Specificity of Herbivore-Induced Hormonal Signalingand Defensive Traits in Five Closely Related Milkweeds(Asclepias spp.)

Anurag A. Agrawal & Amy P. Hastings &Eamonn T. Patrick & Anna C. Knight

Received: 5 December 2013 /Revised: 23 April 2014 /Accepted: 6 May 2014 /Published online: 27 May 2014# Springer Science+Business Media New York 2014

Abstract Despite the recognition that phytohormonal signal-ing mediates induced responses to herbivory, we still havelittle understanding of how such signaling varies among close-ly related species and may generate herbivore-specific in-duced responses. We studied closely related milkweeds(Asclepias) to link: 1) plant damage by two specialist chewingherbivores (milkweed leaf beetles Labidomera clivicolis andmonarch caterpillars Danaus plexippus); 2) production of thephytohormones jasmonic acid (JA), salicylic acid (SA), andabscisic acid (ABA); 3) induction of defensive cardenolidesand latex; and 4) impacts on Danaus caterpillars. We firstshow that A. syriaca exhibits induced resistance followingmonarch herbivory (i.e., reduced monarch growth on previ-ously damaged plants), while the defensively dissimilarA. tuberosa does not. We next worked with a broader groupof five Asclepias, including these two species, that are highlydivergent in defensive traits yet from the same clade. Three ofthe five species showed herbivore-induced changes incardenolides, while induced latex was found in four species.Among the phytohormones, JA and ABA showed specificresponses (although they generally increased) to insect speciesand among the plant species. In contrast, SA responses wereconsistent among plant and herbivore species, showing adecline following herbivore attack. Jasmonic acid showed apositive quantitative relationship only with latex, and this wasstrongest in plants damaged by D. plexippus. Although phy-tohormones showed qualitative tradeoffs (i.e., treatments thatenhanced JA reduced SA), the few significant individualplant-level correlations among hormones were positive, andthese were strongest between JA and ABA in monarch dam-aged plants. We conclude that: 1) latex exudation is positively

associated with endogenous JA levels, even among low-latexspecies; 2) correlations among milkweed hormones are gen-erally positive, although herbivore damage induces a diver-gence (tradeoff) between JA and SA; 3) induction ofcardenolides and latex are not necessarily physiologicallylinked; and 4) even very closely related species show highlydivergent induction, with some species showing strong de-fenses, hormonally-mediated induction, and impacts on her-bivores, while other milkweed species apparently use alterna-tive strategies to cope with insect attack.

Keywords Cardenolides . Coevolution . CommonmilkweedAsclepias syriaca . Jasmonic acid . Latex .Monarch butterflyDanaus plexippus . Phytohormones . Plant-insect interactions

Introduction

Understanding how plants use phenotypic plasticity to re-spond to diverse environmental cues has been a long-standing question in plant evolutionary ecology (Agrawal2001). It generally is assumed that, given the lack of a nervoussystem, plant hormonal responses orchestrate myriad plantresponses, including defense induction to specific herbivores.Even so, with only a handful of phytohormones, they mustfunction together or in particular ratios to dictate distinctresponses (Reymond and Farmer 1998). In particular, theintersection of multiple hormones has played a critical rolein understanding plant responses to important biotic and abi-otic interactions such as herbivory, microbial attack, lightcompetition, and water stress (Cipollini 2004; Moreno et al.2009; Thaler and Bostock 2004; von Dahl and Baldwin 2007).

For plant responses to insect attack, jasmonic acid (JA) isrecognized as the master regulator of induced defense, espe-cially against chewing herbivores (Agrawal 2011; Erb et al.

A. A. Agrawal (*) :A. P. Hastings : E. T. Patrick :A. C. KnightDepartment of Ecology and Evolutionary Biology, and Departmentof Entomology, Cornell University, Ithaca, NY, USAe-mail: aa337@cornell.edu

J Chem Ecol (2014) 40:717–729DOI 10.1007/s10886-014-0449-6

Author's personal copy

2009, 2012; Halitschke and Baldwin 2003). Despite playing arole in a few other known plant functions (Avanci et al. 2010),jasmonates are nearly universally upregulated followingchewing damage, and are known to play a key role in cueinga remarkably diverse set of plant defenses spanning toxins andtrichomes to volatiles and extrafloral nectar (Erb et al. 2009;Halitschke and Baldwin 2003; Heil et al. 2004; Rasmann et al.2009; Thaler et al. 2002; Traw and Bergelson 2003).Nonetheless, it is clear that other plant hormones, namelysalicylic acid (SA), abscisic acid (ABA), and ethylene (ET)play roles as antagonists and synergists in defensive reactions(Erb et al. 2012; Thaler et al. 2012). Thus, the simultaneousexamination of multiple phytohormones during plant defen-sive responses has been advocated as an approach to under-standing specificity in plant-herbivore interactions (Schmelzet al. 2003; Schuman et al. 2009; Thaler et al. 2010).

Our approach to unraveling specificity of plant responsesto herbivory in this study was two-fold. First, we chose towork within a guild of specialized chewing herbivores (acoleopteran, the milkweed leaf beetle Labidomera clivicolisand a lepidopteran, the monarch butterfly Danaus plexippus),native to a wild system (the milkweeds), and sharing a longevolutionary history. The fact that the herbivores are withinthe same guild allows for inferences about specificity ininduced responses that go beyond the impacts of differentfeeding styles. Second, we workedwith a set of closely relatedhost plant species, with known phylogenetic relationships.Thus, we addressed how very closely related species havediverged in their physiological processes, and whether this hasresulted in divergent specificity in plant-herbivore relation-ships. Although such an approach has the drawback of re-duced mechanistic detail compared to a single plant-herbivoreassociation, we are seeking to make broad evolutionary infer-ences about how defense specificity can evolve.

In particular, we have been studying North AmericanAsclepias spp. (Apocynaceae), which are a monophyleticgenus comprising approximately 140 species in the WesternHemisphere (Agrawal et al. 2009; Fishbein et al. 2011;Woodson 1954). Here, we focused on 5 species in a mono-phyletic clade consisting of 9 total taxa (Fig. 1), including thecommon milkweed A. syriaca, which is the best studiedspecies for interactions between monarch butterflies and plantdefense (Bingham and Agrawal 2010; Malcolm et al. 1989;Malcolm and Zalucki 1996; Van Zandt and Agrawal 2004b;Vannette and Hunter 2011). Because all species produce twopotent defenses, latex and cardenolides, neither of which has acredibly hypothesized function other than defense, we have anexcellent model system to address specificity in the context ofevolutionarily relevant interactions (Agrawal and Konno2009; Ali and Agrawal 2012). Furthermore, these speciesare known to have overall divergences in their defensivestrategies, making the comparisons highly informative(Agrawal and Fishbein 2006; Agrawal et al. 2009).

Accordingly, we asked the following questions: 1) Do twophenotypically divergent, but closely related and sympatricAsclepias species differ in the extent to which they exhibitinduced resistance to monarch caterpillars? 2) How variableare five closely related Asclepias species in their induced plantresponses (cardenolides and latex) when attacked by differentspecialist chewing herbivores? 3) To what extent are planthormones (JA, SA, and ABA) linked to the induced re-sponses? and, 4) Are the relationships between plant hor-mones dependent on plant species and the biotic context(i.e., damage by different herbivore species)? In summary,our goal was to develop a predictive framework for howdifferent herbivores induce specific plant defensive responses,and to employ a comparative approach to address how suchinduced responses have evolved.

Methods and Materials

Plant Growth and Herbivores All plants were germinatedfrom wild-collected seed in petri dishes containing moistpaper towels, transplanted into 500 ml plastic pots with Pro-

Fig. 1 Phylogenetic relationships within a monophyletic clade ofAsclepias (Apocynaceae). Shown are all of the extant species in thisclade. Relationships are based on a recent molecular phylogeny (Fishbeinet al. 2011). Arrows and plant images indicate the species studied here.Based on previous work, the five studied species fall among three distinctplant defense syndromes (Agrawal and Fishbein 2006). Photo creditsfrom top to bottom are: Mark Fishbien, Marc Johnson, AAA, AAA,and Ellen Woods

718 J Chem Ecol (2014) 40:717–729

Author's personal copy

Mix BX soil, and fertilized twice with a dilute fertilizer (NPK21:5:20, 150 ppmN). We fully randomized the plants (one perpot) in growth chambers (400μmoles/m2/sec light, 14/8 h L/Dcycle, 28 °C/24 °C temperature cycle), intermixing all plantsfrom a given experiment in a single chamber. Plants weregrown for 1 month (10–12 leaf stage) before experimentation.

Monarchs (D. plexippus L., Nymphalidae) were obtainedfrom a colony maintained variously on Asclepias curassavicaL. or A. syriaca L., and a colony of L. clivicollis Kirby(Chrysomelidae) was maintained on A. syriaca; in both cases,the colonies were established from field collected individualsin the preceding months. Young (2nd instar) Euchaetes egleDrury (Arctiidae) larvae used in the final experiment werecollected from the field on A. syriaca. All experiments wereperformed with larvae.

Induced Resistance Experiment Induced resistance wasassessed on A. syriaca and A. tuberosa L. because thesespecies are closely related, sympatric hosts for monarchs,but have highly divergent defense strategies and relation-ships with herbivores (Agrawal and Fishbein 2006; Haribaland Renwick 1998; Zalucki and Malcolm 1999; Zehnderand Hunter 2007, 2008). While A. syriaca has moderatelevels of leaf cardenolides and latex, and is a primary hostof monarchs, A. tuberosa is nearly devoid of cardenolidesand latex, and is much less frequently attacked in the field.For this experiment, we grew 40 A. syriaca and 40A. tuberosa as above, and plants of the two species wererandomly divided among 2 treatments (control and in-duced) and were intermixed in a single growth chamber.Plants were not caged. Plants in the induced treatmentreceived a freshly hatched monarch caterpillar, which wasallowed to feed for 5 days. Damage was visually assessed tobe consistently 3–5 % leaf damage across both species.After this period, caterpillars were removed, and on thesame day all plants received a freshly hatched monarchcaterpillar as the bioassay. Caterpillars were recollected5 days later and were weighed after being isolated for 3 hrto clear their guts. Total replication was reduced as mortal-ity or loss of bioassay larvae was not trivial (16 %), but thiseffect was not different between species or treatments.

Five Species Experiment To broadly assess differential in-duced responses to herbivory we worked with five closelyrelated Asclepias (Fig. 1) that are all natural hosts of themonarch butterfly: A. syriaca, A. tuberosa, A. obovata Elliot,A. hallii A. Gray, and A. exaltata L.. For this experiment, 40–50 plants of each species was grown from seed and dividedamong undamaged controls, monarch-damaged, andLabidomera-damaged plant (totalN=220 plants). On average,we used 12–15 plants per species for each of the three treat-ments, but replication in the Labidomera treatment was lowerbecause of failure of larvae to eat plants in some cases (lowest

N was 5 in the A. obovata–Labidomera combination).Labidomera is oligophagous, known to eat many Asclepiasspp. In all cases, young larvae (second and third instars wereused to damage plants). Insects were maintained on theplants in mesh cages; after three full days of feeding,the insects were removed and the plants harvested im-mediately afterward (this experiment was conducted in 2blocks that were separated in time by 1 day). Theharvest after 3 days was employed to maximize themeasurement of phytohormonal dynamics (jasmonic ac-id, in particular), which we have shown to peak inmilkweed on day three across a 5 days time course ofsampling. Temporal analyses of phytohormonal dynam-ics were not possible in this study because of highnumber of species-by-treatment combinations.

Cardenolides, Latex, and Hormones At time of harvest, weestimated the percentage of damage, and measured latex exu-dation on plants by cutting the tip off the youngest, fullyexpanded, undamaged leaf and collecting the latex on a pre-weighed 1 cm diam filter paper disc. After absorbing all thelatex (30–60 sec), we placed the disc into a pre-weighedmicrocentrifuge tube. These tubes were reweighed to measurethe amount of wet latex collected. This method is a repeatablemeasure of latex exudation and has been shown to predictresistance to herbivores (Agrawal 2005; Van Zandt andAgrawal 2004b).

Immediately following latex collection, we collected allremaining leaf tissue and split it into 2 equivalent groups forcardenolide analysis and hormone analysis; this tissuecontained both damaged and undamaged leaves. Forcardenolides, the entire sample was weighed, dried, andground to a fine powder. A portion of the sample (100 mg orless for small plants) was then subjected to extraction forcardenolide quantification, following Bingham and Agrawal(2010). Briefly, the dried leaf powder was extracted with1.8 ml methanol (MeOH), spiked with 20 μg digitoxin as aninternal standard, and sonicated for 20 min at 55 ° C in a waterbath. After centrifugation, the supernatant was collected, dried,resuspended in 1 ml MeOH, and filtered through a 0.45 μmsyringe driven filter unit. Fifteen μl of extract were theninjected into an Agilent 1100 series HPLC, and compoundswere separated on a Gemini C18 reversed phase column(3 μm, 150×4.6 mm, Phenomenex, Torrance, CA, USA).Cardenolides were eluted on a constant flow of 0.7 ml/minwith an acetonitrile −0.25 % phosphoric acid in water gradientas follows: 0–5 min 20 % acetonitrile, 20 min 70 % acetoni-trile, 20–25 min 70 % acetonitrile, 30 min 95 % acetonitrile,30–35 min 95 % acetonitrile. UV absorbance spectra wererecorded from 200 to 400 nm by a diode array detector.Peaks with absorption maxima between 214 and 222 nm (cen-tered at 218) were recorded as cardenolides and quantified at218 nm. Concentrations were calculated and standardized by

J Chem Ecol (2014) 40:717–729 719

Author's personal copy

peak areas of the known digitoxin concentration. We calculat-ed total cardenolides (mg/g leaf tissue dry mass), totalcardenolide diversity (i.e., the number of distinct HPLCpeaks), and an index of cardenolide polarity (sum of eachretention time X proportion of the sample for each peak)(Rasmann and Agrawal 2011). Although the impact ofcardenolide diversity is currently unknown, it has been specu-lated this may be important for resistance to insects (Agrawalet al. 2012b; Rasmann and Agrawal 2009). There is someevidence that cardenolide polarity impacts resistance to insects,with less polar compounds having greater toxicity, presumablybecause they can more easily cross cell membranes (Rasmannet al. 2009).

Hormones were quantified using an established liquidchromatography–mass spectrometry (LC-MS) procedure,modified from Thaler et al. (2010). For each plant, 250–400 mg of the freshly-harvested leaf tissue wereweighed and placed directly into a 2-mL screw cap tubein liquid nitrogen and stored at −80 °C. We added 1 mlextraction buffer (isopropanol:water:hydrochloric acid ina 2:1:0.005 solution) to each tube, along with 900 mgzirconia/silica beads (BioSpec, Bartelsville, OK, USA)and an internal standard solution containing d4-SA, d5-JA, and d6-ABA (CDN isotopes, Point-Claire, Canada) .Samples then were extracted in a FastPrep homogenizer(MP Biomedicals, Solon, OH, USA) for two cycles of6.5 m/s for 45 sec. Samples were dissolved in 200 μlmethanol after extraction with dichloromethane and sol-vent evaporation and 15 μl were analyzed on a triple-quadrupole LC-MS/MS system (Quantum Access;Thermo Scientific, Waltham, MA, USA). Analytes wereseparated on a C18 reversed-phase HPLC column(Gemini-NX, 3 μ , 150 × 2.00 mm; Phenomenex,Torrance, CA, USA) using a gradient of 0.1 % formicacid in water (solvent A) and 0.1 % formic acid inacetonitrile (solvent B) at a flow rate of 300 μl/min.The initial condition of 10 % B was kept for 2 min andincreased to 100 % solvent B at 20 min. Phytohormoneswere analyzed by negative electrospray ionization (sprayvoltage: 3.5 kV; sheat gas: 15; auxiliary gas: 15; capil-lary temperature: 350 °C), collision-induced dissociation(argon CID gas pressure 1.3 mTorr [1.3 micron Hg],CID energy 16 V) and selected reaction monitoring(SRM) of compound-specific parent/product ion transi-tions: SA 137→93; d4-SA 141→97; JA 209→59; d5-JA214→62; ABA 263→153; d6-ABA 269→159. The totalconcentration of each hormone per g fresh mass wascalculated by comparing each natural hormone with itslabeled counterpart.

Timing of Induction in A. syriaca and A. hallii Based onsubsequent laboratory observations that induced cardenolideresponses peaked 2 days after the hormonal peak, we

conducted a temporal experiment to examine the in-duced cardenolide response in two of the five focalspecies. We grew 43 A. syriaca and 45 A. hallii asdescribed above, and plants of the two species were ran-domly divided among 2 treatments (control and induced)and all plants were intermixed in a single growth chamber.Plants were not caged. Plants to be induced received afreshly hatched monarch caterpillar, which was allowed tofeed for 3 or 5 days. After 3 days, caterpillars were removedfrom half of the plants across the two species, and leaveswere harvested for cardenolide analysis (as above); theremaining plants and caterpillars were left intact for 2additional days, and then similarly harvested (after 5 daysof feeding damage).

Induction in A. syriaca with 3 Herbivores To further investi-gate specificity of induced responses, we conducted an exper-iment with the common milkweed and three herbivores. Forthis experiment, 10 maternal families (full-sibling seeds froma single fruit, all from a single local population) of A. syriacawere grown as above, and plants were divided among thethree treatments as above (control, beetles, monarchs), as wellas an additional treatment with plants damaged by anotherlepidoptera chewer, the milkweed tussock moth caterpillarEuchates egle (third instar caterpillars were used). Samplesizes were as follows: N=94 for control plants, N=95 formonarchs, N=26 for Euchaetes, and N=35 for Labidomera.Our replication for the 2 latter treatments was limited by theavailability of herbivores. Insects were caged on plants for3 days using spun polyester sleeves, and latex was harvestedas well as tissue for hormone analysis. In this case, we col-lected leaf tissue from either the youngest fully expanded leafalone, or in cases where this was not sufficient, the youngestfully expanded leaf and a developing leaf located above thisleaf. Because leaf position effects on phytohormonal dynam-ics can be substantial, we included the harvest position in ourstatistical models.

Data analysis All data were analyzed using analysis ofvariance in JMP Pro, version 10.0 (SAS Institute, Cary, NC,USA). Data were checked for normality of residuals andoutliers using studentized residuals.

Results and Discussion

Induced Resistance in Asclepias syriaca and A. tuberosa Wefirst conducted an experiment imposing minimal herbivory(<5 % leaf damage) by monarchs to assess induced resis-tance to subsequently feeding monarchs on two defensivelydistinct milkweeds, A. syriaca and A. tuberosa. As has beenpreviously reported, we found induced resistance in A. syriaca,with caterpillars showing a 50 % reduction in growth on

720 J Chem Ecol (2014) 40:717–729

Author's personal copy

previously damaged plants compared to controls (Fig. 2., F1,22=7.09, P=0.014); however, we found no induced resistance onA. tuberosa (Fig. 2., F1,16=0.377, P=0.548).

Specificity in the Induction of Defensive Traits in Five AsclepiasSpecies Constitutive amount of latex exuded and cardenolideconcentrations each varied over five-fold among the fiveAsclepias species studied, and both traits were lowest inA. tuberosa (Fig. 3). Greater than 90 % of the 140 plantsreceiving either Labidomera or monarch herbivory receivedbetween 1 and 5 % leaf tissue damage (with the remainingplants receiving <10 % damage). The amount of latex exudedfollowing tissue damage was inducible in response to herbiv-ory across all plant species (no species-by-treatment interac-tion in Table 1), and was strongest in response to larvalLabidomera attack (a 35 % increase, Fig. 3, mean ± SE mglatex exuded across all species, control 0.83±0.08,Labidomera 1.12±0.09, Monarch 0.88±0.07). A previousstudy also found stronger latex induction in response toLabidomera compared to monarchs for A. syriaca (VanZandt and Agrawal 2004b). While the amount of damageimposed by the two herbivores was comparable, these herbi-vores often damaged milkweeds in different locations, withthe monarchs tending to damage apical leaves andLabidomera tending to damage leaves in the lower half ofthe plant (Van Zandt and Agrawal 2004b).

For total foliar cardenolides, plant responses to herbivorywere variable (significant species-by-treatment interaction inTable 1), with only A. obovata showing a strong increase(≈50 %) in cardenolide concentrations 3 days after caterpillarattack commenced (Fig. 3, Table 2). To dig deeper into theinduced responses, we examined each Asclepias species for

cardenolide composition (using MANOVAwith each distinctcardenolide peak as a response variable), total peak diversity(i.e., number of distinct compounds), and mean cardenolidepolarity. Asclepias syriaca and A. obovata showed significantalterations in cardenolide profiles in response to herbivory asevidenced by a significant treatment effect in a MANOVA(details in Table 2). Yet, we found relatively few other signif-icant impacts of our herbivory treatments, with A. hallii show-ing a 9 % decline in the number of cardenolide peaks inresponse to herbivory, and A. obovata showing a 11 % declinein the mean polarity of the cardenolides (Table 2).

Induced chemical responses have been well-studied inA. syriaca, and we had conducted initial surveys in otherAsclepias species. Our previous work and that of others indi-cated that total cardenolide concentrations often increase 10–30 % in A. syriaca after monarch attack, although genetic andenvironmental conditions can modulate the strength of thiseffect (Ali and Agrawal 2012; Bingham and Agrawal 2010;Mooney et al. 2008; Vannette and Hunter 2011). For the otherfour species studied here, we had preliminary work on theirinducibility (see Table S1 in Rasmann and Agrawal 2011), butit does seem that, at least under some conditions, A. exaltata,A. hallii, and A. obovata showmonarch-induced cardenolides.This is less clear for A. tuberosa, for which cardenolides arebarely detectable, and they are more likely defended by othercompounds (Warashina and Noro 2009).

Because of the above-described variability, and our ownrecent results suggesting that cardenolide induction is mostevident 5 days after monarch feeding, we conducted an addi-tional experiment with A. syriaca and A. hallii. Here, weharvested plants after 3 and 5 days of monarch feeding andassessed foliar cardenolides. Indeed, whereas there was essen-tially no evidence for cardenolide induction after 3 days,induction was apparent 5 days after feeding (Fig. 4). In astatistical model with species, induction treatment, harvestday, and induction-by-harvest day interaction, only species(F1,82=15.374, P<0.001) and the interaction term were sig-nificant (F1,82=6.286, P=0.014). Thus, induction of the de-fensive end-products are most evident after 5 days of monarchfeeding (Fig. 4).

Induced Hormones and Correlations with DefensiveTraits Jasmonic acid and ABA showed strong herbivore-specific and plant species-specific bursts following 3 days ofinsect attack (treatment-by-species interaction in Table 1,Fig. 5). Our previous work indicated that the hormonal re-sponses peak 2–4 days after herbivory commences (Agrawalet al.,. unpublished). For all species except A. hallii, plantsresponded more strongly with JA to monarch feeding than toLabidomera feeding. Conversely, for all species exceptA. syriaca, plants responded more strongly with ABA toLabidomera feeding than to monarch feeding. For SA, re-sponses were consistent among the five plant species (no

Mon

arch

larv

al m

ass

(mg) 50

60

70

Control

Previous monarch damage

0

10

20

30

40

A. syriaca A. tuberosaFig. 2 Induced resistance to monarchs, measured as larval growth fromegg hatch, following initial herbivory on Asclepias syriaca andA. tuberosa. Shown are means ± SE

J Chem Ecol (2014) 40:717–729 721

Author's personal copy

treatment-by-species interaction in Table 1), showing a 14–43 % decline following attack (Fig. 5, mean ± SE ng/g acrossall species, control 433±53, Labidomera 246±64, monarch373±49). Thus, qualitatively speaking, across the fiveAsclepias species, Labidomera induced a stronger set of re-sponses (when considering hormones and end-products col-lectively) than monarchs. In particular, the stronger inductionof latex was associated with an increase in JA and ABA and a

decline in SA (Figs. 2 and 5). An exception to this pattern wasthat despite a stronger latex response to Labidomera, plantstypically showed a stronger burst of JA in response tomonarchs.

To specifically address the hormonal regulation oflatex and cardenolides, we conducted a set of correla-tions between the traits. It is important to conduct suchanalyses while controlling for differences in plant spe-cies, and also in herbivore species, so as not to con-found differences between species or those induced bytreatments with quantitative relationships among individ-ual plants. We found that JA strongly predicted latexexudation quantitatively (Table 3, Fig. 6). In the mon-arch damage treatment, individuals of the five milkweedspecies showed consistently positive correlations be-tween JA and latex. This relationship was more speciesspecific when plants were undamaged or damaged byLabidomera; for example, A. tuberosa (that has thelowest levels of latex) showed no relationship betweenJA and latex in either of those two treatments (Table 3,Fig. 6).

We found no evidence for a quantitative association be-tween JA and foliar cardenolides (Table 3). Additionally, all

Late

x (m

g ex

uded

)C

arde

nolid

es (

mg/

g)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

A. exaltata A. hallii A. obovata A. syriaca A. tuberosa

0

0.5

1

1.5

2

2.5

3

A. exaltata A. hallii A. obovata A. syriaca A. tuberosa

Control

Labidomera

Monarch

Fig. 3 Specificity in herbivore-induced latex exudation and totalcardenolides in five species ofAsclepias. Shown are means ± SE

Table 1 Analysis of variance for effects of Asclepias species, herbivoretreatments (control, Labidomera herbivory, and monarch herbivory), andtheir interaction on two defensive end-products (latex and totalcardenolide concentration), and three phytohormones implicated in de-fensive signaling (N=219)

Species Treatment Species X treatment

Latex 29.20*** 3.29* 1.24

Total cardenolides 59.43*** 0.63 2.72**

Jasmonic acid 12.33*** 28.75*** 2.03*

Salicylic acid 5.40*** 4.08* 1.33

Abscisic acid 81.99*** 5.12** 4.52***

Shown are F-values; bold values are statistically significant at *<0.05,**<0.01, ***<0.001

722 J Chem Ecol (2014) 40:717–729

Author's personal copy

exploratory analyses on the effects of SA and ABA on latexand cardenolides showed no relationship, and these were notfurther examined here (Ps >> 0.1).

Quantitative Relationships Between the Hormones The cor-relations among hormones at the individual plant level re-vealed a striking pattern. Of the six significant or marginallysignificant correlations, all were positive, five involved ABA,and four were in the monarch damage environment (Table 4).Despite the widespread expectation that JA and SA shouldshow antagonistic expression, we only found evidence for thisrelationship at the qualitative level; that is, most speciesshowed a mean increase in JA following damage, but showeda mean decrease in SA compared to undamaged controls(Fig. 5). However, in quantitative correlations between thetwo hormones (within species and treatments), there were nosignificant negative relationships detected (Table 4). Thus, the

strength of JA induction did not correlate with the strength ofSA suppression. As mentioned above, correlations acrossspecies or even treatment groups could provide a false pictureof trade-offs. In Fig. 7, we illustrate this issue by showing datafor Asclepias exaltata, which showed no evidence for a trade-off within treatment groups (Table 4), even though there was aqualitative pattern of a tradeoff across treatments (i.e., inducedplants had higher JA and lower SA, Fig. 5).

At least three explanations may clarify the apparently di-vergent results for whether JA and SA tradeoff. First, inAsclepias spp., JA and SA regulation could well be indepen-dent, and attacking herbivoresmay simply alter the productionof both hormones (adaptive for the plant, adaptive for theherbivore, or incidental). A mechanical damage treatmentwould help resolve whether these dual responses areherbivore-specific, as JA is predicted to increase withmechan-ical damage, but SA is not predicted to change. Second, JAand SA may be antagonistic, but their negative associationmay be dictated by the sequence and timing of attack andthresholds of expression (Koornneef et al. 2008; Mur et al.2006). In other words, at some high levels of JA, SA maydecline (generating the qualitative pattern of induction andsupression), but the negative associations would not necessar-ily be detectable in correlations within treatments where onlyone pathway is primarily triggered. Our own data onA. syriaca plants attacked by caterpillars and aphids showedthat JA and SA only exhibit a negative correlation when bothherbivores attack the plants simultaneously (Ali and Agrawal2014). Finally, if as was suggested for other systems (Leon-Reyes et al. 2009, 2010), ethylene modulates the negative JA–SA interaction, perhaps the induction dynamics of ethyleneare critical to whether the antagonism is evident.

Three Chewing Herbivores Attacking A. syriaca In a finalexperiment, we conducted a larger trial, only with A. syriaca(N=247 plants), and with induction by three chewing

Table 2 Analyses of variance (ANOVA) on the effect of herbivore treatment on cardenolides in Asclepias species

A. exaltata (N=39) A. hallii (N=42) A. obovata (N=40) A. syriaca (N=46) A. tuberosa (N=49)

MANOVAwilks’ λ 0.83 0.36 0.51 0.26 0.78

MANOVA F 0.49 1.16 2.11* 2.55** 1.43

Total cardenolides 0.39 (0.01) 0.83 (0.03) 0.64 (0.04) 0.97 (0.06) 0.24 (0.01)

Total F 0.13 1.01 5.63** 1.65 0.56

Cardenolide richness 5.41 (0.11) 11.72 (0.19) 5.95 (0.19) 8.20 (0.21) 2.84 (0.11)

Richness F 1.45 3.20* 0.95 0.79 0.49

Cardenolide polarity 15.20 (0.11) 18.98 (0.07) 15.78 (0.36) 12.81 (0.17) 18.42 (0.37)

Polarity F 0.10 0.88 3.23* 0.69 1.42

Shown are the analyses for themultivariate analyses of variance (MANOVA) for effects on individual peaks. Additionally, means (and standard error) areshown for total cardenolides (mg/g leaf tissue dry mass), cardenolide richness (number of distinct HPLC peaks), and cardenolide polarity (sum of theretention time X proportion of the sample for each peak) along with univariate ANOVAs for these traits. Bold F-values are statistically significant at*<0.05, **<0.01, ***<0.001

Car

deno

lides

(m

g/g)

0

0.2

0.4

0.6

0.8

1

Day 3 Day 5 Day 3 Day 5

A. syriaca A. hallii

Control

Monarch damage

Fig. 4 Herbivore-induced cardenolides in Asclepias syriaca and A. haliiafter 3 and 5 days of herbivory. Shown are means ± SE

J Chem Ecol (2014) 40:717–729 723

Author's personal copy

herbivores (monarchs, Labidomera, and the specialist mothEuchaetes egle). In addition, we introduced variability interms of the leaf position harvested for hormonal analysis. Inparticular, for plants that lacked enough tissue in the youngestfully expanded leaf, we harvested an additional leaf just above(towards the apex) of the plant. The upper leaves of plantsshowed substantially higher levels of JA and SA, and greaterinducibility of JA (Fig. 8). Although herbivore treatment didnot significantly affect SA in this experiment, there was atrend for lower levels in Labidomera and monarch attackedplants compared to controls (Fig. 8). Corresponding to thestronger impact on JA and SA in Labidomera and monarchattacked plants, these plants also showed the strongest ABA

burst. Latex showed a >30 % increase in all herbivoretreatments (Fig. 8), and latex showed a quantitative(positive) relationship with JA (multiple regression includ-ing treatments and all hormones, JA F1,240 = 7.86,P=0.006). As above, when all five species were consid-ered, all significant correlations between hormones werepositive and were with ABA (Table 5).

Macroevolution of Induced Responses to Herbivory The factthat closely related plant species show variable responses todamage by specialists insects is not surprising, and specificityin the induction response to different herbivore species withina feeding guild has been previously reported for milkweeds

Jasm

onic

aci

d (n

g/g)

Sal

icyl

ic a

cid

(ng/

g)A

bsci

sic

acid

(ng

/g)

0

200

400

600

800

1000

1200

1400

A. exaltata A. hallii A. obovata A. syriaca A. tuberosa

Control

Labidomera

Monarch

0

200

400

600

800

A. exaltata A. hallii A. obovata A. syriaca A. tuberosa

Control

Labidomera

Monarch

0

20

40

60

80

A. exaltata A. hallii A. obovata A. syriaca A. tuberosa

Fig. 5 Specificity of herbivore-induced phytohormones in fivespecies of Asclepias. Shown aremeans ± SE

724 J Chem Ecol (2014) 40:717–729

Author's personal copy

Late

x (m

g ex

uded

)La

tex

(mg

exud

ed)

Late

x (m

g ex

uded

)

Jasmonic acid (ng/g)

CONTROL

LABIDOMERA

MONARCH

Fig. 6 The relationship betweenjasmonic acid and latex exudationin five species of Asclepias.Shown are data from independentplants in three environments(undamaged controls, damagedby Labidomera, and damage bymonarchs)

J Chem Ecol (2014) 40:717–729 725

Author's personal copy

and other plant-insect systems (Bingham and Agrawal 2010;De Moraes et al. 1998; Erb et al. 2012; Van Zandt andAgrawal 2004a, b; Viswanathan et al. 2005). Nonetheless,the fact that close plant relatives, each of which naturallyreceives herbivory by some of the same specialist herbivores,can show qualitatively divergent induced resistance (as in thecomparison ofA. syriaca and A. tuberosa, Fig. 2), is indicativeof induced responses that show substantial evolutionary labil-ity. Phylogenetic analyses have revealed such lability in in-duced plant responses to herbivory for several plant genera,including Gossypium (Malvaceae), Acacia (Fabaceae), andAsclepias (Apocynaceae) (Heil et al. 2004; Rasmann andAgrawal 2011; Thaler and Karban 1997).

Among the five closely related Asclepias studied here,A. tuberosa and A. obovata are sister species (Fig. 1).Despite being highly divergent in latex (the lowest and highestspecies, respectively) and cardenolides (Fig. 3), they showedsimilar phytohormonal dynamics. The lack of induced re-sponse traits measured and induced resistance to monarchsin A. tuberosa, coupled with the highest constitutive salicylicacid among all of the species is worthy of further study.Although little is known about their specific interactions withherbivores and pathogenic microbes, A. tuberosa andA. obovata likely have distinct ecologies based on their habitatrequirements and geographic distributions (Woodson 1954).Asclepias syriaca and A. exaltata are also close relatives, havea similar growth form and geographic distribution, but typi-cally are segregated by habitat. Here again, there is somesimilarity in phytohormonal dynamics, but differentiation indefensive traits. Future work will focus on the ecologicaldrivers and consequences of these evolutionary divergences.

Synthesis and Speculation Here, we have attempted toaddress the potential hormonal underpinnings ofherbivore-specific responses by comparing fiveAsclepias species. A few general patterns emerged.First, overall, plant responses were stronger to a leaffeeding beetle than to caterpillar damage. This resultoccurred in spite of previous research that shows bothinsects are consistently negatively impacted by latex(Agrawal 2005; Van Zandt and Agrawal 2004b;Zalucki et al. 2001), with more variable results regard-ing cardenolides. This finding is important because it

Table 3 Analysis of variance for the quantitative relationship betweenjasmonic acid and latex exudation while controlling for differences be-tween species

Control(N=66)

Labidomera(N=64)

Monarch(N=79)

Latex Species 9.46*** 16.74*** 10.15***

JA 0.21 10.29** 11.66**

Species*JA 2.86* 3.35* 1.27

Cardenolides Species 19.96*** 21.72*** 23.33***

JA 0.35 0.43 0.12

Species*JA 0.24 0.29 1.22

Analyses were conducted on plants with each treatment separately.Shown are F values; bold values are statistically significant at *<0.05,**<0.01, ***<0.001

Table 4 Pearson correlations between phytohormones (jasmonic acid,salicylic acid, and abscisic acid: JA, SA, and ABA, respectively) con-ducted within each Asclepias species-by-treatment combination (N=10–15 for each analysis in each cell)

Control Labidomera Monarch

A. exaltata JA-ABA 0.015 0.028 0.603*

SA-ABA −0.131 −0.128 0.758**

JA-SA −0.500 −0.170 −0.082A. hallii JA-ABA 0.297 0.141 0.149

SA-ABA −0.219 −0.212 −0.013JA-SA −0.082 0.417 0.319

A. obovata JA-ABA −0.245 0.149 0.532*

SA-ABA −0.069 −0.179 −0.128JA-SA −0.044 0.918* 0.077

A. syriaca JA-ABA 0.161 −0.037 −0.019SA-ABA 0.452 0.481† 0.043

JA-SA −0.072 0.228 −0.149A tuberosa JA-ABA −0.179 −0.047 0.646**

SA-ABA −0.048 −0.032 −0.203JA-SA 0.020 0.442 0.131

Shown are correlation coefficients; bold values are statistically significantat †<0.1, *<0.05, **<0.01, ***<0.001

0

400

800

1200

1600

2000

0 1000 2000 3000

Control

Labidomera

Monarch

Jasmonic acid (ng/g)

Sal

icyl

ic a

cid

(ng/

g)

Fig. 7 Trade-offs are often expected in the expression of jasmonic acidand salicylic acid, yet the quantitative comparison of the two hormonesshould only be assessed within species (or genotypes) and within treat-ment groups. Here data are presented for Asclepias exaltata. In this case,pooling data across treatments would have resulted in the false conclusionof a negative correlation between jasmonic acid and salicylic acid

726 J Chem Ecol (2014) 40:717–729

Author's personal copy

leaves open the question of whether monarchs attenuateplant defensive responses, as has been demonstrated in

some other systems (Musser et al. 2005). Indeed, thestronger induced latex to beetles was coupled with re-duced jasmonic acid responses, an unexpected result.

Second, we found more consistent inducible latex by one(Labidomera), but not the other (monarch), insect herbivoreacross all five plant species. This differential response couldbe driven by insect-specific elicitors (i.e., monarch caterpillarsbeing stealthy) or by other differences in their feeding styles orlocations. While Labidomera feeds on older leaves that mightaccumulate lower JA-levels than the younger leaves on whichmonarch larvae prefer to feed, our measures of JA were insystemic apical leaves for all plants. Furthermore, on averagewe found a quantitatively positive relationship between JAand latex across the plant species and treatment combinations,

CONTROL EUCHAETES LABIDOMERA MONARCH

Jasm

onic

aci

d (n

g/g)

Sal

icyl

ic a

cid

(ng/

g)A

bsci

sic

acid

(ng

/g)

Late

x (m

g ex

uded

)

n=246Treatment 21.61***Leaf position (LP)16.57***Treatment X LP 6.86***

n=244Treatment 0.41Leaf position (LP)4.85*Treatment X LP 0.79

n=245Treatment 3.18*Leaf position (LP)0.01Treatment X LP 1.57

n=247Treatment 7.37***Leaf position (LP)<0.01Treatment X LP 0.56

0

50

100

150

200

250

300lower leaves

with apical leaf

0

50

100

150

200

250

300

350

0

5

10

15

20

25

30

35

0

0.2

0.4

0.6

0.8

1

1.2

Fig. 8 Specificity in herbivore-induced foliar hormones and latexexudation in Asclepias syriaca.Phytohormones of replicate plantswere either analyzedwith only theyoungest fully expanded leaf(“lower leaves,” left bar in eachcolored pair), or in cases wherethis was not of sufficient mass,with the youngest fully expandedleaf and a developing leaf locatedabove (“with apical leaf,” righthatched bar in each colored pair).For latex, only one measure wastaken, but leaf position is includedin the model to assess whetherplants that were differentiallyharvested as above haddifferential latex expression.Shown are means ± SE. Samplesizes and results from two-wayANOVAs are shown to the right;Bold F-values are statisticallysignificant at *<0.05, **<0.01,***<0.001

Table 5 Pearson correlations between phytohormones (jasmonic acid,salicylic acid, and abscisic acid: JA, SA, and ABA, respectively) con-ducted within herbivore treatment for Asclepias syriaca

Control Euchaetes Labidomera Monarch

JA-ABA 0.075 0.524** 0.293† 0.379***

SA-ABA 0.122 −0.111 −0.069 0.322***

JA-SA 0.147 −0.096 −0.034 0.232

Shown are correlation coefficients; bold values are statistically significantat †<0.1, *<0.05, **<0.01, ***<0.001

J Chem Ecol (2014) 40:717–729 727

Author's personal copy

despite the lower JA burst and stronger latex response forLabidomera damaged plants. Notwithstanding some varia-tion, this result suggests importantly that although plant spe-cies are consistently responding to herbivores, specificitycould be achieved by modulation of the hormonal response,either induced by the insect or perceived by the plant.

Third, although plant species showed specific cardenolideresponses to herbivores, these effects were rather modest, andonly one species in this experiment (A. obovata) showed astrong induced cardenolide response. We also found no evi-dence for a quantitative association between JA andcardenolides. These results are consistent with our previousresearch, indicating that latex and cardenolides are indepen-dently regulated at the level of plant species, genotype, andresponses to environmental conditions (Agrawal and Fishbein2008; Agrawal et al. 2012a, b; Bingham and Agrawal 2010).Although cardenolide induction may still be regulated by JA,this is more likely operating as a threshold response or acoordinated response with other plant hormones, and possiblyinsect-derived elicitors. Thus far, there have been no investi-gations of the role of ethylene in induced responses of milk-weed, although these are clearly important in other species(Erb et al. 2012; Leon-Reyes et al. 2010), including inducedlatex responses of the rubber tree (Broekaert et al. 1990).

Finally, it may be that correlations between plant hormonesare most strongly exhibited under some conditions. Althoughour treatments generally induced opposing mean responsesbetween JA and SA, these were not evident in quantitativecorrelations. We speculate that a negative correlation betweenJA and SA may be exhibited only when both pathways areelicited (Ali and Agrawal 2014). In the current experiments,JA and ABA were typically positively correlated, and thesewere the strongest when plants were damaged by monarchs.Monarch damage at the apical meristem may induce waterloss (although no leaves were wilted), triggering an ABAresponse. Correlated responses of JA and ABA are frequentlyreported in the literature for other species, although theyappear to be largely independent and reflect the joint waterstress–defense response of plants following herbivory(Lackman et al. 2011; Reymond et al. 2000). For milkweedresponses, we are just beginning to understand how specificinteractions between hormonal signaling may result in altereddefense expression. Given that closely related species haveevolutionarily diverged both in their hormonal signaling aswell as defensive end-products, we speculate that herbivore-driven selection on defense strategies may have contributed tomacroevolution in the milkweeds (Agrawal et al. 2014).

Acknowledgements This study was improved by discussions and in-put from Jared Ali, Rayko Halitschke, Monica Kersch-Becker, GeorgPetschenka, Jennifer Thaler, Tobias Züst, and the Phytophagy Lab atCornell University. We thank Mark Fishbein for help with the phylogenyof Asclepias and Marc Johnson and Peter Van Zandt for collecting

A. obovata seeds. We thank T. Ramsey for help conducting the experi-ments, E. Kearney and C. Creneti for sample preparation, and J. Kerr forhelp with cardenolide analysis. Our research and lab (www.herbivory.com/) is currently supported by NSF-DEB 1118783 and the TempletonFoundation.

References

Agrawal AA (2001) Phenotypic plasticity in the interactions and evolu-tion of species. Science 294:321–326

Agrawal AA (2005) Natural selection on common milkweed (Asclepiassyriaca) by a community of specialized insect herbivores. Evol EcolRes 7:651–667

Agrawal AA (2011) Current trends in the evolutionary ecology of plantdefence. Funct Ecol 25:420–432

Agrawal AA, Fishbein M (2006) Plant defense syndromes. Ecology 87:S132–S149

Agrawal AA, Fishbein M (2008) Phylogenetic escalation and declineof plant defense strategies. Proc Natl Acad Sci U S A 105:10057–10060

Agrawal AA, Konno K (2009) Latex: a model for understanding mech-anisms, ecology, and evolution of plant defense against herbivory.Annu Rev Ecol Evol Syst 40:311–331

Agrawal AA, Fishbein M, Halitschke R, Hastings AP, Rabosky DL,Rasmann S (2009) Evidence for adaptive radiation from a phylogeneticstudy of plant defenses. Proc Natl Acad Sci U S A 106:18067–18072

Agrawal AA, Kearney EE, Hastings AP, Ramsey TE (2012a) Attenuationof the jasmonate burst, plant defensive traits, and resistance tospecialist monarch caterpillars on shaded common milkweed(Asclepias syriaca). J Chem Ecol 38:893–901

Agrawal AA, Petschenka G, Bingham RA, Weber MG, Rasmann S(2012b) Toxic cardenolides: chemical ecology and coevolution ofspecialized plant-herbivore interactions. New Phytol 194:28–45

Agrawal AA, Ali JG, Rasmann S, Fishbein M (2014) Macroevolutionarytrends in the defense of milkweeds against monarchs: latex,cardenolides, and tolerance of herbivory. In: Oberhauser K, AltizerS, Nail K (eds) Monarchs in a changing world: biology and conser-vation of an iconic insect. Cornell University Press (in press)

Ali JG, Agrawal AA (2012) Specialist versus generalist insect herbivoresand plant defense. Trends Plant Sci 17:293–302

Ali JG, Agrawal AA (2014) Asymmetry of plant-mediated interactionsbetween insects from two feeding guilds. Funct Ecol. doi:10.1111/1365-2435.12271

Avanci NC, LucheDD,GoldmanGH, GoldmanMHS (2010) Jasmonatesare phytohormones with multiple functions, including plant defenseand reproduction. Genet Mol Res 9:484–505

BinghamRA, Agrawal AA (2010) Ecological genetics of herbivore-specificinduced defenses in common milkweed. J Ecol 98:1014–1022

Broekaert W, Lee HI, Kush A, Chua NH, Raikhel N (1990) Wound-induced accumulation of messenger-RNA containing a hevein se-quence in laticifers of rubber tree (Hevea brasiliensis). Proc NatlAcad Sci U S A 87:7633–7637

Cipollini D (2004) Stretching the limits of plasticity: can a plant defendagainst both competitors and herbivores? Ecology 85:28–37

De Moraes CM, Lewis WJ, Pare PW, Alborn HT, Tumlinson JH (1998)Herbivore-infested plants selectively attract parasitoids. Nature 393:570–573

Erb M, Flors V, Karlen D, de Lange E, Planchamp C, D’Alessandro M,Turlings TCJ, Ton J (2009) Signal signature of aboveground-induced resistance upon belowground herbivory in maize. Plant J59:292–302

Erb M, Meldau S, Howe GA (2012) Role of phytohormones in insect-specific plant reactions. Trends Plant Sci 17:250–259

728 J Chem Ecol (2014) 40:717–729

Author's personal copy

Fishbein M, Chuba D, Ellison C, Mason-Gamer RJ, Lynch SP (2011)Phylogenetic relationships of asclepias (Apocynaceae) inferred fromnon-coding chloroplast DNA sequences. Syst Bot 36:1008–1023

Halitschke R, Baldwin IT (2003) Antisense LOX expression increasesherbivore performance by decreasing defense responses andinhibiting growth-related transcriptional reorganization inNicotiana attenuata. Plant J 36:794–807

Haribal M, Renwick JAA (1998) Identification and distribution of ovi-position stimulants for monarch butterflies in hosts and nonhosts. JChem Ecol 24:891–904

Heil M, Greiner S, Meimberg H, Kruger R, Noyer JL, Heubl G,Linsenmair KE, Boland W (2004) Evolutionary change from in-duced to constitutive expression of an indirect plant resistance.Nature 430:205–208

Koornneef A, Leon-Reyes A, Ritsema T, Verhage A, Den Otter FC, VanLoon LC, Pieterse CMJ (2008) Kinetics of salicylate-mediatedsuppression of jasmonate signaling reveal a role for redox modula-tion. Plant Physiol 147:1358–1368

Lackman P, Gonzalez-Guzman M, Tilleman S, Carqueijeiro I, Perez AC,Moses T, Seo M, Kanno Y, Hakkinen ST, Van Montagu MCE andothers (2011) Jasmonate signaling involves the abscisic acid recep-tor PYL4 to regulate metabolic reprogramming in Arabidopsis andtobacco. Proc Natl Acad Sci U S A 108:5891–5896

Leon-Reyes A, Spoel SH, De Lange ES, Abe H, Kobayashi M, Tsuda S,Millenaar FF, Welschen RAM, Ritsema T, Pieterse CMJ (2009)Ethylene modulates the role of nonexpressor of pathogenesis-related genes1 in cross talk between salicylate and jasmonate sig-naling. Plant Physiol 149:1797–1809

Leon-Reyes A, Du YJ, Koornneef A, Proietti S, Korbes AP, Memelink J,Pieterse CMJ, Ritsema T (2010) Ethylene signaling renders thejasmonate response of Arabidopsis insensitive to future suppressionby salicylic acid. Mol Plant Microbe Interact 23:187–197

Malcolm SB, Zalucki MP (1996) Milkweed latex and cardenolide induc-tion may resolve the lethal plant defence paradox. Entomol ExpAppl 80:193–196

Malcolm SB, Cockrell BJ, Brower LP (1989) Cardenolide fingerprint ofmonarch butterflies reared on common milkweed, Asclepias syriacaL. J Chem Ecol 15:819–853

Mooney KA, Jones P, Agrawal AA (2008) Coexisting congeners: de-mography, competition, and interactions with cardenolides for twomilkweed-feeding aphids. Oikos 117:450–458

Moreno JE, Tao Y, Chory J, Ballare CL (2009) Ecological modulation ofplant defense via phytochrome control of jasmonate sensitivity. ProcNatl Acad Sci U S A 106:4935–4940

Mur LAJ, Kenton P, Atzorn R, Miersch O, Wasternack C (2006) Theoutcomes of concentration-specific interactions between salicylateand jasmonate signaling include synergy, antagonism, and oxidativestress leading to cell death. Plant Physiol 140:249–262

Musser RO, Cipollini DF, Hum-Musser SM, Williams SA, Brown JK,Felton GW (2005) Evidence that the caterpillar salivary enzymeglucose oxidase provides herbivore offense in Solanaceous plants.Arch Insect Biochem Physiol 58:128–137

Rasmann S, Agrawal AA (2009) Plant defense against herbivory: prog-ress in identifying synergism, redundancy, and antagonism betweenresistance traits. Cur Opin Plant Biol 12(4):473–478

Rasmann S, Agrawal AA (2011) Latitudinal patterns in plant defense:evolution of cardenolides, their toxicity and induction followingherbivory. Ecol Lett 14:476–483

Rasmann S, Johnson MD, Agrawal AA (2009) Induced responsesto herbivory and jasmonate in three milkweed species. JChem Ecol 35:1326–1334

Reymond P, Farmer EE (1998) Jasmonate and salicylate as global signalsfor defense gene expression. Cur Opin Plant Biol 1:404–411

Reymond P, Weber H, Damond M, Farmer EE (2000) Differential geneexpression in response to mechanical wounding and insect feedingin Arabidopsis. Plant Cell 12:707–719

Schmelz EA, Engelberth J, Alborn HT, O’Donnell P, Sammons M,Toshima H, Tumlinson JH (2003) Simultaneous analysis of phyto-hormones, phytotoxins, and volatile organic compounds in plants.Proc Natl Acad Sci U S A 100:10552–10557

Schuman MC, Heinzel N, Gaquerel E, Svatos A, Baldwin IT(2009) Polymorphism in jasmonate signaling partially ac-counts for the variety of volatiles produced by Nicotianaattenuata plants in a native population. New Phytol 183:1134–1148

Thaler JS, Bostock RM (2004) Interactions between abscisic-acid-mediated responses and plant resistance to pathogens and insects.Ecology 85:48–58

Thaler JS, Karban R (1997) A phylogenetic reconstruction of constitutiveand induced resistance in Gossypium. Am Nat 149:1139–1146

Thaler JS, Farag MA, Pare PW, Dicke M (2002) Jasmonate-deficientplants have reduced direct and indirect defences against herbivores.Ecol Lett 5:764–774

Thaler JS, Agrawal AA, Halitschke R (2010) Salicylate-mediated inter-actions between pathogens and herbivores. Ecology 91:1075–1082

Thaler JS, Humphrey PT, Whiteman NK (2012) Evolution of jasmonateand salicylate signal crosstalk. Trends Plant Sci 17:260–270

Traw MB, Bergelson J (2003) Interactive effects of jasmonic acid,salicylic acid, and gibberellin on induction of trichomes inArabidopsis. Plant Physiol 133:1367–1375

Van Zandt PA, Agrawal AA (2004a) Community-wide impacts ofherbivore-induced plant responses in milkweed (Asclepias syriaca).Ecology 85:2616–2629

Van Zandt PA, Agrawal AA (2004b) Specificity of induced plant re-sponses to specialist herbivores of the commonmilkweed, Asclepiassyriaca. Oikos 104:401–409

Vannette RL, Hunter MD (2011) Genetic variation in expression ofdefense phenotype may mediate evolutionary adaptation ofAsclepias syriaca to elevated CO2. Glob Chang Biol 17:1277–1288

Viswanathan DV, Narwani AJT, Thaler JS (2005) Specificity in inducedplant responses shapes patterns of herbivore occurrence on Solanumdulcamara. Ecology 86:886–896

von Dahl CC, Baldwin IT (2007) Deciphering the role of ethylene inplant-herbivore interactions. J Plant Growth Regul 26:201–209

Warashina T, Noro T (2009) 8,14-Secopregnane glycosides from theaerial parts of Asclepias tuberosa. Phytochemistry 70:1294–1304

Woodson RE (1954) The North American species of Asclepias L. AnnMiss Bot Garden 41:1–211

Zalucki MP, Malcolm SB (1999) Plant latex and first-instar monarchlarval growth and survival on three North American milkweedspecies. J Chem Ecol 25:1827–1842

Zalucki MP, Brower LP, Alonso A (2001) Detrimental effects oflatex and cardiac glycosides on survival and growth of first-instar monarch butterfly larvae Danaus plexippus feeding onthe sandhill milkweed Asclepias humistrata. EcologicalEntomol 26:212–224

Zehnder CB, Hunter MD (2007) Interspecific variation within the genusAsclepias in response to herbivory by a phloem-feeding insectherbivore. J Chem Ecol 33:2044–2053

Zehnder CB, Hunter MD (2008) Effects of nitrogen deposition onthe interaction between an aphid and its host plant.Ecological Entomol 33:24–30

J Chem Ecol (2014) 40:717–729 729

Author's personal copy