Growth and nitrogen uptake of CO2-enriched rice under field conditions


Author for correspondence: K. Kobayashi Tel: +81 298 388223 Fax: +81 298 388199


  •  The effects of elevated CO2 are reported here on the uptake of nitrogen (N) and its relationships with growth and grain yield in rice (Oryza sativa).
  •  Using free-air CO2 enrichment (FACE), rice crops were grown at ambient or elevated (c. 300 µmol mol−1 above ambient) CO2 and supplied with low, medium or high levels of N.
  •  For the medium and high N treatments, FACE increased N uptake at panicle initiation but not at maturity. For total dry matter, as well as spikelet number and grain yield, positive interactions between CO2 and N uptake were observed. Furthermore, spikelet number was closely associated with N uptake at panicle initiation.
  •  These results indicate that, to maximize rice grain yield under elevated CO2, it is important to supply sufficient N over the whole season, in order to maintain the enhancement in dry matter production. In addition, N availability must be coordinated with the developmental stage of the crop, specifically to ensure that sufficient N is available at panicle initiation in order to maximize spikelet number and grain yield.


The predicted increase in global atmospheric CO2 concentration (Ca) is expected to increase dry matter (DM) production and yields in C3 agricultural crops (Kimball, 1983; Drake et al., 1997). In terms of both area and tonnage harvested, Oryza sativa. L. (rice) is the most important crop in Asia, providing a significant portion of the dietary needs for this region (Alexandratos, 1995). Studies on rice under controlled environments have found that elevated Ca usually has little effect on individual leaf area but generally increases tiller number, resulting in greater leaf area per plant (Imai, 1995). In combination with increased photosynthesis per unit leaf area (Rowland-Bamford et al., 1991), this results in greater DM production (Baker et al., 1996). The additional tillers also produce more panicles (Ziska et al., 1997), leading to grain yield increases ranging from 5 to 60% (Horie, 1993; Baker et al., 1996; Ziska et al., 1997; Moya et al., 1998). The considerable variation in the magnitude of the response of grain yield to elevated Ca can be attributed to differences in experimental conditions and indicate that factors such as air temperature (Baker et al., 1996; Kim et al., 1996b) and nutrient supply and cultivars (Ziska et al., 1996; Moya et al., 1998) can affect the response of rice to elevated Ca.

There is little information on the interactions between elevated Ca and nitrogen (N) availability in rice. Ziska et al. (1996) grew plants until anthesis inside open top chambers placed within paddies and found that the photosynthetic and DM responses of rice to elevated Ca were greater with higher levels of N supply. Using miniature paddies inside temperature gradient chambers, Kim et al. (unpublished) found increases in DM, yield and N use efficiency with elevated Ca. In both reports, N uptake increased with elevated Ca early in crop development but by the end of the experiment N uptake was similar for both CO2 treatments. When N availability is not limiting, N uptake by crops depends on factors such as root size and physiological activity (Berntson et al., 1998). Elevated Ca can change the size of root systems in rice (Ziska et al., 1997) and in many other species (Rogers et al., 1996; Cotrufo & Gorissen, 1997). This is due mainly to increases in carbon allocation to the roots. However, it is not clear how the effects of elevated Ca on roots influences N uptake and grain yield in rice.

Previous studies investigating the effects of elevated Ca and N supply on rice growth used enclosures to contain the added CO2. Enclosures have been shown to affect microclimatological factors which may influence the response of plants to elevated Ca (McLeod & Long, 1999). The free air CO2 enrichment (FACE) technique allows vegetation to grow under elevated Ca with minimal disturbance. The Japanese Rice FACE project was established with the core objective of investigating the effects of elevated Ca on rice growth, yield, quality and ecosystem processes. Rice crops were grown under two mole fractions of CO2 (ambient and ambient plus 200 µmol mol−1) with three different N applications: the standard used by local farmers (c. 9 g N m−2) and ±50% of the standard rate. We sampled the crops at five different times through the season, and determined the effects of elevated Ca and different levels of N supply on DM production and N uptake. In this paper we present data showing the relationship between crop growth and N uptake and discuss the importance of N availability in determining the response of rice growth and yield to elevated Ca.

Materials and Methods

Experiment site and description

A full description of the Rice FACE facility is provided by Okada et al. (2001). Briefly, the facility is located in northern Japan in an area representative of the agro-climatic region that grows a large proportion of Japan’s rice crop. Plants were exposed in four paddies to elevated CO2 by growing them within 12 m diameter ‘rings’ which sprayed pure CO2 towards the centre from peripheral emission tubes located c. 50 cm above the canopy; these are referred to as FACE plots. In another four paddies, plants were grown under ambient CO2 conditions with no ring structures in place (hereafter called ambient plots). There were four blocks each consisting of a FACE and ambient plot located in paddies having similar soils and agronomic histories. To minimize CO2 contamination, ambient plots were situated at least 90 m from the nearest FACE ring. The target Ca at the centre of the FACE plots was 200 µmol mol−1 above that of ambient. Actual season-long average Ca values in the various FACE subplots (see below) ranged from c. 275–365 µmol mol−1 above ambient in 1998 and 230–300 µmol mol−1 above ambient in 1999. Season-long average Ca values in the ambient plots were 389 ± 39 and 392 ± 45 µmol mol−1 for 1998 and 1999, respectively. The average temperatures during the growing seasons were 19.7 (1998) and 20.8°C (1999) while solar radiation totals were 1638 and 1877 MJ m−2, respectively.

Crop history

In both 1998 and 1999 seeds of rice cv. Akitakomachi were obtained from crops grown locally using agronomic techniques typical of the region. For both years seed was germinated under running water and just after radicle emergence, the seedlings were sown into trays containing fertilizer-impregnated rockwool mats. Trays were placed in plastic chambers fumigated with either ambient air or ambient air plus 200 µmol mol−1 CO2.

Fourteen (1998) and 23 (1999) d after emergence, seedlings were hand-transplanted in groups of three (referred to as a hill) into either FACE or ambient plots on 21 May 1998 and 20 May 1999. Hand transplanting was used to ensure an even number of seedlings per hill and regular hill spacing. Hills and rows were 17.5 and 30 cm apart, respectively (equivalent to 19.05 hills m−2). N was supplied as ammonium sulphate at three application rates: 4 g (low; LN), 8 g (medium; MN) and 12 g N m−2 (high; HN) in 1998, and 4, 9 and 15 g N m−2, respectively, in 1999. To minimize mixing of the paddy water between N treatments, areas receiving LN and HN were separated from the rest of the plot (which received MN) by a 30-cm PVC barrier pushed 10 cm into the ground. N was applied as a basal dressing (63% of the total), at mid-tillering (20%) and at panicle initiation (17%). For all N levels, 30 (1998) and 48 (1999) g P2O5 m−2 and 15 g K2O m−2 were applied in both years. Fields were flooded throughout both seasons except for a 5 d summer drainage in mid-July and from 10 d before harvest. The experimental plants within both FACE and ambient plots were surrounded by machine-transplanted border plants treated similarly to the test plants. With the exception of using hand-transplanting to establish the crops, all other agronomic techniques including cultivation, hill spacing, fertilizer application and crop health measures were similar to those used by the local farmers.

Sampling and harvesting

Plants supplied with MN were sampled (12 hills total) for growth analysis at 25, 53, 81, 109 and 131 d after transplanting (DAT), from three locations in 1998 and on 32, 62, 88, 105 and 123 DAT from two locations in 1999. Sample dates were chosen to coincide as closely as possible with the mid-tillering, panicle initiation (PI), heading, mid-ripening and grain maturity growth stages. In 1999 HN plants were sampled at 63, 89 and 124 DAT and LN plants at 61 and 124 DAT. Four and six hills were sampled for each sampling in 1998 and 1999, respectively. Hills were sampled to minimize disturbance of other hills. To maintain canopy conditions during the first two samplings, in both seasons, hills were replaced with spare hills from the borders; these were not sampled. At each sampling, a block of soil 30 cm wide, 17.5 cm long and 20 cm deep around each individual hill was removed. Hills were separated into living and dead leaf tissue, stems (including leaf sheath), panicles (when present) and roots. For half or one-third of the hills, soil and other extraneous matter was carefully cleaned from the roots under running water and root DM was determined. Green leaf area and the number of tillers and panicles (when present) were also determined. The d. wt of the plant parts was determined separately after drying at 80°C in a forced air oven for 1 wk. After milling (0.5 mm mesh), the N content of all plant parts was determined separately using a micro-Kjeldahl technique (Kjeltech Auto Sampler System 1035, Tecator AB, Höganäs, Sweden).

Yield and its components were determined at grain maturity for 18 (1998) and 20 (1999) hills from one location for both LN and HN treatments. For the MN treatment, 54 hills from three locations (1998) and 60 hills from two locations (1999) were harvested. Panicles and total fertile spikelet number were counted and grain mass determined. Grain yield was expressed as the dry mass of filled grains.

Experimental design

In both years the experiment was a completely randomized block design with two mole fractions of CO2 (ambient and FACE) replicated four times. ANOVA (split-plot) was used to determine differences between treatment means with CO2 level as the main plot and N level as the subplots. All results reported as significant had a P < 0.05.


Nitrogen uptake

Crop N uptake at PI was greater with FACE for all N treatments (Table 1). There was a strong interaction between N supply and CO2 treatment: the increase in N uptake with FACE was only 2% with LN but over 20% with HN. In contrast, at grain maturity, crop N uptake was similar for the two CO2 treatments (Table 1). N uptake increased with N supply, from 10 g N m−2 with LN to nearly 20 g N m−2 with HN.

Table 1.  The effect of low (LN; 4 g N m−2), medium (MN; 9 g N m−2) and high (HN; 12 g N m−2) applications of nitrogen (N) on crop N uptake and root and total dry matter (DM) production at panicle initiation (PI) and grain maturity of rice under ambient (AMB) or free air CO2 enrichment (FACE)
  N uptake (g m−2)Root DM (g m−2)Total DM (g m−2)
Growth stageN treatmentAMB FACEAMB FACEAMB FACE
  1. Anova results for each growth stage are shown with * and ns indicating significance at P < 0.05 and no significance, respectively (1999 data only).

PILN  7.55   7.7245.17 53.01  372.07   513.43
 MN  9.12 10.5345.10 65.67  425.94   618.47
 HN11.37 13.9646.60 60.67  452.53   643.30
AnovaCO2 *  *  * 
resultsN *  ns  * 
 CO2 × N *  ns  ns 
MaturityLN11.65 11.4944.48 58.201375.89 1537.76
 MN13.62 14.1037.94 47.591495.64 1724.11
 HN17.85 19.3041.58 47.511543.54 1839.34
AnovaCO2 ns  *  * 
resultsN *  ns  * 
 CO2 × N ns  ns  ns 

Growth, DM and N use efficiency

At both the mid-tillering and PI stages for the MN treatment, root DM was greater for FACE crops (Fig. 1; Table 1). There was a positive relationship between root DM and crop N uptake across both CO2 treatments and sampling times (Fig. 1). At PI and for all N treatments (Table 1) and levels of N uptake (Fig. 2b), FACE increased total DM production by c. 40%. Crop green leaf area index (GLAI) also increased with increasing N uptake (Fig. 2a), though not to the same extent as total DM (Fig. 2b). At maturity, both root and total DM (Table 1) and panicle DM (Fig. 3b) were greater with FACE. The increase in panicle DM was largely due to an increase in productive tiller number (data not shown). As a result of the greater increase in both DM and panicle production, relative to the increase in N uptake, N use efficiency in FACE crops was c. 10% greater across all levels of N uptake (Fig. 3a,b).

Figure 1.

The relationship between root dry matter (DM) and plant nitrogen (N) uptake at the mid-tillering and panicle initiation stages of rice grown under ambient CO2 (open sqares) and free-air CO2 enrichment (closed squares) conditions in 1998 and 1999. Nitrogen application rates were 8 and 9 g N m−2 in 1998 and 1999 respectively. Data points are the means of each sampling/year; bars are the standard errors of the means. First degree polynomial fitted; R2 is shown.

Figure 2.

The relationships between (a) green leaf area index (GLAI) or (b) total crop dry matter and crop nitrogen (N) content of rice at panicle initiation grown under ambient CO2 (open sqaures) and free-air CO2 enrichment (closed squares) conditions. Data points are from the separate N treatments within the FACE or ambient plots. Second degree polynomials fitted for each CO2 treatment; R2s are shown.

Figure 3.

The relationships between (a) total crop or (b) panicle dry matter (DM) and crop nitrogen (N) content of rice at grain maturity grown under ambient CO2 (open sqaures) and free-air CO2 enrichment (closed sqaures) conditions. Data points are from the separate N treatments within the FACE or ambient plots. Second degree polynomials fitted for each CO2 treatment; R2s are shown.

Grain yield

FACE increased grain yield (Table 2). For MN the increase in yield (14%) with FACE was similar to the increase in DM (15%, Table 1). However, the increases for LN (4%: not significant) and HN (11%) were both less than the increases in DM (12 and 19%, respectively). Across all N treatments harvest index (HI) decreased slightly (3%) with FACE (Table 2). An increase in panicle number, leading to a greater number of fertile spikelets per unit land area, was the most important contributor to the yield increase. The other components of yield, though affected to varying degrees by N supply, were not influenced by FACE. There was a very close relationship between total crop N uptake at PI and the total number of spikelets at grain maturity (Fig. 4). However, the relationship was similar for FACE and ambient crops: for both CO2 treatments an extra 2600 spikelets m−2 were produced for every additional 1 g N m−2 in crop N uptake at PI.

Table 2.  The effect of low (LN; 4 g N m−2), medium (MN; 9 g N m−2) and high (HN; 12 g N m−2) levels of nitrogen (N) application on yield and its components of rice crops grown under ambient (AMB) or free air CO2 enrichment (FACE)
N treatmentCO2Panicle number (m−2)Spikelets panicle−1Spikelet fertility (%)Fertile spikelets (m−2) Individual grain mass (mg)Grain yielda (g m−2)Harvest indexb
  1. Anova results are shown with * and ns indicating significance at P < 0.05 and no significance, respectively (1999 data only). agrain yield is expressed on a dry mass basis; due to rounding values do not calculate exactly. bbecause total DM of the yield determination plants was not measured, harvest index was calculated based on the panicle (Fig. 3) and total dry matter (DM) (Table 1) data from the final destructive sampling.

LNAMB393.4479.7295.429868.2825.93  773.770.52
 FACE405.6178.2396.330595.0926.42  806.660.50
MNAMB454.7484.0094.235905.7624.15  866.720.53
 FACE494.4382.2996.939040.5925.21  983.350.52
HNAMB487.2091.2891.140455.2923.64  944.420.52
 CO2 × Nns*ns*ns*ns
Figure 4.

The relationships between crop nitrogen (N) content at panicle initiation and total spikelet number at maturity of rice crops grown under ambient CO2 (open squares) and free-air CO2 enrichment (closed squares) conditions. Data points are from the separate N treatments and FACE rings or ambient plots. First degree polynomial fitted for both CO2 treatments combined; R2 is shown.


It has been well established that for most C3 crops, the predicted increases in elevated Ca will lead to greater DM production and yields (Kimball, 1983). However, for some crop species the ability to respond and maintain the growth enhancement associated with elevated Ca depends on an adequate supply of nutrients. In species other than rice, elevated Ca has been shown to affect root production and morphology (Berntson et al., 1998), leading to positive, negative or no effects on plant nutrient uptake. For rice grown under field conditions the effects of elevated Ca on the uptake of N and its relationships with growth and grain yield are poorly understood.

In our experiment, for any application rate of N, the total amount of N available over the season was likely to have been similar for both CO2 treatments (assuming similar N dynamics). Because over 80% of the total fertilizer N was applied before PI, N availability was greater early in the season compared with later in the season. During the period from transplanting to PI, crop N uptake was greater with FACE (Table 1, Fig. 2). FACE crops had greater root mass at PI (Table 1); as found in other crops (Rogers et al., 1996). Root DM is an important component of a plant’s ability to take up nutrients (along with other parameters such as root physiological activity, morphology and architecture) and it is probable that N uptake potential was greater in FACE crops. Indeed, we found that at the mid-tillering and PI stages, not only was root DM greater in FACE crops, but across both CO2 treatments there was a positive relationship between root DM and crop N uptake (Fig. 1).

By grain maturity, N uptake was similar for both FACE and ambient crops (Table 1, Fig. 3). Hence, though the available N pool over the season was similar for both CO2 treatments, the FACE crops took up a greater amount of N before PI. A preliminary 15N study supports this hypothesis: before PI, the rate of labelled fertilizer N uptake and its recovery was greater in FACE crops compared with ambient crops (Miura et al., unpublished). By contrast, after anthesis, N uptake rates were similar. Indirect evidence for a larger N pool after PI in ambient crops was the faster rate of crop senescence in FACE crops: dead leaf DM was substantially greater under FACE at each sampling after PI (Kobayashi et al., 1999), indicating that N was likely to have been in shorter supply.

At both PI and grain maturity, the increase in crop DM with FACE was greater than the increase in N uptake (Table 1), leading to greater N use efficiency. This is consistent with previous studies with rice (H. Y. Kim et al., unpublished) and other species (Drake et al., 1997) under controlled environment conditions. FACE crops also appeared to be more responsive to N in terms of grain produced: 22% more grain was produced when N supply increased from 4 to 9 g N m−2 under FACE compared with only 12% under ambient conditions (calculated from Table 2).

Crop DM production is usually positively related to canopy photosynthesis. Since the amount of radiation intercepted by a rice canopy depends largely on the amount of leaf area, the increase in GLAI at PI with FACE in our experiment (Fig. 2a) suggests that radiation interception was also greater up to this stage of growth. For rice in another study (Weerakoon et al., 2000) it was suggested that increased plant DM with elevated CO2 was a result of greater radiation use efficiency (RUE) rather than increases in the amount of intercepted radiation. In our experiment, the magnitude of the increase in DM production (Fig. 2b) with FACE was greater than the increase in GLAI, suggesting that FACE not only increased radiation interception but also RUE.

The magnitude of the increase in panicle DM with FACE was greater with increasing N uptake (Fig. 3b). This increased panicle DM was associated with a greater panicle number (Table 2) as a result of a larger number of productive tillers. Tiller production in rice, which determines panicle number and yield, is strongly dependent on N supply (Yoshida, 1981). Numerous experiments using various controlled environment systems have also demonstrated that elevated CO2 increases tiller and panicle number in rice (Imai et al., 1985; Baker et al., 1996; Kim et al., 1996a; Ziska et al., 1996). In addition to panicle number, spikelets per unit ground area is a critical determinant of grain yield in rice (Yoshida, 1981). In our experiment, for all levels of N supply, the increased spikelet number with FACE resulted mainly from an increase in the amount of crop N uptake at PI (Table 1, Fig. 4). However, as there was no difference between the two CO2 treatments in the number of spikelets produced per unit N taken up, it indicates that the increase in grain yield due to FACE depends primarily on the enhancement of reproductive sink capacity caused by the increased N uptake at PI.

Despite the importance of sink capacity in determining yield, for HN the 15% increase in spikelet number with FACE resulted in only an 11% increase in grain yield. This difference was largely due to a decrease in individual grain mass with FACE (Table 2). There are two possible reasons for this. First, the incidence of panicle-neck blast disease was higher in HN under FACE (data not shown); this disease has been shown to lead to a decrease in the extent of grain filling (Katsube & Koshimizu, 1970). Second, when plants are supplied high levels of N under elevated CO2 the increase in overall spikelet number is frequently due to a greater number of spikelets on secondary rachis-branches. These grains are usually smaller than those on primary branches (Kim et al., 1998).

Overall, the data presented here has important implications with regard to the growth and yield of rice under the Ca conditions predicted for the 21st century. Rice crops took up more N under elevated Ca than ambient crops early in the season (that is before PI). Dry matter production was also higher throughout crop development, resulting in an increase in N use efficiency. To maintain this enhanced DM production, it is likely that rice crops growing under elevated Ca will need a greater total amount of N over the whole season. Therefore, under future elevated Ca conditions, recommended rates of N for rice may need to be higher than current levels, at least for the cultivar and agro-climatic conditions of this experiment. However, higher applications of N frequently lead to greater yield losses in rice due to lodging (Yoshida, 1981). We know little about the effects of both higher CO2 and N on lodging. In our experiment we were not able to determine lodging damage because for both the ambient and FACE treatments, crops supplied at HN were protected against lodging by providing support to the stems. However, Terashima et al. (1995) reported a close relationship between increased root mass and increased resistance to lodging in direct-seeded rice and therefore it is possible that rice crops grown under elevated CO2, with their greater root mass, may be less prone to lodging.

In addition to an increase in recommended fertilizer N rates, the timing and proportions across the season of the N applications may need to be changed. Although the number of spikelets produced per unit N taken up was similar for both CO2 treatments (Fig. 4), because N taken up at PI is an important determinant of final spikelet number (Kobayashi & Horie, 1994; Horie et al., 1997), the greater N uptake in FACE crops resulted in increased yield potential. Therefore, in order to take advantage of this, the proportion of N applied at transplanting and before PI may also need to be higher than current recommendations. In addition, the amount of N supplied after PI should be sufficient to maintain the extra DM produced under the elevated Ca conditions, enabling the additional potential yield to be fulfilled.


The authors thank two anonymous reviewers for their invaluable suggestions for improving the manuscript. This work was supported by the CREST (Core Research for Evolutional Science and Technology) program of the Japan Science and Technology Corporation (JST). We also acknowledge the technical assistance of the Field Management Division of Tohoku National Agricultural Experiment Station, Morioka and the National Institute of Agro-Environmental Sciences, Tsukuba and the statistical advice of Dr Marcia Gumpertz of North Carolina State University.