Physiological and Enzymatic Evaluation of Selected Genotypes of Stevia rebaudiana Bertoni

Abstract

The Germplasm Bank of Stevia rebaudiana segregants of the University of Córdoba is a strategy for the use of genetic variability and efficient crop yield. There are genotypes with important characteristics such as: high tolerance to salt stress and climatic change (high CO₂ in the Earth surface), late flowering, erect architecture, and high contents of steviol glycosides. However, there is a lack of in-depth studies of morphophysiological and biochemical indicators such as gas exchange, chlorophyll a fluorescence, chloroplast pigments, and antioxidant systems, which allow us to develop early selection tools for elite genotypes. The genotypes (L020, Morita II, and L102) were found to have elite characteristics such as high efficiency in water use, excellent biomass production, and a more robust antioxidant system than the genotypes (L057 and L082). The L020 genotype presented the highest content of stevioside and rebaudioside A, followed by the Morita II genotype. We found a close correlation between the electron transport rate and the mechanisms that increase photosystem complexes. In this sense, non-photochemical cooling modulated by the release of heat by the leaves is a fact that is confirmed by the greater activity of the xanthine pool to protect the photochemical complexes in S. rebaudiana.

1. Introduction

Stevia rebaudiana (Bert.) is a perennial herbaceous plant of the Asteraceae family. It was historically cultivated by aboriginal peoples [1], who believed that S. rebaudiana had the sweetest essence. The plant presents diterpenic glycosides 300 times sweeter than sucrose [2], being a powerful natural sweetener that is low in calories, has medicinal and therapeutic properties, and is beneficial to health. More than 144 genotypes are registered in the world, of which the Morita II genotype stands out as the most planted genotype in the world [3]. Yadav et al. [4] described a total of 183 randomly amplified polymorphic DNA (RAPD) markers in S. rebaudiana. Of the markers that showed amplification products, 35.5% detected polymorphic loci, and 62.5% of these marker loci segregated in a 1:1 ratio, indicating a high level of genetic diversity in S. rebaudiana. The high genetic variabilities of genotypes originating in Paraguay and Malaysia were described based on morphological, chemical, and molecular characteristics traits [5]. Studies such as these usually describe varieties or genotypes that are more adapted to certain conditions in a country or region [6].

Due to its industrial importance, Stevia is commercially cultivated in several parts of the world [7]. The Stevia industry is rising rapidly due to the increasing base of health-conscious consumers and preference for natural non-caloric sweeteners [8]. The global Stevia market size is estimated to be valued at US$ 647.7 million in 2022 [9] and is projected to grow at a compound annual growth rate (CAGR) of 8.0% to USD 1.14 billion in 2028 [10]. Stevia is also used extensively in the pharmaceutical and cosmetics industry due to its antimicrobial, antihypertensive, antioxidant, and skin hydration properties. [11]. According to a World Health Organization (WHO) survey [12], approximately 500 million people worldwide will be diabetic by 2030 due to the increased consumption of products sweetened with refined sugar, a market which can be satisfactorily replaced by natural sweeteners, with S. rebaudiana leading the consumer market. Today, China is the largest producer and exporter of Stevia, while Japan is the primary consumer [13].

To date, most of the research on the physiological responses in Stevia rebaudiana plants has concentrated on improving its crop production [14,15,16,17,18], salt content [19,20,21,22], and water tolerance [23] or retarding flowering to provide more time for leaf harvest [24,25,26]. Recently, Pompelli et al. [27] physiologically characterized two genotypes of S. rebaudiana cultivated under a CO₂-enriched-atmosphere, proving that the photosynthetic efficiency of the species can be linked to a photoprotective efficiency and can overcome stomatic, mesophilic, and biochemical barriers that limit photosynthesis. Selection of the S. rebaudiana genotypes and the optimal environmental conditions that elicit higher stevioside production is the main goal. However, the performance of S. rebaudiana grown in land conditions has received meager attention. This research has as its main goal to analyze some morphological, physiological, biochemical, and enzymatic features in order to select a rapid and sensitive measurement tool based on these features to help test and select elite genotypes of S. rebaudiana. Within this assumption, the main goals of this study were the full characterization of a genotype with characteristics similar to the most cultivated genotype in the world (Morita II), presenting an efficiency of water use twice or three times higher than the Morita II genotype, with excellent characteristics of productivity and other genotypes that have leaf yield similar to the control cultivar, and incorporating the photoprotective efficiency of photosystems (PS), which sustains photosynthetic rates twice as high as Morita II genotype. However, the water use efficiency requires improvement.

2. Materials and Methods

2.1. Plant Material and Environmental Conditions

From the 115 segregants produced by self-crossing the Morita II, our research group (unpublished data) initially selected 86 genotypes, which were cultivated in a Biospace where 25 genotypes were selected with agromorphological characteristics of great interest, such as erect growth, large leaf volume [28], flowering delay [24], and high productivity [29]. These 25 genotypes were then cultivated in a type I biospace (1000 m²) [29] for 12 months, where 5 genotypes with very promising characteristics for plant genetic improvement were selected. This study, therefore, begins with these five genotypes with promising characteristics. Thus, 15-day-old cuttings of S. rebaudiana genotypes (L020, L057, L082, L102, and Morita) were transferred to land conditions with a pH of 6.08, organic matter of 1.64, Cation-Exchange Capacity (CEC) of 18.8, and clay loam—silty texture [29] at the experimental farm of the Faculty of Agricultural Sciences at the University of Cordoba, Montería, Colombia (8°47′37″ N; 75°51′51″ W, 15 m a.s.l.), experiencing a mean annual rainfall of 1346 mm, relative humidity of 84%, and mean annual temperature of 27.4 °C [30]. Montería is characterized as a tropical wet climate according to the Köppen climate classification and has a Martonne aridity index of 40.6 [30]. The cuttings were left to grow in this condition for 15 days; plantlets at 30 days were used in this physiological characterization. The plants were irrigated every other day using dripper tubes with a flow of 2.5-L h⁻¹ and nominal diameter of 13 mm, with two drippers per plant, spaced 0.15 m apart to replace the evaporated water measured by the weather station.

2.2. Gas Exchange

The parameters of gas exchange were done on 30-day-old S. rebaudiana plantlets. The leaf gas exchange was determined on the 3rd attached fully expanded leaf from the apex, using a portable open-flow infrared gas analyzer (LI-6400XT; LI-COR Inc., Lincoln, NE, USA). The net photosynthesis (A_N, μmol CO₂ m⁻² s⁻¹), stomatal conductance (g_s, μmol H₂O m⁻² s⁻¹), transpiration (E, mmol H₂O m⁻² s⁻¹), and internal CO₂ concentration (C_i, μmol CO₂ mol⁻¹) were measured at approximately 09:00–11:00 h solar time, under a clear sky and under the leaf irradiance of 1300 μmol photons m⁻² s⁻¹ as 80% of light saturation point, as recently described [27], a fixed CO₂ concentration (C_a) in 400 μmol mol⁻¹, and an air flow of 400 μmol s⁻¹ [31]. To evaluate if the decrease of A_N in S. rebaudiana is limited by stomatic barriers, we applied the concept of a stomatal threshold limitation (SL), calculated according to the following formula: $\mathit{S}\mathit{L}=\mathbf{1}-\frac{{\mathit{C}}_{\mathit{i}}}{{\mathit{C}}_{\mathit{a}}}$ [32], where C_i is an internal CO₂ concentration (in substomatal chamber) and C_a is a standard CO₂, given by a CO₂ cartridge (iSi pure CO₂ cartridge, iSi inspiring solutions, Wien, Austria). From photosynthesis parameters, we calculated the intrinsic water use efficiency (WUEi) and vapor pressure deficit (VPD; kPa).

2.3. Chlorophyll a Fluorescence

In 30 min on dark adapted leaves, the full open oxidized fluorescence was measured at predawn (05:00 a.m.), midday (12:00 p.m.), and last afternoon (07:00 p.m.) using a pocket PEA chlorophyll fluorimeter (Hansatech Instruments, Norfolk, UK). Initially, initial fluorescence (F0) data were sampled at 10 µs intervals for the first 300 µseconds. Afterward, 3500 µmol m⁻² s⁻¹ intensity with a peak wavelength of 627 nm was applied to measure maximum fluorescence (Fm). These data were used to calculate the variable fluorescence as Fv (Fm-F0), Fv:Fm ratio, Fv:F0 ratio, and F0:Fm ratio. In addition, in light-adapted leaves, an actinic PAR (1200 μmol photons m⁻² s⁻¹) was applied for 300 s. Subsequently, a saturating white light pulse of 6000 μmol photons m⁻² s⁻¹ was used to achieve the light-adapted maximum fluorescence (Fm’), and finally a far-red illumination was applied to measure the light-adapted initial fluorescence (F0′). All other parameters were calculated in accordance with Pompelli et al. [33]. The fluorescence in light-adapted leaves was determined with a portable open-flow infrared gas analyzer (LI-6400XT; LI-COR Inc., Lincoln, NE, USA) integrated with fluorescence chamber heads (LI-6400-40; LI-COR Inc.).

2.4. Biochemical Analysis

2.4.1. Photosynthetic Pigments and Xantophyll Measurements

Chlorophyll a + b and total carotenoids were measured in the leaf samples at the same time the gas exchange measurements were performed. In addition, leaves were sampled at approximately predawn (05:00 a.m.) and midday (12:00 p.m.) to measure xanthophyll pool pigments. For this, the leaves were sampled and promptly immersed in liquid nitrogen (N₂) to stop all biochemical reaction and then stored in darkness at −40 °C until use. Chlorophyll (Chl) was extracted with 100% methanol as previously described in Silva et al. [34] and quantified spectrophotometrically according to Lichtenthaler and Buschmann [35] equations, modified by Pompelli et al. [36] to adapt to microplate readings (ThermoScientific™ Multiskan™ GO, Missouri City, TX, USA). Violaxanthin (V), antheraxanthin (A), and zeaxanthin (Z) pigments were extracted with ice-cold 90% acetone, and the homogenate was collected in 2 mL microvials. All samples were then bubbled with gaseous nitrogen, after which they were kept in darkness for 30 min at 4 °C. Subsequently, the homogenate was then used to measure the xanthophyll pigments as fully previously detailed for S. rebaudiana leaves [27].

2.4.2. Protein Analysis

In the methanolic insoluble fraction, used to extract chlorophylls, soluble proteins were analyzed as per the previously tested protocol [37].

2.4.3. Antioxidant Enzyme Activity and Hydrogen Peroxide Analysis

Four attached fully expanded leaves were sampled and promptly immersed in N₂ to stop all biochemical reaction and then stored in darkness at −40 °C until use. Catalase (CAT; EC 1.11.1.6) and ascorbate peroxidase (APX; EC 1.11.1.11) were extracted using extraction buffer containing 50 mM potassium phosphate buffer (pH 7.0), 2 mM EDTA (Sigma-Aldrich Chemical Co, Darmstadt, Germany, part number ED-500G), 20 mM ascorbic acid (Sigma-Aldrich, Darmstadt, Germany, part number A1300000), 10% Tween 20 (Sigma-Aldrich, Darmstadt, Germany, part number 11332465001), and 80% PVPP (Sigma-Aldrich, Darmstadt, Germany, part number 77627). The samples were centrifuged at 15,000× g for 10 min at 4 °C, and the supernatant transferred to new microtubes in an icebox. CAT activity was measured in 50 mM potassium phosphate buffer (pH 7.0) plus 0.1 mM H₂O₂. The reaction was started with the addition of a 30 μL sample, previously tested for its linearity. The absorbance was measured in 240 nm every 15 s, and delta (Δ) Abs was measured after 2 min. Activity measurement was estimated using the H₂O₂ extinction coefficient of 39.4 mM cm⁻¹ and expressed on the basis of CAT units by soluble protein. One unit of CAT was defined as the amount of enzyme required to oxidize 1 μmol of H₂O₂ per minute. APX activity was measured in 50 mM potassium phosphate buffer (pH 7.0), 0.1 mM H₂O₂, and 0.5 mM ascorbic acid. The reaction was started with the addition of 40 μL sample, previously tested for its linearity. The absorbance was measured in 290 nm every 15 s, and Δ_Abs measured after 2 min. Activity measurement was estimated using the ascorbic acid extinction coefficient of 2.8 mM cm⁻¹ and expressed using APX units by soluble protein. One unit of APX was defined as the amount of enzyme required to oxidize 1 μmol of ascorbate per minute.

The hydrogen peroxide (H₂O₂) concentration was measured after extraction with 0.1% trichloroacetic acid (Sigma-Aldrich, Darmstadt, Germany, part number T6399). The samples were centrifuged at 15,000× g for 10 min at 4 °C, and the supernatant transferred to new microtubes in an icebox. The H₂O₂ concentration was estimated in 100 mM potassium phosphate buffer (pH 7.8) plus 1 M KI. The reaction was done with the addition of 40 μL sample, previously tested for its linearity. After 1 h in darkness, the absorbance was measured at 400 nm, and the results were expressed as dry weight (DW).

2.5. Steviol Glycoside Measurements

Steviol glycosides were measured in the leaf samples at the same time the gas exchange measurements were performed. Sixty-day-old leaves were used in this analysis. A protocol described in detail by Rivera-Avilez [24] was used.

2.6. Biomass Quantification

On 60-day S. rebaudiana plantlets, the dry weight of leaf and stalk was measured after 48 h at 70 °C. The DW was measured after dry cooling in an analytical balance (Sartorius Analytical Balance mod. ENTRIS224-1S, Bradford, MA, USA; accurate to 0.1 mg).

2.7. Experimental Design and Statistical Analyses

The experiments were conducted in a completely randomized block design composed of 5 genotypes (L020, L057, L082, L102, and Morita). The full open oxidized fluorescence was analyzed by two-way ANOVA, composed of 5 genotypes (L020, L057, L082, L102, and Morita) and 3 daily time (predawn, midday, and late afternoon) measurements. The VAZ was analyzed by two-way ANOVA, composed of 5 genotypes (L020, L057, L082, L102, and Morita) and 2 daily times (predawn and midday). All other features were analyzed by one-way ANOVA, composed of 5 genotypes (L020, L057, L082, L102, and Morita). All analyzed features were composed of 5 repetitions. The one-way ANOVA was performed to compare medians using an SNK test (p < 0.05) in SigmaPlot for Windows v. 14.0 (Systat Software, Inc., San Jose, CA, USA). The principal component analysis was estimated after multivariate analysis for all analyzed features in Minitab 18.1 (Minitab, Inc., Chicago, IL, USA). Heatmaps were used to compare the mean of each genotype, using the Morita genotype as a reference. After log₂ transformation, the false color method was used, including a color scale. The heatmaps were constructed using Microsoft^® Office 360 (Microsoft Corporation, Redmond, WA, USA) and CorelDRAW Graphics Suite X8 (Corel Corporation, Ottawa, ON, Canada).

3. Results

3.1. Gas Exchange

In respect to net photosynthesis (A_N), all genotypes tested showed superiority to the model genotype (Morita II; A_N = 13.8 μmol CO₂ m⁻² s⁻¹) with A_N 28% (17.6 μmol CO₂ m⁻² s⁻¹ in L082), 40% (19.3 μmol CO₂ m⁻² s⁻¹ in L020 genotype), 46% (20.1 μmol CO₂ m⁻² s⁻¹ in L057 genotype), and 74% (23.9 μmol CO₂ m⁻² s⁻¹ in L102 genotype) higher than the Morita II genotype (Figure 1A). The stomatal conductance (gs) (Figure 1B) showed a similar pattern to A_N, with the following increases: Morita II (341 μmol H₂O m⁻¹ s⁻¹), L082 (773.1 μmol H₂O m⁻¹ s⁻¹), L057 (788.2 μmol H₂O m⁻¹ s⁻¹), and L102 (961.5 mmol H₂O m⁻¹ s⁻¹), which showed an A_N 2.27, 2.31, and 2.82 times higher than the L020 genotype. Beyond these genotypes, the genotype L020 showed a gs of 239.4 μmol H₂O m⁻¹ s⁻¹ or 30% less than that measured in the Morita II genotype. This pattern makes the L020 the best genotype in terms of the intrinsic water use efficiency (WUEi) (Figure 1D). The genotype L020 shows a WUEi twofold higher (80.8 mmol CO₂ mmol⁻¹ H₂O m⁻² s⁻¹) than those measured in the Morita II genotype (40.3 mmol CO₂ mmol⁻¹ H₂O m⁻² s⁻¹). L057, L102, and L082 showed a WUEi of 36.7% (25.5 mmol CO₂ mmol⁻¹ H₂O m⁻² s⁻¹), 38.4% (24.9 mmol CO₂ mmol⁻¹ H₂O m⁻² s⁻¹), and 43.6% (22.7 mmol CO₂ mmol⁻¹ H₂O m⁻² s⁻¹) lower than the Morita II genotype (Figure 1D). The transpiration (E) showed an inverse pattern to those described in WUEi, where L082, L102, and L057 genotypes showed a 1.2-, 1,6-, and 1.9-fold higher E value than those in the Morita II genotype. Elsewhere, the L020 genotype showed an E 36% less than that measured in the Morita II genotype. This fall in E, by decreasing transpiration and carrying loss of latent heat from evaporation, could lead to an increase in leaf temperature; however, this did not happen with any of the treatments, with leaf temperatures not significantly different from Morita II genotype (values not shown).

A probable decrease in gs was not the cause of the decrease in A_N, especially considering that all other genotypes showed higher A_N and gs than the control genotype (Morita II). In addition, the stomatal opening is not likely to be the cause of the decrease in A_N because all genotypes presented a Ci:Ca higher than 0.5 or LS lesser than 0.4 (Figure 1E). However, genotypes L057, L120, and L082 had a respective VPD 1.2% (p = 0.394), 34%, and 39% higher than the Morita II genotype, while the genotype L020 showed a VPD 12.6% lower. In this sense, it should be noted that the increase in VPD consequently leads to a decrease in both A_N and gs. However, the force that the VPD exerts on A_N and gs is not the same as shown in Figure 2. The increase in VPD leads to a decrease in A_N in an inversely proportional correlation of −0.764 (p = 2.94 10⁻⁵), while gs is decreased in −0.730 (p = 3.44 10⁻⁵), a correlation that is also evidenced by the higher determination factor for A_N (0.95) than gs (0.90) (Supplementary Data File S1).

Regarding the strong negative interaction between A_N and gs (−0.970; p = 1.40 10⁻¹⁵), one must understand that under normal conditions, the biggest loss for electrons is the photosynthesis that consumes CO₂ (Figure 3). However, in stressed plants, most of the electrons coming from the electron transport rate (ETR) do not go to A_N, because the stomata inhibit the CO₂ absorption, and these electrons must be consumed in other ways than photosynthesis (for more details see Section 2.2). In this way, we showed that the L102 genotype has a greater use of ETR for P_N; however, this pattern is a water-consuming process, because the higher use of electrons for photosynthesis is done with maximum gs. Contradicting this pattern, we observed that the Morita II genotype has the lowest efficiency for the use of electrons for P_N; however, this pattern shows a lower water-consuming process. The genotype L020, in turn, showed a median efficiency of electrons used at the expense of the water-consuming process than the Morita II genotype. Both L057 and L082 have a median efficiency of electrons for P_N, similar to genotype L020, but with a water-consuming process analogous to the L102 genotype (Figure 2).

3.2. Chlorophyll a Fluorescence, Chlorophyll, and Carotenoid Concentration

The Fv/Fm is a form of the dissipation of the reductor power in the PSII (Figure 4A) and therefore leads to a reduction of gs of the maximum efficiency of the PSII (Φ_PSII; Figure 5C) and to the subsequent reduction of the β-Nicotinamide adenine dinucleotide 2′-phosphate (NADPH) formation, which results in decreased A_N. In this sense, it could be stated that all plants, regardless of the evaluated genotype, are stressed by the supra-optimal amount of electron capture, since the Fv/Fm measured at midday was 3.9%, 5.0%, 5.4%, 6.0%, and 6.5% lower than that measured at predawn, respectively, in L057, L082, L102, Morita II, and L020 genotypes. However, it can be verified that, even if significantly lower, the values measured in the late afternoon were very similar to the values measured in the next predawn time; these values, although statistically different, demonstrate a physiologically insignificant difference. Two methods to analyze the integrity of PSII are the analysis of the Fv:F0 ratio and F0:Fm ratio (Figure 4B,C). At midday, the values of the Fv:F0 ratio ranged from 3.4 (in Morita II genotype) to 5.7 (in L102 genotype); these values had a strong and inverse correlation with the F0:Fm ratio (r = −0.876; p = 3.2 10⁻⁶; Supplementary Data File S1).

At first glance, the physiological characteristics evaluated in photosystem II (Figure 5) seem to resemble a photoinhibitory effect in all genotypes, with the exception of the Morita II genotype. This hypothesis is supported by the strong reduction of light-adapted variable-to-maximum chlorophyll fluorescence (Fv’:Fm’ ratio) and photochemical quenching (qP) (especially in genotype L057, with a reduction of 8.2% and 11.8%, respectively; Figure 5A,D) by the decrease in quantum yield of photosystem II (Φ_PSII), ETR, and fractions of absorbed photosynthetic active radiation used in photochemistry (P) (Figure 5B,C,F) in all genotypes when compared with Morita II, and by the strong increase of non-photochemical quenching (qNP) and fractions of absorbed photosynthetic active radiation dissipated as heat (D) (especially in genotype L057, with an increase of 15.9% and 7.9%, respectively; Figure 5E,G). However, this hypothesis is not supported due to strong increases in chlorophyll concentration (Figure 6A–C) in genotypes L082 (1.4-fold higher), L102 (1.5-fold higher), and L057 (1.8-fold higher) compared to the Morita II genotype. The carotenoid pool (Figure 6D) corroborates the hypothesis of non-stressed plants, since the genotypes L082, L102, L057, and L020 showed an increase in carotenoid synthesis in an order of magnitude of 1.3-, 1.4-, 1.7-, and 2.0-fold higher than values measured in the Morita II genotype.

The photochemical reaction of photosynthesis is directly proportional to the capture of electrons through chlorophylls and the subsequent reduction of β-Nicotinamide adenine dinucleotide 2′-phosphate oxidized (NADP⁺), yielding NADPH, which is consumed in the Calvin-Benson cycle. However, both ETR and Φ_PSII show a pattern that is influenced by the concentration of chlorophylls. Figure 7 shows that increasing the concentration of chlorophylls leads to a decrease in ETR and in the efficiency of using these electrons as a NADP⁺ photoreductor potential. However, the ETR seems to have a greater influence than Φ_PSII, since the slope of the Chl x ETR curve presented an abrupt decrease, while the decrease related to the Φ_PSII was less affected, although both characteristics have a strongly negative influence on the chlorophyll concentration. Figure 7 also shows that genotypes L082 and L102 show a low efficiency in the use of electrons captured by chlorophylls, potentially transported along ETR, and available to reduce NADP⁺. In the opposite way, the Morita II genotype is shown, due to its high efficiency in the use of electrons, to be an NADP⁺ reductor power. The L020 and L057 genotypes are of moderate efficiency.

3.3. Xantophyll Pigments

Before dawn, the concentration of V measured in genotypes L057, L020, L082, and L102 was respectively 3%, 7.2%, 8.5%, and 13% higher than measured in the Morita II genotype (Figure 8A). The V profile measured at midday was completely different, where all genotypes showed concentrations not statistically different between them. Antexanthin measured in the morning showed similar values for all genotypes, with a slight reduction (4.4%) found in genotype L020 in relation to the Morita II genotype (Figure 8B). However, at midday, all the genotypes showed statistically different values in comparison to Morita II genotype. The genotypes L082, L102, and L057 measured respectively 7.7%, 10.7%, and 58.3% more V than the Morita II genotype, at the same time that the L020 genotype showed a moderate reduction of 26.1%. With the exception of genotype L020, all genotypes showed an increase in Z concentration in relation to the Morita II genotype. The genotypes L102, L057, and L082 showed an increase in Z of 24.5%, 61.6%, and 69.6%, respectively. However, at midday, all genotypes showed similar values. Moreover, Z measured at midday was 70.8%, 97.1%, 123.3%, 188.3%, and 261.2% higher than the values measured at predawn, respectively, for genotypes L082, L057, L120, Morita II, and L020 (Figure 8C).

A + Z increased in the same proportion that A and Z increased separately (Figure 8D), a fact confirmed by the strong correlation between A + Z and A (r = 0.915; p = 1.2 10⁻⁹) and Z (r = 0.991; p = 1.1 10⁻²¹). However, the correlation between these factors is stronger at predawn than midday (Supplementary Data File S1). DEPS expressed by the formula $DEPS=\frac{Z+\left(0.5 A\right)}{VAZ}$, evaluated in the predawn, showed that the genotypes L082 and L057 had a DEPS 16.4% and 18% higher, while the L102 genotype showed no difference, and the L020 genotype showed a reduction of 8.3% when compared to the Morita II genotype (Figure 8E). This is different from the pattern shown at midday, where all genotypes showed statistically non-significant values. V + A + Z (VAZ), when correctly expressed with chlorophyll molar weight, showed a pattern similar to DEPS; however, plants evaluated at midday showed a VAZ Chl⁻¹ 47%, 52%, 61%, 62%, and 73% higher in the plants when compared to plants evaluated at predawn (Figure 8F).

In accordance with Figure 5D, all genotypes showed a fraction of absorbed PAR dissipated as D as a way to extinguish the excess electrons that are transported through ETR. Xanthines, especially A and Z, have a great ability to safely dissipate heat and relieve photosystems. Thus, Figure 9 shows the relationship between D and A + Z, as well as D and VAZ Chl⁻¹. The results show that both relationships have a strong correlation between them and D; however, A + Z has a correlation of 0.9605 with D, while VAZ Chl⁻¹ has a correlation of 0.9409. These regressions are supported by Pearson’s correlations, which were, respectively, r = 0.554 (p = 0.0041) and r = 0.723 (p = 3 × 10⁻⁵) for D × A + Z and D × VAZ Ch⁻¹, both measured in the morning. These r values are strongly increased when the correlation is made with the midday dataset: D × A + Z (r = 0.742; p = 2.17 × 10⁻⁵) and D × VAZ Ch⁻¹ (r = 0.965; p = 3.6 × 10⁻¹²).

3.4. Proteins and Antioxidantive System

The concentration of soluble proteins was differentially modulated in the different genotypes. While genotype L102 showed no significant difference from the Morita II genotype, the genotypes L082 and L057 presented 1.5- and 1.6-fold higher protein than Morita II, while the genotype L020 presented 63% less protein than the Morita II genotype, or 76% and 78% less soluble protein, when compared to L082 and L057 genotypes, respectively (

Table 1. Soluble proteins, catalase and ascorbate peroxidase activity, and hydrogen peroxide concentration in genotypes of 60-day-old Stevia rebaudiana. All values denote the mean (±SD) of five repetitions. In each parameter, values followed by different letters are statistically significant at p ≤ 0.05.

Genotypes	Soluble Proteins (g kg−1 DW)	CAT Activity (U CAT min−1 g−1 Protein)	APX Activity (U APX min−1 g−1 Protein)	H2O2 (mM kg−1 DW)
Morita II	139.91 ± 7.12 b	5.54 ± 0.88 c	77.58 ± 5.73 d	27.28 ± 3.57 b
L020	51.80 ± 2.48 c	12.55 ± 3.01 b	107.50 ± 3.07 b	26.53 ± 1.22 b
L057	230.78 ± 7.46 a	7.58 ± 0.34 c	88.47 ± 4.19 bc	57.94 ± 5.11 a
L082	213.41 ± 10.24 a	31.55 ± 2.57 a	97.57 ± 5.12 bc	62.78 ± 0.71 a
L102	136.92 ± 11.72 b	7.78 ± 0.58 c	142.74 ± 5.64 a	66.89 ± 1.75 a

).

Morita II, L057, and L102 genotypes showed similar catalase activity, while the L020 and L082 genotypes showed 2.3- and 5.7-fold higher protein than the Morita II genotype (Table 1). On the other hand, APX activity was increased in all genotypes when compared to the Morita II genotype. Thus, genotypes L057, L082, L020, and L120 showed 1.1-, 1.3-, 1.4-, and 1.8-fold higher APX activity than the Morita II genotype (Table 1).

3.5. Steviol Glycosides

The genotype L020 showed the highest concentration of steviol glycosides (172.6 ± 6.4) among the evaluated genotypes, presenting 143.1%, 82.7%, 79.3%, and 19.4% more steviol glycosides, respectively, in relation to the genotypes L057, L082, L102, and Morita II (Figure 10). This higher concentration of steviol glycosides also translated into a higher concentration of rebaudioside A (105.7 ± 4.5), with the L020 genotype presenting 428.4%, 171%, 140.2%, and 55%, in the same order as described for steviol glycosides. On the other hand, Morita II was the genotype with the highest concentration of stevioside (63.4 ± 1.8), which was 56%, 40.8%, 39.1%, and 19.8% higher than the genotypes of L102, L057, L082, and L102, respectively (Figure 10).

3.6. Biomass Produced after 60 Days

Biomass production is the final result of all the analyses that have been described thus far, as it expresses, in productivity, all the photosynthetic and photoprotective machinery in the production of more biomass. In this sense, it is verified that the L020 genotype accumulated 93% more biomass than the control genotype (Morita II), while the L102, L082, and 1057 genotypes accumulated, respectively, 13%, 26%, and 90% less biomass than the Morita II genotype (Figure 11).

3.7. Multivariate Analysis

The multivariate analysis described by the principal component analysis (PCA) is a great strategy to search and rank similar patterns. The use of a wide variety of features makes PCA more robust; however, some factors tend to weigh less than others, and sometimes this excess of features can lead to a decrease in the sum of PC1 and PC2. Figure 11 shows that the analysis presented in this study explains 91.5% of the variations that may occur in the dataset. The lower axis of the PCA presents the data of gas exchange, chlorophyll a fluorescence, and predawn xanthine pool, while the upper axis of the PCA better explains the pattern of xanthines measured at midday, some chlorophyll a fluorescence data, as well as heat dissipation by means of D and VAZ (Figure 12A). Based on this pattern, it is possible to draw a dendrogram that clusters Morita II with L020 genotypes, as well as another cluster involving L057 and L082 genotypes. A third cluster is formed by the L120 genotype, which has some characteristics that can be shared with the L082 genotype (shading of the groups in Figure 12B).

4. Discussion

Biomarkers are rapid physiological or biochemical assays that can detect a specific condition. In terms of our objectives, they permit us to detect the superiority of one or more characteristics that lead us to infer the agronomic and physiological superiority of these genotypes.

Some gas exchange features are able to be used for specific elite genotypes. Between them, the net photosynthesis (A_N) and stomatal conductance are the most commonly used. Positive modulation of A_N is expected of an elite genotype. S. rebaudiana is considered separately because the total A_N is usually connected to produce more leaf area and thus more biological material in order to produce steviol glycosides. In this sense, studies showing the main objective of promoting leaf yield [38,39,40,41,42] or floral delay [3,24,43,44,45] are interesting in this species. Beyond this, the stomatal conductance flux is an important task, because the stomata opening elicits CO₂ capture as well as water lost to atmosphere, which has an extremally negative water potential. So, in these contexts, the best genotype is the one that is most able to ensure photosynthesis with a low or very low stomatal opening, here described by intrinsic water use efficiency. In this sense, there is no doubt that genotype L020 is the best genotype among all studied genotypes, since its WUEi was 2.0-, 3.1-, 3.2-, and 3.6-fold more efficient in water use when compared with Morita II, L057, L102, and L082 genotypes, respectively. Additionally, the L020 genotype showed a moderate to high efficiency of electron use to drive photosynthesis in the same stomatal conductance base.

Plants absorb solar energy to drive their photosynthetic process. However, while beneficial, this process can be extremely dangerous for plants if the photons absorbed by the chlorophyll molecules are not quickly quenched [46,47,48,49,50]. There are essentially three ways to suppress the energy absorbed by chlorophyll molecules: (1) photons can be used to drive photosynthesis (direct and preferential pathway), and (2) thermal dissipation in the form of heat (D) or (3) re-emitted as light (fluorescence of chlorophyll). These three processes take place in competition, and any increase in the efficiency of one will result in a decrease in the efficiency of the other. A fourth process used by the plant to reduce the photon capture is to activate the photobleaching of chlorophyll [33,46,51,52,53,54,55], promoting the chlorophyllase gene expression [56,57,58]. Less Chl, leads to less captured energy and fewer dangerous molecules able to produce ROS. Thus, measurable physiological features, such as photosynthetic activity and chlorophyll fluorescence data, give us physiological parameters that guide photochemical and photoprotective events. Information about changes in photochemical efficiency and the use of photons can be obtained quickly and safely in order to allow an indirect evaluation of the yield of a crop species, thus serving as a biomarker in integrated genotype selection systems [55,59,60,61,62,63] as well as S. rebaudiana [27]. Fai et al. [62] described that a fluorescence assay could reliably predict the effects of nonphotosynthetic inhibitors on growth and select toxicity in a microalgal species Selenastrum capricornutum. This technique is widely reccomended as a rapid, inexpensive, and reliable method to select an interesting phenomemon [27,51,63] as a select elite genotype [27,51,63].

Another parameter widely used for the selection of elite genotypes is the ability of plants to activate their antioxidant system and make it active [63,64,65,66,67]. In addition, the presence of a more robust antioxidant system and a greater electron transport efficiency are characteristics that are sought as markers of the elite genotypes [27,59,63]. However, biomarkers have yet to be adopted in wide-scale and routine monitoring, despite the advantages that they convey [68]. S. rebaudiana is a fast-growing species with an excellent performance index [24,27], so both features are strictly connected with gas exchange and photochemical extinction events [20,24,27,69,70,71,72]. They could be markers, and as such, useful tools in the rapid selection of elite genotypes.

In the five genotypes studied, we found a significant variation in the content of stevioside and rebaudioside A. However, the L102 genotype presented content of stevioside and rebaudioside A similar to the L082 genotype; the L020 genotype presented the highest content of rebaudioside A. On the other hand, the Morita II genotype had the highest content of stevioside. Stevioside and rebaudioside A confer a higher sweetness than sucrose; however, the latter has a less bitter and metallic taste than the former [73,74]. Therefore, the selection of S. rebaudiana lines with a higher content of rebaudioside A is highly significant to improve sweetness; consequently, the L020 genotype stands out as an elite genotype, as it has a higher content of total glycosides, a higher amount of rebaudioside A, and a lower amount of stevioside, in addition to having the highest leaf yield and zero flowering throughout the study. In this study, we describe a lower concentration of stevioside and rebaudioside A when compared to other articles already published by our work team [24,27]. The reason for the lower concentrations of steviosides and rebaudioside A shown in this study likely comes from the fact that the plants used in this study were in the flowering process, when the vast majority of carbons produced in photosynthesis are used for the synthesis of reproductive organs [75,76] or gibberellin production [23,77]. The confirmation of this hypothesis could also be explained by the greater biomass production in the L020 genotype; the L057, L082, L102, and Morita II genotypes showed a 95%, 61%, 55%, and 48% lower leaf biomass than those of the L020 genotype. While the steviol glycoside levels shown in this work were lower than those published by our team, the steviol glycosides were in agreement with other published papers [26,42,78].

The photolysis of water (Hill reaction; [79]) yields electrons that are transported by PSII protein complexes to ferredoxin NADP reductase, which uses the proton gradient to direct the formation of NADPH, which is rapidly utilized in the Calvin-Benson cycle. However, stomatal closure prevents the uptake of atmospheric carbon, and thus the molecules that would be used to reduce CO₂ to sugars are inhibited or drastically diminished. Without electron transport, the proton-motive force cannot be established; consequently, other components of PSI and PSII remain reduced and, with no other molecule for electron donation, they end up reducing molecular oxygen, giving rise to free radicals (ROS) [80,81]. However, plants have several alternative mechanisms, including cleaning mechanisms or avoidance mechanisms, for dissipating these electrons that reach the PSI. One of the main protection mechanisms is the reduction of the quantum yields of the PSII (herein called the Fv:Fm ratio). Thus, it can be concluded that the reduction of the Fv:Fm ratio is an indication of stress, which can be dynamic or photoinhibitory. It is dynamic when the plants are able to recover full functionality the next day, and photoinhibitory when the plants achieve a stress level that is so great that the defense system fails and proteins or large protein complexes such as PSII protein D1 are damaged [82]. In this study, we showed that in all genotypes a reversible photoinhibition occurred, as denoted by an initial decrease in the Fv:Fm ratio and Fv:F0 ratio and a strong increase of the F0:Fm ratio in plants evaluated at midday, parameters monitored in progression in late afternoon and the next predawn day. Thus, in our study, no photoinhibitory process was detected.

Recently, some scholars have described a characteristic of S. rebaudiana [27,59] where an increase in non-photochemical parameters serves as an efficient mechanism for dissipating excess light energy, thus minimizing ROS generation. Acosta-Motos et al. [59] recently reported that S. rebaudiana presenting high values of D is related to the inability of the plant to protect itself from excess light. However Acosta-Motos et al. [59] did not incorporate the VAZ cycle as a potent heat dissipator, mainly by lumen accidification as a result of photochemistry. This study and other recently published studies [27] consider the VAZ cycle as a potent heat dissipator and describe plants that present high values of A + Z as having an extraordinary capacity to dissipate energy as heat.

In accordance with Cantabella et al. [19], in the short term, it seems that S. rebaudiana can tightly control ROS generation by inducing different antioxidant mechanisms, such as ROS-scavenging enzymes (APX, POX, CAT, SOD). However Cantabella et al. [19] showed that non-stressed S. rebaudiana plants present a CAT activity of 2.1 mmol CAT min⁻¹ mg⁻¹ protein. In this study, we show a higher CAT activity, of 31.55 μmol CAT min⁻¹ g⁻¹ protein or 31.55 nmol CAT min⁻¹ mg⁻¹ protein; thus, our plants did not present any stress signals. The plants showed only 1.5% of the CAT activity described by Cantabella et al. [19] in non-strressed S. rebaudiana plants. A similar problem was identified in the antioxidant potential in S. rebaudiana callus in response to polyethylene glycol, paclobutrazol, and gibberellin treatments [83]. In this study, Hajihashemi et al. [83] described a non-stressed callus of S. rebaudiana CAT activity of ~12 mmol H₂O₂ min⁻¹ mg⁻¹ protein or 12 mol H₂O₂ min⁻¹ g⁻¹ protein, an activity 380-fold higher than values presented in this study. Hajihashemi et al. [84] described S. rebaudiana callus under paclobutrazol treatments with a CAT activity of ~4.5 mmol H₂O₂ min⁻¹ mg⁻¹ protein or ~4.5 mol H₂O₂ min⁻¹ g⁻¹ protein, an activity 143-fold higher than those results shown in our higher data. Saravi et al. [85] describe the effect of salt stress with a co-application of endophytic fungi in non-stressed plants as a CAT activity of 1.06 U CAT min⁻¹ mg⁻¹ protein. A superficial analysis of this value may confuse the reader because this author presented that one unit of CAT was defined by the change of absorbance in 60 s. Change of absorbance is an empirical form to express the enzyme activity, because the aborbance is directly proportional to protein in the reaction media and inversely proportional to the extinction coefficient of absorption (in this specific case, H₂O₂; i.e., 39.4 mmol⁻¹ cm⁻¹). Thus, without knowing the absorbance in each point, it is imposible to compare the values presented in this study with those presented by Saravi et al. [85].

Regarding ascorbate peroxidase, Saravi et al. [85] describes that non-stressed S. rebaudiana plants showed 2.29 U CAT min⁻¹ mg⁻¹ protein; however the manner in which the data is presented is very confusing, because without knowing the specific absorbance, it is imposible to compare our results with the results described in Saravi et al. [85]. Hajihashemi et al. [84] described the S. rebaudiana callus under paclobutrazol treatments as an APX activity of ~0.2 mmol ascorbate min⁻¹ mg⁻¹ protein or ~200 μmol H₂O₂ min⁻¹ g⁻¹ protein. These results are very similar to those presented in this study. A similar situation was described in the response of S. rebaudiana callus to polyethylene glycol, paclobutrazol, and gibberellin treatments [83]. In this study, Hajihashemi et al. [83] describe that the non-stressed callus showed an APX activity of ~0.9 mmol ascorbate min⁻¹ mg⁻¹ protein or 900 μmol ascorbate min⁻¹ g⁻¹ protein, an activity 6.5-fold higher than presented in this study.

Some scholars [59,86,87,88,89,90,91] describe that hydrogen peroxide (H₂O₂) can accumulate in response to a stressor or as an oxidative stress marker. In this study, we associate the H₂O₂ as a cell signaling strategy rather than the consequence of a stress process, as previously reported [92,93,94,95]. It has been established that it acts as a secondary messenger in signal transduction networks. Saravi et al. [85] describe a H₂O₂ concentration of 4.84 nmol H₂O₂ g FW⁻¹ or 4.84 μmol H₂O₂ kg FW⁻¹. At first glance, this value is much smaller than those described in the present study. However Saravi et al. [85] showed results in fresh weight (FW) base. To compare this value with the value presented in our study, the FW may be converted to dry weight (DW). Our research group, after a review of extensive published studies with many S. rebaudiana genotypes [20,24,27,28,29], conclude with a FW to DW conversion coefficient of 4.32 (i.e., 1 g of FW results in 248 mg DW). Then, applying this correction factor to the data of Saravi et al. [85], the H₂O₂ may be computed as 19.55 μmol H₂O₂ kg DW⁻¹, a value similar to those presented in this study. However, the methodology used by Saravi et al. [85] is distinct from the methodology applied in this study, and so the results can not be compared. Similar results were presented by Hajihashemi et al. [83], who describe a H₂O₂ concentration of 1.8 μmol g⁻¹ FW or 1.8 mmol kg⁻¹ FW. Then, applying the conversion coefficient of 4.32, the final results are 7.27 mmol kg⁻¹ DW, a result in growth media of 15% of the values presented in this study. Another example is presented in Hajihashemi et al. [84] where non-stressed S. rebaudiana plants showed a H₂O₂ concentration of 2.5 μmol g⁻¹ FW or 2.5 mmol kg⁻¹ FW, or 29.32 mmol kg⁻¹ DW, after applying the conversion coefficient of 4.32. These last results are in accordance with our results measured in Morita II and L020 genotypes. In addition, the H₂O₂ concentration described in Oryza sativa [96] or Jatropha curcas [90] is lower than that presented in this study; however, the methodology used, the form of expression per unit, and fresh versus dry mass prevent the data presented by these authors from being compared with the data presented in this study. In accordance with Saxena et al. [92], under normal conditions, H₂O₂ emerges as an important factor during many biological processes.

Faced with strongly contrasting data, one might conclude that the data presented in this study are incorrect; however, as shown, these data are generally equivalent to others already published [33,55,97], including S. rebaudiana studies [83,84,98,99,100].

It is important to note that none of the regression analysis in this study resulted from collinear sampling (Supplementary Data File S2), which assumes collinearity to be a sampling artifact or a true reflection of population relationships [101]. According to these authors, collinearity must be considered when data are analyzed with regression analysis, as they can cause large fluctuations with insignificant changes in sampling. Data that result in collinear correlations can cause a misinterpretation, because the closer to collinearity, the more easily the x-axis values can predict the y-axis values, making the regression false and with a high determination index (R2) [102]. The absence of collinearity in all regressions presented in this study demonstrates the cohesion of the data and the ability of the data to estimate certain characteristics in an unbiased way, as reported by Næs and Mevik [102].

The heatmap is a technique widely used to compare profiles against a pattern (Figure 13). As this study aimed to characterize the genotypes and compare them with Morita II genotype, the heatmap technique provides the most interesting results. The direct analysis of the heatmap shows that the L020 genotype has a higher A_N, a strong WUEi, a very extensive concentration of carotenoids, as well as a robust antioxidant system. These characteristics translate into a very impressive yield in dry mass, leading to a putative increase in total stevioside production per plant. On the other hand, the genotype L120 presents a higher A_N in relation to both Morita II and L020 genotypes; however, this is accomplished by increases in both in gs and T, which make the genotype very inefficient in water use (WUEi). A significant increase in the predawn xanthine pool without a significant change in the midday level demonstrates that the genotype was not under any stress. In this sense, the accumulation of 2.5-fold H₂O₂ could be interpreted as a cell signaling strategy rather than the consequence of a stress process. The genotypes L057 and L082, in turn, showed an intermediate A_N between the L020 and L120 genotypes, at the expense of a strong increase in gs and T and, consequently, a significant decrease in WUEi. The reduction in ETR and Φ_PSII may indicate that these genotypes may be experiencing a stress event due to light excess. However, the expressive modulation of the xanthine pool evaluated both at predawn and midday shows that these varieties may provide a more robust system for excess energy dissipation, captured by chlorophylls. In this sense, the increase in CAT and APX activity, analyzed together with the H₂O₂ data, may indicate that these genotypes did not adapt well to the imposed conditions, a fact that translates into a strong reduction in leaf yield in these two genotypes. However, it should be noted that these analyses are preliminary and should not be considered as complete before an analysis of gene expression is conducted in the genotypes both in steady-state conditions, as presented in this study, and in stressful conditions, such as salinity, drought, and waterlogging, which are common in the Caribbean region where these genotypes are cultivated.