The experiment was established in the experimental field area of the Federal University of Santa Catarina (UFSC), near Curitibanos county (27°19′09″S, 50°42′39″W, 836 m a.s.l.), Southern Brazil, where two cultivation systems of yerba mate were set in neighboring plots, with 150 m distance between plots (75 m × 20 m). One system was the anthropized rainforest in recovery enriched with yerba mate in its understory (agroforestry system, AFS), and the second one was the full-sun cultivation (monoculture, MO). The rainforest was composed of native and non-native tree species (A. angustifolia, Ocotea spp., Peltophorum dubium, Luehea divaricata, Cordia ecalyculata, Cordia trichotoma, Eugenia involucrata, Syagrus romanzoffiana, Podocarpus spp., Pinus elliottii, Eucalyptus spp. etc.) and understory vegetation.

The climate is defined as subtropical humid (Cfa) according to the Köppen–Geiger classification, with regularly distributed rains (approximately 1,480 mm annually) and an average air temperature of 20.8°C ± 0.3°C in the warmest month (Alvares et al., 2014). The soil is Dystric Cambisol, and its physico-chemical composition was based on samples collected from 0 to 20 cm depth before the planting of yerba mate in 2018. Soil pH in H₂O and CaCl₂ was 4.8 or 4.15, respectively, sum of bases = 2.02 cmol_c dm⁻³, base saturation = 13.8%, cation exchange capacity at pH 7 = 14.62 cmol_c dm⁻³, effective cation exchange capacity = 4.37 cmol_c dm⁻³, Ca = 1.86 cmol_c dm⁻³, Mg = 0.54 cmol_c dm⁻³, K = 0.16 cmol_c dm⁻³, Al = 2.35 cmol_c dm⁻³, H + Al = 12.78 cmol_c dm⁻³, C = 22.69 g dm⁻³, OM = 39.03 g dm⁻³, Fe = 48.35 mg dm⁻³, Mn = 13.6 mg dm⁻³, Cu = 11.85 mg dm⁻³, Zn = 0.86 mg dm⁻³, and P = 2.52 mg dm⁻³. Fertilization was done in accordance with soil analyses and recommendations for yerba mate planting and post planting (Penteado Junior and Goulart, 2019). In the MO area, 11 g of N and 16 g of P were used per planting hole, respectively. After that, an annual topdressing fertilization with 30 g of P per plant and two annual fertilizations with 20 g of N and 10 g of K per plant were made. In the AFS area, 11 g of N and 22 g of P were used per planting hole. After that, an annual topdressing fertilization with 40 g of P per plant and two annual fertilizations with 20 g of N and 15 g of K per plant were made. Mechanical weed control was done, and there was no pest/disease management.

Young yerba mate clonal plants from the genetic improvement program of Embrapa (Sturion et al., 2017) were planted in June 2018, with 3 m between the rows and 1.5 m between plants in the row. Among 20 clonal and progeny plants, two male (♂) and two female (♀) clones were chosen for evaluation when the plants were 3 years old. Male clones were labeled “1” and “82,” and female clones were labeled “12” and “84.” Three plants of each clone were followed in each cultivation system (MO and AFS). Each of the three replicate plants per clone was arranged in separate rows, respecting randomized experimental design. The selection was based on the presence of 5–6 visible growth units (GU), using the chronological age as a reference. In December 2021 (beginning of the summer rest period), formation pruning was performed at 40 cm of stem height, leaving approximately 10%–20% of photosynthesizing leaf area for plant recovery.

The measurements of light intensity and photosynthetic parameters (light curve parameters and chlorophyll a and b indexes) were performed in four seasons (summer, fall, winter, and spring). Those periods of measurements corresponded to plant growth rhythmicity after the pruning in December 2021: (i) in February 2022 (summer), end of the growth rest 1 (R1) period, (ii) in May 2022 (fall), end of the growth unit 1 formation (GU1), (iii) August 2022 (winter), end of the growth rest 2 (R2) period, and (iv) November 2022 (spring), end of the growth unit 2 (GU2) formation.

Photosynthetic photon flux density (PPFD, µmol photons m⁻² s⁻¹) was measured using silicon pyranometer smart sensors (S-LIx-M003, HOBO, Bourne MA, United States), which capture the sun radiation between 400 nm and 700 nm. The PPFD values were recorded with a data logger (H21-USB Data Logger USB, HOBO, Bourne MA, United States) every 10 s, and the mean values were calculated every 10 min and are here explored as hourly mean values. The silicon sensor was maintained from the summer to the spring season in 2022, at 2 m from the soil. Simultaneously to the measurements of PPFD, the air temperature daily course data were registered in both systems during the four seasons. For that, the U23-003 HOBO 2x external temperature data logger with an incorporated sensor (HOBO, Bourne, MA, United States) was used. The one available data logger was moved from one to the other cultivation system every 2–3 weeks during the experimental period. The hourly mean values and standard errors (SE) of diurnal PPFD and daily temperature course in each of the four seasons (summer, fall, winter, and spring) and in each cultivation system (n = 18–72) were calculated.

Relative chlorophyll (Chl) a and b indexes were measured on recently expanded mature leaves, considering each plant of each clone and gender grown under each cultivation system and during each of the four seasons. A portable chlorophyll meter model CFL 2060 (Falker, Porto Alegre RS, Brazil) was used, showing the relative Chl a and b values (indexes) that are highly correlated to concentrations of pigments (Schlichting et al., 2015). Measurements were taken during the morning hours (9:00 to 11:00 a.m.). The ratio Chl a and b was also calculated.

The responses of CO₂ assimilation to light (A _net/PPFD curves) in yerba mate were measured on the recently expanded mature leaf of each observed plant of each clone and gender, grown under each cultivation system, during each of the four seasons (rhythmic growth periods). The A _net (μmol CO₂ m⁻² s⁻¹) responses were registered considering 23 light intensities (PPFD) reported in µmol photons m⁻² s⁻¹: 1,500, 1,200, 800, 500, 200, 100, 90, 80, 70, 65, 60, 55, 50, 45, 40, 35, 30, 25, 20, 15, 10, 5, and 0 (Supplementary Figure S1A). The 6 cm² cuvette was fitted with a red-blue light source (6400-02B, LI-COR, Lincoln NE, United States), and [CO₂] was set to 420 μL L⁻¹. For the curve, the A _net value at PPFD = 0 μmol m⁻² s⁻¹ was assumed as the dark respiration (R _d, μmol m⁻² s⁻¹, Rakocevic et al., 2024). The rate of respiration in the light (R _L, μmol m⁻² s⁻¹) was calculated based on Kok’s method (Kok, 1948) as the y-axis intercept of a first-order linear regression fitted to plots of A _net/PPFD to measurements made over the range of PPFD from 25 μmol m⁻² s⁻¹ to 65 μmol m⁻² s⁻¹. Gas exchange data were corrected for the increase in intercellular CO₂ concentrations (C _i) with decreasing PPFD, which can result in reduced rates of oxygenation and increased rates of carboxylation for RuBisCO. The correction was applied by adjusting R _L through iteration to minimize the intercept of photosynthetic electron transport (J, μmol m⁻² s⁻¹) as a function of PPFD (Kirschbaum and Farquhar, 1987). J was calculated using the CO₂ compensation point in the absence of R _L (Γ*, μmol m⁻² s⁻¹, von Caemmerer and Farquhar, 1981), as shown in Equation 1: J = 4 A net + R   L C i + 2 Γ * C i − Γ *   (1)

The effects of atmospheric [O₂] or [CO₂] on oxygenation (V _o) and carboxylation (V _c) of RuBisCO at saturating light and also RuBisCO kinetic constants (K _c = 272.38 and K _o = 165.82 μM at 25°C) previously determined by Bernacchi et al. (2002) were used to calculate the maximum rate of carboxylation of RuBisCO (V _cmax, μmol m⁻² s⁻¹), as in Equation 2: V cmax = A net − R L CO   2 − Γ * / CO   2 + K   c   1 + O   2 K   c (2)

Γ* from Equations 1, 2 is dependent on the RuBisCO specificity factor K _o for O₂ partial pressure (von Caemmerer and Farquhar, 1981) and was calculated as in Equation 3: Γ * = 0.5   V omax   K o   O   2 V cmax   K   o (3)

The V _o and V _c also used to compute Γ* at PPFD of 500 μmol m⁻² s⁻¹ [considered as the light saturation point for yerba mate (Rakocevic et al., 2012)], were estimated as shown in Equations 4, 5: V c = CO 2 V   cmax CO   2 + K   c   1 + O   2 K   o (4) V o = O 2 V   cmax O   2 + K   o   1 + CO   2 K   c (5)

The maximum net photosynthesis (A _max, μmol m⁻² s⁻¹)—a proxy of photosynthetic capacity—was calculated from the modeled A’ _net/PPFD curves (Supplementary Figure S1B), as well as the degree of light inhibition of respiration, determined by the R _L/R _d ratio (da Silva et al., 2017; da Silva et al., 2018) and the apparent quantum efficiency of CO₂ assimilation (Ф, μmol CO₂ μmol photons⁻¹). Ф was calculated from the initial linear slope of the modeled light response curves (PPFD <30 μmol m⁻² s⁻¹), while the maximum apparent rate of electron transport (J _max, μmol m⁻² s⁻¹) was considered as the computed value for PPFD at 1,500 μmol m⁻² s⁻¹. Additionally, the ratio between the maximum apparent rate of electron transport and the maximum apparent RuBisCO carboxylation rate (J _max/V _cmax) was calculated.

A complete randomized design with four clones (two females and two males) grown in agroforestry and monoculture situations over four seasons with three replications (plants) was used. Two-way ANOVA was applied to analyze the effects of clones or genders (four clones or two genders) and two cultivation systems (separately for each season) on Chl indexes and photosynthetic traits derived from light response curves. Two families of comparisons/mean contrasts were done: 1) clones and cultivation system effects and 2) genders and cultivation system effects, both calculated separately for each season/growth period. ANOVAs considered mixed linear models (using the lme function from the nlme package) and maximum likelihood to test the significance of studied variables. The Bartlett homogeneity test and the Shapiro normality test were performed for each variable in each season in a model construction. Clones (or genders) and cultivation systems were considered fixed factor effects, while plant number (repetitions) was a random effect observed separately for each season/growth period. If the interaction was not significant, the model reduction was applied (and fitted again). For comparison of variables among the four seasons, one-way ANOVA was performed where season was considered a fixed factor effect, while total plant number was considered as a random effect. All ANOVAs were performed at 95% confidence, followed by a Tukey Honestly Significant Difference test, where p < 0.1 for mean comparisons was considered significant. p-values, estimated means, and standard errors (SE) are shown in the figures. When interaction among factors was significant, only the p-value of interaction was shown, as interactions represent dependence between factors. A second family of comparisons is shown above the figures that compare clone and cultivation system effects during each season/growth period, considering gender and cultivation system effects during each season/growth period. The dynamics of responses over the seasons (one-way ANOVA) are shown as red uppercase letters near season indications. Fisher’s exact test was used to analyze the significance of counting occurrences of SSD over the rhythmic growth (“Yes” for SSD expressed over the rest and growth unit formations and “No” for not expressed). All statistical analyses were performed using R. 4.2.1 software (R Core Team, 2022), using the nlme, emmeans, multcomp, ggplot2, and reshape2 packages.

The mean midday PPFD values at the top of the yerba mate canopy in MO were 1,120 μmol m⁻² s⁻¹, 850 μmol m⁻² s⁻¹, 790 μmol m⁻² s⁻¹, and 950 μmol m⁻² s⁻¹ during the summer, fall, winter, and spring, respectively (Figure 1A). The maximum midday PPFDs registered in MO were 1,760 μmol m⁻² s⁻¹, 1,525 μmol m⁻² s⁻¹, 1,510 μmol m⁻² s⁻¹, and 2,000 μmol m⁻² s⁻¹ registered on open-sky days during the summer, fall, winter, and spring, respectively. The incoming light was drastically reduced in AFS, where midday PPFD values were ∼46 μmol m⁻² s⁻¹, 39 μmol m⁻² s⁻¹, 53 μmol m⁻² s⁻¹, and 75 μmol m⁻² s⁻¹ during the summer, fall, winter, and spring, respectively (Figure 1B). Light availability in the AFS strongly depended on the upper leaf layer distribution in the forest canopy and could show asymmetric diurnal distribution, as in the summer. The incoming midday PPFD in the AFS was only 4%, 5%, 7%, or 8% of the midday PPFD measured in the MO. The mean daily temperatures were 18.9°C in MO and 17.3°C in AFS during the summer, 13.2°C in MO and 11.0°C in AFS during the fall, 13.6°C in MO and 12.4°C in AFS during the winter, and 18.3°C in MO and 15.5°C in AFS during the spring (Figures 1C, D).

The effects of clone, gender, and cultivation systems on the Chl a index were strongly dependent on season (Figure 2A; Supplementary Table 1). The Chl a index was the lowest during the summer rest (R1) and the highest during the spring growth unit (GU2), whereas it was changed by cultivation systems during the summer and fall (R1 and GU1, Supplementary Table 2). During the summer rest (R1), the Chl a index was higher in the AFS than in the MO in the female (♀) clone 84, while the opposite response was observed in ♀ clone 12. The Chl a indexes of male (♂) clones did not vary due to the cultivation system in the summer, but ♂ clone 1 and ♀ clone 84 showed the lowest values under MO. Higher Chl a index values were registered in MO than in AFS during the intensive regrowth period (GU1). During the winter rest (R2), the ♂ clone 1 had a lower Chl a index than the fall other clones. Gender did not show any effect on Chl a index, regardless of the season or cultivation system (Supplementary Table 3).

The highest values of the Chl b index were noticed during the winter rest (R2), decreasing during the fall and spring (GU1 and GU2), and the lowest values were recorded during the summer rest R1 (Figure 2B; Supplementary Table 1). The Chl b index varied due to the cultivation system only during the winter rest (R2), when it was higher under MO than under AFS in ♀ clone 12. During R2, ♀ clone 12 showed a higher Chl b index than the other three clones under MO (Supplementary Table 2). The leaves of ♀ plants had a higher Chl b index than ♂ ones during the R2 and GU2 periods, regardless of the cultivation system (Supplementary Table 3).

The Chl a/b ratio was the highest in the summer rest (R1) and the lowest in the winter rest period (R2) (Figure 2C). This ratio did not differ between the growth unit periods (GU1 and GU2), and it did not vary due to the cultivation system (Supplementary Table 2). However, ♂ clone 1 had higher Chl a/b values than ♀ clone 12 during R2, regardless of the cultivation system. The leaves of ♀ plants had a lower Chl a/b ratio than the leaves of ♂ plants during the R2 and GU2 periods in both cultivation systems (Supplementary Table 3).

The highest maximum net photosynthesis (A _max) values were registered during winter rest (R2) and fall growth unit (GU2) periods, attaining 15.8 μmol m⁻² s⁻¹, and the lowest values were registered during spring growth unit (GU2) and summer rest (R1) periods (Figure 3A; Supplementary Table 1). When considering A _max, there was no interaction between clone and cultivation system or between gender and cultivation system (Supplementary Tables 2, 3). The leaves of yerba mate plants cultivated under MO presented higher A _max values than those under AFS during all four seasons, decreasing by 21%–45% in AFS compared to MO. Clone and gender affected A _max only during the summer rest (R1) and the fall growth unit formation (GU1) periods. During those two periods, the leaves of female plants had higher A _max values than males, with the highest values registered in ♀ clone 84 during the summer, while the lowest ones were registered in ♂ clone 1 during the fall.

The apparent quantum efficiency of CO₂ assimilation (Ф) was higher during the R1 and GU1 than during the R2 and GU2 periods (Figure 3B; Supplementary Table 1). The cultivation systems changed Ф from fall to spring when Ф had been higher in AFS (attaining a mean value of 0.092 μmol CO₂ μmol photons⁻¹) than in MO in all clones (Supplementary Table 2). During the fall growth unit period (GU1), the lowest Ф values were observed in the ♂ clone 1 compared to the three other clones, but no significant gender segregation was noticed. During the winter rest period (R2), the lowest Ф value was estimated in ♀ clone 84 (0.019 μmol CO₂ μmol photons⁻¹) and the highest in ♂ clone 82 (0.057 μmol CO₂ μmol photons⁻¹) under AFS. Under MO and during the winter R2 period, the highest Ф value was estimated in ♂ clone 1, whereas the lowest mean value was found in ♀ clone 12, resulting in lower Ф values in female than male leaves in both light environments (Supplementary Table 3).

The highest values of dark respiration (R _d) were registered during spring under MO (approximately 5.4 μmol m⁻² s⁻¹), whereas the mean values during the other seasons varied between 2 μmol m⁻² s⁻¹ and 3 μmol m⁻² s⁻¹ under MO (Figure 4A). Yerba mate leaves under AFS presented lower R _d values than under MO during all seasons. Clone differentiation in R _d was noticed only during the summer rest period 1 (R1) and spring growth unit 2 (GU2) periods (Supplementary Table 2). The R _d values of ♂ clone 82 and ♀ clone 12 were higher than those of ♂ clone 1 and ♀ clone 84 under both cultivation systems during the summer R1 period. On the other hand, the mean R _d value of ♀ clone 84 was higher than those of ♂ clone 82 and ♀ clone 12 under MO during the spring GU2 period. Such variation in clonal R _d responses indicates that no secondary sexual dimorphism was observed in R _d during the annual experimental period (Supplementary Table 3).

Light respiration (R _L) was lower than R _d but had a similar trend through the seasons (Figure 4; Supplementary Table 1). The only difference between the R _d and R _L trends was observed during the summer rest (R1) period when R _L did not differ among clones (Figure 4B), as happened for R _d (Figure 4A). Yerba mate leaves under AFS presented lower R _L values than under MO during all seasons (Figure 4B). A significant clone effect was registered only during the spring GU2 period when ♀ clone 84 showed the highest R _L under MO conditions. No SSD was observed in R _L during any season (Supplementary Table 3).

The highest values of maximum RuBisCO carboxylation rate (V _cmax) and maximum electron transport rate (J _max) were registered during the fall GU1 period, and the lowest ones during the spring GU2 period (Figure 5; Supplementary Table 1). Yerba mate leaves under AFS presented lower V _cmax than under MO from fall GU1 to spring GU2 period (Figure 5A; Supplementary Table 2). A similar trend was observed for J _max but with differences among clones (Figure 5B; Supplementary Table 2). Clones 1 (♂) and 84 (♀) during the fall and clone 84 (♀) during the spring GU2 had similar J _max values in both cultivation systems, whereas the other two clones showed higher J _max values in MO than in AFS. As found for R _L, no SSD was observed in V _cmax (Figure 5A) or J _max (Figure 5B) during any season (Supplementary Table 3).

The general effect of seasons on R _L/R _d or J _max/V _cmax ratios was not significant during the annual yerba mate growth (Figure 6; Supplementary Table 1), but both ratios were changed by cultivation systems in all growth periods, with the exception of the fall GU1 period (Supplementary Table 2). Considering the cultivation system effect over the seasons, the R _L/R _d was higher in MO than in AFS during the summer and winter rest pauses in all clones (Figure 6A). Clones 1 (♂) and 84 (♀) showed higher R _L/R _d under MO than AFS only during the spring GU2 period. The ♀ clone 12 showed similar R _L/R _d in both cultivation systems, while the ♂ clone 82 showed higher R _L/R _d under AFS than in MO. Such clonal variation suggested that SSD in R _L/R _d was not expressed during the spring GU2. On the other hand, SSD was expressed in R _L/R _d during the winter R2 (Supplementary Table 3), with higher values in males than in females in both cultivation systems, indicating lower inhibition of light respiration in females than in males.

In the summer R1, the J _max/V _cmax ratio was higher in MO than AFS only in ♀ clone 84, while the opposite was found in ♂ clone 82 (Figure 6B; Supplementary Table 2). During the winter R2, J _max/V _cmax was higher in MO than AFS only in ♂ clone 1, with ♀ clone 12 showing an inverse pattern (AFS > MO). During the summer R1, the males expressed a higher J _max/V _cmax ratio than females under AFS but not under MO (Supplementary Table 3).

The expression of SSD in photosynthetic performance was observed in six of eleven observed traits, mostly during rest periods (6/24 for traits with SSD expression) than during growth unit formations (3/24). The SSD expression over annual rhythmic growth was not significantly different between the periods of rest and GU formation (Supplementary Table 4).