Journal of Chemical Ecology https://doi.org/10.1007/s10886-019-01046-8

Mechanisms of Resistance to Herbivores in Isolated Breeding Lineages of pepo

Lauren J. Brzozowski1 & Michael Mazourek1 & Anurag A. Agrawal2

Received: 20 August 2018 /Revised: 5 November 2018 /Accepted: 15 January 2019 # Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract Although crop wild ancestors are often reservoirs of resistance traits lost during domestication, examining diverse cultivated germplasm may also reveal novel resistance traits due to distinct breeding histories. Using ten cultivars from two independent domestication events of Cucurbita pepo (ssp. pepo and texana), we identified divergences in constitutive and induced resistance measured by growth of generalist caterpillars and leaf traits. C. p. texana cultivars were consistently more resistant to Trichoplusia ni and Spodoptera exigua, and this was not due to expected mechanisms including cucurbitacins, nitrogen, sticky phloem sap, or toxicity. Although more susceptible on average, C. p. pepo cultivars showed stronger induced resistance, suggesting a trade-off between constitutive and induced resistance. To test the hypothesis that leaf volatiles accounted for differences in resistance to caterpillars, we devised a novel method to evaluate resistance on artificial diet while larvae are exposed to leaf volatiles. In both subspecies, cultivar-specific induced volatiles that reduced T. ni growth were present in highly inducible cultivars, but absent in those that showed no induction. These results have important agricultural implications as cultivar-specific resistance to caterpillars mirrored that of specialist from field trials. Overall, the eponymous cucurbitacin defenses of the are not the mechanistic basis of differences in constitutive or induced resistance between C. pepo subspecies or cultivars. Instead, deterrent cultivar-specific volatiles appear to provide general resistance to insect herbivores. Divergence during breeding history within and between subspecies revealed this pattern and novel resistance mechanism, defining new targets for plant breeding.

Keywords Cucurbita pepo . Cucurbitacins . Herbivore induced plant volatiles . Plant-herbivore interactions . Trichoplusia ni

Introduction drift (Ladizinsky 2012). The effect of breeding history on herbivore resistance is amplified when crop germplasm is iso- Crop plants are often less resistant to herbivores than their lated in distinct breeding pools, and comparisons between wild ancestors (Chen et al. 2015a; Whitehead et al. 2017), these pools provide opportunities to elucidate mechanisms yet variation for resistance persists within crop germplasm. of resistance. Indeed, tracking plant resistance through several During breeding history, traits impacting herbivore resistance genetic lineages revealed a decline in resistance from wild were altered by direct (natural or human-directed) selection, relatives to landraces to modern cultivars (Dávila-Flores indirect consequences of selection on other traits, and genetic et al. 2013; Rodriguez-Saona et al. 2011; Rosenthal and Dirzo 1997), and occasionally uncovered qualitative losses of major resistance traits between isolated breeding lineages Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10886-019-01046-8) contains supplementary (Rasmann et al. 2005). Studying distinct lineages provides material, which is available to authorized users. insight not only on losses, but other possible outcomes, such as novel resistance traits. * Lauren J. Brzozowski Elucidating resistance mechanisms through comparisons [email protected] of cultivated plants with distinct breeding histories is already situated within the context of crop lineages, and thus may be 1 Section of Plant Breeding, School of Integrative Plant Science, advantageous for further agricultural applications. First, dur- Cornell University, Ithaca, NY, USA ing domestication, human consumers typically selected for 2 Department of Ecology and Evolutionary Biology, and Department palatability and against toxicity (Meyer et al. 2012), presum- of Entomology, Cornell University, Ithaca, NY, USA ably eliminating sources of resistance that would be JChemEcol unacceptable for consumption. Additionally, mechanisms of (Coleoptera: Chyrsomelidae), including Acalymma vittatum effective resistance in crop plants may differ from resistance and Diabrotica spp., strongly prefer C. p. pepo over C. p. traits in their wild relatives. Indeed, some resistance traits are texana cultivars (Ferguson et al. 1983;Hoffmannetal. important in wild plant populations, but neutral or detrimental 1996; Brzozowski et al. 2016). Accordingly, these breeding under cultivation (Gaillard et al. 2018; Turcotte et al. 2014). A lineages provide an opportunity to examine how independent separate suite of resistance traits may be more relevant in domestication events and breeding histories shaped the mech- agricultural contexts because the community and intensity of anistic basis of plant resistance to these, and other, herbivores. pests also differ (Chen et al. 2017). For instance, potential pest Cucurbitacins, intensely bitter and toxic triterpenoids, are pressure from specialist at the onset of domestication an important form of resistance across the Cucurbitaceae to of maize may have disrupted balancing selective forces from generalist herbivores (Da Costa and Jones 1971; Metcalf generalist and specialist herbivores (Gaillard et al. 2018). 1985), yet are tolerated and sequestered as a defense against Evaluation of isolated breeding pools that arose post- predators by some specialist beetles (Ferguson and Metcalf domestication have also provided insight into novel resistance 1985; Metcalf et al. 1980; Metcalf 1986; Tallamy and mechanisms, including plant volatiles. Volatile compounds Krischik 1989). Fruits of wild Cucurbita spp. contain serve as information to herbivores, for instance, a warning of cucurbitacins, but today, Cucurbita spp. food crops lack competition (Bruce and Pickett 2011), or as feeding deterrents cucurbitacins due to the identification of bitter-free mutants (Shiojiri et al. 2006; Veyrat et al. 2016). A notable example is during the domestication process (Nee 1990). The implica- the discovery that independent maize breeding programs in tions of loss of fruit bitterness for leaf-chewing herbivores of North America and Europe diverged in the production of the C. pepo are unknown. While the full molecular pathway of volatile compound, (E)-β-caryophyllene (Degen et al. 2004). cucurbitacin production is yet to be elucidated in C. pepo,loss This volatile is induced in maize roots by the corn rootworm of fruit bitterness does not preclude cucurbitacin production in (Diabrotica virgifera virgifera) and provides critical indirect leaves in other Cucurbitaceae, like cucumber (Cucumis resistance through recruitment of entomopathogenic nema- sativus)(Shangetal.2014;Zhouetal.2016). Indeed, todes (Rasmann et al. 2005). Across plant , substantial cucurbitacins are present in cotyledons of some domesticated variation in plant volatiles exists in cultivated germplasm, and C. pepo (Ferguson et al. 1983),buthavebeenreportedtobe several domesticated plants have greater volatile induction as nil or at low constitutive concentrations in true leaves (Metcalf compared to wild relatives (Rowen and Kaplan 2016). Thus, et al. 1980, 1982; Theis et al. 2014). Herbivory was previously use of multiple cultivars with independent breeding histories shown to induce a substantial increase of a leaf cucurbitacin in provides a structure to identify novel resistance mechanisms, one cultivar of C. pepo (Tallamy 1985), but cucurbitacins have like plant volatiles, some of which could become the target of only been evaluated in a small number of cultivars, and their breeding programs. potential role in resistance and breeding requires additional Cucurbita pepo provides an excellent system to test how investigation. Additional defenses of C. pepo have also been divergent breeding histories in independent lineages may have evaluated in some contexts: deterrent plant volatiles have been led to altered defensive traits. Within C. pepo,therearetwo measured in one cultivar (Peterson et al. 1994), and mucilag- cultivated subspecies, C. pepo ssp. pepo (hereafter C. p. pepo) inous sap has been studied in other Cucurbitaceae (McCloud and C. pepo ssp. texana (syn. C. pepo ssp. ovifera;hereafter et al. 1995; Dussourd 1997), but not C. pepo. Apart from C. p. texana), that have been bred to include multiple market defenses, leaf nutrient content is important for herbivore pref- classes, including pumpkin and zucchini in C. p. pepo,and erence and performance (Awmack and Leather 2002;Mattson acorn squash and summer squash in C. p. texana (Paris 2000). 1980). Nitrogen content was measured in a survey of These subspecies arose from two separate domestication Cucurbitaceae species but was not associated with specialist events (Decker 1988;Nee1990; Sanjur et al. 2002), and cross (A. vittatum) abundance (Theis et al. 2014). breeding is uncommon in modern breeding programs (Gong In this study, we determine how plant resistance mecha- et al. 2012). C. p. pepo is thought to have been domesticated in nisms diverged between the two cultivated C. pepo subspe- central Mexico from a yet unknown wild progenitor (Sanjur cies, representing two independent domestication events et al. 2002), but cultivar development predominately occurred where isolated breeding pools have been maintained. One in Europe starting in the sixteenth century (Paris 2000). In subspecies, C. p. texana, has had continuous interaction with contrast, C. p. texana was domesticated and developed in specialist beetle pests, while the other (C. p. pepo) was geo- the eastern United States and northern Mexico (Decker graphically separated. We thus hypothesized that these distinct 1988; Sanjur et al. 2002;Smith2006). Because C. p. texana breeding histories would lead to disparate defense strategies, has been developed in its center of origin, it exclusively has with different effects for generalists and specialists. We used been exposed to specialist beetle pests endemic to North five cultivars each of C. p. pepo and C. p. texana to test 1) for America (Metcalf 1985) throughout its breeding history. divergence in constitutive and induced resistance to the two North American specialist beetles in the Diabroticina subtribe leaf-feeding generalist caterpillars, Trichoplusia ni and JChemEcol

Spodoptera exigua, as measured by larval performance, and 2) standard greenhouse fertilizing practices (150 ppm 21–5-20 if differences in resistance were associated with secondary NPK fertilizer, (Peters Company, Allentown PA, USA) five metabolites or a suite of other foliar traits. Plant chemical traits times a week), and non-chemical pest control (bio-control) including cucurbitacins and nitrogen were measured and was used as necessary. Lambert LM-111 potting mix complemented by insect growth and behavioral assays. To (Rivière-Ouelle, Québec, Canada) was used in all assays ex- distinguish between deterrence and toxicity, we examined cept the volatile assay, in which McEnroe Organic Lite growth efficiency of caterpillars as well as the impact of con- Growing mix (Millerton, NY, USA) was used to be in com- stitutive and induced volatiles on caterpillar growth. Induction pliance with requirements of a certified organic greenhouse. by caterpillar feeding was also compared to elicitation by Seeds were sown in individual 10 cm diameter pots for assays jasmonic acid, the plant hormone primarily responsible for that required whole plants (whole plant feeding assays), and orchestrating induced resistance. And finally, given that spe- 72-cell flats for assays that used excised plant tissue (mass per cialist beetle associations differentially affected C. pepo unit area consumed of leaf discs, volatile assay). breeding histories, 3) we sought to test how findings from generalist herbivores related to specialist beetle preference. Insects Trichoplusia ni is polyphagous on plants in at least 36 Specifically, we tested the generality of resistance by relating botanical families, including C. pepo and multiple resistance to caterpillars to field preference of a major agricul- Cucurbitaceae (Sutherland and Greene 1984). Spodoptera tural leaf-feeding specialist beetle pest (A. vittatum). In sum- exigua is likewise a generalist herbivore feeding on more than mary, we used the independent domestication events of 20 plant families, also including Cucurbitaceae (Tietz 1972). C. pepo as a means to identify mechanisms of resistance to Prior to conducting the experiments described here, we con- multiple herbivores that may be useful in plant breeding. ducted a feeding trial (in February 2015) to assess degree of feeding on C. pepo, and we observed substantial feeding by both species (data not presented). Spodoptera exigua eggs Methods and Materials were sourced from Benzon Research (Carlisle, PA, USA), and Trichoplusia ni eggs were supplied from a colony at Plant Material Five Cucurbita pepo cultivars (Table 1)each Cornell University (Dr. Ping Wang, Cornell AgriTech, were used from two cultivated subspecies, C. p. pepo and C. p. Geneva, NY, USA). Due to similarity of results between cat- texana (Gong et al. 2012; Paris et al. 2003). Plants were started erpillar species we found at the subspecies level, and high from untreated seed (source, Table 1) in the Cornell University S. exigua mortality in the induced resistance assay, T. ni was Agricultural Experiment Station greenhouses (Ithaca, NY, used in all subsequent assays. For assays where caterpillar USA). In the greenhouses, a 14 hr photoperiod was main- mass was measured, two unfed neonate caterpillars were ap- tained, and the day and night temperatures, were 27 C and plied to each plant or diet. At the conclusion of each assay, all 21 C, respectively. Plants were watered daily, treated with living caterpillars were placed in Eppendorf tubes and frozen

Table 1 Cultivar list Subspecies Type Cultivar name Abbreviation Seed source

C. p. pepo pumpkin Charisma PMRa,f CH Johnny’s Selected Seeds C. p. pepo zucchini Dunjaa,b,d DU Johnny’s Selected Seeds C. p. pepo pumpkin Magic Lanterna,c,e ML Harris Seeds C. p. pepo zucchini Costataa,c,f CO Cornell University C. p. pepo zucchini Reward F1a,c RE Osborne Seed Company C. p. texana acorn Honey Beara,c HB Johnny’s Selected Seeds C. p. texana delicata PMR Bush Delicataa BD Cornell University C. p. texana straightneck Success PMa,b,c,e SP Cornell University C. p. texana crookneck Sungloa,f SU Osborne Seed Company C. p. texana acorn Sweet Rebaa,b,c,d SR High Mowing Organic Seeds

a Included in induced resistance trial b Included in jasmonic acid assays c Included in leaf trenching assay d Included in mass per unit leaf area assays, and volatile assays as most inducible cultivar e Included in mass per unit leaf area assays, and volatile assays as least inducible cultivar f Not included in beetle correlation estimates because defoliation data was not available JChemEcol at −20 C for individual weighing later (AT21 Comparator protocol is included in Supporting File S1.Briefly, Microbalance, Mettler-Toledo, Columbus, OH, USA). In cucurbitacins were extracted from freeze-dried tissue with evaluating leaf trenching, T. ni were raised to second instar methanol, and were then purified with solid phase extraction. on high wheat germ diet (Bell et al. 1981) in a 26 C growth Cucurbitacins were quantified in a triple-quadrupole LC-MS/ chamber before the assay. The same diet was also used in the MS system (Accela-Quantum Access; Thermo Scientific) volatile assay. equipped with a C18 reversed-phase column (Kinetex 2.6 μm EVO C18, 150 × 2.1 mm; Phenomenex). Induced Resistance and Plant Chemistry in the Two Subspecies – Induced Resistance Assay All cultivars were Induced Resistance and Plant Chemistry in the Two grown to assess constitutive and induced leaf chemistry traits Subspecies – Nitrogen Analysis Tissue for nitrogen analysis and plant resistance to T. ni and S. exigua.Thisexperiment was sourced from the same freeze dried tissue used for was conducted in two iterations in a randomized complete cucurbitacin analysis. Tissue was finely ground (120 s at block design with three blocks per iteration. In each block, 27 Hz, MM300, Retsch, Haan, Germany) and submitted to there were seven plants of each cultivar and each plant was the Cornell University Stable Isotope Laboratory (Ithaca, NY, subjected to one of seven treatments (Bt^): (t1) T. ni and (t2) USA) for continuous flow analysis of percent nitrogen and S. exigua induction for chemical analyses, (t3) no herbivory carbon with an elemental analyzer. Per cultivar, five to six control for chemical analyses, (t4) T. ni and (t5) S. eixgua samples of controls and each treatment (T. ni and S. exigua induction to measure the effect on subsequent conspecific prior herbivory) were measured, with the following excep- herbivory, and finally untreated controls for induction by (t6) tions: n = 4 for cv. Success PM – S. exigua; n =3 forcv. T. ni and (t7) S. exigua. Plants induction was achieved by five Sweet Reba – S. exigua, T. ni. days (days 1–5) of herbivory by neonates immediately prior to chemical analyses or measuring effect on subsequent conspe- Induced Resistance and Plant Chemistry in the Two cific herbivory. The effect of induction was measured by cat- Subspecies – Jasmonic Acid Treatment Three cultivars were erpillar mass after five days (days 6–10) of feeding. used to test if jasmonic acid (JA) treatment had similar induc- Seeds were sown in February 2015, approximately two tion effects to prior herbivory. We included two cultivars we weeks prior to treatment to allow for plants to reach the 1–2 had found to be highly inducible, C. p. pepo (cv. Dunja) and leaf stage, and then plants were enclosed in mesh sleeves C. p. texana cultivar (cv. Sweet Reba), by T. ni in previous (30 cm × 18 cm). On the first day of the experiment assays, and a less inducible C. p. texana cultivar (cv. Success (March 2015), plants were infested with T. ni (t1, t4), PM). Seed were sown in September 2015, and appropriate S. exigua (t2, t5), or left as is (t3, t6, t7). On day five, leaf plants were sprayed with jasmonic acid (0.5 mM JA, dis- tissue was collected from plants with treatments for chemical solved in ethanol) ten days after sowing. In one test, JA treated analyses (t1-t3; i.e. cv. Dunja with T. ni feeding, S. exigua plants were compared to plants sprayed with solvent (ethanol) feeding, and no herbivore control). The caterpillars were also alone with two to six replicates per treatment-cultivar. A sep- weighed from those plants, and the plants were discarded. arate set of JA treated plants were compared to plants that Also on day five, caterpillars were removed from the remain- received two neonate T. ni, or nothing (control) with 2 itera- ing T. ni and S. exigua feeding treatments (t4, t5), and tions of 6 blocks (where three blocks were complete, and three weighed. Those plants (t4, t5) were then infested with new were nearly so, typically missing a single treatment-cultivar neonate conspecifics to test the effect of induction by conspe- combination) per cultivar and treatment. For both tests, five cific prior herbivory. At the same time, the no herbivore con- days after the treatments commenced (for JA treatments, trol plants were infested with T. ni (t6), or S. exigua (t7) to there was one spray at the beginning of the five day compare to the effect of prior herbivory. The caterpillars ap- period), initial T. ni were removed from the appropriate plied on day five were allowed to feed until day 10 when they treatments, and then all treatments and control were were removed and weighed. All caterpillars collected from infested with two neonate T. ni, which were allowed plants on day five were analyzed to examine constitutive re- to feed for five days before weighing. sistance in all cultivars (n = 12 per cultivar). Caterpillars col- lected from plants on day 10 were used to examine the degree Feeding and Growth Bioassays – Qualitative Leaf Trenching A of induced resistance (n = 6 per cultivar-treatment subset of six cultivars (three per subspecies, see Table 1)were combination). sown on January 2017, and were grown to the three leaf stage before second instar T. ni were placed on the plants. The plants Induced Resistance and Plant Chemistry in the Two were observed daily for eight days for evidence of trenching, Subspecies – Cucurbitacin Analysis Cucurbitacins were ex- and qualitative notes and photographs were taken. Leaf tracted in a method similar to Theis et al. (2014), and a de- trenching was observed in borealis in response to tailed description of the extraction, quantification and analysis mucilaginous plant sap in C. maxima,wildC. p. texana,and JChemEcol

C. okeechobeensis (McCloud et al. 1995), and also from T. ni was on the bottom of the container, and the leaf was in C. sativus and C. moschata (Dussourd 1997). Observation suspended 10 cm above the diet cup, and was kept in place of T. ni trenching behavior between C. pepo subspecies was by the floral tube. As a check of the effect of the non-plant chosen to test if T. ni exhibited a differential response that may materials, a control treatment with no leaf but all accessories be indicative of differences in plant sap defenses. was also used. After three days, leaves with the T. ni induction treatment were scouted to confirm T. ni presence and feeding, Feeding and Growth Bioassays – Mass Gained per Unit Leaf and T. ni feeding on diet were recovered and weighed. Area To address how larval growth was associated with leaf consumption, which can provide an indication of deterrence Resistance Comparison to A. vittatum To address the gener- versus toxicity, we conducted a bioassay of T. ni on four cul- ality of resistance mechanisms, the mass of T. ni caterpillars tivars. Using the most and least T.ni - inducible cultivars we was compared to previously obtained preference data of a identified in the full cultivar panel from each subspecies specialist herbivore of cucurbit crops, Acalymma vittatum (Table 1), we presented T. ni with a single 10 cm2 leaf disc (Brzozowski et al. 2016). The experiment is detailed in on moistened filter paper in a plastic petri dish. Seeds were Brzozowski et al. (2016), and the objective was to assess sown in March 2016, discs were removed from the newest A. vittatum preference for cultivars in the two subspecies of fully expanded true leaf with a cork borer after two weeks of C. pepo used in this experiment. Briefly, a field choice test growth, and T. ni fed on the discs for five days. Cultivars cv. with n =27cultivars(n =17 C. p. texana,andn =10 C. p. Dunja and cv. Success PM had 20 replicates, cv. Magic pepo) (Brzozowski et al. 2016, Table 2) was conducted in Lantern had 19 replicates, and cv. Sweet Reba had nine repli- 2015 under naturally occurring A. vittatum infestation in cates. Leaf discs were imaged, and area damaged was mea- Freeville, NY. Cultivars were grown with five replicates in sured in imageJ (Schneider et al. 2012). three-plant plots in a randomized complete block design, and Later, the cultivars were grown to measure 10cm2 leaf disc A. vittatum preference was measured as estimated percent leaf fresh and dry weight. Seeds were sown in November 2017, defoliation of plants with one to three leaves (not flowering). and nine samples per cultivar from separate plants were re- Seven of the cultivars used in the field experiment were also moved and weighed two weeks after sowing (HR-120, A&D used in experiments with T. ni (see Table 1). Importantly, in Company, Tokyo, Japan). The discs were lyophilized both experiments, plants were at the same growth stage (1–3 (FreeZone 2.5, Labconco, Kansas City, MO, USA) until dry leaves, non-flowering). To test for C. pepo cross-resistance to and weighed immediately. these herbivores, we determined the correlation between A. vittatum preference and T. ni performance on these Feeding and Growth Bioassays – Volatile Deterrence We test- cultivars. ed if foliar volatiles influenced T. ni feeding on artificial diet using leaf tissue from cultivars we previously found to be the most and least inducible by T. ni from each subspecies (Table 1). This experiment was conducted in two iterations Table 2 Anova table from linear mixed effects model of herbivore mass of a randomized block design. Iteration 1 (August 2017) had in cultivar panel induced resistance assay three complete blocks, and iteration 2 (January 2018) had Insect Effecta,b DF F-value P value seven blocks (where three blocks were complete, and four were nearly so, typically missing a single treatment-cultivar T. ni Subspecies 1 8.241 0.004 combination). Each cultivar had an induction treatment (T. ni Treatment 1 6.133 0.005 feeding on leaf tissue), or control treatment (leaf alone) per Subspecies X Treatment 1 2.344 0.130 block. The seeds were sown approximately 14 days before the Iteration 1 4.834 0.015 assay commenced. Photographs and a detailed description of Residuals 97 the experimental arena are shown in Fig. S4, and described S. exigua Subspecies 1 3.482 0.066 briefly here. In each arena, neonate T. ni were placed on an Treatment 1 0.239 0.626 excess of diet in a small plastic cup covered by cheesecloth. Subspecies X Treatment 1 0.200 0.657 Excised leaves were placed in 9.5 mL floral tubes (Floral Iteration 1 0.135 0.714 Supply, Fruit Heights, UT, USA) filled with water, and refilled Residuals 70 as necessary. An organza mesh bag (SumDirect manufactur- Numbers in bold indicate significant differences at α =0.05 ing, Dongguan, China) was secured around the leaf, and two a Treatment refers to induction by conspecific feeding neonate T. ni were added to leaves of the induction treatments. b The diet cup and excised leaf were placed together in a 1 L There were two nested random effects: randomized complete block nested within iterations of the experiment (Biteration^)(T. ni model plastic container (Clear Lake Enterprises, Port Richey, FL, σ2 =0;S. exigua model σ2 = 0.0002) and cultivar nested within subspe- USA), and closed with a lid with small holes. The diet cup cies (T. ni model σ2 = 0.0104; S. exigua model σ2 = 0.0009) JChemEcol

Statistical Analysis Linear mixed models were used to model variance was used to test significance of fixed effects, except the response variables of caterpillar performance or chemical for in the mass per unit leaf area consumed assay where anal- concentration as a function of plant cultivar and experimental ysis of covariance was used. Tukey’s honest significant differ- parameters. For each caterpillar sample, the mass of the two ence test was used to separate effect levels with the caterpillars was averaged, and the average was used in further ‘TukeyHSD’ function in ‘agricolae’ R package (de analysis. If only one caterpillar was recovered, that mass was Mendiburu 2016). Finally, correlation between A. vittatum used. preference and T. ni mass was calculated using the ‘cor.test’ For the induced resistance and chemical assays, caterpillar Rfunction. mass or chemical concentration, respectively, were modeled with subspecies, treatment, subspecies by treatment interac- tion, and iteration as fixed effects, and cultivar nested within Results subspecies and block nested within iteration were included as random effects. For assays with a subset of cultivars (jasmonic Induced Resistance and Plant Chemistry in the Two acid, mass per unit leaf area consumed, volatiles), caterpillar Subspecies Both generalist caterpillars, T. ni and S. exigua, mass was modeled with cultivar, treatment, and cultivar by showed >40% lower mass after five days of feeding on C. p. treatment interaction as fixed effects. In the jasmonic acid texana cultivars compared to C. p. pepo cultivars (Fig. 1a,c; and volatile assay, iteration was treated as a fixed effect, and T. ni: F1,93 =20.00, P <0.001; S. exigua, F1,89 =10.08, P = block nested within iteration was included as a random effect 0.002). Induced resistance following herbivory reduced for the volatile assay. growth of both species, although the effects were most pro- All statistical analyses were performed in R (R Core Team nounced in C. p. pepo with T. ni. Induced resistance reduced 2016). Linear mixed models were calculated with the ‘lmer’ T. ni mass by 21% in C. p. pepo, and 9% in C. p. texana function in the ‘lme4’ R package (Bates et al. 2015), and linear (Table 2;Fig.1b). For S. exigua, induced resistance reduced models with the ‘lm’ function. In all cases, analysis of mass by 7% in C. p. pepo, but had no effect in C. p. texana

Fig. 1 Mass of generalist caterpillars after feeding on C. pepo subspecies with and without induction by prior conspecific herbivory. Differences in constitutive resistance by subspecies and cultivar are shown for (a) T. ni and (c) S. exigua. The effect of induction is summarized across varieties by subspecies (b, d). Shown are means ±1 SE and asterisks indicate P < 0.05 for the effect of subspecies in ANOVAs described in the text. Cultivar abbreviations are listed in Table 1 JChemEcol

(Table 2; Fig. 1d). Based on these results, we selected the most one non-inducible C. p. texana;Table1;Fig.S1). and least inducible cultivars from each subspecies for further Effects of induction by JA were similar to induction assays (Table 1; Fig. S1). Mean caterpillar mass reduction via prior conspecific herbivory (Fig S2; Table S4), and dis- across both subspecies after induction by prior herbivory tinct from controls (Table S3). was greatest in cv. Dunja (C. p. pepo; 28% decrease; F1,17 = 5.08, P = 0.038), and cv. Sweet Reba (C. p. texana; 26% de- Feeding and Growth Bioassays T. ni exhibited trenching (to crease; F1,10 =5.49,P = 0.041). In contrast, cv. Magic Lantern avoid sticky phloem sap) on all C. pepo cultivars (C. p. pepo) and cv. Success PM (C. p. texana)showedno (Fig. 3), but appeared to avoid feeding on C. p. texana trends of induction, and were chosen as the least inducible cultivars for a longer period of time before trenching (Fig. S1). (LB, personal observation). Cucurbitacins B, D, and E were detected in leaf tissue but To address the nutritional quality or potential toxicity of only in trace concentrations; in the majority of samples, none C. pepo, we conducted an assay to measure larval mass gained of the cucurbitacins reached detectable levels (greater than by T. ni per unit leaf area consumed using four cultivars (two 0.05 ng g−1 dry weight for cucurbitacins B and E, and above of each subspecies, representing extremes of induction 1.0 ng g−1 dry weight for cucurbitacin D), and there was no response Table 1;Fig.S1). T. ni mass on C. p. pepo cultivar pattern of cucurbitacins by subspecies or induction treatment leaf discs was 32% higher than those feeding on C. p. texana

(Table 3). In control plants, mean leaf nitrogen was nearly 9% cultivars (F3,64 =6.77, P < 0.001; Table S5). Similarly, T. ni higher in C. p. texana as compared to C. p. pepo cultivars caterpillars consumed 58% more leaf area of C. p. pepo discs

(F1,51 = 11.60, P =0.001,Table S1). However, following in- (F3,64 = 7.75, P < 0.001, Table S5). However, there was no duction by T. ni herbivory, nitrogen remained constant in C. p. difference between cultivars in T. ni mass attained per unit leaf texana, but dropped 5% in C. p. pepo as compared to controls area consumed (Fig. S3; Table 4). Separately, leaf disc fresh (Fig. 2a; Table S2). Leaf nitrogen also slightly decreased in and dry mass was measured, and fresh mass was significantly

C. p. pepo with induction by S. exigua, and had a smaller different between cultivars (F3,31 = 14.50, P < 0.001; change in C. p. texana as compared to controls (Fig. 2b; Table S5); nonetheless dry mass per area was consistent

Table S2). (F3,31 =1.55,P = 0.211; Table S5). The results did not change We tested whether the effects of induced resistance when we converted area consumed to fresh mass consumed: by prior herbivory could be reproduced by application T. ni caterpillars consumed 65% more fresh leaf mass of C. p. of jasmonic acid (JA) using a subset of cultivars (one pepo discs (F3,64 =13.37,P < 0.001; Table S5), and T. ni mass inducible C. p. pepo, one inducible C. p. texana,and gained per fresh mass consumed did not differ between culti-

vars (F3,64 =0.98,P = 0.41). In summary, these feeding assays Table 3 Leaf cucurbitacin measurements strongly indicate that caterpillar growth was proportional to feeding, and did not appear to be due to leaf toxicity. Cultivar Subspecies Cucurbitacin (ng g−1 dry tissue)a,b c,d In our volatile bioassay, neonate T. ni were fed artificial diet Control / T. ni Induced with exposure to volatiles from an excised leaf of one of four DBE cultivars, representing extremes of induction response in each subspecies (Table 1; Fig. S1; Fig. S4). Mass of T. ni feeding on Charisma PMR C. p. pepo -/- -/- -/- diet was compared between those exposed to constitutive Dunja C. p. pepo -/- -/- -/- plant leaf volatiles and those exposed to plant volatiles in- Magic Lantern C. p. pepo -/- -/- -/- duced by active T. ni feeding. The mass of T. ni feeding on PMR Costata C. p. pepo - / - - / 0.06 0.47 / 1.13 artificial diet was significantly affected by cultivar of the ex-

Reward F1 C. p. pepo -/- -/- -/- cised leaf and induction treatment (F3,57 = 3.82, P = 0.015; Honey Bear C. p. texana -/- -/- -/- Table 5;Fig.4). Volatiles from leaves with active conspecific PMR Bush Delicata C. p. texana -/- 0.07/- -/- feeding significantly lowered mass of T. ni feeding on artificial Success PM C. p. texana -/- -/0.10 -/0.06 diet in the two cultivars previously found to show the stron-

Sweet REBA C. p. texana 5.30 / - - / - - / - gest induction response (C. p. pepo cv. Dunja: F1,16 =5.59, P = 0.031; C. p. texana cv. Sweet Reba: F1,13 =7.44, P = a B ^ - indicates that the particular cucurbitacin was not detected in 0.017; Fig. 4), but not in the two cultivars with weak induction the sample responses (C. p. pepo cv. Magic Lantern: F1,17 =0,P =0.99; b Cucurbitacin I was not detected in any samples, and not presented in this table C. p. texana cv. Success PM: F1,8 =0.80,P =0.40;Fig.4). c BT. ni Induced^ refers to five days of T. ni damage prior to sampling Resistance Comparison to A. vittatum Correlation between d Induced cucurbitacins were additionally measured after S. exigua in cultivars C. p. pepo cv. Charisma PMR, C. p. pepo cv. Dunja, and C. p. adult A. vittatum defoliation previously measured in the field texana cv. Success PM, but none were detected (Brzozowski et al. 2016) and T. ni mass gain on the same JChemEcol

Fig. 2 Leaf nitrogen content in two C. pepo subspecies and induction by prior herbivory by (a) T. ni and (b) S. exigua.Shown are means ±1 SE and the asterisk indicates P < 0.05 for subspecies by induction treatment term in ANOVAs described in Table S2

cultivars measured here (Table 1) was strongly positive C. pepo. While C. p. pepo had lower constitutive resistance, it (Pearson’s r =0.893,df =5,P =0.007;Spearman’srankcor- was more strongly inducible when assayed by measuring growth relation, rho =0.964,P =0.003)(Fig.5). of two generalist, leaf-chewing herbivores (T. ni and S. exigua). Lower resistance to these generalists in C. pepo mirrored greater preference of an important specialist beetle pest, A. vittatum. Discussion Specific analyses of multiple metabolites, including leaf cucurbitacins, did not explain these differences, and a growth Evaluating defensive traits in multiple cultivars from two parallel assay demonstrated that caterpillars gained equal mass per unit domestication events revealed contrasting levels of resistance leaf consumed in cultivars from both subspecies (consistent with and a novel mechanism of plant resistance to insects in the lack of a toxic principle between cultivars). These results

Fig. 3 Examples of semi-circular trenching by second instar T. ni on leaf edge when feeding on (a) Success PM (C. p. texana), (b) Reward (C. p. pepo), and exam- ples of major vein cutting on (c) Honey Bear (C. p. texana), (d) Magic Lantern (C. p. pepo). All cultivars included in this assay are indicated in Table 1 JChemEcol

Table 4 Ancova table for the leaf area consumed covariate for T. ni Plant domestication and breeding trajectory has a range of mass in the mass gained per unit leaf area bioassay effects on herbivore-induced plant volatiles. Overall, recent Covariate Effect DF F-value P value meta-analyses across systems showed that domesticated plants have greater induced volatile production than wild Leaf area consumed Cultivar 3 2.289 0.088 plants (Rowen and Kaplan 2016), although parasitoid and Leaf area 1 37.563 <0.001 predator attraction are not consistently greater in cultivated Cultivar X Leaf area 3 0.692 0.561 plants (Chen et al. 2015b). In maize, there was considerable Residuals 60 genetic variability in induced volatile production in a diverse group of cultivars and wild accessions, and this was not pre- Numbers in bold indicate significant differences at α =0.05 dicted by domestication status or subsequent breeding (Gouinguené et al. 2001). Nonetheless, by examining finer- echo Theis et al. (2014), where they likewise found that scaled contrasts in maize germplasm, it was revealed that plant neither leaf nitrogen nor cucurbitacins (nor other mea- breeding history shaped herbivore-induced root volatiles sured leaf traits) predicted leaf damage by A. vittatum critical for indirect defense: isolated North American on 20 varieties from 12 diverse Cucurbitaceae species. and European breeding programs diverge in production An important result of our work is that volatiles induced by of (E)-ß-caryophyllene (Rasmann et al. 2005), and mod- active T. ni damage in some varieties had a deterrent ern hybrids lack oviposition induced volatiles present in effect on T. ni growth when feeding on a standard diet. landraces (Tamiru et al. 2011). In a more recently do- As discussed below, the impact of plant volatiles on mesticated crop with a known pedigree, cranberry larval feeding has been little studied, and is an area in (Vaccinium macrocarpon), induced volatiles measured need of further investigation. The volatile deterrent ef- following treatment with exogenous jasmonic acid were fect we found was not specific to either C. pepo sub- consistent across pedigree and state of cultivar improve- species, but instead was detected in cultivars that we ment (Rodriguez-Saona et al. 2011). These results indi- found to have the strongest inducible resistance. While cate that while varietal-specific induced volatiles are floral volatile composition in Cucurbita spp. and the common, we may be best able to elucidate changes to implications for specific herbivores and pollinators are well- herbivore-induced plant volatile production by evaluat- characterized (Andersen and Metcalf 1986; Andrews et al. ing clearly defined contrasts of plant breeding lineages. 2007; Theis et al. 2009, 2014), knowledge of leaf volatile Mechanistically, volatile repellents can affect herbivores components is limited. Volatiles from leaf trichomes of through signaling status as a non-host with criterion as subtle one C. p. texana cultivar not used in this study were as one component of a volatile blend, or warning herbivores of shown to have attractant and repellent compounds to the plant defenses, potential competition, or natural enemies (Bruce pickleworm moth (Diaphania nitidalis) (Peterson et al. and Pickett 2011). The impact of plant volatiles on larval feeding 1994). Volatiles from induced by bacterial wilt infection behavior has thus far not received substantial attention, although (Erwinia tracheiphila)inwildC. pepo are also known it may have potential as a resistance mechanism. In maize, indole to attract more A. vittatum than healthy plants (Shapiro is an herbivore induced plant volatile and was recently shown to et al. 2012). Finally, in other Cucurbitaceae, induced reduce feeding and increase mortality of Spodoptera littoralis volatiles have been implicated in the attraction of natural (Veyrat et al. 2016). Deterrent volatiles produced during the enemies of herbivores (Agrawal et al. 2002; Kappers et al. day also drive the nocturnal feeding behavior of Mythimna 2011; Takabayashi et al. 1994). separate on maize (Shiojiri et al. 2006). Overall, deterrent plant volatiles can impact caterpillar feeding behavior, but more work Table 5 Anova table from linear mixed effects model of T. ni mass after feeding on artificial diet while exposed to plant volatiles is needed to understand the mechanistic basis and ecological relevance of such signals. a,b Effect DF F-value P value While we found cultivar-specific induced volatile deter- rence in both subspecies, there are remaining questions about Cultivar 3 5.276 0.003 what mechanisms support the large and repeatable differences Treatment 1 6.939 0.011 in overall resistance between C. pepo subspecies. It is possible Cultivar X treatment 3 3.818 0.015 that induced volatiles are more ubiquitous in one subspecies, Iteration 1 2.136 0.149 and that could be addressed by testing induced volatile deter- Residuals 57 rence in a broader C. pepo panel. We observed no evidence of Numbers in bold indicate significant differences at α =0.05 leaf toxicity, but instead less consumption of C. p. texana a Treatment refers to conspecific leaf feeding cultivars in no-choice assays, which indicates that there may b Effect of block was treated as a nested random effect within experimen- be differences in constitutive deterrent volatiles between sub- tal iteration (σ2 =0.008) species that our volatile assay did not detect. Moving beyond JChemEcol

Fig. 4 Mass of T. ni feeding on artificial diet for three days when exposed to volatiles from leaves alone, leaves with T. ni feeding, or controls (Bempty^). Shown are means ±1 SE and asterisks indicates P < 0.05 for effect of treatment (leaf alone, or with T. ni feeding) in varietal-specific ANOVAs described in the text. # indicates the two cultivars that demonstrated the strongest in- duction response in previous assays

secondary metabolites, there are morphological differences in and induced resistance strategies are apparently less common plant traits between subspecies, including leaf color and in domesticated than wild plants (Kempel et al. 2011). An shape, but their relation to herbivore resistance is unknown exploration of costs of endemic herbivore damage in the re- (Paris et al. 2012). Perhaps examination of these traits, or other gion of domestication and subsequent breeding may provide more multi-functional traits, like leaf water content (Theis insight into whether natural or human-mediated selection may et al. 2014), or leaf turgor pressure (McCloud et al. 1995), have favored one type of resistance over another in each would provide insight into the differences in plant resistance subspecies. in C. pepo. Because C. p. pepo appears to be most palatable for Additional work on mechanisms of preference in C. pepo both generalists assayed and specialist beetles, whatever should also examine causes of divergence in constitutive and mechanism increases C. p. pepo susceptibility to the induced resistance between the independent domestication two generalist herbivores assayed here could be the events. Induction is overall stronger in susceptible C. p. pepo, same as that which increases preference of specialist beetles but there is greater constitutive resistance in the resistant C. p. (Fig. 5) (Chrysomelidae: Galerucinae; Brzozowski et al. 2016; texana. Induced resistance may be favored when herbivore Ferguson et al. 1983;Hoffmannetal.1996). Literature pressure is intermittent and cost of defense is high (Karban on the preference of these beetles has been overwhelm- and Baldwin 1997), although tradeoffs between constitutive ingly associated with cucurbitacin content, where these specialists compulsively feed on and sequester cucurbitacins (Metcalf et al. 1980), and cucurbitacins increase larval perfor- mance (Tallamy and Gorski 1997; Halaweish et al. 1999). However, loss of fruit bitterness is a hallmark of all six do- mestications of the cultivated Cucurbita spp. (i.e. C. pepo, C. moschata, C. maxima,etc)(Decker1988;Nee1990; Sanjur et al. 2002), and we demonstrate here that leaf cucurbitacin content is minimal in domesticated C. pepo. These results are inconsistent with cucurbitacins being the principal determinant of generalist or specialist herbivore be- havior, yet these diverse herbivores still find common ground in increased susceptibility of C. p. pepo. Our results have interesting implications for how specialist and generalist herbivores respond to plant resistance traits in Fig. 5 Correlation of mass of T. ni after feeding for five days on control an agricultural context, and whether there is a dichotomy be- (non-induced) plants compared to percent defoliation by A. vittatum in field plots (Brzozowski et al. 2016) of the same cultivars (Table 1). Each tween them. For instance, while many specialists have point represents the mean of an individual cultivar ±1 SE evolved to overcome specific chemical defenses that are JChemEcol effective against generalist herbivores through sequestration Acknowledgements We thank Wendy Kain and Ping Wang for providing or avoidance (Ali and Agrawal 2012), such Bresistance^ traits T. ni, Georg Jander for providing S. exigua, William Holdsworth for assistance with cultivar selection, Amy Hastings for supporting laborato- may often be lost in the domestication process (Chen et al. ry and greenhouse work, Katalin Boroczky for development of HPLC- 2015a; Whitehead et al. 2017). In the current study, we found MS methods for cucurbitacin detection, Taylor Anderson for de- a similar outcome of the breeding process for well-adapted velopment of cucurbitacin solid phase extraction protocol, and the specialists and non-adapted generalists of C. pepo, indicating Cornell University Agricultural Experiment Station greenhouse staff for providing excellent care of plant material. The manu- substantial cross resistance. script was improved by thoughtful comments from Katja Poveda Overall, this work highlights the benefits of studying and two anonymous reviewers. LB was supported by a Cornell the chemical ecology of diverse pools of cultivated University Presidential Life Science Fellowship (2014-2015) and a germplasm with distinct breeding histories for discover- Seed Matters Graduate Student Fellowship (2015-2019). This work was supported by the United States Department of Agriculture ing new mechanisms of resistance to insect herbivores. National Institute of Food and Agriculture Multi-State Hatch Project Dogma in plant breeding is that we must look to the 1008470, Harnessing Chemical Ecology to Address Agricultural Pest genetic diversity of wild species for traits to introgress and Pollinator Priorities. into elite germplasm to steel it against our most press- ing biotic and abiotic challenges (Dempewolf et al. 2017; Funding LB was supported by a Cornell University Presidential Life Science Fellowship (2014–2015), and a Seed Matters Graduate Student Harlan 1976; McCouch 2004;Rick1978; Tanksley and Fellowship (2015–2019). This work was supported by the USDA McCouch 1997). With this approach, more than 2000 biotic National Institute of Food and Agriculture, Multi-State Hatch Project stress resistance traits have been identified in crop wild rela- 1008470, Harnessing Chemical Ecology to Address Agricultural Pest tives; however, the vast majority of traits identified are for and Pollinator Priorities. disease resistance, and less than one quarter of these target Compliance with Ethical Standards insect pests (Dempewolf et al. 2017). Thus, strategies for breeding for resistance to insect pests must also be in- Conflict of Interest LB and AA declare that they have no conflict of clusive of secondary centers of diversity, and contempo- interest. MM is the co-founder of, but has no financial stake in, Row 7, a rary breeding pools. The context in which these breed- company that sells organic seed. ing pools were developed may also better reflect the context of agricultural plant-herbivore interactions than wild systems, where major secondary metabolites (like References cucurbitacins) may have different effects. In the diverse pools of cultivated germplasm with distinct breeding Agrawal AA, Janssen A, Bruin J, Posthumus MA, Sabelis MW (2002) histories, plant breeders may discover alternative, per- An ecological cost of plant defence: attractiveness of bitter cucum- ber plants to natural enemies of herbivores. Ecol Lett 5:377–385. haps quantitative, and likely smaller-impact resistance https://doi.org/10.1046/j.1461-0248.2002.00325.x traits. Screening material for the most promising, but Ali JG, Agrawal AA (2012) Specialist versus generalist insect herbivores less obvious traits will benefit by being informed by and plant defense. Trends Plant Sci 17:293–302. https://doi.org/10. chemical ecology, and incorporating these traits into 1016/j.tplants.2012.02.006 Andersen JF, Metcalf RL (1986) Identification of a volatile attractant for agroecosystems will require tools from associated disciplines Diabrotica and Acalymma spp. from blossoms of Cucurbita maxima (Brzozowski and Mazourek 2018). Duchesne. J Chem Ecol 12:687–699. https://doi.org/10.1007/ In conclusion, by using two independent domestication BF01012102 events to evaluate divergence in plant resistance traits in Andrews ES, Theis N, Adler LS (2007) Pollinator and herbivore attrac- tion to Cucurbita floral volatiles. J Chem Ecol 33:1682–1691. C. pepo, we identified differences in resistance that spanned https://doi.org/10.1007/s10886-007-9337-7 diverse leaf-chewing herbivores. We found that these Awmack CS, Leather SR (2002) Host plant quality and fecundity differences did not align to major known resistance in herbivorous insects. Annu Rev Entomol 47:817–844. traits, like cucurbitacins, or nitrogen, but instead found https://doi.org/10.1146/annurev.ento.47.091201.145300 Bates D, Maechler M, Bolker B, Walker S (2015) Fitting linear varietal-specific induced volatile deterrents in both sub- mixed-effects models using lme4. J Stat Softw 67(1):–48. species. This work contributes to the growing evidence https://doi.org/10.18637/jss.v067.i01 that plant defenses in the context of cultivated systems Bell RA, Owens CD, Shapiro M, Tardif JR (1981) Development may be distinct from those in the wild (Chen et al. of mass-rearing technology. In: Doane CC, McManus ML (eds) The gypsy moth: research toward integrated pest man- 2017), and has implications for plant breeders. By ex- agement. US Department of Agriculture, Washington, DC, pp ploring defensive novelty through contrasting domestica- 599–633 tions and breeding histories in C. pepo, we found sus- Bruce TJA, Pickett JA (2011) Perception of plant volatile blends by ceptibility in one lineage that persisted through the be- herbivorous insects - finding the right mix. Phytochemistry 72: – havior of diverse herbivores, and discovered a previous- 1605 1611. https://doi.org/10.1016/j.phytochem.2011.04.011 Brzozowski L, Mazourek M (2018) A sustainable agricultural future re- ly unknown defense trait in C. pepo that may provide a new lies on the transition to organic Agroecological Pest management. target for plant breeders. Sustainability 10:2023. https://doi.org/10.3390/su10062023 JChemEcol

Brzozowski L, Leckie BM, Gardner J, Hoffmann MP, Mazourek M Hoffmann MP, Robinson RW, Kyle MM, Kirkwyland JJ (1996) (2016) Cucurbita pepo subspecies delineates striped cucumber Defoliation and infestation of Cucurbita pepo genotypes by beetle (Acalymma vittatum) preference. Hortic Res 3:16028. Diabroticite beetles. HortScience 31:439–442 https://doi.org/10.1038/hortres.2016.28 Kappers IF, Hoogerbrugge H, Bouwmeester HJ, Dicke M (2011) Chen YH, Gols R, Benrey B (2015a) Crop domestication and its impact Variation in herbivory-induced volatiles among cucumber on naturally selected trophic interactions. Annu Rev Entomol 60: (Cucumis sativus L.) varieties has consequences for the at- 35–58. https://doi.org/10.1146/annurev-ento-010814-020601 traction of carnivorous natural enemies. J Chem Ecol 37: Chen YH, Gols R, Stratton CA, Brevik KA, Benrey B (2015b) 150–160. https://doi.org/10.1007/s10886-011-9906-7 Complex tritrophic interactions in response to crop domesti- Karban R, Baldwin IT (1997) Induced responses to herbivory. University cation: predictions from the wild. Entomol Exp Appl 157:40– of Chicago Press, Chicago 59. https://doi.org/10.1111/eea.12344 Kempel A, Schadler M, Chrobock T, Fischer M, van Kleunen M (2011) Chen YH, Shapiro LR, Benrey B, Cibrián-Jaramillo A (2017) Back to the Tradeoffs associated with constitutive and induced plant resis- origin: in situ studies are needed to understand selection during crop tance against herbivory. Proc Natl Acad Sci 108:5685–5689. diversification. Front Ecol Evol 5:1–8. https://doi.org/10.3389/fevo. https://doi.org/10.1073/pnas.1016508108 2017.00125 Ladizinsky G (2012) Plant evolution under domestication. Springer Da Costa CP, Jones CM (1971) Cucumber beetle resistance and Science & Business Media mite susceptibility controlled by the bitter gene in Cucumis Mattson WJ (1980) Herbivory in relation to plant nitrogen content. Annu sativus L. Science 172:1145–1146. https://doi.org/10.1126/ Rev Ecol Syst 11:119–161. https://doi.org/10.1146/annurev.es.11. science.172.3988.1145 110180.001003 Dávila-Flores AM, DeWitt TJ, Bernal JS (2013) Facilitated by nature and McCloud ES, Tallamy D, Halaweish FT (1995) Squash beetle trenching agriculture: performance of a specialist herbivore improves with host- behaviour: avoidance of cucurbitacin induction or mucilaginous plant life history evolution, domestication, and breeding. Oecologia plant sap? Ecol Entomol 20:51–59. https://doi.org/10.1111/j.1365- 173:1425–1437. https://doi.org/10.1007/s00442-013-2728-2 2311.1995.tb00428.x de Mendiburu F (2016) Agricolae: statistical procedures for agricultural McCouch S (2004) Diversifying selection in plant breeding. PLoS Biol 2: research. https://cran.r-project.org/package=agricolae. Accessed 1 e347. https://doi.org/10.1371/journal.pbio.0020347 Sept 2017 Metcalf RL (1985) Plant kairomones and insect pest control. Illinois Nat – Decker DS (1988) Origin(s), evolution, and systematics of Cucurbita Hist Surv Bull 33:175 198 pepo (Cucurbitaceae). Econ Bot 42:4–15. https://doi.org/10.1007/ Metcalf RL (1986) Coevolutionary adaptations of rootworm beetles BF02859022 (Coleoptera: Chrysomelidae) to cucurbitacins. J Chem Ecol 12: – Degen T, Dillmann C, Marion-Poll F, Turlings TCJ (2004) High genetic 1109 1124. https://doi.org/10.1007/BF01638999 variability of herbivore-induced volatile emissions within a Metcalf RL, Metcalf RA, Rhodes AM (1980) Cucurbitacins as broad range of maize inbred lines. Plant Physiol 135:1928– kairomones for diabroticite beetles. Proc Natl Acad Sci 77: – 1938. https://doi.org/10.1104/pp.104.039891 3769 3772. https://doi.org/10.1073/pnas.77.7.3769 Dempewolf H, Baute G, Anderson J, Kilian B, Smith C, Guarino L Metcalf RL, Rhodes AM, Metcalf RA, Ferguson J, Metcalf ER, Lu PY (2017) Past and future use of wild relatives in crop breeding. Crop (1982) Cucurbitacin contents and Diabroticite (Coleoptera: Sci 57:1070–1082. https://doi.org/10.2135/cropsci2016.10.0885 Chrysomelidae) feeding upon Cucurbita ssp. Environ Entomol 11: – Dussourd DE (1997) Plant exudates trigger leaf-trenching by cabbage 931 937. https://doi.org/10.1093/ee/11.4.931 loopers, Trichoplusia ni (Noctuidae). Oecologia 112:362–369. Meyer RS, DuVal AE, Jensen HR (2012) Patterns and processes in crop https://doi.org/10.1007/s004420050321 domestication: an historical review and quantitative analysis of 203 global food crops. New Phytol 196:29–48. https://doi.org/10.1111/j. Ferguson JE, Metcalf RL (1985) Cucurbitacins: plant-derived defense 1469-8137.2012.04253.x compounds for diabroticites (Coleoptera: Chrysomelidae). J Chem Ecol 11:311–318. https://doi.org/10.1007/BF01411417 Nee M (1990) The domestication of Cucurbita (Cucurbitaceae). Econ Bot 44:56–68. https://doi.org/10.1007/BF02860475 Ferguson JE, Metcalf ER, Metcalf RL (1983) Influence of cucurbitacin Paris HS (2000) History of the cultivar-groups of Cucurbita pepo. Hortic content in cotyledons of Cucurbitaceae cultivars upon feeding be- Rev:71–170 havior of Diabroticina beetles (Coleoptera: Chrysomelidae). J Econ Entomol 76:47–51. https://doi.org/10.1093/jee/76.1.47 Paris HS, Yonash N, Portnoy V, Mozes-Daube N, Tzuri G, Katzir N (2003) Assessment of genetic relationships in Cucurbita pepo Gaillard MDP, Glauser G, Robert CAM, Turlings TCJ (2018) (Cucurbitaceae) using DNA markers. Theor Appl Genet 106:971– Fine-tuning the ‘plant domestication-reduced defense’ hypoth- – 978. https://doi.org/10.1007/s00122-002-1157-0 esis: specialist vs generalist herbivores. New Phytol 217:355 ř 366. https://doi.org/10.1111/nph.14757 Paris HS, Lebeda A, K istkova E, Andres TC, Nee MH (2012) Parallel evolution under domestication and phenotypic differentiation of the Gong L, Paris HS, Nee MH, Stift G, Pachner M, Vollmann J, cultivated subspecies of Cucurbita pepo (Cucurbitaceae). Econ Bot Lelley T (2012) Genetic relationships and evolution in 66:71–90. https://doi.org/10.1007/s12231-012-9186-3 Cucurbita pepo (pumpkin, squash, gourd) as revealed by Peterson JK, Horvat RJ, Elsey KD (1994) Squash leaf glandular trichome simple sequence repeat polymorphisms. Theor Appl Genet volatiles: identification and influence on behavior of female 124:875–891. https://doi.org/10.1007/s00122-011-1752-z pickleworm moth [Diaphania nitidalis (Stoll.)] (Lepidoptera: Gouinguené S, Degen T, Turlings TCJ (2001) Variability in herbivore- Pyralidae). J Chem Ecol 20:2099–2109. https://doi.org/10.1007/ induced odour emissions among maize cultivars and their wild an- BF02066246 cestors (teosinte). Chemoecology 11:9–16. https://doi.org/10.1007/ R Core Team (2016) R: A language and environment for statistical com- PL00001832 puting. R Foundation for Statistical Computing, Vienna. https:// Halaweish FT, Tallamy DW, Santana E (1999) Cucurbitacins: a role in www.r-project.org/.Accessed2March2016 cucumber beetle steroid nutrition? J Chem Ecol 25:2373–2383. Rasmann S, Köllner TG, Degenhardt J, Hiltpold I, Toepfer S, Kuhlmann https://doi.org/10.1023/a:1020886210590 U, Gershenzon J, Turlings TCJ (2005) Recruitment of entomopatho- Harlan JR (1976) Genetic resources in wild relatives of crops. Crop Sci 16: genic nematodes by insect-damaged maize roots. Nature 434:732– 329. https://doi.org/10.2135/cropsci1976.0011183X001600030004x 737. https://doi.org/10.1038/nature03451 JChemEcol

Rick CM (1978) Potential improvement of tomatoes by controlled Tallamy DW, Gorski PM (1997) Long- and short-term effect of introgression of genes from wild species. In: proceedings of cucurbitacin consumption on Acalymma vittatum (Coleoptera: the conference on broadening the Genetic Base of. Crops: Chrysomelidae) fitness. Environ Entomol 26:672–677. 167–173 https://doi.org/10.1093/ee/26.3.672 Rodriguez-Saona C, Vorsa N, Singh AP, Johnson-Cicalese J, Szendrei Z, Tallamy D, Krischik VA(1989) Variation and function of cucurbitacins in Mescher MC, Frost CJ (2011) Tracing the history of plant Cucurbita: an examination of current hypotheses. Am Nat 133:766– traits under domestication in cranberries: potential conse- 786. https://doi.org/10.1086/284952 quences on anti-herbivore defences. J Exp Bot 62:2633– Tamiru A, Bruce TJA, Woodcock CM, Caulfield JC, Midega CAO, 2644. https://doi.org/10.1093/jxb/erq466 Ogol CKPO, Mayon P, Birkett MA, Pickett JA, Khan ZR Rosenthal JP, Dirzo R (1997) Effects of life history, domestication (2011) Maize landraces recruit egg and larval parasitoids in and agronomic selection on plant defence against insects: response to egg deposition by a herbivore. Ecol Lett 14: evidence from maizes and wild relatives. Evol Ecol 11:337– 1075–1083. https://doi.org/10.1111/j.1461-0248.2011.01674.x 355. https://doi.org/10.1023/A:1018420504439 Tanksley SD, McCouch SR (1997) Seed banks and molecular maps: Rowen E, Kaplan I (2016) Eco-evolutionary factors drive induced unlocking genetic potential from the wild. Science 277:1063– – plant volatiles: a meta-analysis. New Phytol 210:284 294. 1066. https://doi.org/10.1126/science.277.5329.1063 https://doi.org/10.1111/nph.13804 Theis N, Kesler K, Adler LS (2009) Leaf herbivory increases floral Sanjur OI, Piperno DR, Andres TC, Wessel-Beaver L (2002) fragrance in male but not female Cucurbita pepo subsp. Phylogenetic relationships among domesticated and wild species texana (Cucurbitaceae) flowers. Am J Bot 96:897–903. of Cucurbita (Cucurbitaceae) inferred from a mitochondrial gene: https://doi.org/10.3732/ajb.0800300 implications for crop plant evolution and areas of origin. Proc Natl Theis N, Barber NA, Gillespie SD, Hazzard RV, Adler LS (2014) Acad Sci U S A 99:535–540. https://doi.org/10.1073/pnas. Attracting mutualists and antagonists: plant trait variation explains 012577299 the distribution of specialist floral herbivores and pollinators on Schneider CA, Rasband WS, Eliceiri KW (2012) NIH image to ImageJ: crops and wild gourds. Am J Bot 101:1–9. https://doi.org/10.3732/ 25 years of image analysis. Nat Methods 9:671–675. https://doi.org/ ajb.1400171 10.1038/nmeth.2089 Shang Y, Ma Y, Zhou Y et al (2014) Biosynthesis, regulation and do- Tietz HM (1972) An index to the described life histories, early stages and mestication of bitterness in cucumber. Science 346:1084– hosts of the Macrolepidotera of the Continental United States and 1088. https://doi.org/10.1017/CBO9781107415324.004 Canada. AC Allyn I/II Shapiro L, De Moraes CM, Stephenson AG, Mescher MC (2012) Turcotte MM, Turley NE, Johnson MTJ (2014) The impact of domesti- Pathogen effects on vegetative and floral odours mediate vector cation on resistance to two generalist herbivores across 29 – attraction and host exposure in a complex pathosystem. Ecol Lett independent domestication events. New Phytol 204:671 681. 15:1430–1438. https://doi.org/10.1111/ele.12001 https://doi.org/10.1111/nph.12935 Shiojiri K, Ozawa R, Takabayashi J (2006) Plant volatiles, rather than Veyrat N, Robert CAM, Turlings TCJ, Erb M (2016) Herbivore light, determine the nocturnal behavior of a caterpillar. PLoS Biol 4: intoxication as a potential primary function of an inducible 1044–1047. https://doi.org/10.1371/journal.pbio.0040164 volatile plant signal. J Ecol 104:591–600. https://doi.org/10. Smith BD (2006) Eastern North America as an independent center of 1111/1365-2745.12526 plant domestication. Proc Natl Acad Sci 103:12223–12228. Whitehead SR, Turcotte MM, Poveda K (2017) Domestication impacts https://doi.org/10.1073/pnas.0604335103 on plant–herbivore interactions: a meta-analysis. Philos Trans R Soc Sutherland DWS, Greene GL (1984) Suppression and Management of B Biol Sci 372:20160034. https://doi.org/10.1098/rstb.2016.0034 Cabbage Looper Populations. USDA Tech Bull 1684:1–13 Zhou Y, Ma Y, Zeng J, Duan L, Xue X, Wang H, Lin T, Liu Z, Zeng K, Takabayashi J, Dicke M, Takahashi S, Posthumus MA, van Beek TA Zhong Y, Zhang S, Hu Q, Liu M, Zhang H, Reed J, Moses T, Liu X, (1994) Leaf age affects composition of herbivore-induced Huang P, Qing Z, Liu X, Tu P, Kuang H, Zhang Z, Osbourn A, Ro synomones and attraction of predatory mites. J Chem Ecol 20: DK, Shang Y, Huang S (2016) Convergence and divergence of 373–386. https://doi.org/10.1007/BF02064444 bitterness biosynthesis and regulation in Cucurbitaceae. Nat Plants Tallamy D (1985) Squash beetle feeding behavior: an adaptation 2:1–8. https://doi.org/10.1038/nplants.2016.183 against induced cucurbit defenses. Ecology 66:1574–1579. https://doi.org/10.2307/1938019