Plant material

The plants used corresponded to two V. planifolia chemotypes (QI and QVI) obtained from different genotypes (Genotype VI and Genotype III, respectively) (^{Herrera-Cabrera, Hernández-Ruíz, & Delgado-Alvarado, 2016}), which were collected in acahual and shade mesh systems at locations within the Totonacapan region of Mexico (Table 1). During the flowering period (from April to May 2013), one-meter-long cuttings were collected from the base to the tip to obtain leaves, stems and flowers. In order to obtain fruits with the same state of maturity, the flowers of the two chemotypes were manually pollinated and marked. The fruits were harvested 32 weeks after pollination and were subjected in a uniform way to a traditional curing process (^{Xochipa-Morante, Delgado-Alvarado, Herrera-Cabrera, Escobedo-Garrido, & Arévalo-Galarza, 2016}).

Preparation of extracts for quantification

Total phenolic compounds, total tannins, hydrolysable tannins, condensed tannins and flavonoids: Extracts from each fresh tissue (leaf, stem, flower, green fruit and cured fruit) were prepared in methanol at a concentration of 50 mg·mL^-1. The samples were placed in an ultrasonic bath (AutoScience AS5150B) for 30 min; subsequently, the extracts were filtered and stored in glass vials at -20 °C until analysis.

Saponins: Five 5 mL of distilled water were added to 1 g of fresh sample of each tissue and taken to an ultrasonic bath (AutoScience AS5150B) for a period of 20 min at 10-min intervals.

Total triterpenes: Five mL of 70 % ethanol (v/v) were added to 1 g of fresh sample and left to macerate for 10 h. The extract was centrifuged at 3,344 g for 5 min and the supernatant was evaporated in a rotaevaporator (Heidolph, Laborota 4000); subsequently, the plant residue was washed four times with butanol:water (1:1), where the aqueous phase was discarded and the organic one was evaporated in a rotaevaporator. Ethyl ether was added to the residue and kept at 4 °C for 12 h. Finally, the residue was dried and stored at -20 °C for further analysis.

Prior to quantification, the moisture content of all tissues was determined in a thermobalance (Ohaus MB45) to express the concentration of secondary metabolites based on dry matter (DM).

Quantification of phytochemicals

Total phenolic compounds: The analysis was carried out using the method described by ^{Singleton and Rossi (1965)}, with some modifications. First, 250 μL of 50 % Folin-Ciocalteu reagent (v/v) were added to 50 μL of the extract with methanol and then the mixture was left to stand in the dark for 8 min; subsequently, 1.25 mL of 5 % sodium carbonate (v/v) were added and it was again left to stand for 30 min in the dark at room temperature. The absorbance reading of the mixture was performed at 725 nm in a UV-Vis spectrophotometer (Evolution 300, Thermo Scientific). The results were expressed in mg equivalents of tannic acid per g of DM, for which a standard curve (y = 2.0364x - 0.0016, R² = 0.9942) was obtained with tannic acid (Sigma Aldrich).

Total tannins: They were determined using the method described by ^{Makkar, Bluemmel, Borowy, and Becker (1993)}. Two hundred mg of PVPP (Polyvinylpyrrolidone), 2 mL of methanolic extract and 2 mL of distilled water were placed in a test tube previously lined with aluminum foil. The mixture was vortexed and left to stand for 15 min in the dark at 4 °C. Subsequently, it was again stirred and filtered, and an aliquot of 150 μL was taken and brought to 1 mL with distilled water. Next, 50 μL of this extract were taken, 450 μL of distilled water and 250 μL of 50 % Folin-Ciocalteu reagent (v/v) were added, and it was stirred and left to rest for 8 min, after which 1.25 mL of 5 % Na₂CO₃ (v/v) were added. It was again stirred and left to stand under dark conditions at room temperature for 30 min. Finally, the absorbance reading was performed at 725 nm in a UV-Vis spectrophotometer (Evolution 300, Thermo Scientific). The results were expressed in mg of tannic acid (Sigma Aldrich) per g of DM, based on the equation provided by the standard curve (y = 2.0977x - 0.0241, R² = 0.9919).

Condensed tannins (proanthocyanidins): It was carried out according to the method proposed by ^{Porter, Hrstich, and Chan (1986)}, with some modifications. Three mL of 1-butanol-HCl and 100 μL of 2 % ferric chloride (w/v) in 2 N HCl were added to 500 μL of the methanolic extract. The solution was placed in a water bath at 80 °C for 15 min. Finally, the absorbance at 550 nm was recorded in a UV-Vis spectrophotometer (Evolution 300, Thermo Scientific). The concentration of condensed tannins was calculated according to the following equation and the result was expressed in mg. g^-1 DM:

where A is the absorbance at 550 nm, % DM the percentage of dry matter and 78.26 the correction factor.

Hydrolysable tannins: The content of hydrolysable tannins was calculated by means of the difference between total tannins and condensed tannins. The results were expressed in mg·g^-1 DM (^{García-Ferrer, Bolaños-Aguilar, Lagunes-Espinoza, Ramos-Juárez, & Osorio-Arce, 2016}).

Total flavonoids: They were determined using the methodology proposed by ^{Chang, Yang, Wen, and Chern (2002)}, with some modifications. First, 500 μL of the extract with methanol, 1.5 mL of 80 % ethanol (v/v), 100 μL of 20 % aluminum chloride hexahydrate solution (w/v), 100 μL of 1 M potassium acetate, and 2.8 mL of distilled water were placed in a test tube. The mixture was stirred and incubated for 30 min at room temperature. The absorbance reading was performed at 415 nm in a UV-Vis spectrophotometer (Evolution 300 Thermo Scientific). The results were expressed in mg equivalents of quercetin (Sigma Aldrich) per g of DM, from the equation obtained from the standard curve prepared with quercetin (y = 6.0986x - 0.0004, R² = 0.9952).

Saponins: They were determined by the sulfuric acid-vanillin method described by ^{Hiai, Oura, and Nakajima (1976)}. Five μL of the extract, 95 μL of water, 1 mL of concentrated sulfuric acid and 100 μL of a fresh solution of 8 % vanillin (w/v) in ethanol were mixed in a test tube. Subsequently, the mixture was incubated at 60 °C for 20 min and then placed in an ice bath. The absorbance was read at 544 nm in a UV-Vis spectrophotometer (Evolution 300, Thermo Scientific). As a reference, a standard curve (y = 0.0033x + 0.0786, R² = 0.9919) was obtained with quillaja saponin (Sigma Aldrich). The results were expressed in mg·g^-1 DM.

Total triterpenes: The colorimetric method with vanillin-acetic acid described by ^{Fan and He (2006)} was used, with some modifications. First, 100 μL of 5 % vanillin (w/v) and 400 μL of perchloric acid were added to 0.5 mg of extract; the mixture was kept at 60 °C for 15 min. The tubes were left to cool to room temperature and 2.5 mL of acetic acid were added to them. The absorbance was obtained at 550 nm in a UV-Vis spectrophotometer (Evolution 300 Thermo Scientific). The results were expressed in mg·g^-1 DM from a standard curve (y = 0.0125x - 0.0274, R² = 0.9949) prepared with oleanolic acid (Sigma Aldrich).

Statistical analysis

The effect of plant structure, chemotype and management system was analyzed as a source of variation on the concentration of phytochemicals in V. planifolia. Twenty treatments with nine replicates of different plants were evaluated, with a total of 180 samples. The experimental design used was a factorial arrangement with three factors (two management systems, two chemotypes and five plant structures) (^{Statistical Analysis System [SAS], 2002}) and the comparison of means for each source of variation was performed using Tukey’s test (^{SAS, 2002}). In addition, in order to know the association between environmental and phytochemical (content and type) variables, a canonical correlation analysis was performed according to the PROC CANCORR procedure (^{SAS, 2002}).

Effect of the management system on phytochemicals

Within the management system factor, all phytochemicals showed significant statistical difference (P ≤ 0.05) in their concentration, except for flavonoids that showed no variation between management systems (Table 3). In the acahual system, with 60-70 % intercepted radiation, the vanilla tissues had a higher concentration of total phenolic compounds, total tannins, hydrolysable tannins, condensed tannins and total triterpenes, while in shade mesh, with 30-45 % intercepted radiation, the concentration was only higher in saponins (Table 3), possibly because this condition favors the synthesis of steroidal saponins that are found almost exclusively in monocotyledonous plants (^{Sparg, Light, & van Staden, 2004}), like vanilla. The difference in the concentration of phytochemicals between management systems could be due to particular environmental characteristics, such as temperature, relative humidity, elevation and luminosity (intercepted); even this last factor in some epiphytic species has a great impact on the production of phytochemicals (^{Cach-Pérez, Andrade, & Reyes-García, 2014}).

Although soil nutrient content is a critical factor in the physiological responses of most plants, in species of hemiepiphytic habits (as in vanilla) this is not the case, since this habit allows alternating the functioning of aerial roots and epiphytic habits when there is a nutritional or hydric stress factor, both temporal and spatial. In addition, a large number of species of the family Orchidaceae (V. planifolia) have mycorrhizae in their nutrition system that cushion changes in mineral disposition (^{Canestraro, Mora, & Watkis, 2014}). Therefore, the soil as such only serves the host of the vanilla plants, which suggests that it had little influence on the production of phytochemicals.

Effect of the chemotype on phytochemicals

The results of this study showed that even though the plant structures of both chemotypes had the same phytochemicals, the QVI chemotype showed a greater concentration of total phenolic compounds, total tannins, flavonoids, saponins and, especially, condensed tannins and total triterpenes, which on average had 84 and 45 % higher concentration levels compared to the QI chemotype (Table 3). In this regard, ^{Herrera-Cabrera, Salazar-Rojas, Delgado-Alvarado, Campos-Contreras, and Cervantes-Vargas (2012)} mention that the QVI chemotype has been one of the vanilla germplasm clones most cultivated and used by farmers of the Totonacapan region, while the QI chemotype has been little cultivated and is not of commercial interest.

The chemo-typological variation that exists in vanilla germplasm originated through a process of human selection, based on the aroma and uses that the Totonac producers have given it for more than 250 years (^{Delgado-Alvarado, Salazar-Rojas, & Herrera-Cabrera, 2014}). The use of the QVI chemotype by farmers suggests that, in this material, the synthesis of secondary metabolites that allow the survival of the plant in natural environments and under cultivation is favored. This indicates that this chemotype is more sensitive to stress conditions, so it develops a greater defense mechanism than the QI chemotype.

The adaptive process in relation to secondary metabolites has been observed as a product of coevolution with the enemies of plants (^{Moore, Andrew, Külheim, & Foley, 2014}). Although targeted, intense and permanent selection would force the quantitative variation of phytochemicals to deteriorate, many other selective forces intervene in these compounds, such as herbivores (^{Lankau, 2007}), pathogens, competitors (^{Kliebenstein, Rowe, & Denby, 2005}) and abiotic stresses (^{Burchard, Bilger, & Weissenböck, 2000}).

Effect of plant tissue on phytochemicals

Within plant species, there is a wide variation in the concentration of phytochemicals in tissues depending on the structure of the plant (^{Tahvanainen, Niemelä, & Henttonen, 1991}), which coincides with this study, where the concentration of metabolites was different in relation to the type of plant structure. For example, the leaf had the highest concentration of total tannins (3.84 mg·g^-1 DM) and condensed tannins (2.21 mg·g^-1 DM), while the cured fruits had the lowest concentration of these compounds (Table 3). Generally, these metabolites are found at high concentrations in leaves and stems because they serve as a defense against herbivores, since they have an astringent flavor that causes adverse effects (^{Provenza et al., 1990}).

It was noteworthy that the hydrolysable tannins had the highest value in cured fruits (2.25 mg·g^-1 DM), although it did not differ statistically from those obtained in stem and green fruit (Table 3). It should be noted that the concentration of hydrolysable tannins was obtained by difference, and because it is a gravimetric method it could influence the observed trend (^{Makkar, 2003}).

Although it has been reported that the most abundant tannins in plant tissues are the condensate ones, recently a high concentration of hydrolysable tannins was reported in fruits such as raspberry (3.26 mg·g^-1), blueberry (2.7 mg·g^-1) and pomegranate (1.77 mg·g^-1), among others (^{Smeriglio, Barreca, Belloco, & Trombetta, 2017}). The amount of condensed tannins in vanilla tissues (stems, green fruits and cured ones) is valuable and important for their health effects (^{Olivas-Aguirre et al., 2015}).

In green fruits, the highest amount of flavonoids (11.05 mg·g^-1 DM) was detected (Table 3). In orchid species such as Phalaenopsis spp., the concentration of flavonoids ranges from 1.7 (root) to 4.98 mg·g^-1 (leaf) (^{Minh et al., 2016}), which reveals that vanilla fruits have a significant amount of this phytochemical. However, in cured fruits the concentration of flavonoids was lower, since they are possibly thermosensitive and during curing they degrade (^{Liu, Wang, & Cai, 2015}).

The highest concentration of saponins (2.34 mg·g^-1 DM) was also found in green fruits. Variations in the distribution, composition and amount of saponins in the plant may be due to the different needs they have to protect themselves. In Phytolacca dodecandra and Dioscorea pseudojaponica, during the development of the fruits and tubers, the maximum accumulation of saponins was observed, presumably to avoid loss and thus ensure the maturation of the seeds and the protection of the reproductive organs (^{Lin, Chen, Liu, & Yang, 2009}; ^{Ndamba, Lemmich, & Mølgaard, 1993}). However, in some species the production of saponins was in response to biotic (herbivory or pathogen attack) or abiotic (moisture, light, lack of nutrients and temperature) stress (^{De Costa, Yendo, Fleck, Gosmann, & Fett-Neto, 2013}).

Although in green fruits a concentration greater than 20 mg·g^-1 DM of total phenolic compounds and total triterpenes was detected, in cured fruit they were higher (23.47 mg. g^-1 and 28.77 mg. g^-1, respectively) (Table 3). This is because during fruit curing a complex mixture originated due to the hydrolysis of the precursors (non-volatile glycosylated forms) of the aroma through the activity of the enzyme D-β-glucosidase, which converts them to their volatile forms (^{Odoux, 2000}; ^{Ranadive, 1992}). In this way, in addition to vanillin (which is the most abundant compound, 10 to 20 mg. g^-1), more than 200 components are produced, such as shikimate pathway derivatives, terpenes, furan derivatives, esters, aromatic acids, ketones, phenols, aldehydes, carbonyls and alcohols (^{Sinha, Sharma, & Sharma, 2008}), which provide the flavor and aroma of vanilla.

Relationship between environmental and phytochemical variables

The analysis of canonical correlations identified a high correlation (0.91 and r² = 0.83) between the environmental parameters and the phytochemical content in each management system. The most significant correlations were observed between elevation and the content of total triterpenes (-0.34218***), total tannins (0.34036***) and condensed tannins (-0.4246***), as well as between relative humidity and condensed tannins (0.29691***). The variables relative humidity and intercepted radiation did not show a significant correlation with the production of total phenolic compounds, saponins and total triterpenes (Table 4).

Figure 1 shows that the first two factors of the canonical correlation analysis explained 95.76 % of the total variance. In addition, it shows the structural correlation between environmental and phytochemical variables, where it was distinguished that at higher elevations (Ele) there is a lower content of total and condensed tannins, flavonoids and triterpenes.

Figure 1 Graphical representation of the structural correlation coefficients of the first two factors in the interaction of environmental () and phytochemical () variables of Vanilla planifolia tissues. Elevation (Ele), temperature (Temp), relative humidity (RH), intercepted radiation (IR), total phenolic compounds (TPC), total tannins (TT), hydrolysable tannins (HT), condensed tannins (CT), flavonoids (Flav), saponins (Sap) and total triterpenes (Terp).