Manduca sexta recognition and resistance among allopolyploid Nicotiana host plants

Abstract

Allopolyploid speciation occurs instantly when the genomes of different species combine to produce self-fertile offspring and has played a central role in the evolution of higher plants, but its consequences for adaptive responses are unknown. We compare herbivore-recognition and -resistance responses of the diploid species and putative ancestral parent Nicotiana attenuata with those of the two derived allopolyploid species Nicotiana clevelandii and Nicotiana bigelovii. Manduca sexta larvae attack all three species, and in N. attenuata attack is recognized when larval oral secretions are introduced to wounds during feeding, resulting in a jasmonate burst, a systemic amplification of trypsin inhibitor accumulation, and a release of volatile organic compounds, which function as a coordinated defense response that slows caterpillar growth and increases the probability of their being attacked. Most aspects of this recognition response are retained with modifications in one allotetraploid (N. bigelovii) but lost in the other (N. clevelandii). Differences between diploid and tetraploid species were apparent in delays (maximum 1 and 0.5 h, respectively) in the jasmonate burst, the elicitation of trypsin inhibitors and release of volatile organic compounds, and the constitutive levels of nicotine, trypsin inhibitors, diterpene glycosides, rutin, and caffeoylputrescine in the leaves. Resistance to M. sexta larvae attack was most strongly associated with diterpene glycosides, which were higher in the diploid than in the two allotetraploid species. Because M. sexta elicitors differentially regulate a large proportion of the N. attenuata transcriptome, we propose that these species are suited for the study of the evolution of adaptive responses requiring trans-activation mechanisms.

Polyploid speciation plays a central role in the evolution of plants; as much as 70% of all angiosperm species are thought to have had polyploidization in their lineages (1). Allopolyploid speciation occurs instantly when the genomes of different species combine to produce self-fertile offspring and is a common speciation mechanism in particular taxa. Polyploid speciations are frequently associated with adaptive radiations, with the polyploid taxa exhibiting greater ability to survive under unfavorable conditions, perhaps because of increased heterozygosity (2). However the rapid changes in genomic architecture that result from the combination of new genomes are likely to wreak havoc with adaptations that rely on the trans-activation of many genes.

Evidence is rapidly accumulating that plants “recognize” attack from particular herbivore species and tailor their induced responses accordingly, and that a large fraction of a plant's transcriptome is involved (3–11). However, little to nothing is known about how these complex responses are preserved or altered during polyploid speciation events. Some polyploids are more resistant to herbivore and pathogen attack than are closely related diploids (12), but in autopolyploid complexes of Heuchera grossulariifolia (Saxifragaceae) no clear associations of resistance with ploidy levels have been found (13–15). Consequently, plant polyploidy has not yet been integrated into our understanding of the evolution of insect–plant chemical interactions, although for some plant–herbivore systems, the mechanisms of herbivore recognition and the adaptive tailoring of defense responses are understood in sufficient detail to compare responses across polyploid taxa.

In the genus Nicotiana, allopolyploidy has played an important role in speciation (16). Nicotiana attenuata (Na), a North American species that has been involved in two allopolyploid speciation events in the formation of two other North American species, Nicotiana bigelovii (Nb) and Nicotiana clevelandii (Nc), is also one of the best-studied species with regard to herbivore-recognition mechanisms (17). Cytologically, Na is a 12-paired species (n = 12) and is thought to be the common ancestor during the amphidiploid speciation of Nb and Nc (both 24-paired species; Fig. 1), which Goodspeed (16) deduced from their close similarity in habit, leaf, inflorescence, and trichome morphology, and approximation of “Drosera scheme” pairing in F₁ Nb × Na and Nc × Na hybrids. The other parental line involved in the amphiploid origin of Nb and Nc is an early 12-paired alatoid race, which became the progenitor of the section alatae and is thought to be extinct (16). Goodspeed's phylogenetic hypothesis has recently been tested with molecular techniques and found to be consistent with the available data (18).

Fig. 1.

(A) The two tetraploid (n = 24) species Nb and Nc are thought to have arisen from an allopolyploid speciation of an ancestral diploid (n = 12) Na that hybridized with an extinct diploid alatoid species as proposed by Goodspeed (16) and confirmed by Chase et al. (18). (B) Attack by Manduca sexta larvae is recognized by Na when fatty acid–amino acid conjugates (FACs) in larval oral secretions and regurgitants (OS) are introduced into plant wounds (W) during feeding, resulting in signal crosstalk between jasmonate (JA)- and ethylene (ET)-mediated pathways and the elicitation of direct [nicotine, protease inhibitors (PI), and diterpene glycosides (DTGs)] and indirect [volatile organic compounds (VOCs)] defenses. MJ, methyl jasmonate; PMT, putrescine N-methyltransferase.

Plant–herbivore interactions have been intensively investigated with ecological, chemical, and molecular approaches in the Na–Manduca sexta system (17). From this work, it is clear that Na recognizes feeding by the larvae of its specialist sphingid herbivore, M. sexta, as evidenced by Manduca-induced specific patterns of hormone signaling (JA, ethylene), secondary metabolite accumulation (responsible for both direct and indirect defenses), and gene transcript accumulation. These herbivore-induced responses are different from those induced by mechanical damage or exogenous applications of methyl jasmonate (MeJA), and are elicited by FACs in the OS of the larvae (8, 11, 19) (Fig. 1).

We compare these herbivore recognition and resistance responses observed in the diploid species Na with those observed in the two allopolyploid species Nc and Nb. While Na is found throughout the Great Basin Desert after fires in pinyon–juniper–sage habitats and north along the Sierras into California and Oregon, Nb is found in sandy washes along the California coast, and Nc grows in drier habitats throughout Baja California and southern California (16). Two decades of fieldwork has established that M. sexta larvae are one of the three most abundant and damaging herbivores found on Na. Much less has been published on the herbivore communities of the two tetraploids; however, M. sexta larvae have been found on both species in nature (I.T.B., unpublished observations). Given that the JA burst in response to M. sexta OS applications to mechanical wounds is the signature of herbivore recognition that likely organizes much of the tailoring of the defense responses, we determine whether the three species differ in the timing and magnitude of the JA burst. Because the JA burst occurs locally and is not strongly influenced by leaf ontogeny (20), we measure it in a single, standardized leaf node in all three species. In addition to wounding, OS applications, and herbivore attack, JA applications are also known to elicit the accumulation of herbivore-induced defense metabolites in Na, such as nicotine (9), protease inhibitors (21, 22), VOCs (19), caffeoylputrescine, chlorogenic acid, and DTGs (23). These defense responses exhibit a mixture of systemic and local responses, and we measured them in three phyllotactically adjacent leaves after elicitation, to characterize both systemic and localized elicitation in all three species. Because the indirect defense (VOC release) is a whole-plant response, we measure whole-plant releases. Exogenous application of the methyl ester of JA, MeJA, provides a convenient and reproducible elicitor of insect resistance, and we examine MeJA-elicited resistance to M. sexta attack.

Materials and Methods

Plant Growth. Na seeds originated from a population in Utah (24); Nc and Nb v. bigelovii seeds were kindly supplied by Verne A. Sisson (Oxford Tobacco Research Station, Oxford, NC) and originated from collections made by H. Goodspeed (16). Seeds were sterilized and germinated on agar (for smoke treatment of Na seeds see ref. 25) and after 10 days of growth, planted into soil in Teku pots (Waalwijk, The Netherlands) and once established, transferred to 1-liter pots in soil and grown in the glasshouse at 26–28°C, under 16 h of light supplemented by Philips Sun-T Agro 400- or 600-W sodium lights. Plants in the same stage of rosette growth, 2 weeks after their transfer to 1-liter pots were used in all experiments.

Plant Treatment. Plants were treated with 150 μg of MeJA in 20 μl of lanolin paste (19) applied to two leaves: at nodes 0 (source–sink transition leaf) and one node older (node 1). Controls (lanolin) were similarly treated with 20 μl of pure lanolin. For M. sexta OS-treated plants, leaves at the same two nodal positions per plant were damaged by rolling a fabric pattern wheel over the leaf surface to create a standardized mechanical wound, and 20 μl of OS [diluted 1:10 (vol/vol) with water] from fourth–fifth instar larvae was added to the puncture wounds on each leaf. Controls (water) were wounded and treated with 20 μl of deionized water. Unmanipulated plants (controls) were included in each experiment.

Comparison of Induced Secondary Metabolites. JA burst. Thirty-six plants of each species were selected and randomly assigned to two treatment groups: OS (20 μl of OS) or water (20 μl of deionized H₂O) were added to the lamina of node 1 leaves immediately after three rows of puncture wounds were created with a fabric pattern wheel. The treated leaves were harvested at 0, 0.5, 1.0, 1.5, 3, and 8 h after wounding and treatment, and JA was extracted (three plants per treatment per harvest) for analysis by GC–MS with a doubly labeled internal standard ([1,2-¹³C]JA) as described in ref. 26.

Trypsin Inhibitor (TrypPI). Plants of each species were randomly assigned to the five treatments (three to five plants per treatment and harvest). Leaves at nodes -1, 1, and 2 were harvested (at 1300 hours) at 1, 2, 4, 6, and 8 days after treatment for the lanolin and MeJA treatments. For OS, water, and control treatments, plants were harvested 4 days after treatment. TrypPI concentration was measured as described in ref. 21 and expressed as nmol per mg of protein.

Nonvolatile secondary metabolites. Plants of each species were randomly assigned to MeJA, lanolin, OS, water, and control treatments, each with three to six replicates, and harvested as described for TrypPI measures. Leaf extracts were prepared for analysis of nicotine, rutin, caffeoylputrescine, chlorogenic acid, and DTGs by HPLC as described in ref. 23.

Volatiles. Sixteen plants of each species were randomly assigned to four treatment groups (four replicates each): MeJA, lanolin, OS, and water. Plants were individually placed in 50-liter glass chambers, and the VOCs released were trapped on super Q (Alltech Associates) traps for 6 h, 24 h after elicitation (the time of maximum release after a single elicitation) and measured by GC–MS as described in ref. 19. Germacrene A was confirmed by mass spectra, a diagnostic thermal Cope-rearrangement to β-elemene (27) and retention time of an authentic standard from a liverwort (Frullania macrocephalum) extract (kindly supplied by Jan-Willem de Kraker, Wageningen University, Wageningen, The Netherlands). VOCs were expressed as percentages of peak areas relative to the internal standard, tetralin, per 6 h of trapping per plant.

Herbivory Experiment. Forty plants were randomly assigned to two treatment groups: MeJA and lanolin. Freshly hatched M. sexta L. (Lepidoptera: Sphingidae) larvae (eggs from North Carolina State University Insectary, Raleigh) were placed individually on the node -1 leaf of each plant that had been treated with MeJA or pure lanolin 4 days earlier. Larval mass was measured (to 0.1 mg) on the second, fourth, and sixth day after the start of the experiment.

Statistical Analysis. TrypPI activity and nicotine and rutin data were log transformed, DTG and caffeoylputrescine values were square root transformed, and chlorogenic acid values were inversion transformed before analysis to meet requirements of normality. Differences in JA burst, OS-induced TrypPI, and VOCs were determined by t tests. All other data were analyzed by multivariate ANOVA (MANOVA). If the MANOVA analysis was significant (P < 0.05), univariate ANOVAs for the individual effects and Fisher least significant difference posthoc tests to detect significant differences between groups were conducted. Data were analyzed with STATVIEW (SAS Institute, Cary, NC), and the results are available in Statistical Analysis, which is published as supporting information on the PNAS web site, www.pnas.org.

Results

OS-Elicited JA Burst. Application of M. sexta OS to puncture wounds in Na leaves resulted in a transient JA burst (attaining values that were 30% higher than those of wounded and water-treated leaves), which reached maximum values at 30 min and waned to control levels by 90 min (Fig. 2), as previously described (8, 28). Surprisingly, the JA burst was significantly attenuated in Nc, in which maximum values measured (at 1 h) were half of those observed in Na and Nb. The JA burst observed in Nb, although comparable in magnitude to that of Na, was delayed even longer (by 1 h) than that observed in Nc (delay of 0.5 h in comparison to that of Na). Moreover, the JA burst in Nb, in comparison to those of Na or Nc, did not wane rapidly, and significant differences were still found at the 8-h harvest (Fig. 2).

Fig. 2.

JA concentrations (+1 SE) in Na, Nb, and Nc leaves that were wounded with a fabric pattern wheel and the resulting puncture wounds were immediately treated with 20 μl of either M. sexta OS (•) or deionized water (W, ○) at time 0. Asterisks indicate level of significant differences between members of a pair (*, P < 0.05; **, P < 0.01).

TrypPI. Whereas constitutive levels of TrypPI in untreated control plants (C) did not differ among the three species, the elicited levels did, and averaged across all elicitation treatments, Nb had the highest TrypPI concentrations, followed by Nc, and the lowest in Na. (Fig. 3). OS significantly amplified wound-induced increases of TrypPI concentrations in the treated leaf in both Na (3.4-fold) and Nb (2.8-fold), but not in Nc (1.3-fold) (Fig. 3). In contrast, treatment with MeJA resulted in long-lasting (>8 days; Fig. 8, which is published as supporting information on the PNAS web site) significant local and systemic increases in TrypPI activity in all three species in which the absolute increase in the treated leaf (node 1) was always greater than the systemic response and the systemic response in older sink leaves (node 2) was greater than those in younger source leaves (node -1) (Fig. 3).

Fig. 3.

Mean (+1 SE) TrypPI concentrations of Na, Nb, and Nc leaves growing at nodes -1, 1, and 2, 4 days after leaves at node 0 and 1 were treated with 20 μl of lanolin containing 150 μg of MeJA (MJ) or with 20 μl of pure lanolin (LC), or were wounded and treated with 40 μl of M. sexta OS or 40 μl of deionized water (W) or left unwounded and untreated (C). Asterisks indicate level of significant differences between members of a treatment and control pair (LC vs. MJ and OS vs. W: *, P < 0.05; **, P < 0.01).

Nicotine. MeJA treatment elicited significant long-lasting increases in nicotine levels in all three species, with the largest increase observed in leaves at node -1 in Na and Nb, but in Nc leaves, no consistent ranking was found (Figs. 9 and 10, which are published as supporting information on the PNAS web site). Quantities measured in the three species followed a pattern similar to that observed for TrypPIs: Nb had the highest (5.7- and 2.5-fold higher than those in Na and Nc, respectively) and Nc had intermediate values (2.3-fold higher than those of Na) as averaged across the five harvests from all treatments. OS treatment of puncture wounds did not elicit increases that were significantly higher than those elicited by wounding and water treatments in all three species (Fig. 9), despite the higher elicitation of JA by OS treatment (Fig. 2). As previously described (9, 29), the suppression of wound-induced nicotine increases reflects an OS-elicited ethylene burst that suppresses transcripts of the rate-limiting enzyme in nicotine biosynthesis (putrescine N-methyltransferase; Fig. 1).

DTGs. DTG levels were highest in Na, intermediate in Nb, and not detectable in Nc (Fig. 4). DTGs levels decreased with increasing leaf age in Na but not in Nb. MeJA elicitation significantly increased DTG levels in both Na and Nb both locally and systemically, but the strongest elicitation was observed in the treated leaf (Fig. 4). Wounding and treatment with OS or water did not elicit DTG increases in either Na or Nb.

Fig. 4.

Mean (+1 SE) DTG peak areas of Na and Nb leaves growing at nodes -1, 1, and 2, 4 days after leaves at node 0 and 1 were treated with either 20 μl of lanolin containing 150 μg of MeJA (MJ) or 20 μl of pure lanolin (LC), wounded and treated with 40 μl of M. sexta OS or 40 μl of deionized water (W), or left unwounded and untreated (C). DTGs were not detected in extracts of Nc leaves. Asterisks indicate level of significant differences between members of a treatment and control pair (LC vs. MJ and OS vs. W: *, P < 0.05; **, P < 0.01).

VOCs. Six compounds dominated the headspace of MeJA- or OS-treated plants: cis-β-ocimene, trans-β-ocimene, linalool, α-bergamotene, germacrene A, and cis-jasmone. Qualitative and quantitative differences were found among the three species. Four compounds (not trans-β-ocimene and linalool) were found in the headspace of Na, five (not germacrene A) in Nb, and four (not α-bergamotene and germacrene A) in Nc (Fig. 5). In general, the VOC profile composition was more similar between the two tetraploid species than between the tetraploid and diploid species. MeJA elicitation dramatically increased the release of α-bergamotene and cis-jasmone but not germacrene A and cis-β-ocimene in Na; cis-β-ocimene, trans-β-ocimene, linalool, cis-jasmone (which was not detectable before elicitation), and α-bergamotene in Nb; and cis-β-ocimene (but not trans-β-ocimene), linalool, and cis-jasmone in Nc. OS treatments also increased the amounts of VOCs released in both Na and Nb, but surprisingly, not in Nc. In Na and Nb, OS treatment increased the release of three, albeit different, VOCs: α-bergamotene, germacrene A, and cis-jasmone; and cis-β-ocimene, trans-β-ocimene, and linalool, respectively (Fig. 5).

Fig. 5.

Mean (+1 SE) peak areas of VOCs in the headspace of Na, Nb, and Nc plants sampled for 6 h, starting 24 h after treatment with 20 μl of lanolin containing 150 μg of MeJA (MJ), 20 μl of pure lanolin (LC), or wounding and treatment with 40 μl of M. sexta OS or 40 μl of water (W) to leaves at nodes 0 and 1. Numbers identify compounds: 1, cis-β-ocimene; 2, trans-β-ocimene; 3, linalool; 4, cis-α-bergamotene; 5, germacrene A; and 6, cis-jasmone. Asterisks indicate level of significant differences between members of a treatment and control pair (LC vs. MJ and OS vs. W: *, P < 0.05; **, P < 0.01).

Phenolics. MeJA treatment (but not OS treatment) significantly increased caffeoylputrescine in all species, but the increases (25.6-fold) were larger in Nc than those in Na (10.1-fold) and Nb (2.5-fold). Elicited changes in chlorogenic acid and rutin were complex, with both increases and decreases observed after the different treatments (Table 1, which is published as supporting information on the PNAS web site). Chlorogenic acid concentrations were significantly reduced by OS and W treatments in Na and Nb, but in Nc, only W treatment resulted in significant decreases. MeJA treatment significantly increased levels in older leaves at node 2 in Na and Nc but decreased levels in Nb in all leaves. MeJA treatment decreased rutin levels in all species but in different leaves (at nodes -1 in Na and Nb, but 1 in Nc). OS treatment decreased rutin levels only in the treated leaf of Nc.

Herbivory Experiment. Weight gain was greatest when larvae fed on Nc, followed by those fed Nb, and the least on Na (Fig. 6). By day 6, the masses of caterpillars fed on Na and Nb were only 23.62% and 48.75% of those feeding on Nc, respectively. Interestingly, differences in larval mass between caterpillars fed on MeJA- and lanolin-treated plants were significantly reduced only in Nb-fed larvae at days 4 and 6, although the masses of caterpillars fed both MeJA-treated Na and Nc plants were consistently lower than those fed on lanolin-treated plants from the respective species (Fig. 6).

Fig. 6.

Mean (+1 SE) mass of 12–20 replicate M. sexta larvae fed on individual Na, Nb, and Nc plants, 2, 4, and 6 days after plants were treated with 20 μlof lanolin containing 150 μg of MeJA (MJ) or 20 μl of pure lanolin (LC) to leaves at nodes 0 and 1. Asterisks indicate significant differences between members of a pair (*, P < 0.05).

Discussion

Adaptive phenotypic responses increases the “fit” between organisms and their environment by altering the expression of a large number of genes in response to environmental signals. Different stress factors, such as insect herbivores, pathogens, and abiotic factors, elicit physiological, biochemical, and morphological changes in plants, likely as a result of crosstalk among a large number of signal transduction pathways, which include JA, ethylene, abscisic acid, and salicylates (30). Specific combinations of signals are thought to provide “signature” sets thought to activate an appropriate response to a specific stress. However, little is known whether crosstalk among signaling pathways results in adaptive responses, but work on plant–herbivore interactions provides some of the best examples to date.

When herbivores attack plants, they cause wounding, but a plant's response to herbivore attack in many cases cannot be mimicked by mechanical wounding or simple JA applications, which are thought to mediate many wound responses (30–32). Several different types of elicitors in the OS of herbivorous insects have been reported to alter a plant's wound response, including enzymatic elicitors such as β-glucosidase (33) and glucose oxidase (34) as well as FACs (8, 35). Manduca larvae contain at least 8 FACs in their OS that are necessary and sufficient for the JA burst and VOC release (8) as well as the amplification of TrypPIs (A. Roda, A. Steppuhn, and I.T.B., unpublished results) observed in Manduca-attacked Na plants. Moreover, the two most abundant FACs in M. sexta OS are responsible for 64% of the up-regulated (of 67) and 49% of the down-regulated (of 78) genes that are differentially regulated when M. sexta OS is added to plant wounds (11). Manduca OS is also known to elicit an ethylene burst, which reduces wound-induced nicotine accumulation by down-regulating transcripts of putrescine N-methyltransferase, which catalyzes the key regulatory step in nicotine biosynthesis (9, 29).

Although much remains unknown about the functional significance of these complex alterations, evidence is accumulating that the herbivore-specific increases in TrypPIs and VOCs, as well as the down-regulation of the wound-induced nicotine production, represent adaptive tailoring of the plant's defense response against Manduca attack. The VOC release functions as a potent indirect defense in nature, by attracting the generalist predator Geocoris pallens to feeding larvae (36). This voracious predator is size-selective, preferentially attacking eggs and larvae in the first three instars (37), and the up-regulation of TrypPIs by Manduca attack slows the growth of larvae (22), keeping them in stages that are more vulnerable to the predator. In addition, nicotine, which is sequestered by M. sexta larvae through dietary intake, negatively affects the performance of the parasitoids of M. sexta (38) and hence is coopted for the defense of the herbivore. Hence when Na is attacked by this nicotine-tolerant herbivore, it will likely realize a fitness benefit by the coordinated up-regulation of the predator-attracting VOC release and the amplification of wound-induced TrypPI production, while suppressing wound-induced nicotine production. A recently discovered natural mutant of Na that lacks the ability to produce TrypPIs is also deficient in herbivore-induced VOC release (22). Moreover, nicotine production is costly, requiring 8% of whole-plant nitrogen, an investment that cannot be recouped by metabolism, and is associated with diminished intraspecific competitive abilities for soil nitrogen (17). Hence, when Na is attacked by this nicotine-tolerant herbivore, it will likely realize a fitness benefit from suppressing its induced nicotine production when growing in competition with conspecifics, as it commonly does as a result of its germination behavior that synchronizes growth with the postfire environment (17).

To compare the ability of the diploid species, Na, to recognize attack from M. sexta larvae with that of the two allopolyploid species, Nb and Nc, we examined the timing of the JA burst and the subsequently elicited changes in secondary metabolites when M. sexta OS was applied to mechanically generated wounds on leaves. From these experiments, it was clear that whereas a statistically significant JA burst occurred in all three species, the Nc's JA increase was only half that observed in Na and Nb (Fig. 2). Nc's attenuated JA burst was also associated with a lack of OS-elicited VOC release (Fig. 5), TrypPI increase (Fig. 3), and reduction in chlorogenic acid contents (Table 1), responses that were clearly preserved in Nb. The lack of OS-elicited responses in Nc was not due to an inability of this species to respond because MeJA elicitation resulted in VOC, TrypPI, and chlorogenic acid increases. These results demonstrate that Na's herbivore recognition mechanism has partly been conserved during allopolyploid speciation in Nb, but lost in Nc, a result consistent with Goodspeed's phylogenetic hypothesis that Nb is more closely related to Na than Nc is (16).

Not all OS-elicited responses appear to be lost in Nc, however. The down-regulation of wound-induced nicotine production by OS appears to be retained in all species. The observation that M. sexta OS elicitation did not result in higher nicotine levels compared with water treatments of mechanical wounds (Fig. 9) despite a higher JA pool induced (Fig. 2), as has been observed in another diploid Nicotiana species [Nicotiana sylvestris (39)] suggests that OS-elicited ethylene signaling remains intact in all three species (9). Similarly, some JA-elicited responses are not elicited by OS treatment in all three species. DTGs (Fig. 4), caffeoylputrescine, and rutin (Table 1) were elicited by MeJA treatment but not by OS treatment. For these metabolites, JA elicitation clearly recruits a signal cascade that is not activated by OS treatment.

In addition to the differences in response to OS, the tetraploids also differed in the amount and timing of certain elicited metabolites. The rapid activation and waning of the JA burst observed in Na was both delayed and lasted significantly longer in Nb. As such, Nb's wound-induced JA dynamics are more similar to those observed in Arabidopsis, tomato, and potato (26). In nature, Na grows in close association with sagebrush, Artemisia tridentata Nuttall subsp. tridentata (Asteraceae), which releases MeJA in high, allelopathically active quantities (40, 41), potentially sufficient to influence the defense responses in the neighboring Na plants (42). In contrast, neither Nb nor Nc is commonly found growing in close association with sagebrush, and the rapid endogenous JA dynamics observed in Na may allow it to distinguish endogenous from exogenously derived JA.

The three species also differed in constitutive levels of secondary metabolites in a manner consistent with a gene-dose effect. For example, both tetraploid species had significantly higher nicotine (8.31- and 2.72-fold higher than Na: Fig. 9) and TrypPI (9.18- and 2.63-fold higher than those in Na: Fig. 3) levels, and lower DTGs (Fig. 4), rutin, and caffeoylputrescine (Table 1) levels compared with the level observed in Na. Comparing the two tetraploid species, all of these metabolites, with the exception of caffeoylputrescine, were significantly higher in Nb than in Nc. Differences in ecological habit between the tetraploid and diploid species may select for high nicotine and TrypPI levels. Of the three species, only Na “chases” fires in ecological time by mass-germinating from long-lived seed banks with smoke-related germination cues. By timing its growth with the ephemeral but high-resource postfire environment, Na is commonly exposed to very strong intraspecific competition, which in turn likely selects for rapid growth (17). Because both constitutive and inducible protease inhibitor production (22) and nicotine production (24) are associated with growth and fitness reductions, it is possible that selecting habitats that place a premium on fast growth also selects for low constitutive defense levels. Neither Nb nor Nc times their germination with the postfire environment and therefore they may not be under similarly strong selection for rapid growth and competitive ability.

MeJA-elicitation altered the secondary metabolite profiles in all three species, and by comparing these changes with MeJA-elicited changes in M. sexta larval performance, we could infer their relative influence on larval performance. M. sexta larval growth was highest in Nc, followed by in Nb, and the lowest in Na (Fig. 6). Significant difference in larval mass between caterpillars fed on MeJA- and lanolin-treated plants was found only in Nb, although caterpillars fed on MeJA-treated plants tended to have lower masses than those fed on lanolin-treated plants in Na and Nc. We standardized the MeJA-elicited metabolite changes to the uninduced levels found in Na, which was arbitrarily assigned a value of 1 (Fig. 7) to identify metabolites that correlated with the observed changes in M. sexta larval growth (Fig. 6). This analysis suggested that DTGs, which were highest in Na, intermediate in Nb, and not detectable in Nc, were most strongly associated with larval performance (Fig. 7). However, larval mass did not significantly differ between MeJA- and lanolin-treated Na plants, whereas DTGs did differ. Complex nonlinear interactions between other induced metabolites (nicotine and TrypPIs) and DTGs may influence the relationship with larval mass gain and DTG content. DTGs have been reported to inhibit the larval growth of tobacco budworm larvae, Heliothis virescens (43); however, conclusive evidence that DTGs are directly responsible for the observed effects will require direct manipulations of DTG production in plants. Interestingly, although nicotine (44) and TrypPI (22) have been reported to inhibit larval growth in Na and N. sylvestris, this analysis suggests that neither is as strongly correlated as DTGs are with M. sexta performance.

Fig. 7.

Relative values of nicotine, TrypPI, DTGs, caffeoylputrescine (CP), chlorogenic acid (CA), and rutin in MeJA-treated (solid bars) and lanolin-treated (open bars) Na, Nb, and Nc plants compared with the corresponding chemical concentrations in lanolin-treated Na, which were arbitrarily assigned a value of 1.0.

Although the ecological significance of the loss and retention of herbivore recognition abilities in the two allopolyploid species remains unknown, this species complex may be an ideal system in which to study the retention and modification of transactivated adaptive responses. The changes in secondary metabolites elicited by Manduca OS in Na are accompanied by a large-scale transcriptional change (6, 8, 10, 11), the majority of which can be elicited by only two FACs in Manduca OS (11). Given that these differentially regulated genes are likely dispersed throughout Na's 12 pairs of chromosomes, a small number of FAC-regulated trans-active elements are likely responsible for herbivore recognition in this species. How these putative trans-active elements have retained their ability to respond to OS in Nb, while recruiting different cis-elements [as is suggested by the different spectrum of VOCs released by OS elicitation (Fig. 5)], but losing the ability in Nc, will likely provide important insights into the maintenance or modification of polygenic adaptive traits during allopolyploid speciation. Moreover, given that it is possible to hybridize the different North American Nicotiana species to create artificial tetraploid lines (16), these evolutionary hypotheses are eminently falsifiable.