Species‐specific regulation of herbivory‐induced defoliation tolerance is associated with jasmonate inducibility

Summary Induced changes in root carbohydrate pools are commonly assumed to determine plant defoliation tolerance to herbivores. However, the regulation and species specificity of these two traits are not well understood. We determined herbivory‐induced changes in root carbohydrates and defoliation tolerance in seven different solanaceous plant species and correlated the induced changes in root carbohydrates and defoliation tolerance with jasmonate inducibility. Across species, we observed strong species‐specific variation for all measured traits. Closer inspection revealed that the different species fell into two distinct groups: Species with a strong induced jasmonic acid (JA) burst suffered from a reduction in root carbohydrate pools and reduced defoliation tolerance, while species with a weak induced JA burst maintained root carbohydrate pools and tolerated defoliation. Induced JA levels predicted carbohydrate and regrowth responses better than jasmonoyl‐L‐isoleucine (JA‐Ile) levels. Our study shows that induced JA signaling, root carbohydrate responses, and defoliation tolerance are closely linked, but highly species specific, even among closely related species. We propose that defoliation tolerance may evolve rapidly via changes in the plant's defense signaling network.

Two recent studies have investigated the link between herbivoryinduced changes in root soluble carbohydrate pools and defoliation tolerance.One study found that aboveground herbivory depletes root carbohydrate pools and constrains the regrowth capacity of Manduca sexta-attacked Nicotiana attenuata plants (Machado et al., 2013).Both effects were absent in jasmonate-deficient transgenic plants impaired in the production of plant resistance metabolites, indicating that jasmonate signaling regulates both resistance and tolerance traits and that they might be subject to trade-offs (Machado et al., 2013).In contrast, another study found that Solanum lycopersicum plants regrew better when attacked by M. sexta larvae, despite the herbivoreinduced reduction in root carbohydrates in this species (Gómez et al., 2012;Korpita, Gomez, & Orians, 2014).The contrasting results of these two studies suggest that the connection between herbivoryinduced changes in root carbohydrates and tolerance may differ even between species belonging to the same family.
In this study, we exploited natural variation in plant responses to herbivore attack across the Solanaceae to understand the regulation and species specificity of herbivory-induced tolerance in this plant family (Korpita et al., 2014;Machado et al., 2013;Xu, Zhou, Pottinger, & Baldwin, 2015;Zavala & Baldwin, 2006).To this end, we evaluated changes in leaf jasmonates, root carbohydrate pools, and regrowth capacity in response to Manduca sexta attack in seven solanaceous plant species.Manduca sexta is one of the most damaging specialist herbivores of this plant family and can completely defoliate plants (Campbell & Kessler, 2013;Kessler, Halitschke, & Baldwin, 2004;Yamamoto & Fraenkel, 1960).The obtained results provide insights into the regulation of induced defoliation tolerance and its variability across closely related plant species.

| Plant material
The following plant species were used in this study: Petunia axillaris axillaris, Solanum peruvianum LA2744, Solanum lycopersicum LA2696, Nicotiana miersii, Nicotiana pauciflora, Nicotiana attenuata, and Nicotiana obtusifolia.These species were selected to represent four major Solanaceae clades and cover both intra-and intergenus variation.Nicotiana attenuata seeds were originally collected in Utah (USA).Nicotiana miersii, N. pauciflora, and N. obtusifolia seeds were obtained from the United States Nicotiana germplasm collection.Seeds were multiplied through selfing in the glasshouse.Seeds of S. lycopersicum LA2696 and S. peruvianum LA2744 were initially obtained from the tomato genetics resource center (TGRC) at Davis University in California (USA) and propagated by bulk pollination.Petunia axillaris seeds were derived from a wild accession by self-fertilization.
Nicotiana attenuata seeds were smoke-treated to trigger germination.Approximately 9-10 days later, the seedlings were transferred to Teku pots (Pöppelmann GmbH & Co. KG, Lohne, Germany) filled with chunky sand (Raiffeisen GmbH, Germany) for 12 days before transferring them into 1-L pots filled with sand.Solanum lycopersicum LA2696 and S. peruvianum LA2744 seeds were germinated directly in Teku pots.Between 16 and 20 days later, the seedlings were transferred into 1-L pots filled with sand (Figure 1a).This germination procedure was carried out to synchronize the first day of flowering of all plant species.All plants were grown at 45-55% relative humidity and 23-25°C during days and 19-23°C during nights under 16 hr of light (6 a.m.-10 p.m.) supplied by Master Sun-T PIA Agro 400 or Master Sun-T PIA Plus 600 W Na lights (Philips, Turnhout, Belgium).Plants were watered twice every day by a flood irrigation system and were fertilized daily with 0.5 g/L Ferty B1 and 1.0 g/L Ca(NO 3 ) 2 × 4 H 2 O (Planta Düngemittel, Regenstauf, Germany).
We determined whether the M. sexta-induced changes in regrowth differ between different species that are grown under similar conditions.All plants were treated during the vegetative stage (Figure 1b).Simulated M. sexta attack and the regrowth capacity evaluation were carried out as described (Ferrieri et al., 2015;Machado et al., 2013) (Figure 1b,c).Briefly, a pattern wheel was rolled over the leaf 3-4 times on each side of the midvein and the resulting wounds (W) were immediately treated with 20-30 μl of a 1:5 (v/v) MilliQ water-diluted M. sexta oral secretion (OS) solution.Three to four leaves per plant were treated every time and the treatments were repeated every other day three times to obtain a total of 9-12 treated leaves per plant over 6 days of treatments (Figure 1c).The number of treated leaves and the amount of applied OS to the wounds were kept proportional to the size of the different plant species to standardize induction intensities across plant species and to achieve 30% of the leaves treated on average.Intact plants served as controls (N = 19).Simulated herbivory in N. attenuata changes regrowth capacity in a similar manner as real M. sexta attack (Machado et al., 2013).We used this approach to avoid a bias which may be caused by different feeding intensities of M. sexta on the different species (Boer & Hanson, 1984;Yamamoto & Fraenkel, 1960).All plant species evaluated in this study host M. sexta under natural conditions (Boer & Hanson, 1984;Campbell & Kessler, 2013;Machado, McClure, Hervé, Baldwin, & Erb, 2016;Yamamoto & Fraenkel, 1960).
To specifically evaluate the contribution of belowground tissues to the regrowth capacity of aboveground tissues, all plant species were defoliated 24 h after the last treatment, leaving only the roots and the lowest part of the main stem (0.5-10 cm above the shoot-root junction).Heavy defoliation is commonly observed under natural conditions (Machado, McClure, Hervé, Baldwin, & Erb, 2016).Regrowth was monitored for all species until senescence (Figure 1c) as follows: the number of regrowing leaves was counted, the average rosette diameter was measured, and the cumulative branch length (the sum of the length of all branches) and the number of flowers were quantified.
According to their change in regrowth following simulated M. sexta attack compared to control plants, plants were classified as tolerant or nontolerant.Tolerant species are those whose regrowth capacity was not affected by simulated M. sexta attack.A subset of plants was used for the quantification of root carbohydrates.

| Correlation among M. sexta-induced changes in JA levels, root carbohydrates, and regrowth capacity
To link the different M. sexta-induced phenotypes, we first calculated the magnitude of the M. sexta-induced changes in root carbohydrate pools as the ratio between the total amount of nonstructural root carbohydrates (sum of glucose, fructose, sucrose, and starch) in M. sexta-
Using the constructed phylogenetic tree (Figure S1), we estimated the relationship between herbivory-induced regrowth capacity, root

| The regrowth response upon simulated M. sexta attack is species specific
Across the seven tested solanaceous plant species, we observed neutral to negative regrowth patterns in plants under simulated M. sexta attack compared to nonelicited controls (Figure 2).While simulated M. sexta attack did not affect either the vegetative growth or reproductive output of P. axillaris, S. lycopersicum, N. miersii, and N. obtusifolia, it reduced the leaf number, branch length, and number of flowers in S. peruvianum, N. pauciflora, and N. attenuata (Figure 2).

| The root carbohydrate response upon simulated M. sexta attack is species specific
Similar to the measured regrowth responses, we found neutral to negative effects of simulated M. sexta herbivory on root carbohydrate pools (Figure 3).Total soluble, nonstructural carbohydrate pool-calculated as the total amount of glucose, fructose, sucrose, and starch-in P. axillaris, S. lycopersicum, N. miersii, and N. obtusifolia plants remained unaffected by leaf induction (Figure 3).By contrast, a significant reduction in total root carbohydrates was observed in S. peruvianum, N. pauciflora, and N. attenuata.One exception was N. obtusifolia, in which simulated M. sexta herbivory reduced root glucose and fructose levels, but not total nonstructural carbohydrates due to a slight, nonsignificant increase in starch levels.The only positive response was observed for starch levels in S. lycopersicum, which increased in the roots of leaf-induced plants.Total soluble carbohydrates were not changed in the roots of S. lycopersicum.

| Manduca sexta-induced changes in JA levels and root carbohydrates are correlated across species
Linear correlations revealed a positive relationship between M. sextainduced leaf JA levels and the magnitude of M. sexta-induced changes in root carbohydrate pools (p < .001)and flower production (p = .03)across species (Figure 4a,b).In contrast, M. sexta-induced leaf JA-Ile levels did not correlate with root carbohydrate levels (p = .166) or flower production (p = .676)(Figure 4c,d).Two-dimensional component analysis revealed a clear grouping effect that separated nontolerant from tolerant species according to the magnitude of induced JA, root carbohydrate depletion, and regrowth capacity (Figure 5).
Plant species that induced high levels of JA in response to simulated M. sexta attacked suffered from root carbohydrate depletion and suppression of regrowth, while species that induced low levels of JA did not suffer from carbohydrate depletion or a reduction in flower production (Figure 5).

| DISCUSSION
In this study, we show that herbivory-induced changes in leaf JA, root carbohydrates, and defoliation tolerance are strongly correlated, but highly variable between species.Our findings are consistent with the F I G U R E 3 Impact of M. sexta attack on root carbohydrate pools of seven solanaceous plant species.Average (±SE) glucose (a, f, k, p, u, z, ee), fructose (b, g, l, q, v, aa, ff), sucrose (c, h, m, r, w, bb, gg), starch (d, i, n, s, x, cc, hh) and total nonstructural carbohydrates (e, j, o, t, y, dd, ii We observed a strong grouping effect across M. sexta-induced phenotypes and plant species.The first group consists of plant species which, in response to simulated M. sexta attack, produce high levels of JA, deplete root carbohydrate reserves, and suffer from a reduced regrowth capacity.The second group includes plant species which produce low levels of JA, maintain root carbohydrate reserves, and maintain their regrowth capacity upon simulated M. sexta attack.From a mechanistic point of view, the strong correlation of these three traits across species may be due to several reasons.First, it is likely that the changes in root carbohydrates directly determine a plant's capacity to produce new photosynthetic tissues at the end of the defoliation process.Plants with lower levels of soluble, nonstructural carbohydrates in the roots often regrow smaller shoots (Bokhari, 1977;Lee, Donaghy, Sathish, & Roche, 2009;Machado et al., 2013;Smith & Silva, 1969).
Second, both traits may share a common regulatory basis.Variation in both regrowth and root carbohydrate depletion is tightly associated with jasmonate signaling in N. attenuata (Machado et al., 2013).In both transgenic and naturally occurring jasmonate-deficient N. attenuata lines, neither carbohydrate depletion nor a reduction in regrowth is observed, a result in stark contrast to those of jasmonate-competent plants (Machado et al., 2013).Together with the results presented here, these findings strongly suggest that jasmonates regulate defoliation tolerance by altering root carbohydrate pools and that this mechanism is conserved across different species of the Solanaceae.
Across plant species, we observed high variability in the profiled traits, even within the genus Nicotiana, which included four closely related Nicotiana species.The pattern found here is similar to the pattern of M. sexta-induced defense traits observed in Solanum (Haak, Ballenger, & Moyle, 2014).Regulating tolerance through a major stress hormone pathway may have enabled plants to rapidly change their defoliation tolerance in a changing environment as small genetic changes may have been sufficient to alter this phenotype.More detailed phylogenetic studies will be needed to gain a better picture of the evolution of induced defolation tolerance in the Solanaceae.
A phylogenetic study across 36 Asclepias species uncovered no clear correlation between resistance traits and regrowth ability (Agrawal & Fishbein, 2008), and evidence is emerging that plants may employ mixed strategies to survive herbivore attack (Núñez-Farfán, Fornoni, & Valverde, 2007).At first glance, our results may lead to the conclusion that a negative association between resistance and defoliation tolerance would be very likely for the Solanaceae, as jasmonates regulate resistance traits positively (Jimenez-Aleman, Machado, Baldwin, & Boland, 2017) and defoliation tolerance negatively (Jimenez-Aleman, Machado, Görls, Baldwin, & Boland, 2015;Machado et al., 2013).However, while resistance factors are in large part controlled by JA-Ile, we found strong negative correlations between JA, but not JA-Ile and defoliation tolerance.Could JA predominantly regulate tolerance traits, by, for example, regulating carbon allocation (Machado et al., 2015), while JA-Ile regulates resistance traits?Several studies suggest specific signaling roles of JA (Li et al., plants and the amount in nonattacked plants before defoliation.Second, we calculated the magnitude of the M. sexta-induced changes in flower production of the regrowing plants as a measure of induced defoliation tolerance by determining the ratio between the cumulative amount of flowers produced by induced regrowing plants and noninduced regrowing plants.Third, we classified the plant species into low (<0.7 μg/g FW) or high (>1.3μg/g FW) JA inducers.
Plot 12.0 (Systat Software Inc., San Jose, CA, USA) using analysis of variance (ANOVA).Levene's and Shapiro-Wilk tests were applied to determine error variance and normality.Holm-Sidak post hoc tests were used for multiple comparisons.The effect of simulated herbivory on all vegetative and reproductive parameters measured in regrowing plants was tested by two-way repeated-measures ANOVA with time and treatment as factors.The effect of simulated herbivory on soluble sugars, starch, and total nonstructural carbohydrates was evaluated by two-way ANOVA with plant species and treatment as factors.Datasets from experiments that did not fulfill the assumptions for ANOVA were natural log-, root square-, or rank-transformed before analysis.To account for the relatedness of the different species in the correlations, the phylogenetic relationship among the studied species was determined based on the internal transcribed spacer region of nuclear 5.8S ribosomal DNA (ITS1 and ITS2) and a chloroplastic-tRNA-Leu gene, intergene spacer, and the tRNA-Phe gene (trnLF).The DNA sequences were obtained from GenBank (www. carbohydrates, and jasmonates by phylogenetic generalized least squares (PGLS) models in R (v.3.1.1)using the package caper(Orme et al., 2012; R Development Core Team 2012).

F
I G U R E 4 Manduca sexta-induced changes in jasmonic acid (JA), but not jasmonoyl-L-isoleucine, are positively correlated with regrowth capacity and root carbohydrate pools.Correlation between M. sexta-induced leaf JA and the magnitude of the M. sexta-induced root carbohydrate depletion at the beginning of the regrowth phase (a) and the magnitude of M. sextainduced reduction of flower production by regrowing plants(b).Correlation between the M. sexta-induced leaf JA-Ile and the magnitude of the M. sexta-induced root carbohydrate depletion at the beginning of the regrowth phase (c) and the magnitude of M. sexta-induced reduction of flower production by regrowing plants (d).Correlations were tested using phylogenetic -induced leaf JA-Ile (μg/g FW) Manduca sexta-induced jasmonic acid (JA) levels, total nonstructural root carbohydrates, and regrowth capacity are strongly correlated across species.Grouping effect in the magnitude of M. sexta-induced root carbohydrate depletion at the beginning of the regrowth phase (x-axis) and the magnitude of M. sexta-induced reduction of flower production by regrowing plants (xy-axis) across plant species.Blue circles designate high (more than 1.3 μg/g FW) and gray circles low (<0.7 μg/g FW) JA inducers