Effect of N and P fertilization on the allocation and fixation of photosynthesized carbon in paddy soil

ABSTRACT Potted rice seedlings independently treated with N, P, and NP were continuously13CO2 labeled to investigated the influence of N and P application on the contribution of photosynthesized C to the rhizosphere versus bulk soil and particulate organic matter (POM) versus mineral fraction (MIN). N and NP enhanced net assimilated 13C on day 14 (D14), with maximum C assimilation occurring on day 22 (D22) under NP. Aboveground biomass retained more 13C than belowground biomass for all treatments. 13C incorporation into the rhizosphere exceeded that in bulk soil, with the maximum (6–10%) found under N addition. Newly assimilated 13C incorporated into POM increased in the rhizosphere under N and NP conditions, whereas MIN remained largely unaffected. 13C-MBC proportion in the total microbial biomass C (MBC) pool revealed that N and NP stimulated microbial activity to a greater degree than P. The main portion of 13C in the rhizosphere and bulk soil was found in POM on D14, which decreased over time due to microbial utilization. Contrastingly, root-derived 13C in the MIN remained unchanged between sampling days, which indicates that the stabilization of rhizodeposits in this fraction might be the potential mechanism underlying SOM sequestration in paddy soils.


Introduction
About 75% of the world's rice is produced under paddy conditions (Maclean et al. 2013;Oo et al. 2018), and paddy soil covers a total area of about 161 million ha worldwide . Besides challenges in food supply (Peng, Tang, and Zou 2009a), paddy soils play a crucial role in climate change mitigation, because they can sequester more carbon (C) than upland soils (Tian et al. 2013;Creamer et al. 2016).
Plant rhizodeposits, such as root exudates, mucilage, secretions, sloughed-off root cells/tissues, and senescence-derived compounds (Jones, Nguyen, and Finlay 2009), are among the main pools that link plant derived-C with soil organic matter (SOM), and can contribute up to 11% (about 30% of the C allocated to roots) of the net fixed C in the soil (Jones, Nguyen, and Finlay 2009;Liu et al. 2019a). As rhizodeposits are a highly labile part of plant derived-C (especially exudates), most of them are directly mineralized, whereas only about 5% can accumulate in the SOM (Hütsch, Augustin, and Merbach 2002;Kogel-Knabner 2002). The main location of this accumulated C in upland soils is the pool of particulate organic matter (POM) (up to 54%) or the mineral fraction (up to 46%) (Bradford, Fierer, and Reynolds 2008). The incorporation of root-derived C into these pools in paddy soils can be even greater (Wei et al. 2021) than that in upland soils, because the processes involved in mineralization of organic matter are inhibited under conditions of limited oxygen availability (Devêvre and Horwáth 2000;Olk, Brunetti, and Senesi 2000;Xu et al. 2018), which promotes C sequestration (Mandal et al. 2008). It is also possible that stabilization of rhizodeposited-C occurs in rhizosphere-free soil due to the specific water regime used in paddy soils, which results in migration of C molecules within the soil.
POM is expected to have faster turnover times and a lower capacity for long-term C storage than the mineral-associated pool, which suggests that mineralassociated C is processed more than POM by microorganisms and predominantly contains microbially derived material as a part of its composition (Schlesinger and Lichter 2001;Grandy and Robertson 2007;von Lützow et al. 2007). Therefore, more C associated with the mineral fraction is important for long-term SOC sequestration. Atere et al. (2018) indicated that newly assimilated C or plant residues pass through two stages; i) first, incorporation into macroaggregates and ii) later, redistribution into smaller aggregates belonging to silt-and clay-size classes. However, the possible percentage of root-derived C that can accumulate in POM and mineral fractions of the SOM in paddy soils has not yet been estimated.
Among the factors affecting the amount of rhizodeposits, atmospheric CO 2 concentration (van Ginkel, Gorissen, and van Veen 1997), light and temperature (Hew, Krotkov, and Canvin 1969), fertilizer application Ge et al. 2015;Xiao et al. 2019), soil texture and irrigation management (Atere et al. 2018), tillage (Mo et al. 2019), plant growth stage , plant species (Derrien, Marol, and Balesdent 2004;Mwafulirwa et al. 2016), photosynthetic capacity (Kuzyakov and Cheng 2001), and assimilated 13 C allocated into plant shoots and roots directly drive rhizodeposition (Ge, Luo, and He 2019). Nitrogen (N) and phosphorus (P) are essential plant nutrients, that affect the quality (indirect effects through buildup of more recalcitrant and less biodegradable) and quantity of photosynthesized C at various growth stages by stimulating plant growth (Saggar, Hedley, and Mackay 1997), and they also alter the distribution and fates of photosynthesized C into the shoots, roots, and soil-system (Liu et al. 2019b). Nitrogen fertilizers can nearly double the biomass of rice (Chun et al. 2013) because of the increase in chlorophyll content and the activity of rubisco enzymes, which consequently increases CO 2 assimilation (Xiao et al. 2019) and rhizodeposition by 2-4 times (Ge et al. 2015;Xiao et al. 2019). However, N fertilization can also stimulate mineralization of both rhizodeposits and native SOM, resulting in only 2-33% of the net assimilated C remaining in the SOM composition . Phosphorus also improves the growth of plant roots and increases N uptake (Bowatte et al. 2006), thereby increasing the proportion of root-derived C allocated to the POM fraction by 1-1.5 times during rice growth (Atere et al. 2018). However, what remains unclear is the effect of N and P, and their combined application on the proportion of rhizodeposit-derived C that can be stabilized and fixed in SOC fractions, and the degree to which rhizosphere-free soil may be affected. Thus, the aim of this study was to investigate the allocation and fixation of newly photosynthesized C in response to N and P fertilization in a rice paddy system using the 13 CO 2 continuous labeling approach. We hypothesized that i) N and P addition will stimulate the allocation and fixation of newly assimilated C into the soil during the early plant growth stages with N having a greater effect than P, as it promotes plant biomass and root exudate production, and ii) N and P will promote the allocation and fixation of newly assimilated C into both the POM and mineral fraction in the rhizosphere and bulk soil.

Study site description and soil properties
Soil was collected from the plowing layer (0-20 cm) of soil from a subtropical paddy rice field located at Changsha Research Station, Hunan, China (113°19′E, 28°33′N, 80 m a.s.l). The mean annual temperature and precipitation of the area are 17.5°C and 1,300 mm, respectively. For physicochemical analyses, soil was sieved through a < 4 mm sieve to remove the coarse root residues, and then air-dried at 35 ± 1°C and ground with a mortar and pestle prior to analysis. The soil was classified as a typical Stagnic Anthrosol and had the following texture: sand 28%, silt 66%, clay 6%. The soil contained total organic C, total N, and total P concentrations of 14.26, 1.45, 0.75 g kg, −1 respectively, and 12.75 mg Olsen-P kg −1 with a pH of 5.43 and cation exchange capacity of 7.71 mmol kg 1 .

Experimental design
Air-dried prepared soil (1.26 kg/pot) was used to fill 24 pots with an inner diameter of 11 cm and height of 20 cm. A rhizosphere bag (rhizo-bag with 0.34 kg soil; mesh 30 μm; 3.5 × 15 cm), which limited the passage of the root but permitted the movement of nutrients and water, was placed inside each pot to divide the soil into the rhizosphere and non-rhizosphere sections inside and outside the bag, respectively. Three 25-day-old rice seedlings of Zhongzao 39 (Oryza sativa L.,) were transplanted into the rhizo-bag. Further, basal fertilizers were added to each pot (KCl (80 mg K kg −1 soil), and dicyandiamide (nitrification inhibitor: 26 mg N kg −1 ). Soils in all pots were regularly saturated with distilled water (with a 2-3 cm water layer above the soil surface throughout the rice-growing season) The experimental design was a factorial design with four treatments and three replicates (Control with zero N and P, N with 250 mg N kg −1 added in the form of urea; P with 80 mg P kg −1 added in the form of NaH 2 PO 4 ; and NP with 250 mg N kg −1 plus 80 mg P kg −1 added in the form of urea and NaH 2 PO 4 , respectively.). Thus, the fertilization experiment comprised 12 pots, used for 13 C labeling, and an additional 12 pots without 13 C labeling (controls). The unlabeled controls served as references for the determination of natural 13 C abundance and calculation of 13 C atom percent excess. Therefore, the unlabeled pots were placed outside 10 m away from the labeled treatments in the field and were given the same amount of water and fertilizers.

CO 2 continuous labeling
Immediately after transplanted, rice seedlings were continuously labeled with 13 CO 2 over 22 days in airtight growth chamber systems (80 × 250 × 120 cm) as reported by Ge et al. (2012Ge et al. ( , 2015, which were exposed to natural sunlight in the rice field. A beaker with 1 M NaH 13 CO 3 (50 atoms % 13 C) was kept inside the glass chamber to release 13 CO 2 upon the addition of H 2 SO 4 (10 mL, 0.5 M). 13 CO 2 concentration was constantly maintained between 360 and 380 μL CO 2 L −1 inside the chamber via further reactions or diversion of gas flow through CO 2 traps (1 M NaOH) for absorbing excess CO 2 higher than the desired concentration. The growth labeling chamber was fitted with two temperature/humidity sensors (SNT-96S; Qingdao, China), and two fans for continuous air circulation to keep the temperature inside within 1°C of the ambient temperature via an air-conditioning system.

Sampling and harvesting
Plants and soil were harvested on days 14 and 22 (D14 and D22) after 13 C labeling began. Shoots were cut off at the stem base, and the rhizosphere bags were collected from the pots. Roots were separated carefully from rhizosphere bags, then washed for a minute in 0.01 M CaCl 2 (pH 6.2), and under fresh water to remove adhering soil. Samples were dried at 60°C for 48 h and then the shoots, roots, and soil sub-samples were ball milled and used to estimate the total C (TC), stable C isotope ratio (δ 13 C), total N (TN), and Olsen-P (OP). Meanwhile, fresh subsamples soil were used to measure both types of mineral N (NH 4 + and NO 3 − ), microbial biomass C (MBC), and dissolved organic C (DOC) content (Tian et al. 2013). MBC was determined using the chloroform fumigation-extraction method (Wu et al. 1990). The rhizosphere and bulk soils were used for aggregate fractionation and measurement of TC, δ 13 C, TN, mineral N, and OP. and TC of soil (aggregate) samples, which were analyzed using an isotope ratio mass spectrometer (IRMS, MAT253; Thermo-Fisher Scientific, Waltham, MA, USA), coupled with an elemental analyzer (Thermo Scientific FLASH 2000, USA; Thermo-Fisher Scientific, Waltham, MA, USA).

Soil physical fractionation
Soil organic C (SOC) was fractionated into POM (>53-µm) and MIN (<53-µm) using a method adapted from Cambardella and Elliott (1993). Briefly, 10 g of subsample (of 2-mm sieved and air-dried soil) was shaken for 15 h with 0.5% sodium hexametaphosphate (1:3 ratio), and then, immediately after shaking, the dispersed soil suspension was passed through a 53-µm sieve. The dispersed particles retained on the sieve (>53-µm) were carefully collected and considered as POM fraction; the leftover suspension was centrifuged, and then, the settled soil particles were combined with the sediment and assigned to the MIN fraction. Separated POM and MIN fractions were oven-dried (60°C), homogenized, ground with a mortar and pestle, and then analyzed for TC and δ 13 C.

Calculations and statistical analysis
The amount of photosynthesized 13 C accumulated in the plant organs (whole plant, roots, and shoots) and soil fractions (bulk, rhizosphere, POM, MIN, DOC, and MBC) ( 13 C a ; mg 13 C kg −1 or mg C m −2 ) were calculated using Equation 1: where "l" and "nl" are labeled and non-labeled samples, respectively; "a" refers to a plant organ (whole plant, root, shoot) or a soil fraction (bulk, rhizosphere, POM, MIN, DOC, MBC); and "TC a " is the total carbon content (mg kg −1 ) of a.
Net assimilated 13 C was calculated using Equation 2: The photosynthesized 13 C distribution in plant and soil was calculated as follows: where 13 C proportion(a) is the proportion of total photosynthesized 13 C amount in the plant-soil system represented by a (%); To calculate 13 C amount incorporated into plant organs (whole plant, root, shoot) or soil fractions (bulk, rhizosphere, POM, MIN) in each pot (as mg 13 C kg −1 soil), the following equation was used: where 13 C amount(a) is 13 C amount of a in each pot (mg 13 C kg −1 ); and M a is the mass (kg kg −1 ) of a in each pot. 13 C incorporated into microbial biomass ( 13 C-MBC) was calculated as follows: where (atom 13 C%) indicates atom 13 C% in the fumigated (f) and unfumigated (uf) soil extract, and C is the TC content of the soil extract. "L," and "NL" are labeled, and non-labeled samples, respectively. Thus, the amount of 13 C allocated into MBC was calculated from 13 C in fumigated soil minus that in unfumigated soil divided by a conversion factor of 0.45 (Vance, Brookes, and Jenkinson 1987).
The rhizosphere effect was calculated as the amount of 13 C in rhizosphere soil divided by the amount of 13 C in bulk soil.
The effect of fertilization (four factor levels: control, N, P, and NP) on the contribution of 13 C to soil and plant pools was tested using one-way ANOVAs, and Tukey's post-hoc tests were performed after the residuals were checked for normality and homogeneity.

Effect of fertilization on plant growth
At D14, maximum shoot biomass was found with N and NP fertilization treatments (p < 0.001). While, at D22, highest shoot biomass was found under P and NP treatments. In contrast, root biomass was largely independent of fertilization on D14 (Table  1), and significantly affected by P application on D22 (p < 0. 05). Shoot C followed the same trends as the shoot biomass, and highest values were found on D14 for NP compared to control (1.8-fold that of the control). On D22, the respective increase in shoot C under NP was 2.21-fold that of the control. Nitrogen content increased in the shoots and roots under all treatments compared to the non-fertilized treatment, with the maximum amount found in response to the combined treatment (NP) on D22. Thus, the NP treatment had the greatest effect on shoot growth and C content of the shoot biomass.

Distribution of Photosynthesized 13 C in rice-soil system
Maximum content of net photosynthesized rice 13 C (shoot and root + rhizo and bulk soil) ranged from 2247 to 16,635 mg 13 C m −2 soil during the labeling period, with maximum values observed in response to NP ( Figure S1; P < 0.001). The main fraction of the photosynthesized 13 C was retained in the shoots (55.12-77.40%), whereas only 25-44% C was transferred by the rice plant belowground through rhizodeposition ( Figure 1). Maximum 13 C was found in the shoots in response to NP on D14 (P < 0.001), whereas on D22, the proportion of 13 C increased in response to all other treatments compared to that in the control. In contrast, the roots incorporated the lowest proportion of 13 C under NP on D14. For all treatments (except NP), the proportion of 13 C in the roots decreased between D14 and D22. Two to three times more 13 C was found in the rhizosphere than that in bulk soil, with the maximum proportion (6-10%) observed under rhizosphere soil in response to N treatment on both sampling days (Figure 1), while no effect was observed for the other fertilizer treatments. In bulk soil, a significant proportion of 13 C was found in P (5.88% on D14 and 3.50% on D22) and control (4.1% on D14 and 3.6% on D22) treatments compared to that in N and NP treatments (P < 0.05). Thus, N and NP application promoted the proportion of 13 C assimilated by the shoots, and N fertilization was responsible for the most input of C into the rhizosphere. Table 1. Biomass, total C (TC), total N (TN), and C/N ratio of rice plant roots and shoots on day 14 and 22 of 13 C continuous labeling, and results of the ANOVA used to analyze the effects of N, P, and NP fertilization on these parameters. Values represent the means (±SE), n = 3.

Effect of fertilization on 13 C partitioning between organic matter pools
The proportion of assimilated 13 C allocated into the POM fraction of the rhizosphere soil was 1.5-2 times higher than that assimilated into the MIN on both sampling days, with the maximum proportion observed in response to N and NP application (  with a maximum amount observed in response to P treatment. Thus, the newly fixed C was preferentially incorporated into the POM fraction in response to N treatment, whereas the MIN fraction was largely unaffected by the addition of fertilizers. In rhizosphere soil, the amount of 13 C incorporated into microbial biomass (MBC) was two times higher in case of N and NP than that in case of control and P treatment on D14 ( Figure S2; P < 0.001) and tended to increase 3.5 times on D22 ( Figure S2; P < 0.001). The DOC pool contained 2-5 times less 13 C compared to the MBC for both rhizosphere and bulk soil, with significant amounts found for the rhizosphere soil (except the control) (Figures 3 and 4). The proportion of 13 C-MBC in the total MBC ranged from 0.005% to 0.32%, and was significantly higher under N and NP than that in the other treatments ( Figure S2; P < 0.005), which suggests that the activity of microorganisms is higher under these Proportion of 13 C incorporated into dissolved organic C (DOC) and microbial biomass C (MBC) as % from total assimilated 13 C. Values represent the means ± SE, n = 3. The lower-case letters indicate significant differences between treatments within the same C pool i.e., shoot, root, rhizosphere (rhizo) and bulk soil (bulk) (p < 0.05).
fertilization regimes. The proportion of 13 C in the DOC under N was two times higher than under the other treatments on D14 in rhizosphere soil, and two times higher under NP in bulk soil (Figure 3; P < 0.05). The N treatment induced a higher 13 C-rhizo/ 13 C-bulk ratio compared to that observed in the other treatments on both sampling days (Table 2; P < 0.001). Assimilated 13 C allocated into plant shoots and roots was significantly correlated with that incorporated into the MBC, DOC, SOC, MIN, and POM fractions (Table S3; P < 0. 01).

C allocation into SOM pools
The main primary sources of organic C in soils are plant residues and root exudates released into the rhizosphere (Ge et al. 2013), and the amount of these inputs depends on the life cycle of the plant and soil fertility, which are both strongly affected by the application of N and P fertilizers (Gu et al. 2009;Peng, Tang, and Zou Figure 4. Conceptual 13 C distribution pattern in the soil without fertilization and with N and P fertilization on day 14 (D14) and 22 (D22) of 13 C-CO 2 continuous labeling. Table 2. The rhizosphere effect (rhizo/bulk ratio) of 13 C allocation in soil organic carbon (SOC), dissolved organic carbon (DOC), microbial biomass carbon (MBC), mineral (MIN), and particulate organic matter (POM) fractions following fertilization treatments, and results of the ANOVA used to analyze the effects of N, P, and NP fertilization on these parameters. Values represent the means (±SE), n = 3. 2009b; Chun et al. 2013;Chen et al. 2016;Guo et al. 2019). The application of N, P, and NP increased shoot and root biomass and reduced the C/N ratio of the roots particularly on D22 (Table 1). These effects can be attributed to an increase in chlorophyll content, especially under N application, and thus, rubisco enzyme activity, which promotes CO 2 assimilation and consequently increases plant biomass (Xiao et al. 2019). Besides slowing C turnover in the above ground parts, fertilization can reduce the loss of assimilated 13 C into the soil under N application (Xiao et al. 2019).
The main fraction of the assimilated 13 C was retained in the shoot mass under N and NP (Figure 1; Leake et al. 2006), and less C was transferred to the roots compared to that observed in the case of the unfertilized treatment (control). This follows the resource optimization hypothesis, which indicates that when nutrient availability is increased, plant invest less effort in resource acquisition (Jones, Nguyen, and Finlay 2009). The levels of newly assimilated C derived and fixed in the soil were also enhanced by the N application, particularly during the early growth stage (D14) and decreased with time (D22), which indicates that this fertilizer treatment, as well as C inputs in the form of rhizodeposits, can be used as the main treatment to promote high growth rates in rice. This finding supports our hypothesis that with fertilization, more photosynthetically fixed carbon would be exuded into the soil during the early growth stage, with N having a stronger effect than P. This may be because N fertilization stimulates the release of root exudates into the soil, whereas under P fertilization, the newly assimilated C could be retained for new root production or rapidly mineralized by microorganisms.
The amount of rhizo-derived 13 C decreased from the root surface (rhizosphere soil) to the bulk soil (Figures 1-3), because of the following reasons: i) the limitations of root exudate diffusion (Vale et al. 2005) and ii) the newly assimilated C is taken up and utilized rapidly by microorganisms (Gunina and Kuzyakov 2015;Zhu et al. 2016;Zhao et al. 2019). The relative proportion of 13 C left in the soil also decreased with time ( Figure 1 and Figure S1), as it was decomposed by microorganisms. The fractionation method partitions the SOM pool into two fractions, MIN (<53 µm), composed of particles with large surface area (silt-and claysized particles); and POM (53-200 µm), which is related to macro and microaggregates. As newly photosynthetically fixed carbon is firstly allocated into large particles (macroaggregates) and then later redistributed into the microaggregate fraction (Atere et al. 2018a), it is clear why the POM pool was the dominant fraction that 13 C entered into (Figure 2a-b), and why it decreased with time. In contrast, 13 C associated with the MIN was stable during the experimental period, which shows that C associated with this fraction was more stable with respect to microbial utilization. More 13 C was incorporated into the POM than the MIN fraction under fertilization treatments. This agrees with the results of Yan et al. (2013), who found that the accumulation of SOC and its change in response to fertilization was significantly correlated with the carbon associated with the POM fraction, while that associated with the silt or clay fractions was independent of fertilization. Luo et al. (2018) also found that N application increases the incorporation of newly fixed C into macroaggregates, but not into microaggregates, that is silt, or clay fractions (Luo et al. 2018), as the roots are usually located within large macroaggregates. Besides, roots also stimulate the formation of macroaggregates from microaggregates, silt, or clay fractions, leading to a decrease in newly photosynthesized C incorporated in small fractions. Thus, these results partly support our second hypothesis and show that N and P promote the allocation of newly fixed C into the POM fraction only, while the mineral fraction remains largely unaffected by fertilization.

Effect of fertilization on 13 C allocation into the microbial biomass pool
The proportion of photosynthesized 13 C incorporated into the MBC was 0.35-0.93% on D14 and 0.20-0.85% on D22 in rhizosphere soil, and was very low in the bulk soil except for the control treatment ( Figure 3). These data are in the range reported for pulse labeling of rice, i.e.,, 0.15-0.94% (Lu et al. 2002b). The incorporation of photosynthate-derived C into the MBC of rhizosphere soil was 2-3 times higher under N and NP than that in the control and P treatment on both sampling days ( Figure S2; P < 0.001). Such results are associated with the decrease in the soil C/N ratio under fertilization (Table S2; P < 0.005), which can stimulate microbial activity, and can also be attributed to changes in the C/ N ratio of root exudates, which make them more available to microorganisms (Wei et al. 2017). The proportion of 13 C-MBC in the total MBC pool showed that the activity of microorganisms was significantly higher under N and NP fertilization on both sampling days ( Figure S3; P < 0.005). This shows that ricephotosynthesized C inputs stimulated soil microorganisms, particularly in the rhizosphere (Lu, Watanabe, and Kimura 2004).

Conclusions
Overall, N and NP fertilization enhanced the net photosynthesized rice 13 C in the rice-soil system compared to that of the unfertilized treatment (or control). The main fraction of the photosynthesized rice 13 C (55-77%) was retained in the shoots; however, 25-44% was transferred into the roots (Figure 4). Rhizosphere soil contained 2-3 times more 13 C than bulk soil, which showed that the main accumulation of C in paddy soils occurred in the root zone. The greatest incorporation of 13 C into the MBC pool was found in response to N and NP fertilization (on both D14 and D22), which showed that microorganisms are more active under these conditions. The main proportion of 13 C in the rhizosphere and bulk parts was found in the POM on D14, whereas it decreased over time (on D22) owing to microbial utilization. In contrast, root-derived 13 C in the MIN remained unchanged between sampling dates, which showed that stable rhizodeposits in this fraction can be the potential mechanism underlying SOM sequestration in paddy soils (Figure 4). The main effects of fertilization on the stabilization of newly formed rhizodeposits were observed under NP treatments, indicating that this is a more promising fertilizer treatment for sustaining SOM levels under rice production.