Plant Science 307 (2021) 110890 Contents lists available at ScienceDirect Plant Science journal homepage: www.elsevier.com/locate/plantsci Historic exposure to herbivores, not constitutive traits, explains plant tolerance to herbivory in the case of two Medicago species (Fabaceae) Miquel Capó a, *, Margalida Roig-Oliver a, Carles Cardona b, Joana Cursach a, Jordi Bartolomé c, Juan Rita a, Elena Baraza a a Research Group on Plant Biology under Mediterranean Conditions, Department of Biology, Faculty of Sciences, University of Balearic Islands, Palma, 07122, Spain b Centre Forestal de les Illes Balears, Institut Balear de la Natura, Gremi Corredors, 10, Pol. Son Rossinyol, Palma, 07009, Spain c Small Ruminant Research Group, Department of Animal and Food Science, Universitat Autònoma de Barcelona, Bellaterra, 08193, Spain A R T I C L E I N F O A B S T R A C T Keywords: Mechanisms that allow plants to survive and reproduce after herbivory are considered to play a key role in plant Introduced herbivores evolution. In this study, we evaluated how tolerance varies in species with different historic exposure to her­ Medicago sp. bivores considering ontogeny. We exposed the range-restricted species Medicago citrina and its closely related and Mediterranean island widespread species M. arborea to one and two herbivory simulations (80 % aerial biomass loss). Physiological Range-restricted species and growth parameters related to tolerance capacity were assessed to evaluate constitutive values (without Tolerance herbivory) and induced tolerance after damage. Constitutive traits were not always related to greater tolerance, and each species compensated for herbivory through different traits. Herbivory damage only led to mortality in M. citrina; adults exhibited root biomass loss and increased oxidative stress after damage, but also compensated aerial biomass. Despite seedlings showed a lower death percentage than adults after herbivory in M. citrina, they showed less capacity to recover control values than adults. Moderate tolerance to M. arborea herbivory and low tolerance to M. citrina is found. Thus, although the constitutive characteristics are maintained in the lineage, the tolerance of plants decreases in M. citrina. That represents how plants respond to the lack of pressure from herbivores in their habitat. 1. Introduction chlorophyll fluorescence [10] and a high electron transport rate [11]; these features indicate higher herbivory tolerance. In contrast, high The interaction between plants and herbivores is considered to be a constitutive non-photochemical quenching values are related to a low driver of natural selection [1]. Evolution of the plant defence traits tolerance to herbivory in juvenile plants of endemic species from Hawaii (namely, resistance and tolerance) has enabled plant species to adapt to [11]. Moreover, high oxidative basal stress levels are also associated the presence of herbivores and promoted diversification [1–3]. Plant with low tolerance to herbivory [12]. protection conferred by chemical compounds and physical defence or Moreover, plants have developed mechanisms to compensate for the tolerance strategies are the main mechanisms that support the interac­ loss of biomass by herbivory. As reviewed by Tiffin [13], the responses tion between plants and herbivores [4]. Plant capacity to defend against executed by plants after predation include an increased photosynthetic herbivores is also classified on the basis of constitutive traits—which are rate in the remaining tissues, compensatory regrowth in non-affected present before predation—and induced resistance or tolerance, which meristems, use of storage reserves to generate new biomass, and modi­ appears after herbivory events [5,6]. fication of plant phenology to avoid interactions with herbivores in the Tolerance is considered a good strategy to maintain fitness after future. herbivory [7,8] and includes a wide variety of mechanisms that act to In some cases, it has been described that phylogeny determines the rapidly compensate for the loss of biomass caused by predation, such as impact of plant enemies [14]. But in other cases, is the ecological context growth compensation, an increase of photosynthetic activity or reallo­ (i.e. historic exposure to herbivores) what determines whether plants cation of resources from roots to shoots. Some examples of constitutive can tolerate herbivores [5]. A specific analysis could be necessary to traits related to tolerance are a high net photosynthetic rate [9], high elucidate the role of ecological context on how plants tolerate herbivory. * Corresponding author at: Edifici Guillem Colom Casasnovas Cra, Valldemossa Km 7,5 s/n, 07122, Palma, Balearic Islands, Spain. E-mail address: miquelcaposervera@gmail.com (M. Capó). https://doi.org/10.1016/j.plantsci.2021.110890 Received 15 November 2020; Received in revised form 17 March 2021; Accepted 19 March 2021 Available online 24 March 2021 0168-9452/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M. Capó et al. Plant Science 307 (2021) 110890 Fig. 1. Experimental design. The black horizontal lines indicate the timeline from planting the seeds (top row) until the two harvest events for the seedling (middle row) and adult plant experiments (bottom row). Blue squares indicate when each treatment started. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). The ecological context could be applied to insular endemic plants iso­ obtained soon after damage can vary from the values measured at later lated from predators, which evolved without mammalian herbivores sampling periods [11,21]. Therefore, in order to adequately describe the until the arrival of humans [15]. A trade-off exists between growth and herbivory tolerance capacity of a plant species, it is necessary to plant defence mechanisms [4], which explains the relationship between consider different stages of ontogeny [22,23]. herbivore exposure and plant features that are not solely related to The model species assessed in this work were Medicago critina (Font resistance [3] but also tolerance to herbivory [1]. In fact, the capacity to Quer) Greuter and Medicago arborea L. Unlike most species from the tolerate herbivores can decrease across evolution in species that have same genus, M. citrina and M. arborea are perennial woody shrubs that not coexisted with natural herbivores [16]. However, there is little in­ can grow up to 3 m tall and wide [24]. M. citrina is endemic to a few formation on how historic exposure to herbivores affects the relation­ small islets in the Balearic Sea and close to the Iberian Peninsula Coast, ship between constitutive traits and induced tolerance [17]. In the and a low number of isolated individuals also occur on the Spanish opposite view, some studies have proved that invasive species that have mainland and Western Ibiza [25]. The plant is included in the Spanish lost their herbivores can relocate their defence resources into other Catalogue of Threatened Species as ‘Vulnerable’ (RD 139/2011), and functions in a short-time period [18]. considered one of the most threatened species in the Mediterranean Loss of defence mechanisms in some insular species makes the plants [25]. The presence of M. citrina on other islets or bigger islands is likely more vulnerable to predators than phylogenetically related species that to have been strongly limited by the presence of introduced herbivores, have coexisted with herbivores [15], even though the constitutive traits mainly goats, rats and rabbits [25–28]. Latorre et al. [23] observed that of the species could be similar. Thus, islands and islets provide a suitable this species suffers from the severe combined effects of predation by rats setting to study how the historic absence or presence of herbivores can and rabbits, which damage the early recruitment stages (i.e. seeds, determine plant tolerance to herbivory and whether the historic context seedlings and saplings). At present, due to various rabbit and rat erad­ has a more evident effect on the responses to predation than constitutive ication programs in some islets, the presence of M. citrina has increased tolerance traits. substantially in both the original distribution area and in successfully However, the tolerance capacity of a species is not uniform across translocated zones [25]. In contrast, M. arborea L., a species closely time, since tolerance response varies depending on the life stage of the related to M. citrina, is widespread throughout Europe and is commonly plant [19]. Moreover, Gruntman and Novoplansky [20] demonstrated used as fodder in many areas, indicating mature plants have a good that ontogeny had a strong effect on tolerance to herbivory, predation in capacity to tolerate herbivory [29]. Interestingly, both species have a the vegetative phases induced greater meristem activation whereas different herbivory context: M. arborea is a widely distributed species predation in the reproductive phases induced investment in historically more exposed to herbivory than M. citrina, which is a late-determined reproductive traits such as fruit and seed biomass. In range-restricted species that occurs in free areas of herbivores. Hence, this sense, the responses to herbivory can vary over time, and data we tested the hypothesis that the species under herbivore pressures 2 M. Capó et al. Plant Science 307 (2021) 110890 Table 1 Constitutive values of mechanisms related to tolerance to herbivory for studied species. Seedling Adult Parameters related to tolerance M. citrina M. arborea t-test M. citrina M. arborea t-test Chlorophyll fluorescence (Fv/Fm’) 0.726 (± 0.004) 0.698 (± 0.005) ** 0.736 (± 0.010) 0.728 (± 0.014) ns Lipidic peroxidation (MDA) – – – 18.64 (± 2.40) 44.50 (± 4.01) ** Constitutive non photochemical quenching (NPQ) 0.578 (± 0.023) 0.596 (± 0.034) ns 0.627 (± 0.017) 0.641 (± 0.022) ns Maximum electron transport rate (ETR) 98.26 (± 8.74) 89.08 (± 8.77) ns 144.52 (± 21.42) 116.90 (± 14.41) ns Root C:N ratio 14.86 (± 0.57) 12.60 (± 0.63) * 21.75 (± 1.79) 17.52 (± 0.81) ns Shoot C:N ratio 17.72 (± 0.43) 13.69 (± 0.93) ** 17.36 (± 1.44) 14.75 (± 0.77) ns R:S ratio 0.524 (± 0.033) 0.386 (± 0.024) ** 0.432 (± 0.056) 0.404 (± 0.064) ns Total root biomass (g) 13.91 (± 1.08) 14.05 (± 1.41) ns 181.74 (± 25.27) 100.44 (± 14.22) * Total shoot biomass (g) 19.76 (± 1.73) 23.82 (± 1.39) ns 431.50 (± 37.24) 280.80 (± 38.80) * Nodules biomass (mg) 197.36 (± 47.76) 98.18 (± 16.63) ns – – – Specific leaf area (cm2/g) – – – 105.53 (± 5.96) 159.07 (± 6.925) *** Fv/Fm’ and MDA shown corresponds to the last sampling time (T6). Standard error of each parameter is shown between brackets. Significant differences between species for each life stage are shown with asterisks, and indicate significance strength. * < 0.05. ** < 0.01. *** <0.001. Table 2 transferred to larger pots (6 dm3). Plants were grown outdoors in a cage Temporal variation on chlorophyll fluorescence (Fv/Fm) and lipidic peroxida­ to protect against all types of herbivores and watered daily to field ca­ tion (MDA) during the experiment. pacity during the entire experiment. Fv/Fm MDA Time 58.9 (<0.001***) 24.84 (<0.001***) 2.2. Experimental design Treatment 6.38 (0.001***) 5.00 (0.007**) Species 0.53 (0.464) 4.92 (0.027*) The experiment was performed simultaneously for both species Life stage 0.91 (0.340) – following the same procedures (Fig. 1). The plants were randomly Species × treatment 0.47 (0.610) 9.14 (0.001***) allocated to two life stage groups to evaluate how tolerance varies be­ Treatment × life stage 2.75 (0.064) – Species × life stage 1.50 (0.219) – tween seedlings (7-months-old) and reproductive adults (16-months- old). Numbers indicate Fisher’s F-scores (and p-values). Asterisks indicate The response of seedlings to herbivory was assessed in an experiment significance. * performed during Sep-Dec 2017 (when the first leaves appeared) and the p < 0.05. ** p < 0.01. adult response was assessed in an experiment performed in Jul-Sep 2018 *** p < 0 .001); significant differences are marked in bold. Time was consid­ (when the first flowers appeared). Three herbivory treatments were ered as smooth term in GAMM models. MDA refers to 3,4-methylene applied: (1) no herbivory event as a control (n = 15), (2) removing dioxyamphetamine. biomass once at the beginning of the experiment (1 st cut, n = 15), and (3) removing biomass again 30 days after the first simulation (2nd cut, n present higher herbivory tolerance, although differences in constitutive = 15). Control individuals were used to enable a comparison of consti­ traits related to herbivory tolerance among species were expected to be tutive features at the same time-points as the induced tolerance response low due to their close phylogenetic relationship and their similar plant to simulated herbivory. A severe level of biomass loss was induced at architecture. Also, changes in tolerance would be observed across each simulated herbivory event. Considering that moderate herbivory is ontogeny as observed in many other species. defined as 50 % defoliation [11], we removed 80 % of the total aerial To validate this hypothesis, we aimed to (1) assess whether consti­ part of the plants at each herbivory event (including the leaves and tutive traits are good indicators of plant tolerance to herbivory in two stems) to simulate severe herbivory. To ensure standard regrowth after closely related species with different historical contexts of herbivore damage, a solution of 0.5 mM jasmonic acid was sprayed onto the in­ pressure; (2) evaluate whether induced tolerance after herbivory is cisions after the herbivory simulations [8,11]. All plants were harvested herbivore context-dependent, and (3) determine if the differences in 30 days after the second simulation. However, as some plants died after constitutive traits and induced tolerance between species are main­ simulated herbivory, the sample sizes for evaluation of the growth and tained during ontogeny. biochemical parameters varied among treatments (however, all n ≥ 7). 2. Materials and methods 2.3. Parameters related to physiological and biochemical status 2.1. Plant material We evaluated the parameters related to tolerance to herbivory described by Strauss and Agrawal [1]. Photosynthetic variables were Plants from the widespread M. arborea and the range-restricted assessed via pulse-amplitude modulated chlorophyll fluorometry (Ju­ M. citrina were cultivated from seeds provided by the Centre Forestal nior PAM, Walz) due to its advantages of less time-consuming data de les Illes Balears (IBANAT, Conselleria de Medi Ambient i Territori) at collection compared to gas exchange methods, as explained by Barton Finca Pública de Menut (Mallorca Island, western Mediterranean Basin, [11]. Chlorophyll fluorescence (Fv/Fm’) was assessed a total of seven 39◦ 49′ 44.78′′ N 2◦ 54′ 0.46′′ E). Seeds were germinated on February 2017, times (T0-T6) at 15-day intervals from the beginning of each experiment after 30-min scarification with 98 % sulfuric acid to improve the (before the first herbivory simulation, T0) until harvesting at T6. germination rate (though we found the scarification step was not The electron transport rate (ETR) in response to various photosyn­ necessary for the seeds to germinate). Seeds were planted in forest thetic active radiation (PAR) intensities was evaluated at the end of both germination tablets [30] and when cotyledons emerged, the seedlings experiments prior to harvesting. Light curves were performed over 11 were transferred to 3 dm3 pots. Seven months later, the seedlings were steps (PAR 0–1150 μmol photons m− 2 s− 1). Constitutive non- photochemical quenching (NPQ) at the highest PAR intensity was also 3 M. Capó et al. Plant Science 307 (2021) 110890 Fig. 2. Temporal series of chlorophyll fluorescence (Fv/Fm’) data (top and middle rows) and lipidic peroxidation (MDA) data (bottom row). Data is shown by species (columns) and life stage (rows). Values are mean (± standard error). Control plants are indicated in red circles, one damage treatment event in blue triangles and two damage treatment events in green squares. Blue and green vertical lines indicate the first and second herbivory simulations, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). Leaves were sampled (at the same time that Fv/Fm’ was measured) and Table 3 immediately immersed in liquid nitrogen and stored at − 80 ◦ C until Photosynthetic parameters obtained after light curve fitting of the electron analysis [33]. Approximately 0.1 g of frozen leaves were homogenized transport rate and photosynthetic active radiation (ETR/PAR) curves. in 1 ml of 1 % trichloroacetic acid, centrifuged at 10,000 rpm, and 1.5 ml ETRmax Alpha Ek of 0.5 % thiobarbituric acid prepared in 20 % trichloroacetic acid was added to 0.5 ml of the supernatant. The mixtures were heated at 90 ◦ C Treatment 0.51 (0.599) 1.32 (0.274) 0.08 (0.919) for 20 min, centrifuged at 10,000 rpm and the absorbance of the samples Species 0.32 (0.568) 0.07 (0.789) 0.92 (0.340) Life stage 28.62 (<0.001 58.65 (<0.001 11.25 (0.001 were measured at 532 nm using a Multiskan Sky Microplate Spectro­ ***) ***) **) photometer (ThermoFisher Scientific) and the non-specific absorption Species × treatment 3.15 (0.050*) 5.45 (0.007**) 1.44 (0.247) value at 600 nm was subtracted. The amount of MDA-TBA complex was Treatment × life 0.49 (0.615) 0.32 (0.721) 1.24 (0.299) calculated from the extinction coefficient (Cakmak and Horst, 1991). stage Species × life stage 3.49 (0.067) 0.46 (0.498) 4.48 (0.039*) Numbers indicate Fisher’s F-scores (and p-values). Asterisks indicate 2.4. Growth parameters related to biomass significance. * p < 0.05. Thirty days after the second herbivory simulation and at final har­ ** p < 0.01. vest, the aerial parts of the plants were dried in an oven at 65 ◦ C for 15 *** p < 0 .001); significant differences are marked in bold. days; the roots were also harvested and dried after removing excess soil and water. determined for control plants to evaluate the amount of energy dissi­ The dry weights of the shoots and roots were obtained to evaluate the pated by heat [11]. The maximum electron transport rate (ETRmax), root:shoot ratio (defined as the dry weight of root per dry weight of slope of the light-dependent part of the curve (alpha) and light satura­ shoot). Total dry biomass was also calculated as the sum of the dried tion (Ek) were obtained from the light curves. biomass of the shoots, roots and parts removed during the herbivory Lipidic peroxidation was determined as a proxy of biochemical status simulations. Two grams of roots and shoots from each individual were prior to herbivory [31]. Briefly, the concentration of 3,4-methylene separately ground and passed through a 1-mm mesh sieve. The carbon dioxyamphetamine (MDA) was evaluated in fresh leaves from five in­ and nitrogen contents (i.e. C:N ratio) of each sample were analysed dividuals of each species and treatment in the adult experiments [32]. following the Dumas method [34] using a LECO TruSpec (Leco® 4 M. Capó et al. Plant Science 307 (2021) 110890 Fig. 3. Light-dependent phase of the light saturated electron transport rate (ETR) curves. Data is shown by species (columns) and life stages (rows). Values are mean (± standard error). Control plants are indicated in red circles, one damage treatment event in blue triangles and two damage treatment events in green squares. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). Corporation) to evaluate the ability of plants to store and shunt re­ the treatments. LM models were made, checking that the residuals met sources from the roots to shoots after damage. Root nodules were only the criteria of normality and homoscedasticity through validation quantified in seedlings due to the methodological difficulty of separating graphs. The post-hoc Tukey and pairwise tests were used to identify all nodules from each root in adult plants; the seedling nodules were significant differences, by applying the ‘lsmeans’ function of the ‘lsmeans’ separated from the roots and weighed. package [38]. The same procedures were followed for the nodule dry Finally, five fresh leaves were collected from adult plants at harvest weight data for the seedling dataset and SLA data for the adult plants. and the foliar area was immediately calculated using ImageJ [35]. Leaves were stored at 65 ◦ C for 15 days and weighed to obtain the 3. Results specific leaf area (SLA), which was determined as the foliar area divided by the dry leaf biomass. 3.1. Constitutive traits related to tolerance 2.5. Statistical analysis Comparison of the constitutive traits in the control treatments revealed some significant differences in the analysed variables between Statistical analysis was conducted using R software v3.6.1 [36]. To M. citrina and M. arborea (Table 1). In seedlings, M. citrina exhibited a assess constitutive parameters, data from the control individuals was significant higher chlorophyll fluorescence value, C:N ratio and root: compared between species. The effect of the herbivory simulations was shoot ratio than M. arborea while NPQ, ETR, root, shoot and nodules analysed separately for seedlings and adults. Two sample t-tests were biomass remained no-significant (Table 1). In adult plants, M. arborea used to evaluate the significance of the differences between species for had significantly higher MDA and SLA values, M. citrina had higher root each parameter at each life stage. and shoot biomass and chlorophyll fluorescence, NPQ, ETR, C:N and R:S Generalised Additive Mixed Models were performed using chloro­ ratios remained similar between species (Table 1). phyll fluorescence and MDA as response variables, treatment, life stage and species as categorical parametric variables, time as linear smooth 3.2. Induced tolerance after simulation of herbivory variable and individuals as random factor. Variance of generated models were analysed using the ‘anova’ function of the ‘car’ package [37]. The death rate was used as the first proxy of tolerance to herbivory. Light curves were analysed considering herbivory treatment, life All M. arborea plants in the seedling and adult experiments survived stage and species as explanatory variables and the traits obtained from until harvested. However, in the seedling experiment (Fig. 1), 20 % of the fitted curve (ETRmax, alpha and Ek) as response variables. Linear M. citrina plants died after the second simulation. In the adult experi­ models (LM) were constructed and homoscedasticity and normality as­ ment, 28.5 % of M. citrina plants died after the first simulation and 27.2 sumptions were verified by plotting residuals versus fitted value. % died after the second simulation. The C:N ratio and dry weight of roots and shoots were analysed Chlorophyll fluorescence differed between treatments and over time separately for each species and life stage to evaluate variation between during the experiments as shown in Table 2 and Fig. 2. Chlorophyll 5 M. Capó et al. Plant Science 307 (2021) 110890 Table 4 Growth parameters obtained at the harvest time-point after the herbivory simulations. Response variable Fixed variables d.f., residual F P-value Treatment 2, 48 1.00 0.374 ns Species 1, 48 0.91 0.343 ns Root Life stage 1, 48 162.59 < 0.001 *** Treatment × species 2, 48 8.17 < 0.001 *** Treatment × life stage 2, 48 0.90 0.411 ns Species × life stage 2, 48 1.41 0.239 ns C:N ratio Treatment 2, 48 38.23 < 0.001 *** Species 1, 48 18.92 < 0.001 *** Shoot Life stage 1, 48 0.34 0.562 ns Treatment × species 2, 48 4.51 0.016 ** Treatment × life stage 2, 48 2.41 0.099 ns Species × life stage 2, 48 0.006 0.968 ns Treatment 2, 131 104.97 < 0.001 *** Species 1, 131 0.002 0.987 ns Life stage 1, 131 1023.80 < 0.001 *** Root Treatment × species 2, 131 6.21 0.002 ** Treatment × life stage 2, 131 6.57 0.001 *** Species × life stage 1, 131 39.01 < 0.001 *** Dry weight Treatment 2, 131 8.05 < 0.001 *** Species 1, 131 15.93 < 0.001 *** Life stage 1, 131 846.84 < 0.001 *** Shoot Treatment × species 2, 131 1.13 0.325 ns Treatment × life stage 2, 131 2.19 0.115 Species × life stage 1, 131 42.22 < 0.001 *** Treatment 2, 84 11.44 < 0.001 *** Nodules Species 1, 84 0.34 0.561 ns Treatment × species 2,84 4.00 0.021 * Treatment 2, 84 11.44 < 0.001 *** SLA Species 1, 84 0.34 0.561 ns Treatment × species 2, 84 4.00 0.021 * Asterisks indicate significance strength. * p < 0.05. ** p < 0.01. *** p < 0.001). SLA refers to specific leaf area. Fig. 4. C:N ratio for each species (columns) and life stage (rows) in control plants (Ctrl), and after the single cut simulation (Single) and double cut simulation (Double). Each graph shows the root and shoot data separately. Values are mean (± standard error). Different letters indicate significantly different groups (posterior Tukey-test). 6 M. Capó et al. Plant Science 307 (2021) 110890 Fig. 5. Dry biomass weight data for each species (columns) and life stage (rows) in control plants (Ctrl), and after the single cut simulation (Single) and double cut simulation (Double). Each graph shows the root and shoot data separately. Values are mean (± standard error). Different letters indicate significantly different groups (posterior Tukey-test). fluorescence was reduced by the treatments and over time in both with the exception of M. citrina in the seedling experiment. In contrast, M. citrina and M. arborea. Despite that, the reduction was only detected significant reductions in the shoot C:N ratio were found among treat­ after the first herbivory simulation. After the second herbivory event, ments, with greater reductions in M. citrina than M. arborea and in the Fv/Fm’ values recovered to control values. Specifically, lower values seedlings than adults. Generally, the treatment affected the C:N ratio in than control were observed for chlorophyll fluorescence in M. citrina shoots, but not in roots, with stronger effects observed in M. citrina than seedlings after the first cut, but not after the second. In contrast, no M. arborea. differences were observed among treatments for M. arborea seedlings, Biomass produced by plants (also considering the parts removed even though chlorophyll fluorescence fluctuated over time in both during the herbivory simulations in the total outcome) indicated toler­ species. ance response mechanism varied between both the species and life Lipidic peroxidation increased in the treatment groups after the stages (Fig. 5; Table 4). The roots were severely affected by damage, herbivory simulations, while control plants maintained baseline levels even though only the aerial parts of the plant were removed during the in both species (Table 2). Specifically, the herbivory treatments signif­ herbivory simulations. Seedlings exhibited a greater reduction in root icantly increased the MDA values in M. citrina compared to control biomass than adults in both species. Medicago citrina demonstrated a plants during the whole experiment, whereas treated and control varied response to damage across time. The seedling roots were initially M. arborea plants had similar MDA values (Fig. 2). affected as root biomass decreased after the first cut, and a further The light curves indicated strong variation in all variables between reduction was observed after the second cut. In contrast, adult plants the seedlings and adults. Furthermore, life stages exhibited different maintained root biomass after the first cut, with a reduction in biomass photosynthetic performance in terms of ETRmax and alpha after her­ only observed after the second cut. In M. arborea, seedling root biomass bivory damage (Table 3). decreased in a similar pattern to M. citrina; however, adult M. arborea In seedlings, M. citrina exhibited reduced photosynthetic activity root biomass also reduced after the first cut, but not between the first after the second herbivory damage, whereas M. arborea showed reduced and second cuts. On the other hand, the seedling shoot biomass of both photosynthetic activity after the first herbivory simulation, but recov­ species strongly decreased after the herbivory simulations. The adults of ered after the second damage (Fig. 3). Different patterns of variation in both species resprouted and recovered the initial amount of shoot the light curves of adult plants were observed among treatments for both biomass, with both the single and double damage treatments achieving species. In M. citrina, the treatments led to lower photosynthetic ca­ the same levels as controls. pacity than control plants, whereas M. arborea exhibited increased Root nodules were only sampled during the seedling stage, and their photosynthetic activity after damage (Fig. 3). biomass reduced after the simulations of herbivory in both species Complete analysis of the growth parameters is summarized in (Fig. 6A; Table 4), although the effect was significantly higher in Table 4. The C:N ratio after herbivory varied between species and life M. citrina than M. arborea. Specifically, after the first simulation, root stages, as shown in Fig. 4. The root C:N ratios were not affected by the nodule biomass decreased in M. citrina and was maintained in herbivory simulations and both species showed similar root C:N ratios, M. arborea. After the second simulation, root nodules were almost absent 7 M. Capó et al. Plant Science 307 (2021) 110890 Fig. 6. Nodule dry biomass for seedlings (A) and specific leaf area for adults (B) in controls and the (Ctrl), single cut (Single) and double cut (Double) treatments. Each graph shows the root and shoot data separately. Values are mean (± standard error). Different letters indicate significantly different groups (posterior Tukey-test). in M. citrina and reduced by around 50 % in M. arborea compared to there is a trade-off between growth and defence [40]. Following this control plants. On the other hand, SLA, which was only sampled in the argument, the maintenance of constitutive characteristics that improve adult stage (Fig. 6B; Table 4), increased after damage in both species, the response to herbivores, despite the absence of herbivory pressure, with a greater extent of damage in M. arborea. can be explained by either an absence of a trade-off between growth and the development of such constitutive characteristics, or that these 4. Discussion characteristics are related to tolerance to other types of stress. Indeed, the same constitutive trait could be related to both biotic and abiotic This study partially supports the hypothesis that the range-restricted stresses [41–43]. For example, M. citrina is well adapted to severe species M. citrina has a lower tolerance to herbivory than the wide­ saline-exposure water-stress, which is related with high C:N or root: spread, closely related species M. arborea, independent of ontogeny. shoot ratios [44,45]. Moreover, coastal species usually present lower However, M. citrina exhibited better values for constitutive traits related SLA than their inland relatives, as happened with our study species [46]. to tolerance than M. arborea. Thus, in this case, the relationship between So, some constitutive traits related with herbivory tolerance in other constitutive tolerance traits and the responses to herbivory are only species [1,13], would be related to tolerance to salinity in the case of observed when herbivores are present in as selective pressure, as M. citrina in comparison with M. arborea, which is known to be poorly observed in other cases [1,39] at both the seedling and adult stages. salt-tolerant [47]. As a result, constitutive traits related with resistance to several stress are not always good predictors of herbivory tolerance. Adult plants of M. citrina are a clear example since plants died after 4.1. Are constitutive tolerance traits related to herbivore pressure? damage despite resprouting through nutrient reallocation was expected due to the high nitrogen content of roots. The range-restricted species M. citrina exhibited better values for constitutive traits than the widespread, closely related, M. arborea, including better physiological activity lower oxidative stress and a 4.2. Is induced tolerance to herbivory lower in species with less exposure higher R:S ratio and root C:N ratio. This would suggest that range- to herbivores? restricted species are more tolerant to herbivores [11–13]. Thus, M. citrina may have retained good constitutive traits related to tolerance Medicago citrina and M. arborea responded differently to the her­ to herbivores, even though the species does not coexist with herbivores bivory simulation events. First, a considerable proportion of M. citrina in its natural distribution and it only occurs in isolated areas. individuals died after simulation (20 % of seedlings and 55 % of adults), The lack of herbivores as a selective pressure has been described as a whereas all M. arborea individuals survived. Considering the main traits cause of loss of defence in various species [15], based on the idea that related to compensation [1], M. citrina almost lacks tolerance to 8 M. Capó et al. Plant Science 307 (2021) 110890 Fig. 7. Induced tolerance of Medicago citrina and Medicago arborea in the seedling and adult life stages to simulated herbivory after the first cut and second cut. Negative effects compared to control plants are marked in red arrows; double arrows indicate where the double cut treatment lead to a greater effect than the single cut; green equal symbols indicates where the values were similar in the treatment and con­ trol. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article). herbivory, whereas M. arborea tolerates slight herbivory. For both spe­ in the future [31,56]. Further studies of biochemical disturbances after cies, none of the parameters measured were higher after damage damage are necessary to evaluate how herbivory induces oxidative compared to control plants (Fig. 7), as previously observed for some stress in tolerant plant species, especially in the context of tolerance to herbivory-tolerant species [21,48–50]. Tolerant species exhibit im­ introduced herbivores. In contrast, M. arborea was able to recover provements in many parameters after damage, such increased biomass damage-induced changes in lipidic peroxidation and chlorophyll fluo­ overall [51] or an increase in the C:N ratio of shoots that emerge after rescence, and damages in light caption were detected later in M. arborea herbivory events [13]. However, these effects were not observed in than M. citrina, as also observed in other species such as endemics from either of the studied species, which indicates M. citrina and M. arborea do the Hawaii Islands [11]. not overcompensate after herbivory [52]. Evolution leads to reduce defence allocation in those species that lose Additionally, M. arborea recovered its chlorophyll fluorescence value their natural enemies or that have been introduced in other ecological after damage without increasing lipidic peroxidation and maintained contexts where predators are absent [57]. The present study is an the number of live root nodules, while the opposite trends were observed example of that process since the range-restricted species is less tolerant for all of these variables in M. citrina. Indeed, even though herbivory was than the widespread. only performed on the aerial parts of the plants, indirect biomass loss was observed for the underground parts of both species (i.e. roots and 4.3. Does ontogeny affect tolerance capacity patterns? nodules), which confirms that neither species can use root storage to compensate for herbivory. Medicago citrina was more affected by herbivory than M. arborea in However, the second cut did not synergistically increase the effects both life stages. This lack of tolerance to defoliation in both life stages of herbivory, except for the increase in SLA in M. citrina, which can be may explain why the entire M. citrina populations disappeared from related to optimization of the foliar area and investment of resources for islands and islets when herbivores were introduced [27]. Indeed, the further development [53]. Interestingly, both species were able to dead risk after herbivory is higher in M. citrina than in M. arborea as recover shoot biomass after the first and second damage, even though proved in this study. Adults of M citrina showed a higher death per­ root biomass reduced. This observation would predict tolerance capacity centage than seedlings after herbivory simulations, indicating higher for both species, in which resource allocation from the roots to shoots damage by herbivores. However, evaluating individuals that survived, indicates a mechanism of compensation for herbivory biomass loss [54, seedlings showed less capacity to recover control values than adults, in 55]. Nevertheless, this pattern was not observed for the C:N ratios of terms of root C:N ratio, biomass or even the parameters derived from the either roots or shoots, as the C:N ratios of both plant structures light curves (ETRmax, alpha and Ek), as observed in similar experiments decreased after herbivory. Despite this, we predict the plants could using seedlings of endemic species from Hawaii [11]. Medicago citrina gradually recover over a longer period of time, and probably reach seedlings exhibited a poorer response in terms of root biomass and control values. However, increased lipidic peroxidation indicates regrowth capacity compared to adults from the same species. This long-term damage to biochemical status, and may imply indirect effects observation indicates it would be harder for seedlings to resprout and 9 M. Capó et al. Plant Science 307 (2021) 110890 grow after herbivory [58] but see [21]. Moreover, damage during the References seedling stage can affect the development of further life stages [59]. An incremental increase in tolerance to herbivory has been reported during [1] S.Y. Strauss, A.A. Agrawal, The ecology and evolution of plant tolerance to herbivory, Trends Ecol. Evol. 14 (1999) 179–185. ontogeny [60]. In this study, the seedlings of both species were [2] R. Mauricio, Natural selection and the joint evolution of tolerance and resistance as vulnerable to herbivory, and the adult life stage of both species exhibited plant defenses, Evol. Ecol. 14 (2000) 491–507. improved responses despite many M. citrina adult individuals still died. [3] A.A. Agrawal, M. Fishbein, Plant defense syndromes, Ecology 87 (2006) 132–149. [4] E. Van Der Meijden, M. Wijn, H.J. Verkaar, Defence and regrowth, alternative plant Indeed, death events of M. citrina may plausibly explain the constrained strategies in the struggle against herbivores, Oikos 51 (2016) 355–363. distribution mediated by predation of the seeds and seedlings by intro­ [5] A. Kessler, I.T. Baldwin, Plant responses to insect herbivory: the emerging duced mammalian herbivores [27]. molecular analysis, Annu. Rev. Plant Biol. 53 (2002) 299–328, https://doi.org/ 10.1146/annurev.arplant.53.100301.135207. [6] J. Koricheva, Meta-analysis of sources of variation in fitness costs of plant antiherbivore defenses, Ecology 83 (2002) 176–190, https://doi.org/10.1890/ 4.4. Implications for the conservation of threatened species 0012-9658(2002)083[0176:MAOSOV]2.0.CO;2. [7] A.A. Agrawal, Induced responses to herbivory and increased plant performance, Science (80-.) 279 (1998) 1201–1202, https://doi.org/10.1126/ The evolutionary context of M. citrina and M. arborea may be the science.279.5354.1201. main factor that explains the varied tolerance to herbivory among the [8] S. Rasmann, M.D. Johnson, A.A. Agrawal, Induced responses to herbivory and jasmonate in three milkweed species, J. Chem. Ecol. 35 (2009) 1326–1334, seedling and adult plants of each species. Assessing if endemic plants https://doi.org/10.1007/s10886-009-9719-0. from islands can tolerate herbivory will be essential to evaluate the [9] G. Houle, G. Simard, Additive effects of genotype, nutrient availability Additive impact of introduced mammal herbivores in insular communities. As and type of tissue damage on the compensatory response of Salix planifolia ssp. planifolia to simulated herbivory, Oecologia 107 (1996) 373–378. observed in this study, seedlings are the most threatened stage of the [10] K. Madriaza, A. Saldaña, C. Salgado-Luarte, V.M. Escobedo, E. Gianoli, Chlorophyll population structure and therefore the recruitment of the population fluorescence may predict tolerance to herbivory, Int. J. Plant Sci. 180 (2018) might be severely affected. Also, M. citrina adults die after predation, 81–85, https://doi.org/10.1086/700583. [11] K.E. Barton, Low tolerance to simulated herbivory in Hawaiian seedlings despite despite individuals that survived were able to recover successfully. In induced changes in photosynthesis and biomass allocation, Ann. Bot. 117 (2016) this sense, future conservation strategies must focus on protecting nat­ 1053–1062, https://doi.org/10.1093/aob/mcw021. ural populations from introduced herbivores, and considering this threat [12] G.W. Felton, C.B. Summers, A.J. Mueller, Oxidative responses in soybean foliage to when implementing reintroduction or reinforcing measures. herbivory by bean leaf beetle and three-cornered alfalfa hopper, J. Chem. Ecol. 20 (1994) 639–650, https://doi.org/10.1007/BF02059604. [13] P. Tiffin, Mechanisms of tolerance to herbivore damage: what do we know? Evol. 5. Conclusions Ecol. 14 (2000) 523–536, https://doi.org/10.1023/A:1010881317261. [14] G.S. Gilbert, H.M. Briggs, R. Magarey, The impact of plant enemies shows a phylogenetic signal, PLoS One 10 (2015) e0123758, https://doi.org/10.1371/ Even though its constitutive defence traits predicted M. citrina could journal.pone.0123758. tolerate herbivory, the opposite was observed as the species exhibited [15] L. Bowen, D. Van Vuren, Insular endemic plants lack defenses against herbivores, Conserv. Biol. 11 (1997) 1249–1254, https://doi.org/10.1046/j.1523- poor induced tolerance after damage. Our study is an example of how 1739.1997.96368.x. the herbivore pressure may play a more essential role in determining [16] G. Vourc’h, J.L. Martin, P. Duncan, J. Escarré, T.P. Clausen, Defensive adaptations plant tolerance to herbivory than constitutive defences. Therefore, if of Thuja plicata to ungulate browsing: a comparative study between mainland and island populations, Oecologia 126 (2001) 84–93, https://doi.org/10.1007/ induced tolerance fails to compensate for herbivory, the species must s004420000491. rely on the escape strategy in areas where herbivores are absent, as is the [17] R. Hirata, N. Wasaka, A. Fujii, T. Kato, H. Sato, Differences in flowering phenology, case of the islets where M. citrina occurs. Assessment of ontogeny is architecture, sexual expression and resource allocation between a heavily haired and a lightly haired nettle population: relationships with sika deer, Plant Ecol. 220 crucial when evaluating tolerance to herbivory; seedlings recover slower (2019) 255–266, https://doi.org/10.1007/s11258-019-00910-7. than adults, and M. citrina showed high death percentages in both stages. [18] A. Uesugi, A. Kessler, Herbivore release drives parallel patterns of evolutionary Therefore, lack of herbivory pressure in the historic context leads to a divergence in invasive plant phenotypes, J. Ecol. 104 (2016) 876–886, https://doi. loss of induced tolerance, but not constitutive tolerance traits. org/10.1111/1365-2745.12542. [19] K.E. Barton, M.E. Hanley, Seedling-herbivore interactions: insights into plant defence and regeneration patterns, Ann. Bot. 112 (2013) 643–650, https://doi.org/ Funding 10.1093/aob/mct139. [20] M. Gruntman, A. Novoplansky, Ontogenetic contingency of tolerance mechanisms in response to apical damage, Ann. Bot. 108 (2011) 965–973, https://doi.org/ This work was funded by Project CGL2015-70449-R (Ministerio de 10.1093/aob/mcr204. Economía y Competitividad, Gobierno de España). MC and MR-O were [21] E. Del-Val, M.J. Crawley, Are grazing increaser species better tolerators than decreasers? An experimental assessment of defoliation tolerance in eight British supported by PhD fellowships FPI/1925/2016 through Direcció General grassland species, J. Ecol. 93 (2005) 1005–1016, https://doi.org/10.1111/j.1365- de Política Universitària i Recerca (Govern de les Illes Balears) and 2745.2005.01011.x. European Social Fund; and FPU16/01544 from Ministerio de Economía [22] K.E. Barton, K. Boege, Future directions in the ontogeny of plant defence: understanding the evolutionary causes and consequences, Ecol. Lett. 20 (2017) y Competitividad (MINECO, Spain), respectively. 403–411, https://doi.org/10.1111/ele.12744. [23] S. Ochoa-López, X. Damián, R. Rebollo, J. Fornoni, C.A. Domínguez, K. Boege, Ontogenetic changes in the targets of natural selection in three plant defenses, New Phytol. 226 (2020) 1480–1491, https://doi.org/10.1111/nph.16422. Declaration of Competing Interest [24] A. Juan, Estudio sobre la morfología, variabilidad molecular y biología reproductiva de Medicago citrina (Font Quer) Greuter (Leguminosae), Bases para The authors declare they have no potential conflict of interest su conservación, Universitat d’Alacant - Universidad de Alicante, 2002 https://doi. org/https://www.educacion.gob.es/teseo/mostrarRef.do?ref=388035. [25] E. Laguna, L. Sáez, M. Mus, A. Juan, M.B. Crespo, Medicago citrina, in: S. Pasta, Acknowledgements A. Perez-Graber, L. Fazan, B. Montmollin (Eds.), Top 50 Mediterr. Isl. Plants Updat. 2017, IUCN/SSC/Mediterranean Plant Spec. Gr., Neuchâtel, Switzerland, 2017. [26] G. Bibiloni, J. Rita, Flora y vegetación terrestre. Parq. Nac. Del Arch. Cabrera, The authors are grateful to the CEFOR staff, especially Maite Bover, Esfagnos, Talavera de la Reina, 2000, pp. 51–86. for their valuable help with plant maintenance and harvesting. We also [27] L. Latorre, A.R. Larrinaga, L. Santamaría, Combined impact of multiple exotic herbivores on different life stages of an endangered plant endemism, Medicago thank Alicia V. Perera-Castro for support with analysing the fluores­ citrina, J. Ecol. 101 (2013) 107–117, https://doi.org/10.1111/1365-2745.12005. cence data and Joshua Borràs, Sebastià Perelló and Miquel Àngel Ribas [28] J. Rita, E. Ballesteros, A. Ginés, M. Mcminn, V. Pérez-Mellado, Tejiendo naturaleza: for fieldwork support. Figs. 1 and 7 were generated using Biorender el archipiélago de Cabrera, lugar de encuentro y armonía entre la gea, la fauna y la flora, in: A. Robledo (Ed.), El Parq. Nac. Marítimo-Terrestre Del Arch. Cabrera Un (http://www.biorender.com). Andrea Devlin of Science Editing Experts Paisaje Entre La Tierra y El Mar, IGME & OAPN, Madrid, 2016. proofread the manuscript. Early versions of this article have been [29] A. Robledo, S. Rios, E. Correal, El grupo Medicago arborea en la cuenca improved by comments and suggestions of two anonymous reviewers. mediterranea: I. Origen, distribución y morfología, Pastos 23 (1993) 55–67. 10 M. Capó et al. Plant Science 307 (2021) 110890 [30] C. Cardona, Tolerància germinativa a la salinitat d’espècies litorals de les Illes coastal environments, New Phytol. 160 (2003) 337–347, https://doi.org/10.1046/ Balears, Universitat de les Illes Balears, 2019. https://dspace.uib.es/xmlui/handle/ j.1469-8137.2003.00880.x. 11201/148896. [47] E. Lefi, H. Medrano, J. Cifre, Water uptake dynamics, photosynthesis and water use [31] H. Wei, J. Zhikuan, H. Qingfang, Effects of herbivore stress by Aphis medicaginis efficiency in field-grown Medicago arborea and Medicago citrina under prolonged Koch on the Malondialdehyde contents and the activities of protective enzymes in Mediterranean drought conditions, Ann. Appl. Biol. 144 (2004) 299–307, https:// different alfalfa varieties, Acta Ecol. Sin. 27 (2007) 2177–2183, https://doi.org/ doi.org/10.1111/j.1744-7348.2004.tb00345.x. 10.1016/S1872-2032(07)60048-1. [48] K.N. Paige, T.G. Whitham, Overcompensation in response to mammalian [32] S. Christensen, S. Enge, K.R. Jensen, C. Müller, L.P. Kiær, N. Agerbirk, C. Heimes, T. herbivory: the advantage of being eaten, Am. Nat. 129 (1987) 407–416, https:// P. Hauser, Different herbivore responses to two co-occurring chemotypes of the doi.org/10.1086/284645. wild crucifer Barbarea vulgaris, Arthropod. Interact. 13 (2019) 19–30, https://doi. [49] A.J. Belsky, W.P. Carson, C.L. Jensen, G.A. Fox, Overcompensation by plants: org/10.1007/s11829-018-9633-x. Herbivore optimization or red herring? Evol. Ecol. 7 (1993) 109–121, https://doi. [33] I. Cakmak, W.J. Horst, Effect of aluminium on lipid peroxidation, superoxide org/10.1007/BF01237737. dismutase, catalase, and peroxidase activities in root tips of soybean (Glycine max), [50] A.A. Agrawal, Overcompensation of plants in response to herbivory and the by- Physiol. Plant. 83 (1991) 463–468, https://doi.org/10.1111/j.1399-3054.1991. product benefits of mutualism, Trends Plant Sci. 5 (2000) 309–313, https://doi. tb00121.x. org/10.1016/S1360-1385(00)01679-4. [34] M.E. Ebeling, The dumas method for nitrogen in feeds, J. AOAC Int. 51 (1968) [51] J. Järemo, E. Palmqvist, Plant compensatory growth: a conquering strategy in 766–770, https://doi.org/10.1093/jaoac/51.4.766. plant-herbivore interactions? Evol. Ecol. 15 (2001) 91–102, https://doi.org/ [35] M.D. Abràmoff, P.J. Magalhães, S.J. Ram, Image processing with imageJ, 10.1023/A:1013899006473. Biophotonics Int. 11 (2004) 36–41. [52] J. Bartolomé, E. Baraza, M.C. Espunya, E. Castells, L. Rivera, M.J. Broncano, [36] R Core Team, R: a Language and Environment for Statistical Computing, 2019. Tolerance to severe browsing of three shrub species on Mediterranean islands, Acta [37] J. Fox, S. Weisberg, An R Companion to Applied Regression: Appendices, Robust Oecologica 98 (2019) 30–35, https://doi.org/10.1016/j.actao.2019.05.006. Regres. R., 2014, pp. 1–17, https://doi.org/10.1177/0049124105277200. [53] A.T. Moles, M. Westoby, Do small leaves expand faster than large leaves, and do [38] R.V. Lenth, Least-squares means: the r package lsmeans, J. Stat. Softw. 69 (2016) shorter expansion times reduce herbivore damage? Oikos 90 (2000) 517–524, 1–33, https://doi.org/10.18637/jss.v069.i01. https://doi.org/10.1034/j.1600-0706.2000.900310.x. [39] J. Fornoni, Ecological and evolutionary implications of plant tolerance to [54] B. Li, T. Shibuya, Y. Yogo, T. Hara, Effects of ramet clipping and nutrient herbivory, Funct. Ecol. 25 (2011) 399–407, https://doi.org/10.1111/j.1365- availability on growth and biomass allocation of yellow nutsedge, Ecol. Res. 19 2435.2010.01805.x. (2004) 603–612, https://doi.org/10.1111/j.1440-1703.2004.00685.x. [40] F.J. Messina, S.L. Durham, J.H. Richards, E.D. McArthur, Trade-off between plant [55] M.B. Coughenour, J.K. Detling, I.E. Bamberg, M.M. Mugambi, Production and growth and defense? A comparison of sagebrush populations, Oecologia 131 nitrogen responses of the African dwarf shrub Indigofera spinosa to defoliation and (2002) 43–51, https://doi.org/10.1007/s00442-001-0859-3. water limitation, Oecologia 83 (1990) 546–552, https://doi.org/10.1007/ [41] B. Demmig-Adams, J.J. Stewart, W.W. Adams, Chloroplast photoprotection and the BF00317208. trade-off between abiotic and biotic defense. Adv. Photosynth. Respir, Springer, [56] Y. Tan, L. Bai, L. Xiao, S. Wei, H. Zhao, Herbivore stress by Lygus lucorum inducing Dordrecht, 2014, pp. 631–643, https://doi.org/10.1007/978-94-017-9032-1_28. protective enzyme activity and MDA content on different cotton varieties, Cott. Sci. [42] C.D.G. Harley, Abiotic stress and herbivory interact to set range limits across a two- 5 (2010) (accessed August 21, 2020), http://en.cnki.com.cn/Article_en/CJFDTota dimensional stress gradient, Ecology 84 (2003) 1477–1488, https://doi.org/ l-MHXB201005017.htm. 10.1890/0012-9658(2003)084[1477:ASAHIT]2.0.CO;2. [57] R.M. Keane, M.J. Crawley, Exotic plant invasions and the enemy release [43] P. Pandey, V. Irulappan, M.V. Bagavathiannan, M. Senthil-Kumar, Impact of hypothesis, Trends Ecol. Evol. 17 (2002) 164–170, https://doi.org/10.1016/ combined abiotic and biotic stresses on plant growth and avenues for crop S0169-5347(02)02499-0. improvement by exploiting physio-morphological traits, Front. Plant Sci. 8 (2017) [58] K. Boege, R.J. Marquis, Facing herbivory as you grow up: the ontogeny of 537, https://doi.org/10.3389/fpls.2017.00537. resistance in plants, Trends Ecol. Evol. 20 (2005) 441–448, https://doi.org/ [44] D. Chen, S. Wang, B. Xiong, B. Cao, X. Deng, Carbon/nitrogen imbalance associated 10.1016/j.tree.2005.05.001. with drought-induced leaf senescence in sorghum bicolor, PLoS One 10 (2015), [59] S.K. Wallace, S.D. Eigenbrode, Changes in the glucosinolate-myrosinase defense https://doi.org/10.1371/journal.pone.0137026. system in Brassica juncea cotyledons during seedling development, J. Chem. Ecol. [45] D.E. Karcher, M.D. Richardson, K. Hignight, D. Rush, Drought tolerance of tall 28 (2002) 243–256. fescue populations selected for high root/shoot ratios and summer survival, Crop [60] M.E. Hanley, E.L. Fegan, Timing of cotyledon damage affects growth and flowering Sci. 48 (2008) 771–777, https://doi.org/10.2135/cropsci2007.05.0272. in mature plants, Plant Cell Environ. 30 (2007) 812–819, https://doi.org/10.1111/ [46] C.A. Knight, D.D. Ackerly, Evolution and plasticity of photosynthetic thermal j.1365-3040.2007.01671.x. tolerance, specific leaf area and leaf size: congeneric species from desert and 11