The experiment was performed in the summer of 2015 at the Soybean Free-Air CO₂ Enrichment (SoyFACE) facility, in Champaign, IL, USA (40°2′30.5″N, 88°13′58.8″W, 230 m a.s.l.). Two tobacco cultivars N. tabacum L. cv. Petit Havana (PH) and cv. Mammoth (MM) were used. PH seeds were obtained from the Australian National University while MM seeds with variety name “Kutsaga Mammoth 10” were obtained from the US Nicotiana Germplasm Collection, North Carolina State University. Tobacco seeds were sown in seedling trays (200 cell Speedling Transplant Trays, Speedling, FL, USA) filled with a germination mix soil (Fafard, MD, USA). Germination occurred 7 days after the sowing, on the day of the year (DOY) 173 (Table S1). The environmental conditions in the greenhouse were: natural sunlight, 80% relative humidity (RH), and an average temperature of 26°C during the day and 24°C at night. The seedlings received NPK fertilization (20-10-20 Jack's General Purpose Professional, JR PETERS Inc., PA, USA) and applications of the broad spectrum systemic fungicides (Dithane, Dow Agrochemicals and Terramaster 4EC, Chemtura). After 4 weeks, the seedlings were ~5 cm tall, at which stage they were transplanted into the field.

The field contained four 20 m diameter FACE treatment areas in which [CO₂] was elevated to 600 μmol mol⁻¹ (Morgan et al., 2004). Each of these was paired with a second 20 m diameter area, which served at the ambient [CO₂] (400 μmol mol⁻¹) control of the experimental block. Within each of these areas a plot of 32 m² was allocated to the current experiment (Figure S1). Each plot was split with one half planted with cv. Petit Havana (PH) and the other half with cv. Mammoth (MM). These areas were in turn split into a control (CN) and high nitrogen treatment (HN), giving a split-split-plot design of N × cultivar × [CO₂]. The treatments are denoted as follows: ambient [CO₂] with control nitrogen (AMB CN); ambient [CO₂] with high nitrogen (AMB HN); elevated [CO₂] with control nitrogen (ELE CN); elevated [CO₂] with high nitrogen (ELE HN).

The soil in the field was cultivated to a fine tilth a week before transplantation. The split plots (subplots) were planted at a row spacing of 30.5 cm for PH (6.56 plants/m²) and 38.1 cm for MM (5.25 plants/m²), to allow for the larger size of the latter. Insecticide (Platinum 75SG, Syngenta Crop Protection Inc., NC, USA) was applied on the planting day (DOY 195, Table S1). On the day before field planting (DOY 194, Table S1), the soil inside the experimental plots was fertilized. One half of each cultivar received a standard fertilization with 150 Kg N/ha (CN) and the other 300 kg N/ha (HN) in the form of urea (UREA 46-0-0, J.R. Simplot Company, CA, USA; Figure S1). The N content of the soil before fertilization was on average 17.2 ± 3.4 ppm of NO₃ and 3.0 ± 0.6 ppm of NH₄. Four days after the addition of fertilizer, the soils ontained 55.2 ± 9.9 ppm NO₃ with 36.5 ± 13.4 ppm NH₄ (CN) and 81.8 ± 11.6 ppm of NO₃ with 83.57 ± 28.6 ppm of NH₄ (HN). Soil moisture was maintained during the experiment through drip irrigation to maintain soil volumetric water content between 32 and 40% _v/v throughout the season.

Leaf CO₂ uptake rates (A, μmol CO₂ m⁻² s⁻¹), stomatal conductances (g_s, mol H₂O m⁻² s⁻¹) and ratio of internal [CO₂] (C_i) to ambient [CO₂] (C_a) were determined on four occasions separated by 1 or 2 week intervals across the growing season (Table S1). On each of these days, measurements were made at 09:00, 12:00, and 15:00 h using open gas-exchange systems incorporating a controlled environment leaf chamber that integrated a modulated chlorophyll fluorometer (LI-6400 & LI-6400-40; LICOR, Inc., Lincoln, NE, USA). Before each time-point, the values for the photosynthetically active radiation (PAR; μmol m⁻²s⁻¹) and chamber block temperature were set at the ambient values determined from the SoyFACE weather station (sensors: LI-190; LI-COR, Inc. and HMP-45C; Campbell Scientific, Inc.). Daily air temperature and PAR across the growing season are presented in Figure S2. The reference [CO₂] in the gas-exchange systems was set to 400 μmol mol⁻¹ for the AMB treatments and to 600 μmol mol⁻¹ for the ELE treatments. At each time-point, one of four gas-exchange systems was assigned at random to each of the four blocks, to avoid confounding effects of any undetected difference among gas exchange systems with any difference among blocks. The youngest fully expanded leaves from three plants per subplot were measured. The daily mean for relative humidity inside the leaf chamber was between 57% and 69% for both cultivars. The gas exchange systems were calibrated as described previously (Bernacchi et al., 2006), and the calculation of A, g_s, and C_i/C_a followed the procedures of von Caemmerer and Farquhar (1981).

A vs. C_i response was measured and analyzed following the procedures of Long and Bernacchi (2003). These measurements were also made on the youngest fully expanded leaves in two plants per subplot. This was done three times during the season, at DOY 212, 227, and 240 (Table S1). Because of rain and the difficulty of maintaining constant conditions in the field, the measurements on DOY 227 and DOY 240 were conducted in the field laboratory within the site, rather than in situ. For the field laboratory measurements of the A/C_i curves, leaves were cut predawn and the petioles were recut under water to avoid embolism. All leaves were kept under low light until measured in the laboratory. On DOY 220 these measurements were made in situ in the field and then repeated in the laboratory the next day, and found not to differ (Figure S3). This procedure also avoids the risk of photoinhibition and transient water stress, that can occur under field conditions. All measurements were completed within 8 h of cutting and maximum rates at 8h still equaled or exceeded those measured in the field. For A/C_i measurements, the PAR level was set at 1500 μmol mol⁻¹, leaf temperature at 25°C and relative humidity was set between 60 and 70%. Photosynthesis was induced to steady state in the ambient [CO₂] of 400 μmol mol⁻¹. The chamber [CO₂] was varied through the following step-wise sequence and waiting for steady-state A to be achieved at each step: 400, 300, 200, 100, 400, 400, 500, 600, 800, 1,000, 1,200, and 1,500 μmol mol⁻¹. The maximum rate of Rubisco-catalyzed carboxylation (V_cmax; μmol m⁻²s⁻¹), the regeneration of ribulose-1,5-biphosphate controlled by electron transport rate (J_max; μmol m⁻²s⁻¹) were calculated from the A/C_i response for each individual using the equations from Farquhar et al. (1980) and Bernacchi et al. (2001). Estimates of V_cmax from A/C_i curves are in effect “apparent V_cmax” since they do not take account of mesophyll conductance (g_m).

The g_m (mol m² s⁻¹) and partial preasure of CO₂ inside the chloroplast (C_c; μmol mol⁻¹) were calculate using variable J method as described in Harley et al. (1992) for DOY 240, the only day where chlorophyll fluorescence data was colletect. V_cmax, J_max, and day respiration independent of photorespiration (R_d) were calculated from A/C_c curves as in Harley et al. (1992) and by using a nonlinear analysis with the Marquardt method (PROC NLIN, SAS System 9.4, SAS Institute, Cary, NC, USA; Moualeu-Ngangue et al., 2017). The scaling constant (c) and enthalpies of activation (ΔH_a) for fitting the Michaelis constant of Rubisco for carbon dioxide (K_c; μmol mol⁻¹), inhibition constant (K_o; μmol mol⁻¹), and photorespiratory compensation point (T^*; μmol mol⁻¹) at the measured leaf temperature as well as for normalizing V_cmax and J_max at 25°C were obtained from Sharkey et al. (2007).

Measurements for plant height (cm) were taken at approximately 1 week intervals, from the soil to the tip of the newest leaf in three plants in each subplot. The specific leaf area (SLA, m² kg⁻¹) was calculated from three plants per subplot. Leaf disks of 1.9 cm of diameter were collected at midday on each day of the diurnal measurements of leaf gas exchange measurements. The samples were oven dried to constant weight at 60°C. These leaf disks were subsequently used for CN analysis. The tissue was ground to a powder and a ~2.0 mg sub-sample processed through an elemental analyzer (Elemental Combustion System CHNS-O, Costech ECS 4010, Valencia, CA, USA).

Leaf disk samples (~1.2 cm diameter) were collected from two plants per subplot at noon and dusk on the same days as the diurnal leaf gas exchange measurements. Dawn samples were also collected one of these days, DOY 227 (Table S1). The samples were cut into liquid N in the field and stored at −80°C until analysis. Soluble carbohydrates determination followed the procedures of Ainsworth et al. (2007), excepting the use of an additional ethanol wash. Glucose, fructose, sucrose and starch were expressed as glucose equivalents. Total soluble carbohydrates (TSC) were calculated as the sum of glucose, fructose, and sucrose.

Two harvests were conducted to determine productivity; one in the middle of the season (starting on DOY 222, ~3.5 weeks after field planting), and one just before the PH pods opened (starting on DOY 243, 7 weeks after field planting, Table S1). The above-ground biomass of five plants per subplot was collected in each harvest. The number of leaves, branches and the total leaf area (total LA in cm²; measured using an LI-3100, LICOR, Inc.) for each plant were also determined. The roots of the harvested plants were collected a day after the above-ground biomass harvest. The roots of 3 plants per subplot were taken in the 1st harvest and of 2 plants per subplot in the 2nd harvest. Soil was washed from the roots, and these together with leaves, stems, and floral structures then dried to constant weight at 60°C.

All the variables were analyzed as a split-split-plot design using a mixed model analysis of variance (ANOVA; PROC MIXED, SAS System 9.4, SAS Institute, Cary, NC, USA) with repeated measurements, except for the daily analysis of V_cmax and J_max, which were analyzed without repeated measurements. The fixed effects were time, [CO₂], nitrogen (N), and their interactions. Time was the repeated measurement, and it refers to DOY for the seasonal analysis and to the time-points (3 in each day of measurements) for the daily analysis of the leaf gas exchange variables (A, g_s, and C_i/C_a). The random effect for all the variables was the block. The calculation of the degrees of freedom was done using the Kenward–Roger method. The least square means test (t-test) was used to obtain the pair-wise comparisons, considering the statistical significance as P-value ≤ 0.1. Because of the low replication (n = 4), P ≤ 0.1 was used as the threshold for significance to avoid the risk of type II errors, as described in previous FACE analyses of photosynthesis (e.g., Morgan et al., 2004). The biomass data from one AMB plot was identified as an outlier, so the data from this plot were excluded from the analysis.

Leaf CO₂ uptake (A) of the most recently expanded leaf, averaged over the growing season, was significantly higher in elevated [CO₂] (ELE) regardless of cultivar and nitrogen treatment (+11% in PH and +13% in MM). However, this stimulation by ELE was lost in cv. Petit Havanna (PH) by the last measurement date regardless of N treatment (DOY 239), but not in cv. Mammoth (MM) (Figure 1; Table S2). Stomatal conductance (g_s) was decreased throughout the season by ELE on average by 32% in both cultivars (Figure 1; Table S2). Mesophyll conductance (g_m) decreased 23% under ELE later in the season only in PH (DOY 240; Table S3). Across the season, C_i/C_a was modestly but significantly decreased by ELE in PH (−4.5%; Figure 1; Table S2), but not in MM. There was no effect of HN on A, g_s, and C_i/C_a in either cultivar (Table S2) with the exception of A on DOY 211 in MM and of C_i/C_a on DOY 211 in PH, which slightly increased with HN by 6.5 and 4% respectively.

During two of three sets of A/C_i measurements (Table S4), V_cmax and J_max were down-regulated by elevated [CO₂] in both cultivars. In PH, V_cmax and J_max values obtained from A/Cc curves were, on average, 80% higher and 10% lower, respectively, than those obtained from A/C_i curves on DOY 240 (Tables S3, S4). For MM, V_cmax from A/Cc curves was ~120% higher and J_max did not change compared to the values calculated from A/C_i curves. However, V_cmax and J_max for either A/C_i and A/C_c curves showed that control nitrogen (CN) treatment had a strong down-regulation in PH, where both the initial Rubisco-limited (V_cmax) slope of the response and the upper RuBP regeneration-limited (J_max) portion of the response were lower in the ELE plants of this cultivar (Figure 2; Table S4). This down-regulation in PH was so severe on DOY 240, that A at the growth [CO₂], as indicated by the intercept of the supply and demand functions, was lower in ELE than in AMB (Figure 2). HN only partially ameliorated this (Figure 3). In contrast, only a slight down-regulation was evident in MM (Figures 2, 3; Table S3). The operating C_i (i.e., the C_i obtained when the chamber C_a equals the current atmospheric level of 400 μmol mol⁻¹) in ELE (see the supply function lines for g_s in Figures 2, 3) is on the J_max limited portion of the response indicating that RubP regeneration was the cause of the observed down-regulation of measured A.

Leaf N (g m⁻²) was increased ~10% by the HN treatments in both cultivars. HN had a small but significant effect in leaf C by increasing it 5% in PH and reducing it 6% in MM (Table 1). Leaf C (g m⁻²) was also affected by ELE, which increased by 27% in PH and 31% in MM. Total soluble carbohydrates (TSC; mmol m⁻²) and starch (mmol m⁻²) were significantly higher at noon in ELE compared to AMB, regardless of the cultivar or N treatment. TSC at noon increased by 31% and starch content at noon was almost doubled in both cultivars under ELE (Table 1). By dusk TSC declined by ~50% compared to noon across the treatments while starch remained high in ELE treatments (+46% in PH and +55% in MM). At dawn on DOY 227, TSC and starch were higher in PH in ELE with or without HN compared to control (Table 1).

Specific leaf area (SLA) decreased ~24% in both cultivars grown in ELE (Table 1). Additionally, SLA decreased slightly (−6%) in PH at HN. Height in PH increased under ELE but decreased at HN, resulting in taller plants at ELE CN than ELE HN (Figure S4 and Table S5). In MM, height was affected by the interaction [CO₂] × N such that AMB CN plants were taller than AMB HN and ELE CN plants (Figure S4 and Table S5). No significant effects of ELE or HN were found on the number of leaves or branches for either cultivar (Figure S5 and Table S5). However, there was a significant interaction between [CO₂] and N in the total LA in PH from both harvests (Table S5). Thus, total LA in PH was higher in AMB CN compared to AMB HN on DOY 222, and higher in AMB HN and ELE CN compared to ELE HN on DOY 243 (Figure S5).

The yield component of this crop, leaf biomass, was three times greater in MM by comparison to PH at final harvest (DOY 243), regardless of [CO₂] or N treatment (Figure 4). On this same date, the amount of leaf produced was significantly higher in ELE in both cultivars, but only in CN. Thus, the increase of leaves in PH was 41% and in MM was 65% comparing ELE CN and AMB CN (Figure 4). ELE did not result in significant differences in total biomass in MM by DOY 222 but did by the final harvest on DOY 243 (Figure 4). Curiously, HN had a lesser effect than CN in increasing the total biomass of PH under ELE. Notably, root biomass in ELE HN was less than half of that in CN (−63%; Figure 4).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.