in nymphalid butterßies and a functional juice feeding in Morpho peleides (Nymphalidae, Lepi- model for ßuid uptake in Lepidoptera. Zool. Anz. 243: doptera). J. Insect Behav. 16: 67Ð77. 305Ð312. Krenn, H. W., M.J.B. Eberhard, S. H. Eberhard, A.-L. Hikl, Eberhard, S. H., N. Hrassnigg, K. Crailsheim, and H. W. W. Huber, and L. E. Gilbert. 2009. Mechanical damage Krenn. 2007. Evidence of protease in the saliva of the to pollen aids nutrient acquisition in Heliconius butterßies butterßy Heliconius melpomene (L.) (Nymphalidae, Lep- (Nymphalidae). Arthropod Plant Interact. (doi: 10.1007/ idoptera). J. Insect Physiol. 53: 126Ð131. s11829-009-9074-7). Eberhard, S. H., H. L. Nemeschkal, and H. W. Krenn. Linskens, H. F., and J. Schrauwen. 1969. The release of free 2009. Biometrical evidence for adaptations of the sal- amino acids from germinating pollen. Acta Bot. Neerl. 18: ivary glands to pollen feeding in Heliconius butterßies 605Ð614. O’Brien, D. M., C. L. Boggs, and M. L. Fogel. 2003. Pollen (Lepidoptera: Nymphalidae). Biol. J. Linn. Soc. 97: feeding in the butterßy Heliconius charitonia: isotopic 604Ð612. evidence for essential amino acid transfer from pollen to Galetto, L., and G. Bernardello. 2005. Nectar, pp. 261Ð313. In eggs. Proc. R. Soc. Lond. B Biol. 270: 2631Ð2636. A. Dafni, P. G. Kevan, and B. C. Husband [eds.], Penz, C. M., and H. W. Krenn. 2000. Behavioral adaptations Practical pollination biology. Enviroquest Ltd., Cam- to pollen-feeding in Heliconius butterßies (Nymphalidae, bridge, ON, Canada. ): an experiment using Lantana ßowers. J. In- Gilbert, L. E. 1972. Pollen feeding and reproductive biology sect Behav. 13: 856Ð880. Heliconius of butterßies. Proc. Natl. Acad. Sci. U.S.A. 69: Ribeiro, J.M.C. 1995. Insect saliva: function, biochemistry 1403Ð1407. and physiology, pp. 74Ð97. In R. F. Chapman and G. de Gilbert, L. E. 1991. Biodiversity of a Central American Heli- Boer [eds.], Regulatory mechanism in insect feeding. conius community: pattern, process and problems, pp. Chapman & Hall, New York. 403Ð427. In P. W. Price, T. M. Lewinsohn, G. W. Fer- Stanley, R. G., and H. F. Linskens. 1965. Protein diffusion nandes, and W. W. Benson [eds.], PlantÐanimal interac- from germinating pollen. Physiol. Plant 18: 47Ð53. tions: evolutionary ecology in tropical and temperate regions. Wiley, New York. Received 18 May 2009; accepted 8 September 2009.