Triose phosphate utilization in leaves is modulated by whole-plant sink–source ratios and nitrogen budgets in rice

Abstract Triose phosphate utilization (TPU) is a biochemical process indicating carbon sink–source (im)balance within leaves. When TPU limits leaf photosynthesis, photorespiration-associated amino acid exports probably provide an additional carbon outlet and increase leaf CO2 uptake. However, whether TPU is modulated by whole-plant sink–source relations and nitrogen (N) budgets remains unclear. We address this question by model analyses of gas-exchange data measured on leaves at three growth stages of rice plants grown at two N levels. Sink–source ratio was manipulated by panicle pruning, by using yellower-leaf variant genotypes, and by measuring photosynthesis on adaxial and abaxial leaf sides. Across all these treatments, higher leaf N content resulted in the occurrence of TPU limitation at lower intercellular CO2 concentrations. Photorespiration-associated amino acid export was greater in high-N leaves, but was smaller in yellower-leaf genotypes, panicle-pruned plants, and for abaxial measurement. The feedback inhibition of panicle pruning on rates of TPU was not always observed, presumably because panicle pruning blocked N remobilization from leaves to grains and the increased leaf N content masked feedback inhibition. The leaf-level TPU limitation was thus modulated by whole-plant sink–source relations and N budgets during rice grain filling, suggesting a close link between within-leaf and whole-plant sink limitations.


Introduction
CO 2 response curves of leaf photosynthesis as obtained from gas exchange analysis are typically described by the canonical Farquhar-von Caemmerer-Berry (FvCB) biochemical photosynthesis model (Farquhar et al., 1980).This model predicts that leaf photosynthetic rates under the current environmental conditions are determined by two main parameters, Rubisco carboxylation capacity (V cmax ) and maximum linear electron transport rate (J max ).The transition from Rubisco limitation to electron-transport limitation occurs around a leaf internal CO 2 level when the equivalent ambient-air CO 2 concentration (C a ) is about 400 μmol mol −1 (e.g.Mathan et al., 2021).However, the internal CO 2 inside a leaf can attain a high level at which leaf photosynthesis is limited by a third parameter, T p , the rate of triose phosphate utilization (TPU).Until now, compared with the first two biochemical components (i.e.V cmax and J max ), T p has received less attention since observable TPU limitations only occur occasionally and are highly variable, depending on species, genotype, growth conditions, and measurement conditions (e.g.Kumarathunge et al., 2019).Also, the TPU-limited condition is observed only temporarily, because it can be removed quickly as other parameters like V cmax and J max may be regulated to a level where T p is no longer 'apparently' limiting (McClain et al., 2023).Nevertheless, with the increase of atmospheric CO 2 (to roughly 600 μmol mol −1 ; Lombardozzi et al., 2018), TPU limitation will probably become increasingly important for predicting photosynthesis and yield.
TPU refers to the rate at which triose phosphates exit from the photosynthetic Calvin-Benson-Bassham (CBB) cycle and are used as sugar precursors for processes like the synthesis of sucrose and starch.As triose phosphates are phosphorylated carbon, any TPU requires the returns of inorganic phosphate (P i ) to the chloroplast, since the quantity of phosphate in the chloroplast is finite and under tight homeostasis (McClain and Sharkey, 2019).A limitation of TPU on photosynthesis is triggered when carbon exports from (with the accompanying P i import to) the cycle cannot keep pace with carbon fixation, which is in essence a local sink-source disequilibrium.TPU limitation causes unresponsiveness to CO 2 (Sharkey, 1985) or sometimes reversed sensitivity of photosynthesis to increasing CO 2 (e.g.von Caemmerer and Farquhar, 1981;Harley and Sharkey, 1991).The TPU limitation is considered to be a biochemical mechanism for sink limitation on photosynthesis, expressed at a sub-foliar scale (Sharkey, 2019).This differs from the sink (panicles) limitation on source (leaves) activity agronomists commonly define at the whole-plant or crop scale.While sink limitations at sub-foliar and whole-plant scales are not necessarily independent (Yin et al., 2022), few studies have been conducted on their connections.
Agronomists commonly manipulate sink-source ratios at the whole-plant scale by pruning leaves or panicles during the grain filling process, which modifies the transport of assimilates between source and sink organs (Li et al., 2017;He et al., 2019).Many reports (Rossi et al., 2015;Dingkuhn et al., 2020;Fabre et al., 2020) have also demonstrated that genotypes with larger crop carbon sink capacity can benefit more from future CO 2 -rich climate.Conversely, a smaller sink (e.g.small-panicle genotypes, or plants with panicles pruned) reduces phloem loading, forcing assimilate accumulation in leaves or stems that may exert a feedback inhibition on leaf photosynthetic source activity (Burnett et al., 2016;White et al., 2016), and even on TPU (Fabre et al., 2019).
Carbon metabolism generally interacts with nitrogen (N) assimilation.In fact, Harley & Sharkey (1991) hypothesized that the reversed sensitivity of photosynthesis to increasing CO 2 under TPU limitation resulted from diverting a fraction of N-containing glycine from the photorespiratory pathway and this glycine is used elsewhere for other amino acid or protein synthesis.Glycine is derived from the photorespiratory glycolate carbon; normally 25% of the glycolate carbon is lost as CO 2 as a result of glycine decarboxylation, and the remaining 75% is recycled to glycerate and further to 3-phosphoglycerate to rejoin the CBB cycle (Supplementary Fig. S1).With the exit of glycine, the P i normally used in converting glycerate to 3-phosphoglycerate is made available for phosphorylation, thereby stimulating photosynthesis (Harley and Sharkey, 1991).Busch et al. (2018) extended this hypothesis by considering the exit of both glycine and serine from the photorespiratory pathway.The export of N from the photorespiratory pathway requires de novo N assimilation and amino acid synthesis (Supplementary Fig. S1).They proposed that a large proportion of N assimilation in leaves accomplished via the photorespiratory pathway is innately linked with TPU.The exported amino acids represent an additional sink for carbon and decrease P i consumption for phosphorylating glycerate to 3-phosphoglycerate, thereby explaining the increased photosynthetic rate with decreasing CO 2 levels (with increasing photorespiration) within the TPU-limited range (Busch et al., 2018;Yin et al., 2021).The extent of increase in photosynthesis depends on the proportion of glycolate carbon exported from the photorespiratory pathway.It is conceivable that the proportion of amino acid carbon export may be associated with the availability of NO 3 − for N assimilation.Crop N assimilation occurs throughout the life cycle, but varies with NO 3 − availability and leaf N content.We hypothesize that values of a parameter related to TPU limitation (i.e. the fraction of the glycolate carbon exported from the photorespiratory pathway) increase with leaf N content, and thus vary among growth stages and N treatments.
In this study, we aim to (i) quantify how TPU-limited photosynthetic carbon uptake is affected by nitrogen assimilation via the photorespiratory pathway; and (ii) analyse how the local sink-source mechanisms of TPU limitation within leaves are regulated by the whole-plant physiological source and sink relationships.We address these aims in the context of our recent effort to examine the impact of leaf-colour modification on photosynthesis, using rice genotypes of different leaf colour based on our previous finding that genotypic leaf yellowness affects leaf photosynthetic rate via several mechanisms including altered leaf morphology (Zhou et al., 2023).As adaxial and abaxial photosynthetic rates are known to differ (e.g.Soares et al., 2008), gas exchange measurements with the local illumination on either leaf side may result in a varied leafscale source activity but with a constant sink demand.Thus, in our experimental set-up, we use three means to vary carbon sink-source ratios (genotypic leaf-colour variants, panicle pruning, and adaxial versus abaxial illumination while measuring gas exchange), and two means to vary nitrogen status (N treatments and plant developmental stages).In addition, we use 21% O 2 versus 2% O 2 conditions to alter photorespiration, likely modifying the amount of glycine and serine export.We hoped to obtain information on how TPU limitation is related to photorespiration-associated N assimilation and affected by altered sink-source ratios within the leaf and at the wholeplant scale.

Plant material and growth conditions
Rice (Oryza sativa L.) materials were based on two background genotypes: japonica type cv.Wuyunjing 3 (WYJ) and early indica type cv.Zhefu 802 (ZF).Both were modified by radiation mutagenesis with 60 Co γ-rays, and the yellower-leaf variants were identified from a larger population of phenotypes.These genotypes showed stability of the lines over generations (Zhou et al., 2023).Hereafter, yellower genotypes are denoted as Y and the wild type as control (C).
Two experiments were conducted in a climate-controlled glasshouse in Wageningen, the Netherlands in 2019 and 2022.The growth conditions were the same as described in our earlier study (Zhou et al., 2023): incident global radiation outside the greenhouse was kept within 400−500 W m −2 (resulting in a photosynthetic photon flux density measured at plant height of ~500 μmol m −2 s −1 ), temperature was set at 26 °C for the 12-h light period and at 23 °C for the 12-h dark period, the CO 2 level was about 400 μmol mol −1 , and the relative humidity was 65−75%.The 2019 experiment was to examine if genotypes differ in the extent of photosynthetic differences between adaxial and abaxial illumination.
In the 2022 experiment, nitrogen supply and panicle pruning treatments were added as factors.Nitrogen was applied as urea at two levels: N1 (in total 0.7 g urea per pot) and N2 (in total 1.4 g urea per pot).All pots were evenly divided into four blocks (corresponding to four experimental replicates), and each block contained 64 pots, representing all combinations of four genotypes, two nitrogen levels, and two pruning levels.The four pots per treatment combination were used for measurements at three developmental stages (see below), with one pot as reserve in case of plant damage.Plants to be pruned were randomly selected and pre-labelled, and panicles of these plants were pruned at the moment when the first panicle of the plant had emerged from the flag-leaf sheath.This operation lasted a week to ensure that no new heads were produced from any culms.

Leaf photosynthesis measurements
Pre-labelled and fully expanded main-stem leaves in each experimental treatment per replicate were measured using an open-path gas exchange system integrated with a fluorescence chamber head (LI-COR 6800; LI-COR Inc., Lincoln, NE, USA) to simultaneously obtain gas exchange and chlorophyll fluorescence parameters.All measurements were carried out at a leaf temperature of 25 °C and a vapour pressure difference of 1.0−1.6 kPa between the leaf and air outside of the leaf, with a flow rate of 400 μmol s −1 .
For both 2019 and 2022 experiments, measurements were conducted on the same leaf segment at both adaxial and abaxial sides.It should be noted that measurements on 'adaxial' or 'abaxial' sides always integrated the gas exchange occurring on both sides, as both sides were exposed to the chamber air.However, the light was only received by the side that faced the light source.Strictly speaking, it was not the gas exchange measurement but the light orientation that varied, causing inverted light gradients through the leaf.
For the 2019 experiment, measurements were conducted only at the tillering stage.Light and CO 2 response curves were measured at the same position at both adaxial and abaxial sides of the leaves.The curves for net photosynthetic rate (A) response to incident irradiance (I inc ) were obtained with I inc in a decreasing series of 2000,1500,1000,500,280,150,100,80, and 50 μmol m −2 s −1 (6−8 min per step), while maintaining ambient CO 2 level (C a ) at 400 μmol mol −1 .The CO 2 response curves were measured at I inc of 1000 μmol m −2 s −1 , with the C a steps of 400,250,150,80,50,400,400,400,650,1000, and 1500 μmol mol −1 (3−5 min per step; note that using the three repeated 400 μmol mol −1 steps was merely to re-adapt leaves, and the data from these three points were excluded in the analysis).Both curves were measured at ambient O 2 (21%) level.To estimate day respiration (R d ) and establish a calibration factor (s) that converts chlorophyll fluorescence-based electron transport efficiency of photosystem II (PSII) into linear electron transport rate (see Yin et al., 2009), we also conducted half of the light response curve (with I inc being 280, 150, 100, 80, and 50 μmol m −2 s −1 ) under non-photorespiratory conditions (2% O 2 combined with C a at 1000 μmol mol −1 ).These low light levels were applied to ensure that data for calibration were within the range where A is limited by electron transport.The low O 2 level was realized by using a cylinder containing a gas mixture of 2% O 2 and 98% N 2 .
For the 2022 experiment, photosynthesis was measured on both adaxial and abaxial leaf surfaces at three stages: tillering stage (TS), flowering stage (FS), and grain-filling (~15 days after flowering (DAF)).The panicle-pruned plants were only measured at grain filling because these plants were supposed to function the same as the non-pruned plants at tillering and flowering.As this experiment was meant to examine the TPU limitation, only CO 2 response curves (where A is likely limited by TPU) were measured at I inc of 1500 μmol m −2 s −1 under both 21% and 2% O 2 conditions, with the C a in an increasing series: 400, 500, 600, 700, 800, 900, 1000, 1200, 1400, 1600, and 1800 μmol mol −1 .These C a levels were chosen to ensure that part of the curve could reach the TPUlimited range.As with the 2019 experiment, we additionally measured the light-response curve with I inc being 300, 150, 100, 80, and 40 μmol m −2 s −1 under non-photorespiratory conditions to estimate R d and s.
For each irradiance or CO 2 step in both experiments, F s (the steadystate fluorescence) was recorded after A reached the steady state.The maximum fluorescence (Fʹ m ) was determined using a three-phase flash method (Loriaux et al., 2013): each phase went through a duration of 300 ms, and flash intensity of 6500 μmol m −2 s −1 in the second phase was attenuated by 40%.The apparent operating photochemical efficiency of PSII was assessed from chlorophyll fluorescence measurements: (Genty et al., 1989).
All gas exchange data were corrected for any small basal leakage of CO 2 into and out of the leaf cuvette, based on measurements on boiled leaves across the CO 2 levels, and intercellular CO 2 levels (C i ) were then re-calculated.

Leaf SPAD and nitrogen content
All leaf segments used for measuring photosynthesis curves were cut out and used immediately to measure the leaf area with a LI-3100 area meter (LI-COR) and the values for SPAD indicating chlorophyll content (SPAD-502, Minolta Camera Co., Japan).SPAD was measured at both adaxial and abaxial sides of these leaf segments.Leaf materials were then oven-dried at 70 °C for 48 h to constant weight.Specific leaf area (SLA, m 2 kg −1 ) was calculated as the leaf area to dry leaf mass ratio.Each leaf segment was ground into powder in a 2-ml centrifuge tube, which was used to measure the N concentration by an element analyser based on the micro-Dumas combustion method.Specific leaf nitrogen (SLN, g N m −2 ) was then calculated.

Plant growth measurements
At grain-filling stage in the 2022 experiment, the aboveground parts were sampled and separated.Dry weight of each part was determined after oven drying at 75 °C for 72 h to constant weight.The leaf samples were ground into powder, which was then assessed for nitrogen concentration with a Kjeldahl apparatus (Kjeltec 8400, Foss Corp., Germany).Total leaf-nitrogen per pot was calculated by leaf nitrogen concentration multiplied by total leaf dry weight.We counted the fertile spikelet number for each culm, and then measured the flag leaf area (just after photosynthesis measurement) and total leaf area by a LI-3100 area meter (LI-COR).Following Fabre et al. (2020), the ratio of flag leaf area (source) to the fertile spikelet number of the panicle (sink) on the culm was used as an indicator of the single-culm sink-source ratio, while total spikelet number divided by total leaf area of the whole plant was used as an indicator of the whole-plant sink-source ratio.

Estimating photosynthetic parameters
We estimated parameters of the FvCB model (Farquhar et al., 1980), which expresses net photosynthetic rate (A) as the minimum of the Rubisco carboxylation-limited rate (A c ), electron-transport limited rate (A j ), and the TPU-limited rate (A p ): For A c : where C c and O are the chloroplast partial pressures of CO 2 and O 2 , respectively, V cmax is the maximum rate of Rubisco activity for carboxylation, and K mC and K mO are Michaelis-Menten constants of Rubisco for CO 2 and O 2 , respectively.Γ * is the CO 2 compensation point in the absence of day respiration (R d ), described by: Γ * =0.5O/S c/o , where S c/o is the relative CO 2 /O 2 specificity factor for Rubisco.Values of these Rubisco parameters vary significantly, depending on techniques used to measure them; here, we used the representative values of Rubisco parameters measured in vitro at 25 °C by Cousins et al. (2010) for wheat: i.e. 291 μbar for K mC , 194 mbar for K mO , and 3.022 mbar μbar −1 for S c/o , given that values of these Rubisco parameters are believed to be conserved among C 3 species (von Caemmerer, 2000).For A j : where J is the potential linear electron transport rate supporting the CBB cycle and the photorespiratory cycle.J can be calculated using the calibration factor s, incident irradiance (I inc ), and fluorescence-based photochemical efficiency of PSII (Φ 2 ) as: J = sI inc Φ 2, where parameters s and R d can be estimated from the slope and intercept of a linear plot of A j against (I inc Φ 2 /4) measured under non-photorespiratory conditions (Yin et al., 2009).The calculated J can be fitted according to: where J max is the maximum value of J under saturated light; κ 2LL represents the conversion efficiency of incident light into J at strictly limiting light; and θ is a dimensionless convexity factor for the response of J to I inc , and here a common value of 0.76 for θ was adopted for all rice genotypes from Zhou et al. (2023).
For A p , the widely used algorithm (Harley and Sharkey, 1991;von Caemmerer, 2000) assumes that glycine is taken out from the photorespiratory pathway.However, this algorithm does not consider the required change of the CO 2 compensation point, as a result of the glycine export, to (1−α G )Γ * (Busch et al., 2018;Yin et al., 2021, also see Supplementary Fig. S1; where α G is the proportion of glycolate carbon exported from the photorespiratory pathway in the form of glycine).Model fitting results of Busch et al. (2018) suggested that the proportion of glycolate carbon exported as glycine is lower than the proportion exported as serine.Isotope-labelling measurements (Abadie et al., 2016;Fu et al., 2022) more convincingly confirmed little export in the form of glycine.As serine export causes no change in the CO 2 compensation point, here for the purpose of simplicity, we assumed only the serine export, for which the model becomes (Yin et al., 2021; see their Equation 17b): where α S is the proportion of glycolate carbon exported from photorespiratory pathway in the form of serine (with 0≤α S ≤0.75).This guarantees the same term in the numerator, (C c −Γ * ), which is consistent with Rubisco-or electron transport-limited forms.Such consistency simplifies the modelling algorithms for the next steps of analysis (Equations 5, 6).The simple model in Equation 4 also generates the TPU-limited rate A p similar to the full model with both glycine and serine export if the total fraction of glycolate carbon export remains the same (Yin et al., 2021).Note that the coefficient in front of the term for the proportion of glycolate carbon export in the denominator of Equation 4is 4, whereas this is 3 in the commonly used old equation assuming the glycine exit (von Caemmerer, 2000;Ellsworth et al., 2015;Busch and Sage, 2017;Kumarathunge et al., 2019; also see Supplementary Fig. S1).
As a result, the old equation, when applied to fit gas-exchange data, overestimates the glycolate carbon export fraction by a factor of 4/3 (Yin et al. 2021).
The method of Harley et al. (1992) was first applied to examine whether mesophyll conductance g m varied with intercellular CO 2 level (C i ) or I inc , and we found that g m is variable and declines with increasing C i or with decreasing I inc , with g m =0 as I inc approaches to zero (results not shown).To describe this pattern of variable g m , we used an equation of Yin et al. (2009): where parameter δ represents the carboxylation resistance to mesophyll resistance ratio (Yin et al., 2020).Then, this Equation 5 was combined with Equations 2, 3a, and 4, and C c was replaced by (C i −A/g m ) to solve for A (Yin et al., 2020): where where Note when calculating A in Equation 6, the minus sign in front of the √ b 2 − 4ac term was applied for either the A c -or A j -limited part while the + sign was required for A p -limited part (see Yin et al., 2020).
For analysing the 2019 data where light-response curves were measured, we first estimated J max by fitting equation 3b to data points of light response of J derived from chlorophyll fluorescence parameters Φ 2 (i.e.J=sI inc Φ 2 ).Then photosynthetic parameters δ, V cmax , T p , and α S can be estimated simultaneously by fitting combined Equations 1, 3b, and 6 to all CO 2 exchange data from both light-and CO 2 -response curves.For data from 2022, as only CO 2 response curves were measured yet starting with C a from 400 μmol mol −1 onwards that only covered A j -and A p -limited parts, we thus combined J=sI inc Φ 2 and A j and A p parts of Equation 6 to estimate parameters δ, T p , and α S simultaneously.
Once photosynthetic parameters were estimated, the transition point from A j -to A p -limited rates can be solved.We also estimated the transition point by solving the second-order polynomial regression equations that were fitted to A j and A p ranges, respectively, of A-C i curves.The estimated threshold C i was highly consistent (see Results); thus, we used the polynomial-based values for showing the transition.

Statistical analyses and curve fitting
Simple linear regressions were conducted using Microsoft Excel.Nonlinear regressions were performed using the Gauss method in PROC NLIN of SAS (SAS Institute Inc., Cary, NC, USA).An analysis of variance (ANOVA) of multiple experimental factors (i.e.genotype, adaxial versus abaxial, pruning, N level, stage), and their interaction effects on each parameter was performed in the 2022 experiment.A multiple comparison of means was then performed using the LSD (least significant difference) test.

Effect of adaxial versus abaxial measurements on leaf source activity
In the 2019 experiment, the two Y-variant genotypes exhibited an opposite trend in leaf photosynthetic rate (A): relative to their control (C) genotypes, a decrease in A and estimated parameters (J max , V cmax , and T p ) was obtained in WYJ-Y whereas an increase in these parameters was observed in ZF-Y (Fig. 1).In addition, the parameter α S was also altered by the Y modification: it became lower (P<0.05),particularly in the WYJ background.
The difference in photosynthetic rates between the two sides of the same leaf depended on genotypes.For light response curves (Fig. 1A-D), a great reduction (ca.8−10%) in A at a light intensity of 2000 μmol m −2 s −1 (A 2000 ) was observed on the abaxial side in C genotypes, resulting in a lower estimated J max on the abaxial surface compared with that on the adaxial surface.In contrast, the light response curves were similar on both sides of leaves in Y-variant genotypes (difference of A 2000 less than 2%).Similar patterns were observed for CO 2 response curves (Fig. 1E-H), with greater differences between adaxial and abaxial values for parameters V cmax , T p , and α S in C genotypes than in Y-variant genotypes.

Overview of A-C i curves from the 2020 experiment
Given the above differences between adaxial and abaxial measurements in the 2019 experiment, we continued in the 2022 experiment to use adaxial versus abaxial measurements for all nitrogen×pruning combinations as a means to manipulate within-leaf sink-source ratios.In addition, measurements were made for three different stages and at two O 2 levels.All the A-C i curves obtained are shown in Supplementary Fig. S2.
Differences in measured A-C i curves and in the estimated A j -to-A p transition point between 21% and 2% O 2 (Supplementary Fig. S2) agreed with those theoretically expected for TPU limitation either with (Supplementary Fig. S2A) or without (Supplementary Fig. S2B) glycolate carbon exit from the photorespiratory pathway (Harley and Sharkey, 1991;Busch et al., 2018).Thus, combined data from the two O 2 levels were fit to estimate TPU parameters.

Estimated triose phosphate utilization capacity
Values of T p estimated for pruned and non-pruned plants under two nitrogen levels (N1 and N2) and at three growth stages in the 2022 experiment are shown in Fig. 2. As expected, the rate of TPU (T p ; Fig. 2A-D; Supplementary Table S1) and photosynthetic rate at a light intensity of 1500 μmol m −2 s −1 (A 1500 ; Supplementary Table S2) increased with the addition of N fertilizer and decreased with advancing growth stage.Significant effects mainly occurred from flowering onwards.At 15 DAF, there were no significant effects of panicle pruning on T p except for an increase of T p in WYJ-C at N1 (Fig. 2A).In line with the results of the 2019 experiment (Fig. 1), values of T p from the adaxial measurements were generally higher than those from the abaxial measurements (Fig. 2E).

Estimated proportion of photorespiratory carbon exited as serine
The effects of N treatments on the values of α S are shown in Fig. 3.The high N level generally increased α S (P<0.05).But growth-stage effects were more complex, as α S varied more among growth stages in C genotypes than in Y variants (Fig. 3A-D).The estimated α S of WYJ-C declined significantly (P<0.05) at 15 DAF after an increase at FS, whereas that of ZF-C decreased along all growth stages.The estimated α S was negatively correlated with SLA, an indicator of leaf thinness (Supplementary Fig. S3).
Unlike T p , α S was greatly decreased by pruning in all genotypes, especially at N2.There was a significant interaction between pruning and N level on α S (P<0.001;Supplementary Table S3).In addition, compared with T p , α S differed more between adaxial and abaxial measurements (by ~24%), with larger differences in C genotypes than in Y-variant genotypes, especially at TS and FS stages (Fig. 3E).No interactions of measurement side with N level or growth stage were found for α S (Supplementary Table S3).

Correlations between triose phosphate utilization parameters and leaf nitrogen content
In general, T p and α S were positively correlated with SLN across all N levels and growth stages, but correlations were genotype-dependent (Fig. 4).Given the smaller α S values in Y genotypes (Figs 1, 3), slopes were smaller and correlations were poorer, particularly when measuring the abaxial leaf side, compared with the C genotypes.
The threshold C i values (at which TPU became limiting) estimated by the two methods were highly consistent (Supplementary Fig. S4).The threshold C i increased with advancing growth stage.It was negatively correlated with T p (R 2 =0.51,P<0.001; Fig. 5A) and with SLN (R 2 =0.42,P<0.001; Fig. 5B).

Triose phosphate utilization-limited photosynthesis in relation to whole-plant sink limitation
The variation in A 1500 of each genotype-nitrogen combination, either across stages or across pruning levels, was positively correlated with T p for adaxial (Supplementary Fig. S5A) or abaxial surfaces (Supplementary Fig. S5B) of leaves.A positive linear correlation of T p with single-culm sink-source ratio (R 2 =0.69;Fig. 6) or whole-plant sink-source ratio (R 2 =0.60;Supplementary Fig. S6) was observed at the grain-filling stage.Note that data points presented on the line x=0 in these figures for panicle-pruned plants (thus, panicle sink was zero) were in a similar range of T p (ca. 2-9 μmol m −2 s −1 ) of those plants without pruning.This suggests that, overall, there was little observable effect of panicle pruning on T p .We then compared the correlation between T p and SLN for non-pruned and panicle-pruned plants separately (Fig. 7A-D).The pruned plants still had close T p correlations with SLN, but the relationship deviated from (e.g. the slope became smaller in most cases) that of the non-pruned plants.
Sink limitation caused by panicle pruning also exerted significant impacts on whole-plant traits, such as increased leaf area, leaf and stem dry weight, and total leaf nitrogen (Table 1).As a result of increased N accumulation in the flag leaf (Fig. 7E-H), panicle pruning lowered leaf photosynthetic nitrogenuse efficiency (PNUE) for each combination of genotype and N level (Fig. 7I-M; P<0.001, Supplementary Table S3).

Discussion
In response to sink-source (im)balance, plants can adjust physiological processes at different scales, and these scales can be inter-connected involving dynamic feedback.We hypothesized that the sub-foliar sink-source (im)balance involving triose phosphate utilization (TPU) is regulated by whole-plant sink-source relationships.We used yellower-leaf (Y) modification and adaxial abaxial illumination, and 21% versus 2% O 2 gas mixture to alter the leaf-scale source activity.Panicle pruning was used to alter whole-plant sink/source ratios.The factorial design involving these two scales enabled linking them.By observing plants under different N supply conditions and at three growth stages, we introduced additional variation of source and sink capacity and enabled the establishment of parameter relationships, notably including TPU-limited photosynthesis and photorespiration-associated N assimilation.

Adaxial versus abaxial measurements on leaf-colour genotypes as a means to alter leaf source activity
In the 2019 experiment using rice genotypes of contrasting leaf colour, a TPU limitation on leaf photosynthesis was observed (Fig. 1).The yellower-leaf variants (Y) differed from their control (C) genotypes in photosynthetic capacity (i.e.A 2000 ) and underlying parameters (J max , V cmax , and T p ), and the difference was mostly expressed when comparing response curves measured under adaxial versus abaxial illumination (Fig. 1).Zhou et al. (2023) demonstrated differences in leaf photosynthetic capacity between C and Y genotypes were associated with intra-leaf photosynthetic N reallocation of the surplus N resources liberated by decreased investment in chlorophyll.However, modifying leaf colour changed not only intra-leaf N partitioning but also leaf morphology, such as increased SLA (i.e.thinner leaves) in Y-genotypes (Supplementary Table S2).
This genotypic difference in leaf thickness probably contributed to our result that when light illuminated the abaxial surface, the decrease in leaf photosynthetic capacity parameters was small in Y genotypes but greater in C genotypes (Fig. 1).This suggests that our C genotypes, like plants in previous reports (e.g.Soares et al., 2008), had an adaptive advantage to the adaxial illumination, which is the predominant condition occurring in the field.As the carbon sinks of the leaf and the whole plant were unchanged, altered leaf photosynthetic capacity via adaxial versus abaxial illumination during gas exchange measurement will alter the sink-source ratio, and this was particularly the case in the C genotypes.

Effects of altered sink-source ratios on leaf triose phosphate utilization limitation
The occurrence of the TPU limitation requires a high photosynthetic rate (Yang et al., 2016).We chose a saturating light intensity (1500 μmol m −2 s −1 ) and CO 2 levels up to 1800 μmol mol −1 for measuring A-C i curves in the 2022 experiment.In our study, A 1500 was highly correlated with T p in all cases (Supplementary Fig. S5), reflecting a parallel change between them.Y modification reduced leaf T p and A 1500 in WYJ background but increased them in ZF background (Fig. 2A-D; Supplementary Tables S1, S2); adaxial illumination gave higher T p than abaxial, especially for C genotypes (Fig. 2E; Supplementary Table S1).Strong correlations between T p and SLN in each genotype (Fig. 4) suggested that N always exerted a positive effect on T p .We also found that the higher the leaf N content (resulting in higher A 1500 ), the lower was the threshold C i where leaf photosynthesis became TPU-limited (Supplementary Fig. S2), as shown by the negative relationship between the threshold C i and T p or SLN (P<0.001;Fig. 5).This relationship was built from data across three growth stages, reflecting dynamic changes in the TPU limitation during rice development.
During the vegetative phase, there is no sink demand from panicles or grains, and all assimilates are used for vegetative organ growth (i.e.absence of a dominant sink like the panicle).In this phase, growing leaves rich in N resources serve as both source and sink organs.The high photosynthetic potential of young plants feeds a plastic (partly facultative) demand exerted by organ development, probably explaining why TPU limitation occurred at a lower C i level (ca.400 μmol mol −1 ; Fig. 5).
After flowering, carbon assimilates are exclusively used to support grain growth, and N resources are mobilized from vegetative organs, particularly from leaves (Sinclair and de Wit, 1975).The resulting decrease in A 1500 and T p (Fig. 2; Supplementary Tables S1, S2), coupled with increased sink demand, alleviated the extent of TPU limitation.
Our study provides insights on how altered sink-source relationships influence T p and thus leaf source activity.The estimated T p and the C i threshold for the onset of TPU limitation were associated with leaf N content across growth stages.Thus, leaf N content not only determines leaf photosynthetic capacity (Nakano et al., 1995), but also modulates TPU limitation in response to sink-source imbalance.

Nitrogen assimilation increases photosynthesis under potential triose phosphate utilization limitation
Assimilating NO 3 − via exporting glycolate carbon from the photorespiratory pathway in the form of amino acids can contribute to photosynthetic carbon uptake (Busch et al., 2018;Busch, 2020;Fu et al., 2022).Our A-C i curves, indicating A was higher under 21% O 2 than under 2% O 2 conditions, confirmed this-although this advantage from photorespiration diminished with advancing growth stages (Supplementary Fig. S2).The increase in A largely depends on the extent of glycolate carbon exported from the photorespiratory pathway (Bauwe et al., 2010;Busch et al. 2018;Yin et al. 2021; also see Supplementary Fig. S2A, B).Fu et al. (2022) demonstrated that the carbon flow out of the pathway was primarily in the form of serine, with a proportion of 23-41% in tobacco plants.Our estimates of parameter α S from model analysis (up to 0.37;  Supplementary Table are in line with the measured values of Fu et al. (2022) as well as with the modelling results of Busch et al. (2018).Based on modelled α S , we further assessed the effect of various experimental factors on photorespiration-associated N assimilation.We found that similar to the effect on T p , leaf N content is also critical to α S , as evidenced by correlations between α S and SLN (Fig. 4I-P).However, this positive effect of N level disappeared under Y modification (Supplementary Table S3), since both Y genotypes had lower α S than their corresponding C genotypes (Figs 1, 3; Supplementary Table S1).In conjunction with this, there was a negative relationship between α S and SLA (Supplementary Fig. S3), suggesting that the thinner leaves associated with Y modification may have limited NO 3 − assimilation.This limiting effect was enhanced by illuminating the abaxial surface of leaves during measurements (Fig. 3E; Supplementary Table S3).In our experiment, urea was used as N source in fertilizing plants.Busch et al. (2018) found that α S was higher in sunflower plants fertilized with NO 3 − -N than those with NH 4 + -N.Whether this is also the case when applied to our rice plants of different leaf colours remains to be investigated.Previous studies (Reich et al., 2006;Bloom et al., 2012Bloom et al., , 2014) ) reported that elevated growth CO 2 (resulting in increased source) inhibited the NO 3 − assimilation in C 3 plants.Here, we found a similar result with a significant drop in α S for the plants after panicle pruning (resulting in decreased sink) (Fig. 3).However, the underlying mechanisms might differ.Elevated growth CO 2 not only decreased photorespiration but also diluted leaf N concentration via increased biomass (Yin, 2013;Igarashi et al., 2021).In contrast, panicle pruning resulted in an increased N content in leaves (Table 1; Fig. 7E-H).Nevertheless, despite higher leaf N, a lower α S (thus, probably less N assimilation) was observed in leaves of panicle-pruned plants, suggesting a dominating role of whole-plant sink-source ratio in the control of leaf N metabolism.

Feedback inhibition of whole-plant sink limitation on leaf photosynthesis
It has long been observed that the plant sink-source ratio can affect leaf photosynthesis (Tanaka and Fujita, 1974;Crafts-Brandner and Egli, 1987;Arp, 1991;Li et al., 2015;Aslani et al., 2020).In line with this, studies indicated that a larger sink capacity can increase the effect of elevated atmospheric CO 2 on photosynthesis and yield (Erice et al., 2011;Hasegawa et al., 2013;Rossi et al., 2015;Kikuchi et al., 2017;Dingkuhn et al. 2020).Our finding of a positive correlation between T p and the sink-source ratio for single culm (Fig. 6) or for the whole plant (Supplementary Fig. S6) confirmed the role of sink size in modulating source activity of leaves.The reduced T p under sink limitation is probably associated with the accumulation of sucrose in leaf photosynthetic tissues (Fabre et al. 2019(Fabre et al. , 2020)).This may give a signal regulating the activity of sucrose-phosphate synthase (SPS) through transduction of SnRK1 protein kinases (Halford and Hey, 2009), and the SPS feedback inhibition on sucrose synthesis can decrease export of triose phosphates from the chloroplast (McClain and Sharkey, 2019).Our results also showed that panicle pruning dramatically increased the dry matter accumulation in leaves and stems (Table 1).
Given these considerations, one would expect that for the 15 DAF stage, panicle-pruned plants would have lower T p compared with intact plants where the TPU limitation was alleviated.Surprisingly, distinct from previous findings (Fabre et al., 2019;Nomura et al., 2022), T p was little affected under this apparently sink-limited state (Figs 2, 6).In the study of Fabre et al. (2019), elevated CO 2 (increased source) coupled with pruning treatment (decreased sink) imposed a dual effect on T p , which might contribute to more significant sink limitation and feedback effect in their study than in our study.However, the mechanism behind this difference could also be attributed to a larger amount of N accumulated in the leaves of our panicle-pruned plants (Fig. 7E-H).After removal of all panicles, no N was required to be remobilized from leaves to support grain growth and more N remained in leaves.Leaf N is the most important resource for the photosynthetic machinery (Nakano et al., 1995), as shown by our close relationship between T p and SLN, even in pruned plants (Fig. 7A-D).Thus, we posit that the feedback inhibition of sink limitation on T p still occurred in our panicle-pruned plants, but the inhibitory effects from sink limitation were offset by the increased SLN as a consequence of the smaller N remobilization from leaves, thereby resulting in no observable impact of pruning on T p in leaves.The increased SLN did not result in higher N assimilation as α S was lower in panicle-pruned plants (Fig. 3; Supplementary Table S1); as a result, these plants might not benefit from amino acid export for higher CO 2 uptake rates, but instead, had similar or lower A 1500 compared with the control plants (Supplementary Table S2).Because of the increased SLN (Fig. 7E-H) accompanied with no or little increase in A 1500 , leaf PNUE decreased in the panicle-pruned plants (Fig. 7I-M).Taking these together, our results suggest that T p and leaf photosynthesis during grain filling are controlled by the whole-plant sink demand and N budgets.

Concluding remarks
We have shown that the differences in photosynthetic rates between adaxially and abaxially illuminated leaves and the extent of such differences depends on leaf colours of the genotypes.Based on this finding, we further assessed how TPU limitation of photosynthesis, involving photorespiration-associated N assimilation, can be affected by altered sink-source ratios at different scales.We found: (i) Higher leaf N (observed at early growth stages or high N inputs) caused TPU limitation at relatively low intercellular CO 2 concentration.(ii) The proportion of photorespiratory glycolate-carbon exported as serine (α S ) was positively correlated with leaf N content, suggesting that photorespiration was involved in leaf N assimilation more in high-N than low-N leaves.However, values of α S were smaller in yellower-leaf genotypes, under illumination of abaxial leaf sides, and when panicles were removed.(iii) Absence of observable effect of panicle pruning on the rate of TPU (T p ) suggested that the feedback inhibition of photosynthesis by whole-plant sink limitation was offset by the positive effect of increased leaf N content, caused by lesser N remobilization.
Thus, our results revealed the crucial role of leaf N in affecting photosynthetic parameters T p and α S , and thus TPU limitation of photosynthesis.Likewise, N resources modulate the link between leaf-level TPU limitation and whole-plant sink limitation during rice grain filling.The latter link could be

Fig. 1 .
Fig. 1.Effects of adaxial versus abaxial measurement on the photosynthetic parameters in four rice genotypes (data measured at the tillering stage in the 2019 experiment).(A-D) Light-response curves of photosynthesis (A) at the CO 2 concentration of 400 μmol mol ─1 .(E-H) CO 2 -response curves at the light intensity of 1000 μmol m ─2 s ─1 , for rice control (C) genotypes (circles) and their yellower-leaf (Y) variant genotypes (triangles).Data shown as the mean of four replicates (±SE) for each genotype, with black symbols representing measurement on adaxial surface of leaves and red symbols representing measurement on abaxial surface of leaves.WYJ and ZF are the abbreviations of two genetic backgrounds Wuyunjing 3 and Zhefu 802, respectively.For (A-D), the curves are drawn from Equation 6 using fitted parameter values.The estimated maximum linear electron transport under saturating light (J max , μmol m ─2 s ─1 ), net photosynthesis rate under light intensity of 2000 μmol m ─2 s ─1 (A 2000 , μmol m ─2 s ─1 ), and the percentage difference in the J max and A 2000 (calculated as [(Adaxial−Abaxial)/Adaxial]×100) are listed.For (E-H), the curves representing A c -(dashed curve), A j -(dotted curve), and A p -limited (full curve) parts are drawn from Equation 6 using fitted values of the parameters: the estimated maximum rate of Rubisco carboxylation (V cmax , µmol m ─2 s ─1 ), rate of triose phosphate utilization (T p , µmol m ─2 s ─1 ), and the proportion of glycolate carbon exported from the photorespiratory pathway in the form of serine (α S ).The different letters indicate statistical significance at the P<0.05 level for the estimated parameters between adaxial and abaxial measurements, and the asterisks represent significant differences (P<0.05) between C genotype and its Y variant.

Fig. 2 .
Fig. 2. Effects of altered sink-source ratios on parameter T p (the 2022 experiment).(A-D) The rate of triose phosphate utilization (T p , based on measurements on the adaxial leaf surface) for rice control (C) genotypes and their yellower-leaf (Y) variant genotypes of intact or panicle-pruned plants at tillering (TS), flowering (FS), and 15 days after flowering (DAF) stages under low-nitrogen (N1, white bars) and high-nitrogen (N2, dashed bars) levels.The value of each bar representing the mean ±SE of four replicates was estimated by fitting curves to CO 2 exchange data (see Supplementary Fig. S2).For intact plants (no pruning), different letters indicate statistical significance at the P<0.05 level between three stages within each genotype-nitrogen combination, and the asterisks in black represent significant differences (P<0.05) between N1 and N2 levels within each genotype and stage.The asterisk in red represents significant difference (P<0.05) for a given genotype-nitrogen combination between pruned and un-pruned plants at 15 DAF stage.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.(E) Comparisons of the values of T p measured at adaxial surface versus those measured at abaxial surface.The percentage is the average of adaxial relative to abaxial parameters and the diagonal line is the 1:1 line.Data represented by different colours and symbols are from C genotypes (circles) and Y-variant genotypes (triangles) of intact (open symbols) and panicle-pruned (filled symbols) plants at TS (blue), FS (red), and 15 DAF (green) stage.Each point represents the mean of three or four replicates.

Fig. 3 .
Fig. 3. Effects of various growth stages and nitrogen levels on parameter α S , the proportion of glycolate carbon exported from the photorespiratory pathway in the form of serine (the 2022 experiment).(A-D) α S (based on measurements on the adaxial leaf surface) for rice control (C) genotypes and their yellower-leaf (Y) variant genotypes of intact or panicle-pruned plants at tillering (TS), flowering (FS), and 15 days after flowering (DAF) stages under low-nitrogen (N1, white bars) and high-nitrogen (N2, dashed bars) levels.The value of each bar representing the mean ±SE of four replicates was estimated by fitting curves to CO 2 exchange data (see Supplementary Fig. S2).For intact plants (no pruning), different letters indicate statistical significance at the P<0.05 level between three stages within each genotype-nitrogen combination, and the asterisks in black represent significant differences (P<0.05) between N1 and N2 levels within each genotype and stage.The asterisks in red represent significant differences (P<0.05) for a given genotype-nitrogen combination between pruned and un-pruned plants at 15 DAF stage.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.(E) Comparison of the values of α S measured at adaxial surface versus those measured at abaxial surface.The percentage is the average of adaxial relative to abaxial parameters and the diagonal line is the 1:1 line.Data represented by different colours and symbols are from C genotypes (circles) and Y-variant genotypes (triangles) of intact (open symbols) and panicle-pruned (filled symbols) plants at TS (blue), FS (red), and 15 DAF (green) stage.Each point represents the mean of three or four replicates.

Fig. 4 .
Fig. 4. Relationship between photosynthetic parameters and leaf nitrogen content (based on measurements on the un-pruned plants in the 2022 experiment).(A-H) Relationship between triose phosphate utilization rate (T p ) and specific nitrogen content (SLN).(I-P) Relationship between the proportion of glycolate carbon exported from photorespiratory pathway in the form of serine (α S ) and SLN.Data represented by different colours and symbols are from tillering (TS, blue), flowering (FS, red), and 15 days after flowering (DAF) (green) stage under low-nitrogen (N1, open symbols) and highnitrogen (N2, filled symbols) levels, with circles for rice control (C) genotypes and triangles for their yellower-leaf (Y) variant genotypes.Linear regressions were fitted for each genotype with four or five replicates across two nitrogen levels and three stages.The significance of each correlation is shown by asterisks: *P<0.05,**P<0.01,***P<0.001.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.

Fig. 5 .
Fig. 5.The threshold C i values in relation to leaf physiological parameters.(A) Relationship between threshold C i and the rate of triose phosphate utilization (T p ) (data based on measurements on the adaxial surface of leaves in the 2022 experiment).(B) Relationship between threshold C i and specific leaf nitrogen (SLN).The threshold C i represents the transition point where the photosynthesis-limiting process changed from electron transport to TPU, derived from CO 2 response curves (see Supplementary Fig. S2).Data represented by different colours and symbols are from tillering (TS, blue), flowering (FS, red), and 15 days after flowering (DAF) (green) stage under low-nitrogen (N1, open symbols) and high-nitrogen (N2, filled symbols) levels, with circles for rice control (C) genotypes and triangles for their yellower-leaf (Y) variant genotypes.Linear regressions were fitted for overall data with the significance of each correlation shown by asterisks: ***P<0.001.

Fig. 6 .
Fig. 6.Relationship between the rate of triose phosphate utilization (T p , based on measurements on the adaxial leaf surface) and single-culm sink-source ratio.Here, following Fabre et al. (2020), the ratio of flag leaf area (source) to the fertile spikelet number of the panicle (sink) on the culm was used as an indicator of the single-culm sink-source ratio (also see the text).Data are for rice control (C) genotypes (circles) and yellower-leaf (Y) variant genotypes (triangles) from grain-filling stage under low-nitrogen (N1, open symbols) and high-nitrogen (N2, filled symbols) levels in the 2022 experiment, with cv.Wuyunjing 3 (WYJ) in black and cv.Zhefu 802 (ZF) in red.For those plants with panicle pruning, we define their sink-source ratio to be zero, so all their data points fall on the y-axis.Linear regression was fitted for data (representing no pruning) with the significance of the correlation shown by asterisks: ***P<0.001.

Fig. 7 .
Fig. 7.The effects of panicle pruning on leaf photosynthetic parameters (based on measurements on the adaxial leaf surface in the 2022 experiment).(A-D) Relationship between the rate of triose phosphate utilization (T p ) and specific leaf nitrogen (SLN) for the intact and panicle-pruned plants.(E-H) Effect of panicle pruning on SLN.(I-M) Leaf photosynthetic nitrogen-use efficiency (PNUE) at tillering (TS), flowering (FS), and 15 days after flowering (DAF) stage (see Supplementary TableS2for the definition of PNUE).Linear regressions in (A-D) were fitted for each genotype with four or five replicates under two nitrogen levels; the significance of each correlation is shown by asterisks: *P<0.05,**P<0.01,***P<0.001.Data in (E-M) represent the mean ±SE of four replicates; the asterisks represent significant differences (P<0.05)within each genotype-nitrogen combination between unpruned and pruned plants.The data in (A-H) are from the 15 DAF stage, and the data in (A-D) and (I-M) represent the values for rice control (C) genotypes (circles) and their yellower-leaf (Y) variant genotypes (triangles) of the intact (black) and panicle-pruned (red) plants under low-nitrogen (N1, open symbols) and highnitrogen (N2, filled symbols) levels.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.
Fig. 7.The effects of panicle pruning on leaf photosynthetic parameters (based on measurements on the adaxial leaf surface in the 2022 experiment).(A-D) Relationship between the rate of triose phosphate utilization (T p ) and specific leaf nitrogen (SLN) for the intact and panicle-pruned plants.(E-H) Effect of panicle pruning on SLN.(I-M) Leaf photosynthetic nitrogen-use efficiency (PNUE) at tillering (TS), flowering (FS), and 15 days after flowering (DAF) stage (see Supplementary TableS2for the definition of PNUE).Linear regressions in (A-D) were fitted for each genotype with four or five replicates under two nitrogen levels; the significance of each correlation is shown by asterisks: *P<0.05,**P<0.01,***P<0.001.Data in (E-M) represent the mean ±SE of four replicates; the asterisks represent significant differences (P<0.05)within each genotype-nitrogen combination between unpruned and pruned plants.The data in (A-H) are from the 15 DAF stage, and the data in (A-D) and (I-M) represent the values for rice control (C) genotypes (circles) and their yellower-leaf (Y) variant genotypes (triangles) of the intact (black) and panicle-pruned (red) plants under low-nitrogen (N1, open symbols) and highnitrogen (N2, filled symbols) levels.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.
Fig. 7.The effects of panicle pruning on leaf photosynthetic parameters (based on measurements on the adaxial leaf surface in the 2022 experiment).(A-D) Relationship between the rate of triose phosphate utilization (T p ) and specific leaf nitrogen (SLN) for the intact and panicle-pruned plants.(E-H) Effect of panicle pruning on SLN.(I-M) Leaf photosynthetic nitrogen-use efficiency (PNUE) at tillering (TS), flowering (FS), and 15 days after flowering (DAF) stage (see Supplementary TableS2for the definition of PNUE).Linear regressions in (A-D) were fitted for each genotype with four or five replicates under two nitrogen levels; the significance of each correlation is shown by asterisks: *P<0.05,**P<0.01,***P<0.001.Data in (E-M) represent the mean ±SE of four replicates; the asterisks represent significant differences (P<0.05)within each genotype-nitrogen combination between unpruned and pruned plants.The data in (A-H) are from the 15 DAF stage, and the data in (A-D) and (I-M) represent the values for rice control (C) genotypes (circles) and their yellower-leaf (Y) variant genotypes (triangles) of the intact (black) and panicle-pruned (red) plants under low-nitrogen (N1, open symbols) and highnitrogen (N2, filled symbols) levels.WYJ and ZF are the abbreviations of two genetic backgrounds: cv.Wuyunjing 3 and cv.Zhefu 802.
the analysis of variance, the significance is shown by asterisks: *P<0.05,**P<0.01,***P<0.001,according to the LSD test.Note, we use 'genotype' rather than 'Y modification' as a fixed factor mainly because Y modification produces different effects on leaf photosynthetic physiology between WYJ (cv.Wuyunjing 3) and ZF (cv.Zhefu 802) backgrounds (see Results).

Table 1 .
Values (means ±SE among four replicates) of rice and source organs for single culm and whole plant in the control (C) genotypes and the yellower-leaf variant (Y) genotypes at 15 d after flowering stage in the 2022 experiment.