Resource Allocation Among Cladodes of Opuntia Robusta From East-central Mexico

The Optimal Defense Hypothesis (ODH) predicts that younger, more valuable plant organs should be better defended. We tested this hypothesis in Opuntia robusta Wendl. since its sequential, hierarchical and segmented architecture permits the consideration of a possible reallocation of secondary metabolites among cladodes with different age. We performed a eld study taking samples of vegetative tissues from cladodes of different orders during eight months, and a eld experiment, by covering either apical or basal cladodes with a fabric. We determined the content of 4-hydroxybenzoic acid (4-HBA), chlorogenic acid (CGA), quercetin (QUE) and salicylic acid (SA). 4-HBA and CGA followed the predictions of ODH. QUE followed an inverse tendency. SA did not show any tendency related either with the position of the cladode or the treatment (light vs. shade) however, its concentration was positively correlated with the concentration of 4-HBA. As we detected SA only in a low proportion of cladodes and 4-HBA in all cladodes, we hypothesize the conversion of the latter metabolite to the former one. Contrary to ODH, CGA presented lower concentrations in apical than in basal cladodes when co-occurred with SAL, and QUE was transferred from apical (younger) to basal (older) cladodes. In this study, we found contradictions in the premises and outcomes of both the Resource Availability Hypothesis (RAH) and the ODH, suggesting that a new hypothesis concerning the plant defense against stress factors should be proposed.


Introduction
Most plants grow under strong herbivory pressure (Coley and Aide 1991;Coley and Barone 1996). To arrest the effect of herbivores, plants produce a wide array of chemical defenses. Different hypotheses have aimed to explain the patterns of defense against herbivores in plants.
For example, Optimal Defense Hypothesis (ODH) (McKey 1974) considers that more expensive plant tissues (or, more expensive sexual forms) can be more defended than cheaper ones as a results of optimal energy allocation to defense vs. growth, reproduction and other physiological processes. If so, an optimal pattern of allocation should change with the grazing rate of a given tissue in the absence of defense, with the cost of the allocation of a given defense to that tissue, and with the value of this tissue for the reproductive success of the plant (McKey 1974). For example, if the relative cost of tissue production is low (per tissue biomass unit), then a loss of some amounts of poorly defended but cheap tissue can be easily compensated by tissue regrowth. Also, younger tissues might be expected to be better defended than the older ones, since the former have a higher photosynthetic potential that allows plant to increase the allocation of resources to reproductive tissues, and in turn, to increase the reproductive success, and thus, their loss implies stronger tness impairment (McKey 1974;Meldau et al. 2012;van Dam et al. 1994). On the other hand, the Resource Availability Hypothesis (RAH) (Coley et al. 1985), forecasts that natural selection favors slow-growing plants with high levels of defense in less productive environments and faster-growing plants with low levels of defense in high-resource environments. Some predictions of both hypotheses, ODH and RAH are not mutually exclusive, because following RAH the biomass of plants growing in less productive environments is more valuable in terms of both energy and tness, as it is more di cult to recover due to a lower assimilation rate. On the other hand, the two hypotheses lead to contradictory predictions because fast-growing young plants, that generally have more resources available for growth and have a higher photosynthetic potential, should be better defended following ODH but less defended following RAH. Both hypotheses may be applied also to plants' organs assuming that they may grow on either more or less productive plant (a plant is an "environment" for its parts).
Addressing predictions of ODH and RAH, we studied the production of defense substances in Opuntia robusta Wendl. (Cactaceae), an endemic succulent from semiarid highlands of Meridional Plateau, Mexico, which forms three types of populations: strictly hermaphrodite, dioecious (masculine and feminine plants), and trioecious. In this species, hermaphrodite individuals have bisexual owers, and the other sexual forms, unisexual ones (only the organs of one sexual form are functional) (del Castillo and Argueta 2009). The species chosen by us is a convenient study organism to address ODH and RAH since it grows in a modular way by producing well-de ned body units called cladodes, which form easily distinguishable age classes. Importantly, the existence of sequential, hierarchical and segmented architecture of individual plants allows considering a possible reallocation of secondary metabolites among cladodes with different age. Very little is known about secondary metabolites in cladodes of O. robusta. Guevara-Figueroa et al. (2010) found 10 different secondary metabolites, including 4hydroxy benzoic (4-HBA) and salicylic acid (SA). Gonzalez-Ponce et al. (2016) found quercetin (QUE) in fruits. As far as we know, only one study con rmed the presence of cyanogenic acid (CGA) in O. robusta (Janczur et al. 2020b). There is no study on either the possible defensive function or on ecological aspects of secondary metabolites in this species (but see Janczur et al. (2014) and Janczur et al. (2020b)). Studies of other plant species showed defensive function of some of the metabolites found in Opuntia species, i.e. CGA combated arthropods (Erb et  Higher sunlight intensity, which is one of the indicators of higher environmental productivity, is expected to promote less intense chemical response to herbivory, since at better productivity it is easier to recover lost tissues (Coley et al. 1985), and is expected to promote more intense production of metabolites that protect a plant against UV radiation. So, a plant should solve a dilemma of how to distribute resources among physiological processes, whose effects are frequently opposite. The solution of such disjunctive should produce trade-offs between the proportions of resources allocated in different physiological processes.
In our previous study we found that the tissue of non-parental cladodes of the individuals of Opuntia robusta, contained a higher concentration of total phenolic compound than cladodes that bore sprouts of new cladodes (parental cladodes) (Janczur et al. 2014). Here, we studied whether the differences in the incidence of four secondary metabolites in cladodes of O. robusta are a result of their transfer between cladodes from different orders, or cladodes from each order produce these metabolites independently. The direction of a potential ow of secondary metabolites might have important implications for testing plant defense hypotheses. If younger, faster growing cladodes present a higher concentration of secondary metabolites, it would be in accordance with ODH and not with RAH that predicts lower concentration of chemical defenses in faster growing plants. If lower level, light-exposed cladodes, contain either lower or similar concentrations of secondary metabolites than higher-level shaded cladodes, it means that secondary metabolites are transferred from the former to the latter cladodes that is, the plant "takes decision" to send them in that direction, because secondary metabolites are important for the protection of young and thus photosynthetically more valuable tissues. This would be a straightforward argument in support of ODH. If parental shade-exposed cladodes have higher or similar concentrations of secondary metabolites than daughter cladodes, it would mean that the latter photosynthesized the metabolites and sent them to the parental cladodes. The lack of differences may also mean that the secondary metabolites are stable over time, however, in this case, the transfer of a stable secondary metabolite to another compartment containing a secondary metabolite will produce increase of concentration in the latter compartment. If the secondary metabolites are unstable, their concentration or incidence should decrease under shaded conditions: if not, it means that a light-exposed cladode transfers them to the shade-exposed one. If secondary metabolites are produced to deal with UV damage, their concentration should be higher in  (2007) and (Manetas 2003). However, if a shaded cladode, either parental or daughter to a lighted cladode, contains either a higher or equal concentration of a secondary metabolite, it means that the production of this secondary metabolite is not a response to UV.
Ultimately, our study aimed at resolving whether the differences in the probability of incidence or the concentration of secondary metabolites in the cladodes of O. robusta result from their reallocation between cladodes of different levels. We explored the following questions: 1) Do younger O. robustae organs contain a higher concentration (or, higher incidence) of secondary metabolites? 2) Do plants possess a mechanism that allows them the transference of secondary metabolites from older to younger organs?

Methods And Materials
Study Species. Our eld study was conducted on hermaphrodite individuals of Opuntia robusta (Cactaceae). The subgenus of Platyopuntia that this species belongs to, develop photosynthesizing shoots called cladodes, covered by spines that grow on highly specialized branches called areolae (Muñoz-Urias et al. 2008). This plant produces dark-red fruits that contain avonoids and betalains (Chavez-Santoscoy et al. 2009). Concerning the sex ratio, three kinds of population have been found: strictly hermaphrodite, dioecious (males and females), and trioecious. In the population we are actually working on, we found hermaphrodites, females and very scarce males (approx. 1-2%), so it can be considered subgynodioecious. The owering times of different sexual forms overlapped. Even when female owers were larger when measured in the same day, it was probably an effect of their earlier blossoming. The blossoming begins as early as in late February, spans through June and differs among zones. In the study area, we found the rst fruits in April and the last in November (Janczur et al. 2014 Plant Choice for Field Study. We used the same eld-work design as that used in Janczur et al. (2020b). During eldwork, we chose in each plant branches with the largest number of cladodes. We sampled the vegetative tissues from plants during eight months: on 10 March, the second group on 12 April, and so on, through 10 May, 14 June, 12 July, 10 August, 13 September, and 11 October, 2014 (Janczur et al. 2020a).
Plant Choice for Field Experiment. To test whether hermaphrodites of O. robusta relocate some secondary metabolites between apical and basal cladodes, and to test whether such relocation is light dependent, we covered some basal and apical cladodes of selected plants, to reduce photosynthetic activity of the covered parts ( Fig. 1A and B). The experiment involved plants of similar sizes and plant branches with similar number of cladode orders. To exclude the effects of cladode size and light intensity on the secondary compound production in the vegetative tissues, we found two branches of similar lengths on each plant, with the plain surface of both basal and apical cladodes turned in the same direction with respect to sun. As not all the plants possessed these characteristics on branches with the same cladode orders, the corresponding shaded and lighted cladodes were mostly of 3rd and 4th order, but in few cases, of 2nd and 3rd, or of 4th and 5th orders. We chose only non-ligni ed cladodes. We randomly assigned the branches for shaded basal and lighted apical cladode and vice versa treatments: that is, both combinations were placed on the same plant. To test the similarity of the cladodes' size and form for both experimental orders, we carried out a slope comparison for the relationship between cladode length and cladode width, between the basal and apical cladodes, using Wald test (W T -test statistic); when slopes were similar, we tested the signi cance of differences between intercepts; we used a numerical algorithm (Warton et al. 2006b) applied in SMATR software (Falster et al. 2003). We did not nd signi cant differences between slopes for this relationship between cladode levels (Likelihood ratio = 1.048, df = 1, P = 0.31).
The Effect of Shading Fabric on Cladode Temperature. To test whether the type of fabric used to shade cladodes had an effect on temperature near cladode surface, we performed an experiment: we measured the temperature on the surface of uncovered cladodes, and on cladodes covered with black cotton coarse fabric or with black cotton coarse fabric lined with white tulle, for twelve hours (8 a.m. through 7 p.m.). We used HOBO data logger model UA-002-64 Pendant Temp/Light. The distributions of the hour-dependent environmental temperatures for non-covered cladodes, and for cladodes covered with cotton coarse fabric or with cotton coarse fabric covered with white tulle were not signi cantly different from normal distributions at P ≤ 0.05. The paired t-test showed signi cant differences between the temperatures of non-covered cladodes and cladodes covered with black cotton coarse fabric (t = 2.88, P = 0.015), and non-signi cant differences between the temperatures of non-covered cladodes and cladodes covered with black cotton coarse fabric lined with white tulle (t = 0.84, P = 0.42). Given this result, we used the black cotton coarse fabric lined with white tulle for eld experiment ( The Effect of Shading on pH. Following the actually existing knowledge, a more acid average pH should occur in shaded cladodes due to a higher malic acid concentration in the shortage of light (Bieto and Talón 2008). This is a precursor of carbohydrates in the plant's tissue. That is because in darkness, CAM plants, as is the case of O. robusta, maintain their stomata open, while at full light they are closed in order to avoid desiccation. CAM plants absorb CO 2 from the air and x it as oxalic acid at night through PEP carboxilase. A second step in CAM metabolism is the transformation of oxalic acid into malic acid through malate deshidrogenase dependent of NAD + . Malic acid formed actively during the night is accumulated with high concentrations in vacuoles that may occupy 95% of cellular volume. At predawn, acid concentration may be very high (more than 0.2 M). At full light (stomata closed), malic acid is decarboxilated to pyruvate, forming CO 2 inside mesophyll cells that are processed by RUBISCO to form two molecules of 3-PGA (3 Phosphogliceric acid) that is the precursor of carbohydrates. We performed an experiment to examine whether cladode covering by fabrics played a similar role as natural shading.
Additionally, to show if pH is either different or similar in covered as in uncovered cladodes. If it is similar this means that cladodes can transfer the malic acid between them or that their photosynthetic activity is similar. We found in the eld 20 plants that bore branches with at least 3 orders of cladodes. The plants were different from those used in both eld experiment and eld study, but two treatments were the same as those from the experimental design of the eld experiment. If they bore more than one three-order branches, we randomly chose only one of them. We assigned a chosen branch from a selected plant to a given experimental group: 1 -Uncovered apical cladode, covered basal cladode, 2 -Covered apical cladodes, uncovered basal cladode, 3 -Uncovered apical and basal cladodes, and 4 -Covered apical and basal cladodes ( Fig. 1; Supplementary Table S2 and S3). On average, branches from each group were located on 6 plants. We covered the cladodes from the experimental groups for three days (January 3 through 5, 2020). During a sunny day, January 5, 2020, we took tissues for pH estimation using a stainless steel punch (Ø = 0.5 cm) to remove approx. 1 g tissue sample, without going through the cladode's width. We walked from one plant to another taking the samples from the cladode side opposite to the direction of the sun. As the plants were distributed over a large area, it was impossible to take the samples from different plants at the same time, so some combinations of treatment and cladode positions lacked some repetitions in time. However, the combinations of pH and sampling hour for different experimental groups encompassed the time interval of sample taking in similar way for all groups (Online Resource S2 and S3 on Janczur et al. (2020c)). We macerated each sample of vegetative tissue in a porcelain mortar. After maceration of one sample and before the next maceration, we rinsed the mortar with distillated water. We measured the pH of each sample using the Sper Scienti c 850062 LCD Test Tube pH Pen. We put UA-002-64 data logger on each cladode, to record the temperature and luminosity during the experiment.
Sampling of the Vegetative Tissue for Secondary Metabolite Determination. In both, eld study and eld experiment, of each plant, we removed three samples of vegetative tissue from cladodes belonging to each order in eld experiment, and to the apical and sub-apical orders, in eld experiment, using a stainless steel punch (Ø = 0.5 cm). We perforated the mid-section of the arc delimited by the border of the upper quarters of the cladodes, approximately 1 cm away from the edge (each sample contained the cuticle from both sides of the cladode). Also, we took two samples, approx. 10 cm away from the midsection. We placed the samples packed in Ziploc bags in a cooler with ice, and transported them to the laboratory in a portable refrigerator at − 20 °C. At the laboratory, they were stored at − 40 °C until extraction (Janczur et al. 2020b).
Extraction of Secondary Metabolites. We cut 2 g of each sample from one side of the sample containing the cuticle. We homogenized it with 35 ml of the methanol 100%, and ultrasonicated them in 6 l bath during 30 min at room temperature. We ltered the methanol extracts obtained, placed them in amber bottles to avoid degradation by light, and stored them at − 20 °C until processed (Janczur and González Camarena 2018).
Determination of Secondary Metabolites with HPLC. We followed the procedure described elsewhere (Janczur and González Camarena 2018). We used standards for salicylic acid (SA), 4-hydroxybenzoic acid (4-HBA), chlorogenic acid (CGA), and quercetin (QUE), all supplied by Sigma-Aldrich, and we obtained the following calibration curves: y i = 1109.4x i + 481.67, y i = 296.01x i + 133. 74 Statistical Analysis. To assess the effect of shading and sampling time on the pH of the cladodes' vegetative tissue, we used the generalized linear mixed effect model with the identity link function, Gaussian response variable distribution (pH), and independent variables, treatment (covered, uncovered), cladode position (apical, basal), and nested plant ID, using GLIMMIX procedure of the SAS software. To t the curves out of the data obtained out of the GLIMMIX procedure, we used the PLM procedure of the SAS software (SAS Institute Inc. 2016; SAS Institute Inc. 2017).
We performed the analyses of the secondary metabolites from the "cladode-point-of-view". To test the effect of the number of cladodes above a given cladode and the cladode order above soil level, on the probability of detection of cladodes bearing a given secondary compound, we performed a logistic regression using generalized linear mixed models (GLMM) with binomial error distribution and "logit" as link function in R using the glmmTMB package (Brooks et al. 2017). We tested the effect of month of study, the number of cladodes above a given cladode, and the cladode order above soil level, on the probability of detection of cladodes bearing a given secondary metabolite, for female and hermaphrodite cladodes, for the data obtained in a parallel study of us (Janczur et al. 2020a): a combination of levels of a larger number of factors allowed us for a more accurate estimation of the probabilities of occurrence of each secondary metabolite. With the model obtained in this analysis we modeled the probabilities to nd a given metabolite in a given cladode, for hermaphrodite individuals. Then, for these probabilities, we tted a Linear Mixed-Effects Model using the nlme package (Pinheiro et al. 2018), to estimate the differences between months of study, the number of cladodes above a given cladode, and the cladode order above soil level. We treated the month as discrete variables, and the other traits, as continuous variables, since the latter data were ordinal (Sperandei 2014). Models were tted using in R (RCoreTeam 2020; RStudioTeam 2020).
For the experiment, we applied an ANOVA model with the variables "conditions" (basal or apical cladode) and treatment (covered or uncovered) and the interaction between them. For 4-HBA and CGA, we compared the concentrations of these metabolites. As the data for QUE and SA contained many zeros, we used the proportion of cladodes bearing these metabolites, instead. We compared the means for the main variables as well as for all the combinations condition × treatment (SAS Institute Inc. 2017).
We related the concentrations of different secondary metabolites using standardized major axis functions (SMA) ( . When the number of cladodes containing a given metabolite was very low, we included into analysis only the cladodes containing the substances. We performed the outlier analysis to exclude the atypical concentrations (Barnett and Lewis 1994). To test the similarity of slopes of the relationships between either concentrations of two metabolites or cladode position/treatment, we carried out a slope comparison using Wald test; when slopes were similar, we tested the signi cance of differences between intercepts applied in SMATR software (Falster et al. 2003;Warton et al. 2006a). These analyses were performed in R (RCoreTeam 2020; RStudioTeam 2020).

Results
The Effect of Cladode Traits and Sampling Month on the Presence of Secondary Metabolites in Vegetative Tissue. We found a signi cant effect of May, July, August, and September on the proportion of cladodes with 4-HBA: it was signi cantly higher in March than in July, August, and September; higher in May than in September; higher in June than in July, August, and September; higher in April than in August and September; higher in October than in July, August, and September (Fig. 3A). May, August, and September had a signi cant effect on the proportion of cladodes with CGA: it was higher in August than in March, April, June, July, and October; higher in May than in March, June, July; lower in September than in March, April, May, and August; higher in October than in June and September ( The effect of shading on pH of the vegetative tissues -The effect of temperature, humidity and sampling hour were correlated, so we adjusted the model for the daytime as the independent variable. The effects of branch type (treatment) and of cladode condition (covered/uncovered) were non signi cant. Rather daytime had an affect on pH of the cladodes' vegetative tissue: pH decreased monotonically until 1 p.m., approximately at 1 p.m. reached its minimum (the highest acidity) and then grew monotonically until approx.  Table 3).
There was a tendency towards a higher CGA concentration in both lighted and apical cladodes (Fig. 7 A and B). Light increased CGA concentration in apical cladodes with respect to both lighted basal (Fig. 7 C), shaded apical cladodes (Fig. 7 D), and shaded basal cladodes  Table 3).
The proportion of shaded cladodes did not differ signi cantly from the proportion of lighted cladodes bearing QUE (Fig. 8 A). Rather the position of a cladode determined the presence of QUE in their tissues: basal cladodes bore QUE with higher probability than apical cladodes (only two cladodes bore this secondary metabolite) (Fig. 8 B). Shading of basal cladodes increased the probability of production of QUE with respect to lighted (Fig. 8 C and E) and to shaded apical cladodes (Fig. 8 E and H). Shading did not change the proportion of cladodes bearing QUE either in apical (Fig. 8 D) or in basal cladodes (Fig. 8 G), and increased this proportion in basal cladodes with respect to shaded apical cladodes (Fig. 8 F and H; Table 3).
Contrary to the eld study, in the experiment we found SA in a few samples, however the proportion of cladodes bearing it was not associated either to cladode position of treatment ( Fig. 9 A through B). Even when the differences between the combinations of cladode positions and experimental treatments were not signi cant, some tendencies could be observed. For example, light increased the difference in the proportion of cladodes bearing this metabolite between apical and basal cladodes ( Fig. 9 C and D), because it increased the proportion of apical cladodes, and decreased the proportion of basal cladodes bearing it. The proportion of cladodes bearing SA was higher for lighted apical than for shaded basal, and higher for shaded apical than for lighted basal ( Fig. 9 E and F). It was also higher in shaded basal than in lighted basal cladodes (Fig. 9 H). It was similar in shaded apical and shaded basal cladodes ( Relationship between the concentrations of secondary metabolites -Because of the presence of many null concentration values for SA, we analyzed only the SA-positive data, assuming that to test the relationship between two metabolites at least one of them should be present in the tissues. We found positive and signi cant relationships between the concentrations of 4-HBA and SA for all data, and for lighted cladodes. We removed from the analysis two outliers. This relationship for shaded cladodes was non-signi cant (Fig. 10 A). The difference among slopes of SMAs for lighted and shaded cladodes was on the edge of signi cance. The relationship between the concentrations of these two metabolites was also positive for both apical, and basal cladodes, however both SMA equations were not signi cant, even when they explained 36% and 75% of the variability, respectively (Fig. 10 B; Table 4; Online Resource S8 on Janczur et al. (2020c)).
We found a positive relationship between the concentration of 4-HBA and CGA for all shaded and lighted apical cladodes. We removed from the model two outliers (Fig. 10 C and D). This relationship was positive for all apical cladodes and negative for basal cladodes (Fig. 10 E). The difference between slopes for apical and basal cladodes was non-signi cant (Table 4). In basal cladodes, the relationship between the concentrations of these two metabolites was negative and independent of light treatment, even when the dispersal of the data was too large and thus, the relationship was statistically non-signi cant ( The relationships between the concentrations of SA and CGA were negative for all lighted vs. shaded and for apical vs. basal cladodes (one outlier was removed from the analysis). All relationships were non-signi cant (Fig. 10 G and H). Additionally, in apical cladodes, the concentrations of CGA were very low and the concentration of SA decayed very quickly with a small increase of CGA, or, in other words, the concentration of CGA decreased very slowly with the increase of the SA concentration. The concentrations of CGA for apical cladodes did not overlap with the concentrations of SA for basal cladodes: the slope for the apical cladode differed signi cantly from the slopes for basal and pooled cladodes (Fig. 10 H; Table 4; Online Resource S10 on Janczur et al. (2020c)).
The relationship between the concentrations of QUE and 4-HBA was negative and non-signi cant for all lighted and shaded basal cladodes (Fig. 11 A; Table 4; Online Resource S11 on Janczur et al. (2020c)).
The relationship between the concentration of QUE and CGA was positive and non-signi cant for lighted basal and pooled (lighted and shaded) cladodes. We removed from the analysis two outliers (Fig. 11 B; Table 4). This relationship for shaded basal cladodes was negative and non-signi cant (Fig. 11 B; Table 4). The slopes for the relationships for shaded cladodes differed signi cantly from the slopes for the lighted and for the pooled (lighted and shaded) basal cladodes (Table 4; Resource S12 on Janczur et al. (2020c)). We compared here only the basal cladodes for the relationships involving QUE, because only two apical cladodes bore this secondary metabolite (Resource S5 on Janczur et al. (2020c)).

Discussion
Even when the comparisons of cladodes' conditions and positions for 4-HBA and CGA were statistically not signi cant, there were some tendencies in the differences between cladode state and position that could be observed. For example, the results suggest that lighted apical cladodes produced 4-HBA and sent it to basal cladodes when they could not produce it by themselves (i.e. when shaded). That is because the incidence of this secondary metabolite was similar in lighted apical cladodes and in shaded basal cladodes, and similar in shaded apical and in lighted basal, but lower than in the former situation. small, but some less common higher concentrations allowed extending the prediction function for such concentrations. Low concentrations of 4-HBA for the relation SA vs. 4-HBA and moderately low concentrations of this secondary metabolite for the relation CGA vs. 4-HBA were associated with higher concentrations of either SA or CGA in lighted than in shaded cladodes, and vice versa for higher concentrations of 4-HBA. Low or moderately low concentrations of 4-HBA were referred to the interval of all the data for this secondary metabolite.
Higher slope of the relationship CGA vs. 4-HBA for lighted than for shaded apical cladodes means that for most concentrations of 4-HBA, lighted cladodes produced higher concentrations of CGA than shaded cladodes. This result, together with a negative non-signi cant relationship between these two metabolites in basal cladodes and with the observed decrease of CGA concentration in basal cladodes is parallel with the tendency obtained in the comparison of the concentrations of CGA for different cladodes states that suggested an autonomous production of this secondary metabolite by apical cladodes, without a ow from neither lighted nor shaded basal cladodes. The lack of signi cant differences between cladode states in the comparison of CGA concentration, for example between lighted apical vs. shaded apical cladodes, was probably an effect of the fact that for lower or higher 4-HBA concentration, the relationship between the concentrations of both metabolites was inverse and thus, during the comparisons of the CGA concentrations between different cladodes states, these concentrations were averaged for high and low 4-HGA concentrations.
An analogous phenomenon existed for the relationship between SA and 4-HBA, con rming the observed tendency for higher concentration of SA in lighted apical than in lighted basal, and in lighted apical than in shaded apical cladodes were mainly the outcome of the autonomous production of this secondary metabolite by apical cladodes.
Considering that 1) cladodes are growing more slowly at least during the event of shading and thus, their tissue is more di cult to recover from damage (they are more valuable from both ODH and RAH point of view, and thus, should be better defended); 2) lighted cladodes are more valuable in terms of higher photosynthetic activity (they are more valuable from the ODH point of view and should be better defended) but at the same time are growing faster and thus, their tissue is easier to recover from damage (they are less valuable, from the RAH point of view and should be less defended); and provided that 3) the relationships between pairs of secondary metabolites have different slopes for different cladodes' states, this is di cult to nd a congruence between the concepts 1) and 2) and the predictions of either RAH or ODH.
The results of the experiment concerning QUE were concordant with the general tendency of the results of the eld study, since basal cladodes from the experiment and cladodes from the eld study bearing more daughter cladodes above, that is, the older ones contained this metabolite with higher probability. Both results show that rather the basal position of a cladode determined a higher probability of presence of QUE in its vegetative tissue than the access to light. However, higher and statistically signi cant probability of detection of QUE in shaded basal cladodes when apical cladodes were lighted, compared to lighted basal cladodes when apical cladodes were shaded and to lighted basal cladodes when apical cladodes were also lighted, suggesting that shaded cladodes received QUE or its precursor from apical cladodes. Since the lighted condition of basal cladodes did not increase the proportion of either shaded or lighted apical cladodes bearing QUE, it probably meant that basal cladodes did not transfer QUE or its precursors to apical cladodes. As the proportion of basal cladodes with QUE was lower under light than in the shade, it meant that 1) QUE was transferred from basal (penultimate order) to the cladodes parental to them, or, 2) QUE was unstable in the light and was decomposed. Similar pH in shaded and lighted cladodes suggests that differences in the probability of detection of QUE were not an effect of higher concentration of the malic acid, precursor of most secondary compounds (Bieto and Talón 2008). This result reinforces the conclusion about the production of QUE in apical cladodes and its transfer to the basal cladodes. Higher incidence of this metabolite in older cladodes obtained in eld study, and the existence of signi cant differences between some combinations of treatment and position were a result of QUE transfer from higher to lower order cladodes that decreased the variability between cladodes and thus, smaller slope of the relationship QUE incidence vs. cladode age. This metabolite was probably not produced to promote cell growth, since higher growth intensity occurs in apical, and thus, younger cladodes: in such case the ow of the secondary metabolite should occur in the opposite direction. Neither was it produced to protect the plant from UV radiation or to rest the effect of free radicals: in both cases, QUE should be moved to both apical and lighted cladodes. These conclusions are parallel with the conclusions of our previous eld study in which we did not nd in hermaphrodite individuals signi cant relationships between the incidence of QUE and of meteorological variables that affect both cell growth (air temperature and humidity) and the response against abiotic factors like UV (Janczur and González Camarena 2018). Independently of the proximate causes, the results of both eld study and eld experiment do not support the predictions of the ODH, since we could not show the existence of signi cantly higher QUE incidence in younger cladodes even when a mechanism of transport of this secondary metabolite between cladodes existed. They rather support RAH, since slower-growing plant parts in our study had high levels of QUE. This was an analogous result to that predicted by RAH for plants growing in less productive environments (that is, slower-growing). The results showed that the mechanism of the plant is to maintain either similar incidence of this metabolite in cladodes from all levels or to maintain its higher incidence in older cladodes, rather than to transfer them to the youngest cladodes. This is why we proposed in our previous study the use of a relative value of plants' organs (in this case cladodes) in terms of tness rather than their age (Janczur et al. 2020b). The hierarchical and frequently unilinear architecture of the individuals of O. robusta in the population studied here makes the plants sensitive to damage. A loss of lower-order cladodes (closer to soil) brings about a loss of all higher-order cladodes, where reproductive biomass is produced. So, the defense of older cladodes would be justi ed from an evolutionary standpoint. However, we are aware that in plants with more complex architecture, it is di cult to decide which vegetative organ contributes more in terms of tness: probably, plant parts that bear reproductive organs. In Opuntia robusta cladodes of any age can potentially produce owers, however, reproduction is much more probable on apical cladodes.
No study exists concerning the movement of avonoids between cladodes of any species of Opuntia genus. There is a controversy concerning the possibility of transfer of phenolic compounds from the site of their manufacture to other tissues (Buer et al. 2007;Waller and Nowacki 1978), however it is known that the synthesis of the avonoid biosynthetic enzyme is light dependent (Jenkins et al. 2001;Kubasek et al. 1998;Pelletier and Shirley 1996). Some researchers state that they are synthesized in the cells in which they accumulate and serve local functions (Peer et al. 2001); however, the results from the study of Buer et al. (2007) showed in Arabidopsis thaliana that avonoids can move long distances from the light-grown shoot tissues to the shade-grown roots. The authors also suggested the active movement of avonoids (among them QUE) because of their tissue-speci c location, their cell-to-cell movement, and their unidirectional movement when applied midroot. If the movement is diffusion mediated, the avonoid distribution is less localized and their movement is bidirectional. Our results showed that the movement of QUE was rather unidirectional and light dependent. The stability of QUE is low and its concentration is inversely correlated to temperature and to pH: its half-life decomposition may last less than one hour at high temperature and high pH, and as long as few hours at low temperature and lower pH (Liu and Zhao 2019; Wang and Zhao 2016). Therefore, it had to either disappear or decrease its concentration (incidence) in the shade, although this did not occur in our study. This reinforces our prediction that in our study, QUE was transferred from lighted apical cladodes to shaded basal cladodes.
In eld study, higher order cladodes or the ones that hosted less orders of daughter cladodes contained CGA with higher probability than cladodes basal to them. On the other hand, in the experiment, light increased the difference in the concentration of this metabolite with respect to basal cladodes (both, shaded and lighted), but this effect was statistically not signi cant. This result showed that apical cladodes produced CGA autonomously, and did not receive it from the basal cladodes. Signi cant negative relationship between the cladode age (number of daughter cladodes above) and the incidence of CGA obtained in the eld study is an effect of the lack of relocation of this secondary metabolite between cladode orders, autonomous production of this compound by higher-order cladodes showed in the experiment, and of low stability of this compound showed in other studies (Gil and Wianowska 2017). Higher CGA concentration in younger (higher-order) cladodes in eld study was not an effect of a relocation of this metabolite from lower order to higher order cladodes; that is, plants did not have a "policy" to protect more photosynthetically valuable plant parts, but rather these parts had a "policy" of not sending this metabolite to lower level cladodes. The results of the relationships between the concentrations of CGA and other secondary metabolites con rmed the conclusion from the comparisons between cladodes' states. For example, at states that promote a lower photosynthesis rate (basal, shaded, or shaded basal cladodes), the concentration of CGA either decreased with the increase of, or was non-signi cantly related with, the other secondary metabolites.
The outcome of this result is concordant with the predictions of ODH although the rhetoric is different: plant cladodes did not defend more valuable cladodes, but rather, these cladodes defended themselves. It was also possible that the plant used CGA to arrest the negative effect of either UV or free radicals, because of a tendency of higher CGA concentration in lighted cladodes, but in such case also basal cladodes should be protected, since a negative effect of this stress factor on basal cladodes would affect apical cladodes, and thus, future reproductive biomass. Also, in eld study, higher concentration of CGA in younger cladodes could be an effect of the participation of this metabolite in cell  (Boerjan et al. 2003;Gamborg 1967). As in our previous study (Janczur et al. 2020b) we did not nd a relationship between the incidence of CGA or meteorological factors that might affect growth (temperature, humidity) and increase the negative effect of UV (radiation intensity), we suggest that in this population of O. robusta, this metabolite plays a defensive role. In any case, the response to non-biotic stress factors should be similar to the response to biotic factors, that is, "respond when you are attacked". Our study showed that, contrary to the predictions of ODH, when CGA coexisted with SA, older tissues contained higher concentrations of this metabolite. We are not sure why it occurred, however this result was contradictory to both the results of eld study and the non-signi cant tendency in the experiment, although as the coexistence of SA and CGA was non-frequent, these data did note affect the general pattern. Independently of a low dataset for the relationship between the concentrations of SA and CGA, we can conclude that SA was not an elicitor of the CGA. Larger dataset would show if the coexistence of these two secondary metabolites is associated to a lack of relationship between them or may otherwise suggest that they are competing for the same precursor. High percentage of model explanation for basal cladodes suggests that a small increase of the sample will bring about a signi cant relationship between the concentrations of these two substances.
The positive cladode-position-independent relationship between the concentrations of 4-HBA and SA in eld experiment was probably an effect of the conversion of the former into the latter. If it had been a case of a promoting action of the SA, this metabolite would have been more common and 4-HBA an either less or equally common metabolite. We are not sure why we did not nd SA in the eld study and found it during experiment. If it had been an elicitor for the production of other substances in response to the stress factor (fabric bag covering cladodes) as mentioned in several studies (Beckers and Spoel 2006;Mendoza et al. 2018;Sudha and Ravishankar 2003), it would have had a positive relationship with other secondary metabolites. However, it had negative non-signi cant relationship with some secondary metabolites as determined in this study. The model of positive relationship between the concentrations of SA and 4-HBA explained a considerable proportion of variance although the number of data points was low. Additionally, three out of ve regression lines were signi cant. The nonsigni cant relationships also explained a considerable proportion of variance (36%, 39%, and 75%) suggesting that additional data had brought about a statistical signi cance.
It is not clear why 4-HBA was converted into SA in only few samples (if it was a case), since the pathway involving 4-HBA as a precursor of the latter compound is very common (Lee et al. 1995). Another pathway is also possible, i.e. phenylalanine ◊ trans-cinnamic acid ◊ benzoic acid ◊ SA, together with a parallel pathway phenylalanine ◊ trans-cinnamic acid ◊ 4-coumaric acid ◊ 4-HBA. Following this pathway, the synthesis of 4-HBA may compete for precursor (trans-cinnamic acid) with SA. As SA is known mainly for its signaling properties in the induction of pathogenesis-related proteins (Bennett and Wallsgrove 1994;Pierpoint 1994;Pieterse et al. 2012) and cases of signaling function of 4-HBA are scarcely known (Schnitzler et al. 1992;Tan et al. 2004;Ulmasov et al. 1994), it is possible to put forward the following hypotheses: 1) 4-HBA was produced in O. robusta for defensive purposes; SA was produced only in few cladodes either because the response against herbivores was not SA-signaling-dependent or there were no other stress factors requiring SA signaling; 2) The production of 4-HBA was higher in priority than that of SA synthesis, thus, sharing the same precursor, the plant lacked su cient resources to produce both of them. If the conversion of 4-HBA to SA in the cladodes were a phenomenon, it would mean that the latter secondary metabolite was not produced in response to UV, since in either the lighted or the apical cladodes, the concentration of SA increased slower than in either the shaded or the basal cladodes. This effect should be con rmed in larger sample: because of a low incidence of this secondary metabolite, only a few cladodes had high concentration of 4-HBA.
In the present study 4-HBA and QUE showed dynamics in hermaphrodite cladodes during eight-months study period. Both metabolites showed a nearly monotonic dynamics: the incidence of the 4-HBA decreased from March through September with the exception of October, and the incidence of QUE increased from March through October, with the exceptions of April and September. This result is parallel with the result of the eld experiment that showed a negative non-signi cant relationship between these two metabolites. This dynamic occurred probably because QUE was either relocated among cladodes from different ages or cladodes stopped producing them. This reallocation may have produced the following effect: if both mutually dependent cladodes contained a given metabolite, a complete relocation from an apical to a basal cladode would bring about a lower proportion of apical cladodes bearing that given secondary metabolite. It would produce the same effect as an incomplete relocation from apical to basal cladode, together with the lack of production of new secondary metabolite and/or, decomposition of the remainder metabolite. The transfer of QUE from apical to basal cladodes shown in eld experiment was concordant with this result. Higher QUE incidence in older cladodes shown in eld study points out that during the season the incidence of this metabolite increased because the proportion of older cladodes bearing it also increased. As the tendency of 4-HBA to move in the same direction was slight in the eld experiment, this dynamics was probably produced by the higher mobility of QUE and the dynamics of the incidence of 4-HBA followed inversely this tendency because of the reasons already described (i.e. competition for the same precursor). Independently of the proximate causes, this result is not concordant with ODH because the proportion of cladodes bearing QUE was lower when tissues were younger and it increased throughout season when tissue was older. At the same time, it was concordant with RAH, because the prevalence of QUE was higher when tissues' growth rate was lower because they were older. The contradiction was inverse for 4-HBA with respect to ODH and RAH. The probability of detection of CGA in cladodes of O. robusta did not follow the predictions of any of the hypotheses mentioned, because it did not follow monotonic seasonal dynamics. Its dynamics were mainly a result of an atypically high prevalence in August and low prevalence in September: its incidence in other months did not follow a clear tendency. This result suggests either 1) a lower mobility and/or, 2) a higher stability of this metabolite in the tissues of O. robusta. As the study of Gil and Wianowska (2017) evidenced a low stability of this metabolite, and our experiment showed that CGA is not transferred from apical to basal cladodes, the atypical CGA incidences occurred because of a higher insulation in August and lower insulation or decomposition of this compound in September.
Even when several studies showed the involvement of SA in resistance and tolerance to many abiotic factors (i.e. ozone, UV radiation, heat, cold (Dempsey and Klessig 2017;Horváth et al. 2007; Rivas-San Vicente and Plasencia 2011; Yuan and Lin 2008)) in our study the incidence of SA was similar at light than in the shade. It meant that the production of this secondary metabolite was not a response against short-term action of UV. However, this metabolite could probably take part in tissue growth, since the proportion of apical cladodes was statistically nonsigni cantly higher in apical cladodes. It is di cult to conclude whether the lack of signi cant differences in SA incidence between cladode positions and experimental treatments was a result of the lack or of only a low-intensity transport of this secondary metabolite between cladodes from different levels, since this process has not been studied in Cactaceae. There is no unique transport mechanism in different plant species ( Takanaga et al. 1994). In our study, there was no signi cant difference in pH between either the apical and basal or covered and uncovered cladodes. This lack of differences in pH was probably responsible for the lack of signi cant differences in the incidence of this SA in apical/basal and shaded/lighted cladodes. In other words, there existed only a reduced transfer of SA between cladodes, because of the lack of difference in pH between cladodes. As SA is known for its elicitor action for other secondary metabolites, the absence of biotic stress was probably responsible for the small proportion of cladodes bearing it. As experimental treatments were a kind of stress, the presence of tulle bags exerted a stressful effect on cladodes, similar to that exerted by non-biotic factors. No one out of the three other secondary metabolites studied here responded to the presence of SA in a manner that could be perceived as elicitation. Even more, the concentrations of CGA and SA were associated inversely, probably due to the competition for the same precursor, i.e. phenylalanine (Chen et al. 2009;Dempsey et al. 2011;Tuan et al. 2014).
The negative (non-signi cant) relationship between the concentrations of QUE and 4-HBA may be explained in two ways: 1) They shared of a common precursor for the synthesis of both secondary metabolites (Bennett and Wallsgrove 1994) that is, enhanced synthesis of 4-HBA brought about a decrease of the concentration of QUE (or, vice versa), or 2) The negative relationship is a result of the fact that high concentrations of QUE corresponded to basal (mainly, shaded basal) cladodes, and that high concentrations of 4-HBA corresponded to lighted apical and to shaded basal cladodes. Analogously, the negative (non-signi cant) relationships between the concentrations of QUE and CGA for shaded basal cladodes may be explained by 1), or by the fact that for shaded basal cladodes, high concentrations of QUE cooccurred with low concentrations of CGA. On the other hand, in lighted basal cladodes low and high concentrations of both QUE and CGA cooccurred. As these relationships were non-signi cant, it is di cult to discuss the representativeness of this result at population level. However, it showed that both secondary metabolites responded to the presence of light in independent manner and were an effect of very short exposure to experimental treatments. To the best of our knowledge, CGA is not an elicitor of the secondary metabolites studied here, since only one case of an elicitor activity of QUE is actually known (Mahady and Beecher 1994), so a possible positive relationship between the concentrations of QUE and CGA in lighted cladodes was rather an effect of an independent production of secondary metabolites as a response against different stress factors.
QUE and SA did not co-occur. This means that SA did not promote the production of QUE. Neither did it promote the production of CGA, since the relationship between theses two metabolites was negative and statistically not signi cant. It rather means that these two secondary metabolites competed for the same precursors.
Regardless of the proximate factors that explain the relationships between the concentrations of different secondary metabolites studied here (i.e. elicitor effect, or competition for the same precursor) either the incidence or the concentration of different secondary metabolites maintain a rather complex, frequently negative, relationship with the analogue variables of other secondary metabolites, and thus, this is di cult to expect a congruent pattern of their presence in younger or faster-growing plants' vegetative tissues. Additionally, there does not exist a general pattern of the between-cladode dynamics for different secondary metabolites. Rather, each of them is produced for a particular purpose and its incidence or dynamics depend on the goal it was synthetized for, and of either the competition for the same precursor or a possible elicitor function. Probably, the whole response (all chemical and physical defenses) of a plant against the action of herbivores should be studied to propose a new hypothesis concerning plant defense. However, this is di cult to achieve, since a matrix of metabolites and of the herbivores against which these metabolites were produced should be studied in such case. A reductionist approach may be used, when a speci c secondary metabolite is used for a speci c herbivore. Neither does there exist a uni ed hypothesis concerning plant's response to the joint action of herbivores and abiotic stress such as UV. If a secondary metabolite protects a plant against both factors, its concentration should increase in the organs where the stress factor is exerting the main damage. However, if different substances sharing the same precursor are used for two different kinds of defenses (i.e. herbivory and UV), the investment to herbivory may take away resources from the defense against the negative effect of UV or vice versa. In such case, the resources will be allocated to the process that affects more severely the reproductive success.
Since in the experiment the cotton coarse fabric did not change signi cantly cladodes' temperature, and the pH among the covered and uncovered cladodes did not differ, it meant that covered cladodes did not open their stomata, because the temperature of cladodes' surface was similarly high during daytime as exposed to the environmental temperature. So, the concentrations of the malic acid were similar in both experimental treatments and in both cladodes' positions. Therefore, we consider that any difference or the lack of differences in secondary metabolite concentration between treatments was an effect of either metabolite transfer between cladodes and not of a different concentration of the precursor (malic acid).

Conclusions
The ODH predicts that younger, more valuable plant organs should be better defended. In this study, 4-HBA and CGA followed the predictions of this hypothesis. QUE followed an inverse tendency. We did not nd SA in the eld study even when we found it in the eld experiment. SA did not show any tendency related either with the position of the cladode (apical vs. basal) or the treatment (light vs. shade) however, its concentration was positively correlated with the concentration of 4-HBA. As we detect SA only in a low proportion of cladodes and 4-HBA in all cladodes, we hypothesize the conversion of the latter metabolite to the former one. For the same reason, SA probably did not serve as elicitor. Also, contrary to the prediction of ODH, CGA presented lower concentrations in apical than in basal cladodes when co-occurred with SA. The incidence of QUE did not hold the predictions of the ODH, since in the eld experiment this secondary metabolite was transferred from apical to basal cladodes. This transfer was probably responsible for the existence of a tendency for a higher incidence of QUE in older than in younger cladodes obtained in the eld study. A lower incidence of CGA in older cladodes was an effect of an autonomous lightdependent production of this secondary metabolite in apical cladodes and its quick decomposition in basal cladodes. Also, the results of eld experiment suggested an autonomous synthesis of SA in apical cladodes, without relocation from basal to apical cladodes. High mobility of QUE between cladodes, a tendency of 4-HBA to be transferred between cladodes, and an autonomous way of CGA production in apical cladodes showed in eld experiment together with its quick decomposition rate shown in other studies that occurred in basal cladodes, explained why in the eld study carried out in another study of ours (Janczur et al. 2020b), 4-HBA and QUE presented a statistically signi cant, almost monotonic dynamic over eight-months, contrary to CGA.
In the present study, we found contradictions in the premises and outcomes of both the RAH and the ODH and suggests that a new hypothesis concerning the plant defense against stress factors should be proposed.