“Nitrogen demand, supply, and acquisition strategy control plant responses to elevated CO2 at different scales”

Plants respond to elevated atmospheric CO2 concentrations by reducing leaf nitrogen content and photosynthetic capacity – patterns that correspond with increased net photosynthesis rates, total leaf area, and total biomass. Nitrogen supply has been hypothesized to be the primary factor controlling these responses, as nitrogen availability limits net primary productivity globally. Recent work using evo-evolutionary optimality theory suggests that leaf photosynthetic responses to elevated CO2 are independent of nitrogen supply and are instead driven by leaf nitrogen demand to build and maintain photosynthetic enzymes, which optimizes resource allocation to photosynthetic capacity and maximizes allocation to growth. Here, Glycine max L. (Merr) seedlings were grown under two CO2 concentrations, with and without inoculation with Bradyrhizobium japonicum, and across nine soil nitrogen fertilization treatments in a full-factorial growth chamber experiment to reconcile the role of nitrogen supply and demand on leaf and whole-plant responses to elevated CO2. After seven weeks, elevated CO2 increased net photosynthesis rates despite reduced leaf nitrogen content and maximum rates of Ribulose-1,5-bisphosphate (RuBP) carboxylase/oxygenase (Rubisco) carboxylation and electron transport for RuBP regeneration. Effects of elevated CO2 on net photosynthesis and indices of photosynthetic capacity were independent of nitrogen fertilization and inoculation. However, increasing nitrogen fertilization enhanced positive effects of elevated CO2 on total leaf area and total biomass due to increased nitrogen uptake and reduced carbon costs to acquire nitrogen. Whole-plant responses to elevated CO2 were not modified by inoculation across the nitrogen fertilization gradient, as plant investment toward symbiotic nitrogen fixation was similar between CO2 treatments. These results indicate that leaf nitrogen demand to build and maintain photosynthetic enzymes drives leaf photosynthetic responses to elevated CO2, while nitrogen supply regulates whole-plant responses. Our findings build on previous work suggesting that terrestrial biosphere models may improve simulations of photosynthetic processes under future novel environments by adopting optimality principles.


Introduction
Terrestrial ecosystems are regulated by complex carbon and nitrogen cycles.As a result, terrestrial biosphere models, which are beginning to include coupled carbon and nitrogen cycles (Shi et al., 2016;Davies-Barnard et al., 2020;Braghiere et al., 2022), must accurately represent these cycles under different environmental scenarios to reliably simulate carbon and nitrogen fluxes (Oreskes et al., 1994;Prentice et al., 2015).While the inclusion of coupled carbon and nitrogen cycles in terrestrial biosphere models was intended to improve model reliability, large uncertainty in the role of nitrogen availability and nitrogen acquisition strategy on leaf and whole plant responses to increasing atmospheric CO2 concentrations persists (Arora et al., 2020;Davies-Barnard et al., 2020;Kou-Giesbrecht et al., 2023), contributing to widespread divergence in future carbon and nitrogen flux simulations across terrestrial biosphere models (Hungate et al., 2003;Friedlingstein et al., 2014;Zaehle et al., 2014;Wieder et al., 2015;Meyerholt et al., 2020).
Despite consistent plant responses to elevated CO2 documented across experiments, mechanisms that drive these responses remain unresolved.Some have hypothesized that plant responses to elevated CO2 are constrained by nitrogen availability, as net primary productivity is limited by nitrogen availability globally (Vitousek & Howarth, 1991;LeBauer & Treseder, 2008).The progressive nitrogen limitation hypothesis predicts that elevated CO2 will increase plant nitrogen uptake to support greater net primary productivity, which will cause nitrogen availability to decline over time (Luo et al., 2004).The hypothesis predicts that this response should increase growth and net primary productivity under elevated CO2 over short time scales that dampen with time as nitrogen becomes progressively more limiting and stored in longerlived tissues.Growth responses to elevated CO2 expected from the progressive nitrogen limitation hypothesis have received some support from free-air CO2 enrichment experiments (Reich et al., 2006;Norby et al., 2010), though these patterns are not consistently observed (Finzi et al., 2006(Finzi et al., , 2007;;Moore et al., 2006;Liang et al., 2016).
Assuming positive relationships between soil nitrogen availability, leaf nitrogen content, and photosynthetic capacity (Field & Mooney, 1986;Evans, 1989;Evans & Seemann, 1989;Walker et al., 2014;Firn et al., 2019;Liang et al., 2020), the progressive nitrogen limitation hypothesis implies that reductions in nitrogen availability over time might explain why C3 plants exhibit decreased leaf nitrogen content and photosynthetic capacity under elevated CO2.
However, results from free-air CO2 enrichment experiments show that reductions in leaf nitrogen content and photosynthetic capacity under elevated CO2 are decoupled from changes in nitrogen availability (Crous et al., 2010;Lee et al., 2011;Pastore et al., 2019).Additionally, variance in leaf nitrogen and photosynthetic capacity across environmental gradients tends to be more strongly determined through aboveground growth conditions that set demand to build and maintain photosynthetic enzymes than through changes in soil resource availability (Dong et al., 2017(Dong et al., , 2020(Dong et al., , 2022a;;Smith et al., 2019;Smith & Keenan, 2020;Paillassa et al., 2020;Peng et al., 2021;Querejeta et al., 2022;Westerband et al., 2023;Waring et al., 2023).These patterns indicate that leaf photosynthetic responses to elevated CO2 may be a product of altered leaf nitrogen demand to build and maintain photosynthetic enzymes and may not be as strongly linked to changes in nitrogen availability.
Eco-evolutionary optimality theory provides a framework for understanding how leaf photosynthetic responses to elevated CO2 may be determined through demand to build and maintain photosynthetic enzymes (Harrison et al., 2021).Merging photosynthetic least-cost (Wright et al., 2003;Prentice et al., 2014) and optimal coordination (Chen et al., 1993;Maire et al., 2012) theories, eco-evolutionary optimality theory posits that reduced leaf nitrogen allocation under elevated CO2 is the downstream result of a stronger downregulation in the maximum rate of Ribulose-1,5-bisphosphate (RuBP) carboxylase/oxygenase (Rubisco) carboxylation (Vcmax) than the maximum rate of electron transport for RuBP regeneration (Jmax), which reduces leaf nitrogen demand to build and maintain photosynthetic enzymes.Optimal leaf nitrogen allocation to photosynthetic capacity allows plants to make more efficient use of available light while avoiding overinvestment in Rubisco, which has high nitrogen and energetic costs of construction and maintenance (Evans, 1989;Sage, 1994;Evans & Clarke, 2019).Such optimal leaf nitrogen allocation responses to elevated CO2 increases photosynthetic nitrogen-use efficiency and allows increased net photosynthesis rates to be achieved through increasingly equal co-limitation of Rubisco carboxylation and electron transport for RuBP regeneration (Chen et al., 1993;Maire et al., 2012;Wang et al., 2017;Smith et al., 2019).The expected optimal leaf response to elevated CO2 has received some empirical support (Crous et al., 2010;Lee et al., 2011;Smith & Keenan, 2020;Harrison et al., 2021;Dong et al., 2022b;Cui et al., 2023), though no studies have connected these patterns with concurrently measured whole-plant responses.
The eco-evolutionary optimality hypothesis deviates from the progressive nitrogen limitation hypothesis by indicating that photosynthetic responses to elevated CO2 are driven by leaf nitrogen demand to build and maintain photosynthetic enzymes and are independent of changes in soil nitrogen supply.However, the eco-evolutionary optimality hypothesis does not discount the role of soil nitrogen availability on whole-plant responses to elevated CO2, where the expected optimal strategy in response to elevated CO2 is to allocate surplus nitrogen not needed to satisfy leaf nitrogen demand toward the construction of a greater quantity of optimally coordinated leaves and other plant organs.Thus, whether the supply-driven progressive nitrogen limitation hypothesis or demand-driven eco-evolutionary optimality hypothesis controls plant responses to elevated CO2 may be a matter of scale, where leaf photosynthetic responses to elevated CO2 are determined through demand to build and maintain photosynthetic enzymes and whole-plant responses to elevated CO2 are regulated by changes in nitrogen supply.
Plants allocate carbon belowground in exchange for nutrients through different nutrient acquisition strategies, including direct uptake pathways or symbioses with mycorrhizal fungi and symbiotic nitrogen-fixing bacteria (Gutschick, 1981;Smith & Read, 2008).Carbon costs to acquire nitrogen, or the amount of carbon allocated belowground per unit nitrogen acquired, vary in species with different nitrogen acquisition strategies and are dependent on environmental factors such as atmospheric CO2, temperature, light availability, and nutrient availability (Brzostek et al., 2014;Terrer et al., 2018;Allen et al., 2020;Eastman et al., 2021;Perkowski et al., 2021;Lu et al., 2022;Peng et al., 2023).Therefore, nitrogen acquisition strategy cannot be ignored when considering effects of nitrogen availability on plant responses to elevated CO2.To date, few studies account for acquisition strategy when considering the role of nitrogen availability on leaf and whole-plant responses to elevated CO2 (e.g., Terrer et al., 2016Terrer et al., , 2018;;Smith & Keenan, 2020).Such studies found that nitrogen acquisition strategies with reduced carbon costs to acquire nitrogen may buffer the effect of nitrogen limitation at the whole-plant level (Terrer et al., 2018), but leaf-level responses remain inconsistent (Terrer et al., 2018;Smith & Keenan, 2020).
Here, we conducted a growth chamber experiment using Glycine max L. (Merr.)seedlings grown under full factorial combinations of two CO2 concentrations, two inoculation treatments, and nine soil nitrogen fertilization treatments to reconcile the role of nitrogen supply and demand on plant responses to elevated CO2.We used this experimental setup to test the following hypotheses: (1) Following the demand-driven eco-evolutionary optimality hypothesis, elevated CO2 will downregulate Vcmax more strongly than Jmax, increasing Jmax:Vcmax and allowing increased net photosynthesis rates to approach equal co-limitation of Rubisco carboxylation and electron transport for RuBP regeneration.Leaf photosynthetic responses to elevated CO2 will be independent of nitrogen fertilization and inoculation treatment and will correspond with increased photosynthetic nitrogen-use efficiency.
(2) Following the supply-driven nitrogen limitation hypothesis, positive effects of elevated CO2 on total leaf area and total biomass will be enhanced with increasing nitrogen fertilization due to increased plant nitrogen uptake and reduced carbon costs to acquire nitrogen.Inoculation with symbiotic nitrogen-fixing bacteria will enhance positive growth responses to elevated CO2, though these responses will only be apparent under low nitrogen fertilization levels where individuals will have increased investment in nitrogen acquisition through symbiotic nitrogen fixation.
Before planting, all G. max seeds were surface sterilized in 2% sodium hypochlorite for 3 minutes, followed by three separate 3-minute washes with ultrapure water (MilliQ 7000; MilliporeSigma, Burlington, MA USA).Subsets of surface-sterilized seeds were inoculated with Bradyrhizobium japonicum (Verdesian N-Dure™ Soybean, Cary, NC, USA) in a slurry following manufacturer recommendations (3.12 g inoculant and 241 g ultrapure water per 1 kg seed).
Seventy-two pots were randomly planted with surface-sterilized seeds inoculated with B. japonicum, while the remaining 72 pots were planted with surface-sterilized uninoculated seeds.

Growth chamber conditions
Plants were randomly placed in one of six Percival LED-41L2 growth chambers (Percival Scientific Inc., Perry, IA, USA) over two experimental iterations due to chamber space limitation.Two iterations were conducted such that one iteration included all plants grown under elevated CO2 plants, and the second iteration included all plants grown under ambient CO2.
Daytime growth conditions were simulated using a 16-hour photoperiod, with incoming light radiation set to chamber maximum (mean±SD: 1230±12 μmol m -2 s -1 across chambers), air temperature set to 25°C, and relative humidity set to 50%.The remaining 8-hour period simulated nighttime growing conditions, with incoming light radiation set to 0 μmol m -2 s -1 , chamber temperature set to 17°C, and relative humidity set to 50%.Transitions between daytime and nighttime growing conditions were simulated by ramping incoming light radiation in 45minute increments and temperature in 90-minute increments over a 3-hour period (Table S2).
Plants grew under average (± SD) daytime light intensity of 1049±27 μmol m -2 s -1 , including ramping periods.In the elevated CO2 iteration, plants grew under 24.0±0.2°Cduring the day, 16.4±0.8°Cduring the night, and 51.6±0.4% relative humidity.In the ambient CO2 iteration, plants grew under 23.9±0.2°Cduring the day, 16.0±1.4°Cduring the night, and 50.3±0.2%relative humidity.Within each experiment iteration, any differences in climate conditions across the six chambers were accounted for by shuffling the same group of plants throughout the growth chambers.This process was done by iteratively moving the group of plants on the top rack of a chamber to the bottom rack of the same chamber, while simultaneously moving the group of plants on the bottom rack of a chamber to the top rack of the adjacent chamber.Plants were moved within and across chambers daily during each experiment iteration.

Leaf gas exchange measurements
Leaf gas exchange measurements were collected on the seventh week of development, before the onset of reproduction.All gas exchange measurements were collected on the center leaf of the most recent fully expanded trifoliate leaf set using LI-6800 portable photosynthesis machines configured with a 6800-01A fluorometer head and 6 cm 2 aperture (LI-COR Biosciences, Lincoln, NE, USA).Specifically, net photosynthesis (Anet; μmol m -2 s -1 ), stomatal conductance (gsw; mol m -2 s -1 ), and intercellular CO2 (Ci; μmol mol -1 ) concentrations were measured across a range of atmospheric CO2 concentrations (i.e., an Anet/Ci curve) using the Dynamic AssimilationÔ Technique.The Dynamic AssimilationÔ Technique corresponds well with traditional steady-state Anet/Ci curves in G. max (Saathoff & Welles, 2021).Anet/Ci curves were generated along a reference CO2 ramp down from 420 µmol mol -1 CO2 to 20 µmol mol -1 CO2, followed by a ramp up from 420 µmol mol -1 CO2 to 1620 µmol mol -1 CO2 after a 90-second wait period at 420 µmol mol -1 CO2.The ramp rate for each curve was set to 200 μmol mol -1 min -1 , logging every five seconds, which generated 96 data points per response curve.All Anet/Ci curves were generated after Anet and gsw stabilized in a LI-6800 cuvette set to a 500 mol s -1 flow rate, 10000 rpm mixing fan speed, 1.5 kPa vapor pressure deficit, 25°C leaf temperature, 2000 μmol m -2 s -1 incoming light radiation, and initial reference CO2 set to 420 µmol mol -1 .Snapshot Anet measurements were extracted from each Anet/Ci curve, both at a common CO2 concentration, 420 µmol mol -1 CO2 (Anet,420; μmol m -2 s -1 ), and under each individual's growth CO2 concentration, 420 and 1000 µmol mol -1 CO2 (Anet,growth; μmol m -2 s -1 ).Dark respiration (Rd; μmol m -2 s -1 ) measurements were collected with the same leaf used to generate Anet/Ci curves following at least 30 minutes of darkness.Measurements were collected on a 5second log interval for 60 seconds after the leaf stabilized in a LI-6800 cuvette set to a 500 mol s -1 flow rate, 10000 rpm mixing fan speed, 1.5 kPa vapor pressure deficit, 25°C leaf temperature, and 420 µmol mol -1 reference CO2 concentration (regardless of CO2 treatment), with incoming light radiation set to 0 μmol m -2 s -1 .A single dark respiration value was determined for each leaf by calculating the mean dark respiration value across the logging interval.

A/Ci curve-fitting and parameter estimation
Anet/Ci curves were fit using the 'fitaci' function in the 'plantecophys' R package (Duursma, 2015).This function estimates the maximum rate of Rubisco carboxylation (Vcmax; µmol m -2 s -1 ) and maximum rate of electron transport for RuBP regeneration (Jmax; µmol m -2 s -1 ) based on the Farquhar et al. (1980) biochemical model of C3 photosynthesis.Triose phosphate utilization (TPU) limitation was included as an additional rate-limiting step in all curve fits after visually observing clear TPU limitation for most curves.All curve fits included measured dark respiration values.As Anet/Ci curves were generated using a common leaf temperature (25°C), curves were fit using Michaelis-Menten coefficients for Rubisco affinity to CO2 (Kc; μmol mol -1 ) and O2 (Ko; mmol mol -1 ), and the CO2 compensation point (Γ * ; μmol mol -1 ) reported in Bernacchi et al. (2001).Specifically, Kc was set to 404.9 μmol mol -1 , Ko was set to 278.4 μmol mol -1 , and Γ * was set to 42.75 μmol mol -1 .For clarity, Vcmax, Jmax, and Rd estimates are referenced throughout the rest of the paper as Vcmax25, Jmax25, and Rd25.

Leaf trait measurements
The leaf used to generate Anet/Ci curves and dark respiration measurements was harvested immediately following gas exchange measurements.Images of each focal leaf were curated using a flat-bed scanner to determine fresh leaf area using the 'LeafArea' R package (Katabuchi, 2015), which automates leaf area calculations using ImageJ software (Schneider et al., 2012).
Post-processed images were visually assessed to check against errors in the automation process.
Each leaf was dried at 65°C for at least 48 hours and subsequently weighed and ground until homogenized.Leaf mass per area (Marea; g m -2 ) was calculated as the ratio of dry leaf biomass to fresh leaf area.Leaf nitrogen content (Nmass; gN g -1 ) was quantified using a subsample of ground and homogenized leaf tissue through elemental combustion analysis (Costech-4010, Costech, Inc., Valencia, CA, USA).Leaf nitrogen content per unit leaf area (Narea; gN m -2 ) was calculated by multiplying Nmass and Marea.Photosynthetic nitrogen-use efficiency (PNUEgrowth; µmol CO2 g -1 N s -1 ) was estimated as the ratio of Anet,growth to Narea.
Chlorophyll content was extracted from a second leaf in the same trifoliate leaf set as the leaf used to generate Anet/Ci curves.A cork borer was used to punch between 3-5 0.6 cm 2 disks from the leaf.Images of each set of leaf disks were curated using a flat-bed scanner to determine wet leaf area, again quantified using the 'LeafArea' R package (Katabuchi, 2015).Leaf disks were shuttled into a test tube containing 10 mL dimethyl sulfoxide, vortexed, and incubated at 65°C for 120 minutes (Barnes et al., 1992).Incubated test tubes were vortexed again before being loaded in 150 μL triplicate aliquots to a 96-well plate.Dimethyl sulfoxide was loaded in each plate as a single 150 μL triplicate aliquot and used as a blank.Absorbance measurements at 649 nm (A649) and 665 nm (A665) were recorded in each well using a plate reader (Biotek Synergy H1; Biotek Instruments, Winooski, VT USA), with triplicates averaged and corrected by the mean of the blank absorbance value.Blank-corrected absorbance values were used to estimate Chla (μg mL -1 ) and Chlb (μg mL -1 ) following equations from Wellburn (1994): and Chla and Chlb were converted to mmol mL -1 using the molar masses of chlorophyll a (893.51g mol -1 ) and chlorophyll b (907.47 g mol -1 ), then added together to calculate the total chlorophyll content in dimethyl sulfoxide extractant (mmol mL -1 ).Total chlorophyll content (mmol) was determined by multiplying the total chlorophyll content in dimethyl sulfoxide by the volume of dimethyl sulfoxide extractant (10 mL).Area-based chlorophyll content (Chlarea; mmol m -2 ) was then calculated by dividing the total chlorophyll content by the total area of the leaf disks.
Subsamples of ground and homogenized leaf tissue were sent to the University of California-Davis Stable Isotope Facility to determine leaf δ 13 C and δ 15 N using an elemental analyzer (Elementar vario MICRO cube elemental analyzer; Elementar Analysensysteme GmbH, Langenselbold, Germany) interfaced to an isotope ratio mass spectrometer (PDZ Europa 20-20 Isotope Ratio Mass Spectrometer, Sercon Ltd., Chestshire, UK).Leaf δ 13 C was used to estimate the time-integrated ratio of leaf intercellular CO2 concentration to atmospheric CO2 concentration (χ, unitless) using leaf δ 13 C and chamber air δ 13 C following Farquhar et al. (1989): where Δ 13 C represents the relative difference between leaf δ 13 C (‰) and air δ 13 C (‰), and is calculated as: δ 13 Cair is the chamber δ 13 C air fractionation, a represents the fractionation between 12 C and 13 C due to diffusion in air, assumed to be 4.4‰, and b represents the fractionation caused by Rubisco carboxylation, assumed to be 27‰ (Farquhar et al., 1989).δ 13 Cair was quantified in each chamber by collecting air samples in triplicate for each CO2 treatment using a 20 mL syringe (Air-Tite Products Co., Inc., Virginia Beach, VA, USA).Each air sample was plunged into a manually evacuated 10 mL Exetainer (Labco Ltd., Lampeter, UK) and sent to the University of California-Davis Stable Isotope Facility, where δ 13 Cair was determined using a gas inlet system (GasBenchII; Thermo Fisher Scientific, Waltham, MA, USA) coupled to an isotope ratio mass spectrometer (Thermo Finnigan Delta Plus XL; Thermo Fisher Scientific, Waltham, MA, USA).
δ 13 Cair for each CO2 treatment was estimated by calculating the mean of the triplicate δ 13 Cair samples within each chamber, then calculating the mean δ 13 Cair across all chambers.Specifically, δ 13 Cair was -8.81‰ for the ambient CO2 treatment and -5.95‰ for the elevated CO2 treatment.
Finally, the percent of leaf nitrogen acquired from the atmosphere (%Ndfa; %) was estimated using leaf δ 15 N and the following equation adapted from Andrews et al. (2011): was not calculated within each unique nitrogen fertilization-by-CO2 treatment combination, as previous studies suggest decreased reliance on nitrogen fixation with increasing nitrogen fertilization (e.g., Perkowski et al., 2021).

Whole-plant measurements
Seven weeks after experiment initiation and immediately following gas exchange measurements, all individuals were harvested, and biomass of major organ types (leaves, stems, roots, and nodules when present) were separated.Fresh leaf area of all harvested leaves was measured using a LI-3100C (LI-COR Biosciences, Lincoln, Nebraska, USA).Total fresh leaf area (cm 2 ) was calculated as the sum of all leaf areas, including the leaf used to collect gas exchange data and the leaf used to extract chlorophyll content.All harvested material was dried in an oven set to 65°C for at least 48 hours to a constant mass, weighed, and ground to homogeneity.Leaves and root nodules were ground using a mortar and pestle, while stems and roots were ground using an E3300 Single Speed Mini Cutting Mill (Eberbach Corp., MI, USA).Total biomass (g) was calculated as the sum of dry leaf, stem, root, and root nodule biomass.Carbon and nitrogen content was measured for each organ type through elemental combustion (Costech-4010, Costech, Inc., Valencia, CA, USA) using subsamples of ground and homogenized organ tissue.
The ratio of root nodule biomass to root biomass was calculated as an additional indicator of investment toward symbiotic nitrogen fixation.
Following Perkowski et al. (2021), carbon costs to acquire nitrogen were quantified as the ratio of belowground carbon biomass to total nitrogen biomass (Ncost; gC gN -1 ).Belowground carbon biomass (Cbg; gC) was calculated as the sum of root carbon biomass and root nodule carbon biomass.Root carbon biomass and root nodule carbon biomass were calculated as the product of the organ biomass and respective organ carbon content.Total nitrogen biomass (Nwp; gN) was calculated as the sum of total leaf, stem, root, and root nodule nitrogen biomass.Leaf, stem, root, and root nodule nitrogen biomass was calculated as the product of the organ biomass and respective organ nitrogen content.This calculation does not account for additional carbon costs associated with respiration, root exudation, or root turnover, and therefore may underestimate carbon costs to acquire nitrogen (Perkowski et al., 2021).

Statistical analyses
Uninoculated plants that had substantial root nodule formation (root nodule biomass: root biomass values greater than 0.05 g g -1 ) were removed from analyses under the assumption that A series of linear mixed-effects models were built to investigate the impacts of CO2 concentration, nitrogen fertilization, and inoculation on G. max leaf nitrogen allocation, gas exchange, whole-plant growth, and investment in nitrogen fixation.All models included CO2 treatment as a categorical fixed effect, inoculation treatment as a categorical fixed effect, and nitrogen fertilization as a continuous fixed effect, with all possible interaction terms between all three fixed effects also included.Models accounted for climatic differences between chambers across experiment iterations by including a random intercept term that nested the starting chamber rack by CO2 treatment.Models with this independent variable structure were created for each of the following dependent variables: Narea, Marea, Nmass, Chlarea, Anet,420, Anet,growth, Vcmax25, Jmax25, Jmax25:Vcmax25, Rd25, PNUEgrowth, χ, total leaf area, total biomass, Ncost, Cbg, Nwp, %Ndfa, rood nodule biomass: root biomass, root nodule biomass, and root biomass.
Shapiro-Wilk tests of normality were used to assess whether linear mixed-effects models satisfied residual normality assumptions.All models that did not satisfy residual normality assumptions satisfied such assumptions when response variables were fit using either a natural log or square root data transformation (Shapiro-Wilk: p>0.05 in all cases).Specifically, models for Narea, Nmass, Chlarea, Anet,420, Anet,growth, Vcmax25, Jmax25, Jmax25:Vcmax25, Rd25, PNUEgrowth, χ, total leaf area, and Ncost each satisfied residual normality assumptions without data transformation.
Models for Marea, total biomass, and Cbg satisfied residual normality assumptions with a natural log data transformation, while models for Nwp, root nodule biomass: root biomass, root nodule biomass, root biomass, and %Ndfa satisfied residual normality assumptions with a square root data transformation.
In all models, we used the 'lmer' function in the 'lme4' R package (Bates et al., 2015) to fit each model and the 'Anova' function in the 'car' R package (Fox & Weisberg, 2019) to calculate Type II Wald's χ 2 and determine the significance (α=0.05) of each fixed effect coefficient.We used the 'emmeans' R package (Lenth, 2019) to conduct post-hoc comparisons using Tukey's tests, where degrees of freedom were approximated using the Kenward-Roger approach (Kenward & Roger, 1997).Trendlines and error ribbons representing the 95% confidence intervals were drawn in all figures using 'emmeans' outputs across the range in nitrogen fertilization values.All analyses and plots were conducted in R version 4.1.0(R Core Team, 2021).Model results for χ, Cbg, Nwp, root nodule biomass: root biomass, root nodule biomass, and root biomass are reported in the Supplemental Material (Tables S3-S6; Figs.S3-S6).p<0.05 in all cases).These responses resulted in a stronger reduction in Narea and Nmass and a stronger increase in Marea under elevated CO2 with increasing nitrogen fertilization than ambient CO2 (Fig. S1).Nitrogen fertilization did not modify reductions in Chlarea due to elevated CO2
An interaction between inoculation and CO2 (p<0.05;tests (α=0.05).P-values less than 0.05 are in bold.A superscript " a " is included after trait labels to indicate if models were fit with natural log-transformed response variables.Key: df=degrees of freedom, χ 2 =Wald chisquare test statistic, Narea=leaf nitrogen content per unit leaf area (gN m -2 ), Nmass=leaf nitrogen content (gN g -1 ), Marea=leaf mass per unit leaf area (g m -2 ).
Inoculation did not modify Anet,420 responses to elevated CO2 (CO2-by-inoculation interaction: p>0.05).However, an interaction between CO2 and inoculation (p<0.05;Table 2) indicated that inoculated plants experienced a stronger increase in Anet,growth due to elevated CO2 (38% increase; Tukey test of the CO2 effect in inoculated plants: p<0.001) compared to uninoculated plants (26% increase; Tukey test of the CO2 effect in uninoculated plants: p<0.05).An interaction between nitrogen fertilization and inoculation (p<0.001 in both cases; Table 2) indicated that positive effects of increasing nitrogen fertilization on Anet,420 and Anet,growth (p<0.001 in both cases; Table 2; Fig. 2a-b) were stronger in uninoculated plants than inoculated plants (Tukey test comparing the nitrogen fertilization-trait slope between inoculation treatments: p<0.001 in both cases).Elevated CO2 decreased Vcmax25 and Jmax25 by 16% and 10%, respectively, increasing Jmax25:Vcmax25 by 8% (p<0.05 in all cases; Table 2; Fig. 2c-e).Vcmax25, Jmax25, and Jmax25:Vcmax25 responses to elevated CO2 were not modified by nitrogen fertilization (CO2-by-nitrogen fertilization interaction: p>0.05 in all cases; Table 2; Fig. 2c-e) or inoculation (CO2-byinoculation interaction: p>0.05 in all cases; Table 2).An interaction between nitrogen fertilization and inoculation (p<0.05 in both cases; Table 2) indicated that positive effects of increasing nitrogen fertilization on Vcmax25 and Jmax25 (p<0.001 in both cases; Table 2) and negative effects of increasing nitrogen fertilization on Jmax25:Vcmax25 (p<0.001;Table 2) were driven by uninoculated plants (Tukey test of the nitrogen fertilization-trait slope in uninoculated plants: p<0.001 in all cases), as there was no effect of nitrogen fertilization on Vcmax25, Jmax25, or Jmax25:Vcmax25 in inoculated plants (Tukey test of the nitrogen fertilization-trait slope in inoculated plants: p>0.05 in all cases).
There was no effect of CO2 concentration on Rd25 (p>0.05;Table 2).An interaction between nitrogen fertilization and inoculation (p<0.001;Table 2) indicated that the positive effect of increasing nitrogen fertilization on Rd25 (p<0.05;Table 2) was driven by uninoculated plants (Tukey test of the nitrogen fertilization-Rd25 slope in uninoculated plants: p<0.001), as there was no effect of nitrogen fertilization on Rd25 in inoculated plants (Tukey test of the nitrogen fertilization-Rd25 slope in inoculated plants: p>0.05).
This pattern was driven by a negative effect of increasing nitrogen fertilization on PNUEgrowth (p<0.001;Table 2) that was stronger under elevated CO2 than ambient CO2 (Tukey test comparing the nitrogen fertilization-PNUEgrowth slope between CO2 treatments: p<0.05).An interaction between nitrogen fertilization and inoculation (p<0.001;Table 2; Fig. 3) indicated that the negative effect of increasing nitrogen fertilization on PNUEgrowth was driven by inoculated plants (Tukey test of the nitrogen fertilization-PNUEgrowth slope in inoculated plants: p<0.001), as there was no effect of nitrogen fertilization on PNUEgrowth in uninoculated plants (Tukey test of the nitrogen fertilization-PNUEgrowth slope in uninoculated plants: p>0.05).

Whole-plant traits
Elevated CO2 increased total leaf area and total biomass by 51% and 102%, respectively (p<0.001 in both cases; Table 3).Positive effects of elevated CO2 on total leaf area and total biomass were enhanced with increasing nitrogen fertilization (CO2-by-nitrogen fertilization interaction: p<0.001 in both cases; Table 3; Fig. 4a-b) but not inoculation (CO2-by-inoculation interaction: p>0.05 in both cases; Table 3).An interaction between nitrogen fertilization and inoculation (p<0.001 in both cases; Table 3) indicated that positive effects of increasing nitrogen fertilization on total leaf area and total biomass (p<0.001 in both cases; Table 3) were stronger in uninoculated plants than inoculated plants (Tukey tests comparing the nitrogen fertilization-trait slopes between inoculation treatments: p<0.05 for both traits).Elevated CO2 increased Ncost by 62% (p<0.001;Table 3), a pattern that was not modified by nitrogen fertilization (CO2-by-nitrogen fertilization interaction: p>0.05;Table 3).An interaction between CO2 and inoculation (p<0.05;Table 3) indicated that the positive effect of elevated CO2 on Ncost was stronger in uninoculated plants (99% increase; Tukey test evaluating the CO2 effect on Ncost in uninoculated plants: p<0.001) than inoculated plants (21% increase Tukey test evaluating the CO2 effect on Ncost in inoculated plants: p<0.05).An interaction between nitrogen fertilization and inoculation (p<0.001;Table 3) indicated that the negative effect of increasing nitrogen fertilization on Ncost (p<0.001;Table 3) was stronger in uninoculated plants (Tukey test comparing the nitrogen fertilization-Ncost slope between inoculation treatments: p<0.001).A three-way interaction (p<0.001;Table 3) indicated that interactions between nitrogen fertilization and inoculation were stronger under elevated CO2 than ambient CO2.This pattern was driven by greater Ncost in uninoculated plants grown under elevated CO2 and low nitrogen fertilization than any other CO2-by-inoculation treatment combination under low nitrogen fertilization (Tukey test comparing Ncost in uninoculated individuals grown under elevated CO2 and 0 ppm N to all other CO2-inoculation treatment combinations grown under 0 ppm N: p<0.001 in all cases; Fig. 4c).Ncost was generally reduced in inoculated plants (p<0.001;Table 3).Negative effects of increasing nitrogen fertilization and inoculation on Ncost were driven by stronger positive effects of each treatment on Nwp than Cbg, while positive effects of elevated CO2 on Ncost were driven by stronger positive effects on Cbg than Nwp (Table S4; Fig. S4).tests (α=0.05).P-values less than 0.05 are in bold and p-values between 0.05 and 0.10 are italicized.A superscript " b " after trait labels indicates if models were fit using square root transformed variables.Key: df=degrees of freedom, χ 2 =Wald chi-square test statistic, total leaf area (cm 2 ), total biomass (g), carbon cost to acquire nitrogen (gC gN -1 ), %Ndfa=percent leaf nitrogen content fixed from the atmosphere (%).

Discussion
Glycine max seedlings were grown under two CO2 concentrations, two inoculation treatments, and nine soil nitrogen fertilization treatments in a full-factorial growth chamber experiment to reconcile the role of nitrogen supply, demand, and acquisition strategy on leaf and whole-plant responses to elevated CO2.
Results revealed that elevated CO2 increased Anet,growth despite reduced Narea, Vcmax25, and Jmax25.Larger reductions in Vcmax25 than Jmax25 increased Jmax25:Vcmax25, while respective increases and decreases in Anet,growth and Narea increased photosynthetic nitrogen-use efficiency.Effects of elevated CO2 on Anet,growth, Vcmax25, Jmax25, and Jmax25:Vcmax25 were similar across the nitrogen fertilization gradient, suggesting that leaf photosynthetic responses to elevated CO2 were decoupled from changes in nitrogen supply.Instead, increased Jmax25:Vcmax25 under elevated CO2 indicated that plants responded to increasing atmospheric CO2 concentrations by allowing enhanced net photosynthesis rates to be achieved by approaching equal co-limitation of Rubisco carboxylation rate-limited photosynthesis and electron transport for RuBP regeneration ratelimited photosynthesis (Chen et al., 1993;Maire et al., 2012).These responses supported our hypothesis that leaf photosynthetic responses to elevated CO2 would be driven by leaf nitrogen demand to build and maintain photosynthetic enzymes and would be independent of nitrogen supply.Leaf photosynthetic responses to elevated CO2 corresponded with increased total leaf area and total biomass, patterns that were enhanced with increasing nitrogen fertilization and associated with increased nitrogen uptake efficiency.These results supported our hypothesis that whole-plant responses to elevated CO2 would be constrained by nitrogen supply.However, contrasting our hypothesis, inoculation did not modify whole-plant responses to elevated CO2 due to similar plant investment in symbiotic nitrogen fixation between CO2 treatments.
Combined, results indicate that nitrogen supply and demand were each important factors that determined plant responses to elevated CO2 -leaf nitrogen demand to build and maintain photosynthetic enzymes drove leaf photosynthetic responses to elevated CO2, while nitrogen supply constrained whole-plant growth responses to elevated CO2.These findings support leaflevel patterns expected from eco-evolutionary optimality theory, suggesting that terrestrial biosphere models may improve simulations of leaf photosynthetic processes under future novel environments by considering frameworks that adopt optimality principles (Smith & Keenan, 2020;Harrison et al., 2021;Luo et al., 2021).Below, we expand and contextualize these conclusions and suggest their implications for terrestrial biosphere model development.
Nitrogen supply and demand regulate leaf and whole-plant responses to elevated CO2 at different scales Leaf photosynthetic responses to elevated CO2 were consistent with previous studies that have investigated or reviewed leaf responses to elevated CO2 (Drake et al., 1997;Makino et al., 1997;Ainsworth et al., 2002;Ainsworth & Long, 2005;Ainsworth & Rogers, 2007;Crous et al., 2010;Lee et al., 2011;Smith & Dukes, 2013;Poorter et al., 2022), and follow patterns expected from eco-evolutionary optimality theory (Chen et al., 1993;Wright et al., 2003;Maire et al., 2012;Prentice et al., 2014;Wang et al., 2017;Smith et al., 2019;Smith & Keenan, 2020;Harrison et al., 2021).Positive effects of elevated CO2 on Anet,growth and Jmax25:Vcmax25 and negative effects of elevated CO2 on Vcmax25 and Jmax25 were similar across the nitrogen fertilization gradient, indicating that leaf photosynthetic responses to elevated CO2 were decoupled from changes in nitrogen supply.Increased Jmax25:Vcmax25 and photosynthetic nitrogen-use efficiency under elevated CO2 provide strong support for the idea that leaves were downregulating Vcmax25 in response to elevated CO2 such that enhanced net photosynthesis rates approached becoming equally co-limited by Rubisco carboxylation and RuBP regeneration (Chen et al., 1993;Maire et al., 2012;Smith & Keenan, 2020).These patterns suggest that leaf photosynthetic responses to elevated CO2 were likely the result of reduced demand to build and maintain photosynthetic enzymes, following patterns expected from eco-evolutionary optimality theory (Harrison et al., 2021;Dong et al., 2022b).
Whole-plant responses were also consistent with previous studies that have investigated or reviewed whole-plant responses to elevated CO2 (Makino et al., 1997;Ainsworth et al., 2002;Hungate et al., 2003;Ainsworth & Long, 2005;Norby et al., 2010;Smith & Dukes, 2013;Poorter et al., 2022).Greater whole-plant growth under elevated CO2 was associated with greater carbon costs to acquire nitrogen through stronger increases in belowground carbon allocation than whole-plant nitrogen uptake.These patterns indicate that plants grown under elevated CO2 supported greater total leaf area and total biomass through increased plant nitrogen uptake, though at reduced nitrogen uptake efficiency.Unlike leaf photosynthetic responses to elevated CO2, positive whole-plant responses to elevated CO2 were enhanced with increasing nitrogen fertilization, supporting our hypothesis that nitrogen supply would constrain whole-plant responses to elevated CO2 (Hungate et al., 2003;Luo et al., 2004;Finzi et al., 2007).Positive effects of increasing nitrogen fertilization on total leaf area and total biomass were associated with reductions in carbon costs to acquire nitrogen, a pattern that was driven by stronger increases in whole-plant nitrogen uptake than belowground carbon allocation (Perkowski et al., 2021).While reductions in carbon costs to acquire nitrogen due to increasing nitrogen fertilization were similar between CO2 treatments, increasing nitrogen fertilization increased whole-plant nitrogen uptake more strongly under elevated CO2.This pattern, coupled with similar effects of nitrogen fertilization on belowground carbon allocation responses to elevated CO2, indicated that stronger growth responses to elevated CO2 with increasing nitrogen fertilization were likely driven by enhanced nitrogen uptake efficiency.These findings suggest that positive short-term effects of nitrogen supply on whole-plant responses to elevated CO2 are linked to reduced costs of acquiring nitrogen and increased nitrogen uptake efficiency, supporting conclusions from Terrer et al. (2018).
Our findings indicate that nitrogen supply and demand could each explain plant responses to elevated CO2, though these factors operated at different scales.Specifically, photosynthetic responses to elevated CO2 were determined through reduced leaf nitrogen demand to build and maintain photosynthetic enzymes.Reduced leaf nitrogen demand resulted in a shift in nitrogen allocation to photosynthetic enzymes independent of soil nitrogen supply that increased photosynthetic nitrogen use efficiency and allowed net photosynthesis rates to occur by approaching optimal coordination of Rubisco carboxylation-limited and RuBP regenerationlimited photosynthesis.Whole-plant responses to elevated CO2 were enhanced with increasing soil nitrogen supply.Interestingly, optimized nitrogen allocation to photosynthetic capacity may have resulted in nitrogen savings at the leaf level that could have maximized nitrogen allocation to growth.These results suggest that plants grown under elevated CO2 responded to increased nitrogen supply by increasing the number of optimally coordinated leaves and that the downregulation in photosynthetic capacity under elevated CO2 was not a direct response to changes in nitrogen supply.

Inoculation with symbiotic nitrogen-fixing bacteria does not modify leaf or whole-plant responses to elevated CO2
Inoculation increased Narea, Anet,420, Anet,growth, Vcmax25, Jmax25, photosynthetic nitrogen-use efficiency, total leaf area, and total biomass, and decreased Jmax25:Vcmax25 and Rd25.These patterns support previous literature suggesting that species that form associations with symbiotic nitrogen-fixing bacteria often have increased leaf nitrogen content, photosynthetic capacity, and growth compared to species that do not form such associations (Adams et al., 2016;Bytnerowicz et al., 2023).Positive effects of inoculation on leaf and whole-plant traits were strongest under low nitrogen fertilization and rapidly diminished with increasing nitrogen fertilization as investment in symbiotic nitrogen fixation decreased (Andrews et al., 2011;Friel & Friesen, 2019;McCulloch & Porder, 2021;Perkowski et al., 2021), supporting the idea that nitrogen fixation is a nutrient acquisition strategy that may confer competitive benefits for nitrogen-fixing species growing in low soil nitrogen environments (Rastetter et al., 2001;Vitousek et al., 2002).Interestingly, inoculation did not modify effects of elevated CO2 on Vcmax25, Jmax25, Jmax25:Vcmax25, photosynthetic nitrogen-use efficiency, total leaf area, or total biomass.These patterns corresponded with null effects of elevated CO2 on %Ndfa and the ratio of root nodule biomass to root biomass, suggesting that null inoculation effects on plant responses to elevated CO2 were primarily due to similar plant investments toward symbiotic nitrogen fixation between CO2 treatments.We observed these patterns regardless of nitrogen fertilization level, contrasting our hypothesis that inoculation would enhance whole-plant responses to elevated CO2 under low nitrogen fertilization where individuals were expected to be invested more strongly in symbiotic nitrogen fixation.These patterns also contrast previous work showing that inoculated G. max is generally more responsive to increasing atmospheric CO2 concentrations (Ainsworth et al., 2002) and that plant investment toward symbiotic nitrogen fixation tends to be greater under scenarios that increase whole-plant demand to acquire nitrogen (Taylor & Menge, 2018;Friel & Friesen, 2019;McCulloch & Porder, 2021).

Implications for future model development
Many terrestrial biosphere models predict photosynthetic capacity through parameterized relationships between Narea and Vcmax (Rogers, 2014;Rogers et al., 2017), which assumes that leaf nitrogen-photosynthesis relationships are constant across growing environments.Our results build on previous work suggesting that leaf nitrogen-photosynthesis relationships dynamically change across growing environments (Smith & Keenan, 2020;Luo et al., 2021;Dong et al., 2022b;Waring et al., 2023), as elevated CO2 reduced leaf nitrogen content more strongly than it increased Anet,growth and decreased Vcmax25 and Jmax25.Additionally, positive effects of increasing nitrogen fertilization on indices of photosynthetic capacity were only apparent in uninoculated plants, as there was no effect of nitrogen fertilization on Vcmax25 or Jmax25 in inoculated plants.
Positive effects of increasing nitrogen fertilization on Narea and Chlarea were also markedly weaker in inoculated plants compared to uninoculated plants.These patterns indicate that leaf nitrogen-photosynthesis relationships are context-dependent on nitrogen acquisition strategy, may only be constant in environments where nitrogen supply limits leaf physiology, and will likely shift in response to increasing atmospheric CO2 concentrations.Terrestrial biosphere models that predict photosynthetic capacity through parameterized relationships between Narea and Vcmax (e.g., Kattge et al., 2009;Walker et al., 2014) may risk overestimating photosynthetic capacity, therefore net primary productivity and the magnitude of the land carbon sink, under future novel growth environments.
Our results demonstrate that optimal resource allocation to photosynthetic capacity defines leaf photosynthetic responses to elevated CO2 and that these responses are independent of nitrogen supply.Current approaches for simulating photosynthetic responses to CO2 in terrestrial biosphere models with coupled carbon and nitrogen cycles often invoke patterns expected from progressive nitrogen limitation, where photosynthetic responses to elevated CO2 are modeled as a function of positive relationships between nitrogen availability and leaf nitrogen content.Our results contradict this framework, suggesting that photosynthetic responses to elevated CO2 are driven by optimal nitrogen investment to satisfy leaf nitrogen demand to build and maintain photosynthetic enzymes.Optimality models that use principles from optimal coordination and photosynthetic least-cost theories (Wang et al., 2017;Stocker et al., 2020;Scott & Smith, 2022) are capable of capturing responses to CO2 independent of nitrogen supply (Smith & Keenan, 2020;Harrison et al., 2021), suggesting that including optimality frameworks in terrestrial biosphere models may improve the accuracy by which photosynthetic processes are simulated in response to increasing atmospheric CO2 concentrations.
Previous work has highlighted the fact that pot experiments restrict belowground rooting volume and may alter plant allocation responses to environmental change (Ainsworth et al., 2002;Poorter et al., 2012).In this study, the ratio of pot volume to total biomass was greater under elevated CO2 and increased with increasing nitrogen fertilization such that several treatment combinations exceeded values recommended by Poorter et al. (2012) to avoid growth limitation imposed by restricted pot volume (<1 g L -1 ; Table S6; Fig. S6).While pot size may have limited plant responses to elevated CO2, similar responses to elevated CO2 have been observed using field measurements that do not restrict belowground rooting volume (Bernacchi et al., 2005;Crous et al., 2010;Lee et al., 2011;Pastore et al., 2019;Smith & Keenan, 2020).
Additionally, there was no apparent saturating effect of increasing fertilization on total biomass, belowground carbon biomass, or root biomass under conditions where biomass: pot volume ratios exceeded 1 g L -1 (e.g., individuals of either inoculation status grown under high fertilization and elevated CO2), which might be expected if pot volume had limited plant growth.
The lack of such responses indicate that the pot volume used in this study (6 L) was sufficient to avoid growth limitation.

Conclusions
Our results indicate that nitrogen supply and demand each helped explain G. max responses to elevated CO2, though operated at different scales.Supporting eco-evolutionary optimality theory, leaf photosynthetic responses to elevated CO2 were independent of soil nitrogen supply and ability to associate with symbiotic nitrogen-fixing bacteria and were instead driven by leaf nitrogen demand to build and maintain photosynthetic enzymes such that net photosynthesis rates approached optimal coordination.Supporting the progressive nitrogen limitation hypothesis, whole-plant responses to elevated CO2 were enhanced with increasing nitrogen fertilization due to increased plant nitrogen uptake efficiency coupled with possible cascading effects of nitrogen savings at the leaf level that may have maximized nitrogen allocation to whole-plant growth.However, inoculation did not modify whole-plant responses to elevated CO2, as plants invested similarly in symbiotic nitrogen fixation between CO2 treatments.Results suggest that plants grown under elevated CO2 responded to increased nitrogen supply by increasing the number of optimally coordinated leaves and that the downregulation in photosynthetic capacity under elevated CO2 was not modified by changes in nitrogen supply.
The differential role of nitrogen supply on leaf and whole-plant responses to elevated CO2 coupled with dynamic leaf nitrogen-photosynthesis relationships across CO2 and nitrogen fertilization treatments suggests that terrestrial biosphere models may improve simulations of )where δ 15 Ndirect refers to the δ 15 N value from plants that exclusively acquired nitrogen via direct uptake, δ 15 Nsample refers to an individual's leaf δ 15 N, and δ 15 Nfixation refers to the δ 15 N value from individuals that were entirely reliant on nitrogen fixation.δ 15 Ndirect was calculated as the mean leaf δ 15 N of uninoculated individuals within each unique nitrogen fertilization-by-CO2 treatment combination.Any individual with visual evidence of root nodule formation or nodule initiation was omitted from the calculation of δ 15 Ndirect.δ 15 Nfixation was calculated within each CO2 treatment using the mean leaf δ 15 N of inoculated individuals that received 0 ppm N. δ 15 Nfixation plants were either incompletely sterilized or were colonized by symbiotic nitrogen-fixing bacteria from neighboring plants in the chamber.This decision resulted in the removal of sixteen plants from the analysis: two plants in the elevated CO2 treatment that received 35 ppm N, three plants in the elevated CO2 treatment that received 70 ppm N, one plant in the elevated CO2 treatment that received 210 ppm N, two plants in the elevated CO2 treatment that received 280 ppm N, two plants in the ambient CO2 treatment that received 0 ppm N, three plants in the ambient CO2 treatment that received 70 ppm N, two plants in the ambient CO2 treatment that received 105 ppm N, and one plant in the ambient CO2 treatment that received 280 ppm N.

Table 3
Effects of CO2 concentration, inoculation, and nitrogen fertilization on whole-plant growth, carbon costs to acquire nitrogen, and investment toward symbiotic nitrogen fixation * * Significance determined using Type II Wald χ 2