TRADE-OFFS IN ANTIHERBIVORE DEFENSES IN Piper cenocladum: ANT MUTUALISTS VERSUS PLANT SECONDARY METABOLITES

Transcription

1 Journal of Chemical Ecology, Vol. 27, No. 3, 2001 TRADE-OFFS IN ANTIHERBIVORE DEFENSES IN Piper cenocladum: ANT MUTUALISTS VERSUS PLANT SECONDARY METABOLITES LEE A. DYER, 1,2, * CRAIG D. DODSON, 2 JON BEIHOFFER, 3 and DEBORAH K. LETOURNEAU 2,4 1 Department of Ecology and Evolutionary Biology Tulane University New Orleans, Louisiana Western Colorado Center For Tropical Research Mesa State College Grand Junction, Colorado US EPA, National Enforcement Investigation Center Bldg 53, Box 25227, DFC Lakewood, Colorado Department of Environmental Studies University of California Santa Cruz, California (Received November 16, 1999; accepted November 14, 2000) Abstract Ant plant mutualisms may provide indirect evidence for costs of antiherbivore defenses when plants demonstrate trade-offs between allocating resources and energy into ant attractants versus chemical defenses. We tested the hypothesis that ecological trade-offs in defenses are present in Piper cenocladum. This plant possesses two distinct defenses: food bodies that attract predatory ants that destory herbivore eggs and amides that deter herbivores. Previous studies have demonstrated that the food bodies in P. cenocladum are an effective defense because the ants deter herbivory by specialist herbivores. Amides in other Piper species have been shown to have toxic qualities, but we tested the additional hypothesis that these amides have an actual defensive function in P. cenocladum. To test for ecological trade-offs between the two putative defenses, fragments of P. cenocladum were examined for the presence of amides both when the plant was producing food bodies and when it was not producing food bodies. Plants with active ant colonies had redundant defenses, producing food bodies and high levels of amides at the same time, but we detected a trade-off in that they had significantly lower levels of amides than did plants with no ants. To test for the defensive value * To whom correspondence should be addressed / 00/ $19.50/ Plenum Publishing Corporation

2 582 DYER, DODSON, BEIHOFFER, AND LETOURNEAU of P. cenocladum amides, we used an ant bioassay and we examined herbivory results from previous experiments with plants that had variable levels of amides. These tests demonstrated that amides are deterrent to omnivorous ants, leaf cutting ants, and orthopterans. In contrast, the resident Pheidole bicornis ants are effective at deterring herbivory by specialist herbivores that oviposit eggs on the plant but not at deterring herbivory by nonresident omnivores. We concluded that although both amides and food body production appear to be costly, redundancy in defenses is necessary to avoid damage by a complex suit of herbivores. Key Words Piper cenocladum, ant plants, chemical defense, trade-offs, amides, herbivory, predation, Costa Rica, mutualism. INTRODUCTION Defenses against herbivores are assumed to be costly in terms of fitness-enhancing functions (Cates and Orians, 1975; Levin, 1976; Fox, 1981; Gould, 1983, 1988; Gershenzon, 1994; Sagers and Coley, 1995; Elle et al., 1999), but these costs are presumably outweighed by the benefits associated with lower levels of herbivory (Zangerl and Bazzaz, 1992; Simms, 1992). As a result of these costs, ecological and evolutionary defensive trade-offs are expected when resources are limited (Rehr et al., 1973; Simms, 1992). For example, a plant that invests heavily in trichomes as a defense should have low levels of chemical defenses compared to a similar species that has a low density of foliar trichomes or compared to an individual of the same species that has decreased its investment in trichome production. Such trade-offs can be difficult to detect (Simms, 1992), and some authors have even found that allocation of resources to multiple defenses is common in plants that need to defend against many different types of herbivores (e.g., Lindroth and Hwang, 1996). While the presence of trade-offs is consistent with the hypothesis that defenses are costly, allocation of resources to multiple defenses, or redundancy in defenses (Romeo et al., 1996), does not imply absence of cost nor does it exclude the possibility of ecological trade-offs. Limited resources can still cause reduction of one redundant defense in response to an increase in another. Nevertheless, studies uncovering redundant defenses in ant plant mutualisms have been used to argue that trade-offs between antiherbivore defenses do not occur and that the defenses are not costly because plants are able to allocate resources and energy into chemical defenses as well as ant attractants (biotic defenses). Steward and Keeler (1988) found that the mean number of different indole alkaloid compounds was not lower in species of Ipomoea that produce extrafloral nectaries compared to nonnectary plants; these results have been used as weak evidence that neither alkaloids nor nectaries are costly (Simms, 1992). Letourneau and Barbosa (1999) found that a mechanical defense (trichomes) in Endospermum

3 TRADE-OFFS IN Piper DEFENSES 583 was induced both in the presence and absence of a biotic defense (ant defenders); this result implies that the mechanical defense is costly because it is induced (Baldwin, 1998), but that there is no trade-off between the mechanical and biotic defense. Janzen (1973) found that individuals of five Cecropia species found in areas without mutualistic Azteca ants did not produce mullerian bodies for these ants; he concluded that this biotic defense was metabolically expensive, but he did not examine the system for trade-offs with production of chemical defenses. Only one study with ant plant mutualisms has indicated that a trade-off may exist between biotic and chemical defenses, suggesting that both types of defenses are costly. Rehr et al. (1973) found that Acacia species without mutualistic Pseudomyrmex ants had leaves with cyanogenic glycosides, whereas a species with resident ants did not have cyanogenic glycosides. Furthermore, diets made from the nonant species adversely affected the feeding efficiency of Prodenia eridania (Noctuidae) larvae. These results have been used as evidence to support the hypothesis that there are costs associated with biotic defenses as well as chemical defenses (Simms, 1992). However, HCN did not affect the larvae in this study (Rehr et al., 1973), thus their conclusions rely on the assumption that the nonant Acacias contained an unidentified defensive compound that affects herbivores; it is equally plausible that nutritional factors other than plant defenses affected larval feeding efficiency. We tested the general hypothesis of trade-offs in energy and resource allocation to competing defenses using an ant plant that facultatively produces food bodies in the presence of a specific ant species and that varies in levels of defensive secondary compounds. Piper cenocladum (Piperaceae) is an understory shrub that produces lipid- and amino acid-rich food bodies for a resident ant, Pheidole bicornis (Formicidae: Myrmicinae), which protects the plants from specialist herbivores (Letourneau, 1983). Plants in the Piperaceae commonly have high levels of secondary compounds that have antifeedant activities (reviewed by Parmar et al., 1997), and P. cenocladum has high levels of three amides that may be deterrent to herbivores (Dodson et al., 2000; Dyer and Letourneau, 1999a). We conducted a study to test the hypothesis that production of food bodies and secondary compounds is metabolically expensive so that high investment in one defense will lead to absence or lower levels of the other defense. Regardless of the presence of trade-offs, the fact remains that this plant species has redundancy in its antiherbivore defenses; thus we also tested the hypothesis that the multiple defenses are necessary because amides in P. cenocladum deter those consumers that are not deterred by ants: generalist herbivores and omnivores. METHODS AND MATERIALS Study System. Piper cenocladum is an understory shrub (usually less than 4 m tall) common in lowland wet forests in Costa Rica (Burger, 1971). The species

4 584 DYER, DODSON, BEIHOFFER, AND LETOURNEAU reproduces both vegetatively through layering (fallen shrubs root adventitiously) or fragmentation (petioles and twigs break off and root) and through seed (Gartner, 1989; Greig, 1993). The plant is commonly found as a shrub as well as a fragment (Gartner, 1989), and the biology of the two architectures can be very different (Dyer and Letourneau, 1999b). The leaves are large (most leaves fall within the range of cm 2 when fully expanded) and long-lived (approximately 2 years), and opalescent food bodies are produced on the adaxial surface of sheathing leaf bases (petiolar cavities) when occupied by Pheidole bicornis ants. The herbivores most commonly found feeding on Piper ant plants at our study site are specialist lepidopterans and coleopterans. Feeding damage from generalist herbivores, including leafcutter ants (Hymenoptera: Formicidae: Atta cephalotes) and orthopterans (Orthoptera: Tetigoniidae, Acrididae: Microtylopteryx hebardi, and Eumasticidae: Homeomastax robertsi) is occasionally found. The main herbivores that eat leaf tissue are geometrid moth larvae (Lepidoptera: Geometridae: Cambogia sp. and Eois sp.), skippers (Lepidoptera: Hesperiidae: Quadrus cerealis), saddleback caterpillars (Lepidoptera: Limacodidae), weevils (Coleoptera: Curculionidae: Ambates spp.), and at least 10 species of flea beetles (Coleoptera: Chrysomelidae: Physimera spp.) (Marquis, 1991; Dyer, Letourneau, and G. Gentry unpublished data). Pheidole bicornis is a small, dimorphic species that occupies Piper cenocladum, harvests the food bodies produced by the plant, and removes insect eggs, some vines, and small phylloplane particles from the leaves (Risch et al., 1977; Letourneau, 1983, 1998). Production of food bodies by the plant is induced by an unknown compound produced by the ants (Dodson et al., unpublished data), and the food bodies are produced continually as a source of nutrition for the ants, in contrast to the occasional herbivores ants collect from the plant surface (Letourneau, unpublished data). Other species of Pheidole and other ants are occasionally found in the petiolar cavities of P. cenocladum, but they do not induce food body production by the plant. The phytochemistry of the family Piperaceae and the genus Piper is well documented. Members of the genus Piper are known to produce alkaloids, aromatic hydrocarbons, oxygenated cyclohexanes, terpenes, chalcones, flavones, phenyl propenes, lignans, neolignans, and amides of a characteristic type sometimes referred to as Piper amides (reviewed by Parmar et al., 1997). These amides contain a phenyl moiety with a variable length carbon side chain (typically with at least one unsaturation) ending in a carbonyl carbon. The nitrogencontaining portion of the amide is derived from piperidine, pyrrole, or an isobutyl group and may contain an unsaturation and/ or a carbonyl group. Piper cenocladum contains several amides at high concentrations (up to 0.58% dry weight): piplartine, 4 -desmethyl piplartine, and cenocladamide (Dodson et al., 2000). To our knowledge, none of the other Piper ant plants (P. sagittifolium C.DC., P.

5 TRADE-OFFS IN Piper DEFENSES 585 obliquum Ruiz & Pavon, and P. fimbriulatum C.DC.) have been investigated for secondary compounds. A closely related species without ant mutualists, Piper imperiale, also contains high levels of amides (C. Dodson and J. Searcy, unpublished data). Many Piper amides that have been investigated for biological activity have been demonstrated to be insecticidal (Gbewonyo et al., 1993; Su and Horvat, 1981; Miyakado et al., 1989) or deterrent to leaf-cutting ants (Capron and Wiemer, 1996). One of the amides in P. cenocladum, piplartine, is known to be cytotoxic in vitro (Duh et al., 1990). Study Site. Plant and insect collection areas were located at the La Selva Biological Station, Heredia Province, Costa Rica, located at N, W at ca. 100 m elevation on the Caribbean slope (Hartshorn and Hammel, 1994). The natural history of this lowland wet forest is described in great detail in McDade et al. (1994). We used plant fragments from previous experiments that had been established on a hectare of primary forest at about 1400 m on the Sendero Jaguar trail at La Selva. The soil at this site is the poorest of soil types where P. cenocladum grows at La Selva (Sollins et al., 1994), and the understory vegetation is similar to other primary and secondary forest areas where shrubs and fragments are found. Quantification of Amides. We examined amides in experimental fragments. Plant samples were prepared as described in Dodson et al. (2000) and included 1- g replicate samples from individual plant fragments from La Selva. We harvested leaves from a two-year experimental study where fragments had been established in the forest (see Dyer and Letourneau, 1999a, for complete methods). The leaves were harvested from five samples of P. cenocladum fragments with ants (each sample included a homogenate of plants, with a mean of 4.4 plants per sample) and five samples of P. cenocladum fragments from that ants had been excluded (a mean of 3.8 plants per sample). We chose most recently expanded leaves that were similar in size and that had low levels of herbivory (<10%) and low epiphyll loads. Ant exclusion was accomplished by applying to the petioles one to two drops of dilute Diazinon insecticide (0.85 mg wettable powder per liter of distilled water) every two to three months for 1.8 years (see Dyer and Letourneau, 1999a for methods). Experiments examining the effects of these dilute insecticide applications indicated no measurable effects of the insecticides on plant biomass, herbivory, epiphyll cover, and other variables (Letourneau and Dyer, 1998; Letourneau, Dyer, and G. Vega, unpublished observations). When the plant fragments were harvested, mean number of ants per plant per sample was 4 ± 3 SE for the ant exclusion plants and 129 ± 56 SE for the unmanipulated ant plants. None of these 10 samples were manipulated in any other way (i.e., no fertilizer was added to the soil, light was not manipulated). Leaves were dried at room temperature, ground, and extracted with 95% ethanol; the amides being isolated, as well as most putative Piper plant defenses, are stable at room temperature (Dodson et al., 2000; Parmar et al., 1997). The

6 586 DYER, DODSON, BEIHOFFER, AND LETOURNEAU crude extract residue was resuspended in 3 : 1 water ethanol, which was then exhaustively extracted with chloroform and quantitatively analyzed by GC-MS using commercially available piperine as an internal standard at the 80 mg/ ml level. Five point calibrations (50, 100, 200, 300, and 500 mg/ ml) were prepared using synthetic piplartine (r ) and synthetic 4 -desmethylpiplartine (r ). Cenocladamide is unstable in solution, so its structural isomer, 4 - desmethylpiplartine was used as a standard, thus all concentrations we report for cenocladamide are estimates (Dodson et al., 2000). We used multivariate analysis of variance (MANOVA) to examine the effects of ant exclusion on the concentrations of the three amides. We used profile analysis (sensu Tabachnick and Fidell, 1996) to detect any differences in responses of the three amides to the ant exclusion. The Wilks lambda statistic was used for all hypothesis tests. Assays for Other Compounds. To examine the possibility of trade-offs between ant defense and other chemical defenses, we assayed for other common defensive compounds. TLC and GC-MS of all crude residues and 1 H NMR of the other fractions obtained in previously described chromatography (Dodson et al., 2000) were conducted to detect phenyl propanoids, aromatic hydrocarbons, oxygenated cyclohexanes, lignans, and neolignans in bulk samples of naturally occurring shrubs with and without ants. A separate sample of crude extract was screened for alkaloids (differential ph partitioning and TLC with iodoplatinic acid visualization). Deterrence of Amides to Omnivores and Generalist Herbivores. Leaves from bulk shrub samples with high (1.36% dry weight) versus low (0.41% dry weight) levels of total amides were tested for palatability to insects using a bioassay with the giant tropical ant, Paraponera clavata (Formicidae: Ponerinae). The two distinct levels of amides were obtained by using naturally occurring shrubs without ants (high amides) and shrubs with ants (low amides). The Paraponera bioassay has successfully identified deterrent compounds in caterpillars and herbivore toxins in plants such as nicotine and solapalmatine in normal plant concentrations (Dyer, 1995, 1997; Dodson, Dyer, and G. Gentry, unpublished data). The bioassay only uncovers very deterrent compounds because the ants are less selective than other arthropods with respect to diet (Dyer, 1997), and extracts that deterred the ants were even more deterrent to wasps and bugs (Dyer, 1997). Complete methods for the bioassay are in Dyer (1995). Briefly, 15 g dry weight of each plant treatment (ants versus no ants) was extracted in 250 ml methanol. A 6.7% solution of this extract in 20% (by weight) sugar water and a control (6.7% methanol in 20% sugar water) were offered simultaneously to ants in 2.4-ml microcentrifuge tubes. Tubes were weighed before and after being exposed to foraging ants for 1 hr. A total of 150 extracts (each offered with a control tube) per plant treatment were offered to 15 colonies found between 2000 and 3200 m along the Sendero Tres Rios trail at La Selva. Adjusted consump-

7 TRADE-OFFS IN Piper DEFENSES 587 tion differences (sensu Dyer, 1995) were calculated for each offering. This value compares consumption of extract versus control vials, adjusting for total consumption; for unpalatable extracts, a higher positive adjusted consumption difference indicates greater unpalatability. Mean adjusted consumption differences for the two plant treatments (high versus low amides) were compared using a t test (N 15 colonies). We conducted new analyses on folivory data from concurrent experiments (Dyer and Letourneau, 1999a,b) to test the hypothesis that P. cenocladum amides are deterrent to generalist herbivores. In one experiment (Dyer and Letourneau, 1999a,b) all leaves from 47 experimental plant fragments with ants and all leaves from 44 fragments from which ants had been removed were examined for damage from various orthopterans. This is the same experiment from which leaves were harvested from fragments for the chemical analysis. In a second experiment (Dyer and Letourneau, 1999b), all leaves from 40 shrubs with ants and all leaves from 37 shrubs from which ants had been excluded were examined for damage from Atta cephalotes and various orthopterans. We used chi-square statistics to test hypotheses of association between presence of ants and presence of herbivore damage. RESULTS Assays with the shrub samples for potential defenses other than amides were negative. TLC and GC-MS of all crude residues and 1 H NMR of the other fractions obtained in the previously described chromatography (Dodson et al.) indicates the absence of phenyl propanoids, aromatic hydrocarbons, oxygenated cyclohexanes, lignans, and neolignans. The screen for alkaloids was also negative. Plant fragments from which ants were excluded had significantly higher concentrations of total amides (Table 1, F 1,8 8.3, P 0.02), and the profile analysis revealed that the magnitude of this increase did not differ between the three amides (P > 0.05 for all comparisons). For plant fragments maintained with insecticide applications to remove ant colonies, overall amide concentrations were 0.2 times greater than in plants with ants, which is a smaller difference than that reported for naturally occurring shrubs (3.3 times greater levels of amides in shrubs without ants) (Dodson et al., 2000). The overall levels of amides in fragments appear to be higher than those reported for naturally occurring shrubs (Dodson et al., 2000), but other differences between the fragments and shrubs were not controlled (i.e., many of the shrubs came from a different site). The bioassays demonstrated that leaves from shrubs without ants were significantly more unpalatable than leaves from shrubs with ants. The mean adjusted

8 588 DYER, DODSON, BEIHOFFER, AND LETOURNEAU TABLE 1. CONCENTRATIONS OF Piper AMIDES IN P. cenocladum EXPERIMENTAL FRAGMENTS WITH AND WITHOUT Pheidole bicornis ANTS a Concentration (% dry mass) Compound Ants present Ants excluded Piplartine desmethylpiplartine Cenocladamide Total amides a Exclusion of ants was accomplished by applying dilute insecticide (Dyer and Letourneau, 1999a). consumption difference was for ant leaves and for nonant leaves (t , P 0.008). Both of these values are relatively high, and the value for nonant leaves is not significantly different from plants with high levels of toxic compounds such as nicotine from cigarette tobacco (Dodson, Dyer, and G. Gentry, unpublished data). Orthopteran damage was found on 18 of 91 P. cenocladum plants in the fragments experiment, and 78% (14) of these were plants with ants (x , P 0.01). In the shrub experiment, orthopteran damage was found on 43 of 77 plants, and 27 (63%) of these were plants with ants (x , P 0.03). Damage from the leaf cutting ant, A. cephalotes, is generally rare on P. cenocladum, but for the shrub experiment, 5 of 77 plants had ant damage, and all of these were plants with ants (x , P 0.03). DISCUSSION A trade-off was evident between biotic and chemical defenses in experimental fragments of P. cenocladum. Overall levels of amides were higher when ants were either absent or at very low numbers in P. cenocladum. These experimental data corroborate correlational data that also demonstrated a defensive trade-off: naturally occurring shrubs without ants had 3.3 times higher concentrations of total amides than shrubs with ants (Dodson et al., 2000). Such a trade-off suggests that both biotic and chemical defenses are costly. If both defenses were low in material and energy costs, the plant would experience minimum tissue loss from herbivores by having both the ants and high levels of the amides, since it is likely that together they protect the plant against a broad range of different herbivores. Letourneau (1983; also see Letourneau and Dyer, 1998) has demonstrated that Pheidole bicornis ants cause lower lepidopteran folivory by removing eggs from the surface of the leaf. However, there is no evidence that these ants are able to decrease levels of herbivory by orthopterans and Atta cephalotes

9 TRADE-OFFS IN Piper DEFENSES 589 (leaf cutting ants), since they are too small to deter such large herbivores. Piper amides could deter orthopteran herbivory by decreasing their feeding efficiency and deter A. cephalotes herbivory (Capron and Weimer, 1996) through some other mechanism such as antifungal properties. Examination of leaves from previous experiments (Dyer and Letourneau, 1999a,b) corroborated the hypothesis that P. cenocladum amides are deterrent to generalist herbivores. Damage from leaf cutting ants and orthopterans was more common on plants with ants, and we have demonstrated that P. cenocladum shrubs and fragments with ants have lower levels of amides than those without ants. Furthermore, the bioassays demonstrate that plants without ants are more unpalatable than plants with ants. For the orthopteran data and the bioassay data, the plants were the same plants that we analyzed for amides, with ant-exclusion plants having higher levels of amides; thus, higher amide concentrations are present in plants that are avoided more by orthopterans and ants. However, orthopteran damage was still low when ants were present, and the levels of amides in ant-containing plants were still high enough to deter ants in the bioassay, so it is likely that plants are protected from generalists by amides whether or not they have ants the protection is simply enhanced when the levels of amides increase. These results support recent hypotheses that stress that importance of redundancy in defenses when faced with a diverse herbivore assemblage (Romeo et al., 1996). Since each mole of each amide contains a mole of nitrogen, the concentrations found in both our manipulated plants and the naturally occurring shrubs represent a considerable investment (Gershenzon, 1994) by these plants, which are nutrient limited (Dyer and Letourneau, 1999a). Results from other studies with P. cenocladum suggest that not only are amides costly, but they are more expensive to synthesize than food bodies. Plants grown in an experiment where all herbivores were excluded had a greater total biomass when ants were present and the plants were presumably producing lower levels of amides (Dyer and Letourneau, 1999a). Piper amides (Dodson et al., 2000) and food bodies (Risch and Rickson, 1981) both represent considerable investments of nitrogen and carbon, but it is possible that the costs of biosynthetic machinery, storage structures and enzymes, and overall investment of raw materials (reviewed by Gershenzon, 1994) for the amides is higher per plant than for the food bodies. One alternative mechanism for this result, that the ants provide nutrients to the plants in the form of nitrogenous wastes, has not been supported by experiments and field observations (Letourneau, 1983). It has not escaped our notice that other secondary metabolites may also increase when ants are excluded. We have eliminated this possibility for all probable candidates except the terpenes and flavanoids. We cannot completely rule out the possibility that greater levels of herbivory on the ant-excluded plants induced increased levels of amides in those plants. However, the correlational results for shrubs (Dodson et al., 2000) only

10 590 DYER, DODSON, BEIHOFFER, AND LETOURNEAU included plants with low levels of herbivory (less than 10%) that had the same leaf size, levels of damage, and epiphyll load. For the fragments, which had lower levels of herbivory for ant versus nonant plants (Dyer and Letourneau, 1999a), the differences in amide concentrations were not as great as they were in the shrubs, which suggests that induction was not an important mechanism in our results. It is also possible that stressing the plants by adding insecticide to the petioles induced an increase in defensive compounds, but we have evidence that Diazinon has minimal physiological effects on P. cenocladum (Letourneau and Dyer, 1998; Dyer and Letourneau, 1999b), and we did not add insecticides to any of the shrub petioles. Thus, in this ant plant system, there appear to be trade-offs between investment in chemical defenses and ant attractants. Both shrubs and fragments utilize the two defenses and both exhibit the trade-off. Redundancy in defenses is also important to deter a diverse suite of herbivores, and the plant will invest in both defenses if possible. Acknowledgments This study was funded by NSF grant DEB to D.K.L., faculty research grants from the UCSC Academic Senate and Social Sciences Division to D.K.L., and an NSF REU grant to D.K.L., L.A.D., and C.D.D. A portion of the study was funded by an Earthwatch grant and a National Geographic grant to L.A.D. We thank the Organization for Tropical Studies for use of their facilities and for valuable logistical help. Excellent technical assistance was provided by G. Vega, R. Krach, J. Sorensen, G. Gentry, J. Searcy, J. Jay, Z. Wright, and 12 Earthwatch volunteers. We are grateful to G. Gentry for comments that improved previous drafts of the manuscript. REFERENCES BALDWIN, I. T Jasmonate-induced responses are costly but benefit plants under attack in native populations. Proc. Natl. Acad. Sci. U.S.A. 95: BURGER, W Flora Costaricensis. Field. Bot. 35: CAPRON, M. A., and WIEMER, D. F Piplaroxide an ant-repellent piperidine epoxide from Piper tuburculatum. J. Nat. Prod. 59: CATES, R. G., and ORIANS, G. H Successional status and the palatability of plants to generalized herbivores. Ecology 56: DODSON, C. D., DYER, L. A., SEARCY, J., WRIGHT, Z., and LETOURNEAU, D. K Cenocladamide, a dihydropyridone alkaloid from Piper cenocladum. Phytochemistry 53: DUH, C. Y., WU, Y. C., and WANG, S. K Cytotoxic pyridine alkaloids from Piper aborescens. Phytochemistry 29: DYER, L. A Tasty generalists and nasty specialists? A comparative study of antipredator mechanisms in tropical lepidopteran larvae. Ecology 76: DYER, L. A Effectiveness of caterpillar defenses against three species of invertebrate predators. J. Res. Lepid. 35:1 6. DYER, L. A., and LETOURNEAU, D. K. 1999a. Relative strengths of top-down and bottom-up forces in a tropical forest community. Oecologia 119: DYER, L. A., and LETOURNEAU, D. K. 1999b. Trophic cascades in a complex, terrestrial community. Proc. Natl. Acad. Sci. U.S.A. 96: ELLE, E., VAN DAM, N. M., and HARE, J. D Cost of glandular trichomes, a resistance character in Datura wrightii Regel (Solanaceae). Evolution 53:22 35.

11 TRADE-OFFS IN Piper DEFENSES 591 FOX, L. R Defense and dynamics in plant-herbivore systems. Am. Zool. 21: GARTNER, B. L., Breakage and regrowth of Piper species in rain forest understory. Biotropica 21: GBEWONYO, W. S. K., CANDY, D. J., and ANDERSON, M Structure activity relationships of insecticidal amides from Piper quieneese. Root Pestic. Sci. 37: GERSHENZON, J The cost of plant chemical defense against herbivory: A biochemical perspective, pp , in E. Bernays (ed.). Insect Plant Interactions, Vol. V. CRC Press, Boca Raton, Flordia. GOULD, F Genetics of plant herbivore systems: Interactions between applied and basic study, pp , in R. Denno, and B. McClure (eds.). Variable Plants and Herbivores in Natural and Managed Systems. Academic Press, New York. GOULD, F Genetics of pairwise and multispecies plant-herbivore coevolution, pp , in K. C. Spencer (ed.). Chemical Mediation of Coevolution. Academic Press, San Diego, California. GREIG, N Predispersal seed predation on five Piper species in a tropical rainforest. Oecologia 93: HARTSHORN, G. S., and HAMMEL, B. E Vegetation types and floristic patterns, pp , in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn (eds.). La Selva. Ecology and Natural History of a Neotropical Rainforest. University of Chicago Press, Chicago. JANZEN, D. H Comments on host-specificity of tropical herbivores and its relevance to species richness, pp , in V. H. Heywood (ed.). Taxonomy and Ecology. Academic Press, London. LEVIN, D. A The chemical defenses of plants to pathogens and herbivores. Annu. Rev. Ecol. Syst. 7: LETOURNEAU, D. K Passive aggression: An alternative hypothesis for the Piper Pheidole association. Oecologia 60: LETOURNEAU, D. K Ants, stem-borers, and fungal pathogens: Experimental tests of a fitness advantage in Piper ant-plants. Ecology 79: LETOURNEAU, D. K., and BARBOSA, P Ants, stem borers, and pubescence in Endospermum in Papua New Guinea. Biotropica 31: LETOURNEAU, D. K., and DYER, L. A Experimental test in lowland tropical forest shows top-down effects through four trophic levels. Ecology 79: LINDROTH, R. L., and HWANG, S. Y Diversity, redundancy, and multiplicity in chemical defense systems of aspen, pp , in J. T. Romeo, J. A. Saunders, and P. Barbosa (eds.). Phytochemical Diversity and Redundancy in Ecological Interactions. Plenum Press, New York. MARQUIS, R. J Herbivore fauna of Piper (Piperaceae) in a Costa Rican wet forest: Diversity, specificity, and impact, pp , in P. W. Price, T. M. Lewinsohn, G. W. Fernandes, and W. W. Benson (eds.). Plant Animal Interactions: Evolutionary Ecology in Tropical and Temperate Regions. John Wiley & Sons, New York. MCDADE, L. A., BAWA, K. S., HESPENHEIDE, H. A., and HARTSHORN, G. S La Selva. Ecology and Natural History of a Neotropical Rainforest. University of Chicago Press, Chicago. MIYAKADO, M., NAKAYAMA, I., and OHNO, N Insecticidal unsaturated isobutylamides, pp , in J. T. Arnason, B. J. R. Philogene, and P. Morand (eds.). Insecticides of Plant Origin. American Chemical Society, Washington, D.C. PARMAR, V. S., JAIN, S. C., BISHT, K. S., JAIN, R., TANEJA, P., JHA, A., TYAGI, O. D., PRASAD, A. K., WENGEL, J., OLSEN, C. E., and BOLL, P. M Phytochemistry of the genus Piper. Phytochemistry 46: REHR, S. S., FEENY, P. P., and JANZEN, D. H Chemical defence in Central American non-antacacias. J. Anim. Ecol. 42:

12 592 DYER, DODSON, BEIHOFFER, AND LETOURNEAU RISCH, S. J., and RICKSON, F. R Mutualism in which ants must be present before plants produce food bodies. Nature 291: RISCH, S., MCCLURE, M., VANDERMEER, J., WALTZ, S Mutualism between three species of tropical Piper (Piperaceae) and their ant inhabitants. Am. Midl. Nat. 98: ROMEO, J. T., SAUNDERS, J. A., and BARBOSA, P. (eds.) Phytochemical Diversity and Redundancy in Ecological Interactions. Plenum Press, New York. SAGERS, C. L., and COLEY, P. D Benefits and costs of defense in neotropical shrub. Ecology 76: SIMMS, E. L Costs of plant resistance to herbivory, pp , in R. S. Fritz, and E. L. Simms (eds.). Ecology and Evolution of Plant Resistance. University of Chicago Press, Chicago. SOLLINS, P., SANCHO, M. F., MATA Ch.R., SANFORD, R Soils and soil process research, pp , in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn (eds.). La Selva. Ecology and Natural History of a Neotropical Rainforest. University of Chicago Press, Chicago. STEWARD, J. L., and KEELER, K. H Are there trade-offs among antiherbivore defenses in Ipomoea (Convolvulaceae)? Oikos 53: SU, H. C. F., and HORVAT, R Isolation, identification, and insecticidal properties of Piper nigrum amides. J. Agric. Food Chem. 29: TABACHNICK, B. G., and FIDELL, L. S Using Multivariate Statistics, 3rd ed. Harper Collins, New York. ZANGERL, A., and BAZZAZZ, F. A Theory and pattern in plant defense allocation, pp , in R. S. Fritz and E. L. Simms (eds.). Ecology and Evolution of Plant Resistance. University of Chicago Press, Chicago.