Genotypic differences in grain yield, and nitrogen absorption and utilization in recombinant inbred lines of rice under hydroponic culture

Abstract

To examine the possibility of breeding higher-yielding cultivars with a high nitrogen use efficiency (NUE) and to provide simple criteria for the selection and breeding of high-yielding cultivars with a high NUE and useful information for the mapping of quantitative trait loci (QTL) controlling NUE, recombinant inbred lines (RILs) of rice were subjected to hydroponic culture in 2000 and 2001. Grain yield and yield components, total nitrogen absorption (NTA), pre-heading and post-heading nitrogen absorption (pre-NA and post-NA, respectively), and nitrogen use efficiency for grain yield (NUEg) were analyzed. Transgressive segregations for grain yield and yield components, NTA and NUEg were observed. RILs with a high grain yield tended to show a larger number of panicles m⁻² and of spikelets per panicle, and a higher filled grain percentage than RILs with a low grain yield. Increasing NTA and NUEg resulted in increases in the number of spikelets m⁻² and grain yield, while increasing NTA resulted in a decrease in NUEg. The contribution ratio of the number of panicles m⁻² to grain yield was higher than that of the other yield components in both 2000 (31.8%) and 2001 (32.2%). In both years, the relative contribution ratios of NUEg and NTA to grain yield were approximately 56% and 43%, respectively. At the same high-yielding level, there was a significant difference in the NTA and NUEg values. These results suggest the possibility of breeding high-yielding cultivars with higher NUEg.

Key words:

• grain yield
• nitrogen use efficiency
• nitrogen absorption
• recombinant inbred line
• yield components

INTRODUCTION

Nitrogen is one of the most important nutrients for growth regulation and grain production of rice in paddy fields (Mastsushima 1976; Murayama 1967, 1979; Tanaka et al. 1964; Wada 1969). It is generally recognized that all high-yielding rice varieties respond well to applied N, requiring large amounts of N fertilizer (Tanaka et al. 1964; Yoshida 1981). However, heavy application of chemical fertilizers not only increases the cost of rice production, but also causes agro-environmental pollution, including eutrophication of rivers and lakes and nitrate pollution of surface and underground water, and contributes to the greenhouse effect through the release of NOx gases from paddy soils (Duxbury et al. 1993; Galbally et al. 1987; Murayama 1979). Therefore, many studies have been conducted to improve N use efficiency through agronomic management, for example, by controlling the timing, rate, placement and source of fertilizers (Cassman et al. 1994; Datta and Buresh 1989; Fillery and Vlek 1986; Schnier et al. 1990). However, the acceptance and adoption of N management strategies, which have often been associated with high labor requirements, had been discouraging (Singh et al. 1998). Recently, field experiments have revealed the existence of genetic variability for N use efficiency in rice (Broadbent et al. 1987; Inthapanya et al. 2000; Singh et al. 1998; Tirol-padre et al. 1996), and the possibility of improving N use efficiency in rice through genotype selection (Koutroubas and Ntanos 2003).

The genetic basis of nitrogen use efficiency has not been fully elucidated because of its complexity and quantitative nature. For quantitative trait loci (QTL) analysis of such a trait, DNA markers and genome mapping techniques have become powerful tools (Tanksley 1993). To date, QTL technology has been suitable for studies on important agricultural traits in rice, such as a yield and yield-related traits, plant height and heading (Cao et al. 2001). Another important aspect is the detailed evaluation of the traits of each genotype. To minimize the effect of heterozygosity, the use of doubled haploid lines or recombinant inbred lines (RILs) is preferable because the traits can be measured in replicated experiments and the effect of environmental factors can also be minimized.

We analyzed the characteristics of nitrogen absorption and nitrogen use efficiency in relation to grain yield among RILs derived from a cross between Zhenshan 97 and Minghui 63, parents of Shanyou 63, a famous F₁ hybrid rice from China. In the current study, the RILs were cultured using the same nutrient solution and under the same environmental conditions because it is difficult to assess genotypic differences in a paddy field. This information should also be useful for the selection and breeding of high yielding rice cultivars with high N use efficiency and for the mapping of QTL controlling NUEg.

MATERIALS AND METHODS

Rice materials

A total of 118 (F₁₂) and 191 (F₁₃) RILs were subjected to hydroponic culture in 2000 and 2001, respectively. They were developed using the single-seed descent method from a cross between the cultivar Zhenshan 97 (ZX 97, an inbred indica and the female parent of Shanyou 63 [SY 63], the traditional hybrid rice grown in China) and the cultivar Minghui 63 (MH 63, an inbred indica and the male parent of Shanyou 63) to the F₁₂/F₁₃ generations. MH 63 is one of the most important restorer lines in hybrid rice production in China, expressing abundant agronomically desirable traits such as superior grain quality and resistance to various types of bacterial blight caused by Xanthomonas oryzae pv. oryzae and blast caused by the heterothallic ascomycete fungus Magnaporthe grisea (Peng and Zhang 1999). ZX 97 is an elite maintainer line of hybrid rice extensively used in rice production in China (Liu et al. 2003).

Experimental design and growth conditions

Hydroponic culture was carried out outside at Yangzhou University, China, in the rice-growing seasons of 2000 and 2001. Eight concrete ponds (880 cm in length × 135 cm in width × 50 cm in depth) were connected to each other using iron tubes (10 cm in diameter) running along the bottom. The ponds were covered with concrete planks (135 cm in length, 16.7 cm in width), each containing 14 holes (4 cm in diameter) for seedling fixation. One 2-leaf-old seedling per hole (62 hills m⁻²), grown for 20 days in a soil seedbed, was fixed into the holes with a sponge on 2 June in each year. Twenty-eight seedlings were prepared for each RIL material. The nutrient solution was formulated according to the method described by Yoshida et al. (1976). Ten mg L⁻¹ nitrogen (Urea-N) with 250 mg L⁻¹ Mg²⁺, 34 mg L⁻¹ K⁺, 50 mg L⁻¹ Fe³⁺, 2.86 mg L⁻¹ BO³⁻ ₄, 0.22 mg L⁻¹ Zn²⁺, 2.21 mg L⁻¹ Mn²⁺, 0.08 mg L⁻¹ Cu²⁺, 0.02 mg L⁻¹ MoO²⁻ ₄ and 41.86 mg L⁻¹ Ca²⁺ were added to the ponds at the time of transplanting and replaced at 10-day intervals. During the rice-growing season, the weather was unstable across the 2-year period. Mean temperature was higher from 1–10 July and 21–31 July in 2001 compared with 2000, and vice versa from 11–20 July and 1–20 August. Sunshine hours were longer from 21–30 June, 1–10 September and 21–30 September in 2001 compared with 2000, and vice versa from 11–20 June, 11–20 July, 11–20 September and 1–10 August.

Sampling and measurements

Two plants from each RIL were sampled at heading and maturity, respectively, separated into leaf blades, leaf sheaths plus stems, roots and panicles, and milled to determine N content using the Kjeldahl distillation method. At maturity, 10 plants were sampled to determine the number of panicles m⁻², the number of spikelets per panicle, filled grain percentage and 1000-grain weight. Filled grain was selected using tap water with a gravity of 1.00 g cm⁻³ and the grain (unhulled) yield was adjusted to a moisture content (MC) of 13.5%. A further five plants were sampled to determine dry weight after drying at 85°C to a constant weight.

Traits

Calculations for the traits included dry weight and nitrogen content determinations and the traits were defined as follows: (1) total nitrogen absorption (NTA g m⁻²) = plant nitrogen content at maturity, (2) pre-heading nitrogen absorption (pre-NA g m⁻²) = total plant nitrogen absorption from transplanting to heading, (3) post-heading nitrogen absorption (post-NA g m⁻²) = NTA minus pre-NA, (4) grain yield (g m⁻²) = unhulled grain weight, (5) nitrogen use efficiency for grain yield (NUEg gGW gN⁻¹) = grain yield divided by NTA.

Statistical analyses

Hierarchical cluster analysis procedure was used in the current study to identify relatively homogeneous groups of cases (or variables) based on selected characteristics using an algorithm that started with each case (or variable) in a separate cluster and combined clusters until only one was left (SPSS, Chicago, IL, USA, 1998).

All data were subjected to one-way analysis anova to determine the differences among the clusters.

The calculation of the contribution ratio of each component was as follows: data analysis involved linearising the multiplicative relationships using logs and then determining the contribution of each component trait to the sum of squares of the resultant trait. The sum of the cross products of each component trait by the resultant trait (Σx_iy_i) divided by the resultant trait (Σy² _i) gave the relative contribution of each component variable to the resultant variable (Moll et al. 1982).

RESULTS

Growth duration

Heading of the RILs occurred from 26 July to 2 September (56–92 days after transplanting [DAT]) in 2000 and from 29 July to 6 September (59–96 DAT) in 2001. Maturity occurred from 25 August to 25 September (80–115 DAT) in 2000 and from 24 August to 27 September (83–117 DAT) in 2001.

Distribution of the mean values of grain yield and yield components, NTA and NUEg

The distribution of the mean values of grain yield and yield components, NTA and NUEg in the RILs is shown in Table 1. Significant genetic variations in these measured traits were detected among the parents in 2000 and among RILs in 2000 and 2001, respectively, except for the 1000-grain weight in the RILs. Grain yield and yield components, NTA and NUEg values of MH 63 were significantly higher than those of ZX 97. The values of the grain yield, number of spikelets per panicle and NUEg of the F₁ hybrid (SY 63) were significantly higher than those of both parents. Transgressive segregation of yield and yield components, NTA and NUEg in the RILs in 2000 was also observed, and the mean values of these traits were closer to those of MH 63 than ZX 97. Moreover, a number of the RILs in 2000 showed higher values than SY 63.

Grain yield and yield components

Grain yield and yield components are listed in Table 2. According to the hierarchical cluster analysis procedure, the grain yield of the RILs was classified into 6 clusters (A–F) from high to low, respectively. Significant differences (P = 0.05) in grain yield were observed between adjacent clusters from A to F. The coefficient of variation (CV) of the lowest grain yield was comparatively higher than that of the medium and higher grain yields. The range of each yield component in the RILs was approximately the same between years, whereas the mean number of spikelets per panicle and filled grain percentage were higher in 2000 than in 2001. The mean values of all the yield components tended to increase with increasing grain yield in both years.

In both years, grain yield was significantly correlated with the number of panicles m⁻², the number of spikelets per panicle, and the filled grain percentage. In both years, grain yield was also closely correlated with the number of spikelets m⁻², which is a product of the number of panicles m⁻² and the number of spikelets per panicle, the latter mainly determining the number of spikelets m⁻² (Table 3).

The results of multiple regressions between the resultant trait (grain yield Y₁) and the component traits (number of panicles m⁻², X₁; number of spikelets per panicle, X₂; filled grain percentage, X₃) are shown in Table 4. Eighty percent of the variation in grain yield in both years was explained by these three traits, the contribution ratio of each trait was 31.8%, 26.0% and 24.4% in 2000 and 32.2%, 25.5% and 24.1% in 2001, respectively.

Relationship between grain yield and growth duration

The duration of growth from transplanting to heading showed a significant positive correlation with grain yield in both years, although the correlation coefficients were comparatively low (r ₀₀  = 0.271, P = 0.01; r ₀₁  = 0.222, P = 0.05, respectively). In contrast, the correlation between the growth duration from transplanting to maturity and grain yield was not significant in either year (r ₀₀  = 0.010, r ₀₁  = −0.032, respectively).

Nitrogen absorption and NUEg

The NTA and NUEg values among the RILs for each cluster of grain yield are listed in Table 2. NTA was significantly different among all the RILs (P = 0.01) in both years. The mean value of NTA was 16.9 g m⁻² with a CV of 24.0% and 15.3 g m⁻² with a CV of 22.1% in 2000 and 2001, respectively. A wide variation range in NTA values was observed among RILs in the same cluster, except for cluster A, in 2000. In general, the higher the mean grain yield, the higher the mean value of NTA.

NTA could be divided into two components, pre-NA and post-NA. Pre-NA ranged from 7.5 to 24.6 g m⁻² (CV,

Table 1 Grain yield and yield components, NTA and NUEg of Shanyou 63 (SY 63) and parents (ZX 97 and MH 63) in 2000 and of the RILs in 2000 and 2001

Traits	2000			RILs
	Female parent (ZX 97) Mean ± SD	F1 hybrid (SY 63) Mean ± SD	Male parent (MH 63) Mean ± SD	2000 (F12)				2001 (F13)
				Mean ± SD	Sig.	Max.	Min.	Mean ± SD	Sig.	Max.	Min.
Grain yield (g m^−2)	128.0 a ± 0.4	909.0 c ± 40.3	754.0 b ± 80.1	634.0 ± 167.9	0.000	1080.0	207.0	533.0 ± 143.7	0.000	957.0	178.0
No. panicles m^−2	168.0 a ± 8.8	317.3 b ± 8.8	369.1 b ± 34.2	285.2 ± 46.4	0.000	496.0	198.4	297.6 ± 52.3	0.000	477.4	179.8
No. spikelets panicle^−1	59.9 a ± 4.0	108.8 c ± 3.1	74.9 b ± 3.5	104.1 ± 20.2	0.009	157.5	62.3	91.5 ± 19.8	0.000	152.8	46.8
Filled grain percentage (%)	51.8 a ± 0.5	86.4 b ± 3.4	86.1 b ± 4.3	77.4 ± 11.4	0.000	95.5	30.4	70.5 ± 10.9	0.000	91.3	32.7
1000-grain weight (g)	21.1 a ± 0.0	26.3 b ± 0.2	27.4 b ± 1.0	23.9 ± 2.7	0.055	30.3	17.4	24.7 ± 2.9	0.193	31.8	17.3
NTA (g m^−2)	4.3 a ± 0.1	18.2 c ± 0.9	17.6 bc ± 4.3	16.9 ± 4.06	0.000	28.1	9.0	15.3 ± 3.38	0.000	24.2	6.9
NUEg (gGW gN^−1)	29.6 a ± 0.7	49.8 c ± 4.7	42.8 b ± 5.3	37.9 ± 7.2	0.000	55.8	14.1	35.6 ± 9.4	0.000	65.9	13.8
NTA, total nitrogen absorption; NUEg, nitrogen use efficiency for grain yield; RILs, recombinant inbred lines; Sig, significant difference value. Values followed by the same letter in a column are not significantly different (P = 0.05).

Table 2 Grain yield and yield components, NTA and NUEg in each cluster of the RILs grown in 2000 and 2001

Cluster^†	N	Yield^†		No. panicles (m^−2)		No. spikelets panicle^−1		1000-grain weight^‡ (g)		Filled grain percentage (%)		NTA (g m^−2)		NUEg (gGW gN^−1)
		Mean	CV (%)	Mean	CV (%)	Mean	CV (%)	Mean	CV (%)	Mean	CV (%)	Mean	CV (%)	Mean	CV (%)
2000
A	7	1028 a	3.4	353.4 a	8.0	116.5 a	8.2	26.1 a	11.7	84.1 a	6.4	25.1 a	6.8	41.1 a	6.4
B	18	835 b	7.3	328.6 a	19.0	116.6 a	21.2	23.5 bc	11.8	83.6 a	7.9	21.0 b	15.5	40.6 a	14.9
C	40	654 c	5.3	285.2 b	12.9	103.7 ab	16.6	24.0 bc	10.3	82.0 a	9.5	16.4 c	13.6	40.3 a	12.1
D	38	543 d	7.1	272.8 bc	10.4	98.6 bc	17.8	24.3 ab	10.2	73.5 b	13.7	15.3 d	20.9	36.8 b	20.2
E	13	414 e	8.8	248.0 c	12.3	100.2 ab	24.3	22.5 c	12.6	67.9 b	17.5	13.7 e	19.4	31.5 c	23.2
F	2	224 f	10.6	279.0 bc	1.6	85.1 c	29.3	21.9 c	15.7	39.7 c	33.2	13.3 e	42.6	18.6 d	33.8
Total	118	634	26.5	285.2	16.1	104.1	19.4	23.9	11.2	77.4	14.8	16.9	24.0	37.9¡¡	19.0
2001
A	13	825 a	6.8	303.8 ab	20.4	112.3 a	16.0	26.2 a	11.2	82.2 a	8.9	17.4 ab	14.7	48.0 a	12.3
B	22	696 b	3.9	322.4 a	12.8	102.7 a	17.5	24.3 ab	12.8	77.2 ab	10.4	17.6 a	16.0	40.4 b	15.7
C	63	584 c	5.8	310.0 a	16.5	92.4 b	19.8	25.1 ab	11.3	73.2 b	10.1	16.2 b	18.5	37.4 c	21.1
D	59	479 d	6.9	285.2 b	16.5	90.5 b	20.1	24.5 ab	11.9	69.3 c	13.9	14.3 cd	23.5	35.2 c	23.7
E	15	375 e	4.7	272.8 bc	16.3	78.5 c	18.3	24.2 ab	10.0	64.7 c	12.2	13.4 de	19.4	29.1 d	20.4
F	19	272 f	17.8	254.2 c	18.0	74.9 c	22.0	24.0 b	10.5	54.4 d	20.1	12.7 e	21.9	22.3 e	24.1
Total	191	533	26.9	297.6	17.8	91.5	21.6	24.7	11.6	70.5	15.4	15.3	22.1	35.6	26.5
CV, coefficient of variance; N, number of lines; NTA, total nitrogen absorption; NUEg, nitrogen use efficiency for grain yield. Values followed by the same letter in a column are not significantly different (P = 0.05). ^†Clusters A to F represent high to low grain yield, respectively.
^‡Moisture content is 13.5%.

17.6%) with a mean value of 13.2 g m⁻² and from 6.4 to 19.4 g m⁻² (CV, 19.6%) with a mean value of 11.9 g m⁻² in 2000 and 2001, respectively. Post-NA ranged from 0.1 to 14.5 g m⁻² (CV, 81.0%) with a mean value of 3.8 and from 0.1 to 12.3 g m⁻² (CV, 78.3%) with a mean value of 3.4 g m⁻² in 2000 and 2001, respectively. The mean values of pre-NA were higher than those of post-NA, but the variation range in post-NA among the RILs was wider than that of pre-NA. The results obtained from multiple regression analysis indicated that the contribution ratios of pre-NA and post-NA to NTA were 76.7% and 23.3%, respectively, in 2000 and 78.7% and 21.3%, respectively, in 2001 (Table 4).

There was considerable variation in the NUEg values among the clusters, with a mean value of 37.9 gGW gN⁻¹ in 2000 and 35.6 gGW gN⁻¹ in 2001. The CV was 19.0% in 2000 and 26.5% in 2001 (Table 2). The CVs of NUEg in each cluster tended to be higher with decreasing grain yield in both years. The higher the mean grain yield, the higher the mean NUEg value.

Relationships among grain yield and yield components, nitrogen absorption at different rice growth stages and NUEg

Relationships among grain yield and yield components, nitrogen absorption at different growth stages and NUEg are presented in Table 3. The NTA values were strongly correlated with the number of panicles m⁻², number of spikelets m⁻², pre-NA and post-NA and grain yield in both years. NUEg showed a significant positive correlation with the number of spikelets per panicle, number of spikelets m⁻², filled grain percentage and grain yield as well as a significant negative correlation with NTA in both years.

Results obtained from multiple regression analysis indicated that 97.8% and 94.3% of the variation in grain yield in 2000 and 2001 was explained by NUEg and NTA, respectively. The contribution ratios of these traits were 56.3% and 43.7%, respectively, in 2000, and 56.6% and 43.4%, respectively, in 2001 (Table 4).

Fig. 1 shows the relationships between the amount of NTA, NUEg in 2000 and 2001, respectively. NUEg showed a significant negative correlation with NTA (r ₀₀  = −0.278, P = 0.01; r ₀₁  = −0.361, P = 0.01, respectively). This correlation was strong at the same yielding level. In the current study, for example, along the high iso-grain yield line at 800 g m⁻² among the RILs belonging to cluster B in 2000 and cluster A in 2001, the highest nitrogen use efficiency line showed a minimum NTA (14.9 g m⁻²) with a maximum NUEg (51.3 gGW gN⁻¹) in 2000; these values were 13.2 g m⁻² and 61.6 gGW gN⁻¹, respectively, in 2001. In contrast, the lowest nitrogen use efficiency line showed a NUEg of 32.8 gGW gN⁻¹ with an NTA of 28.1 g m⁻²; these values were 37.0 gGW gN⁻¹ and 23.4 g m⁻², respectively, in 2001.

Table 3 Correlation coefficients between grain yield and yield components or NTA and NUEg of RILs grown in 2000 and 2001

	No. panicles (m^−2)	No. spikelets panicle^−1	No. spikelets (m^−2)	Filled grain percentage (%)	1000-grain weight (g)	Pre-NTA (g m^−2)	Post-NTA (g m^−2)	NTA (g m^−2)	NUEg (gGW N^−1)	Grain yield (g m^−2)
2000
No. panicles m^−2	1.000
No. spikelets panicle^−1	−0.216*	1.000
No. spikelets m^−2	0.492**	0.728**	1.000
Filled grain percentage (%)	0.177 ns	−0.153 ns	−0.005 ns	1.000
1000-grain weight (g)	0.075 ns	−0.384**	−0.282**	−0.001 ns	1.000
Pre-NTA (g m^−2)	0.417**	0.075 ns	0.373**	0.081 ns	0.300**	1.000
Post-NTA (g m^−2)	0.201*	0.169ns	0.272*	0.227*	0.075 ns	−0.185*	1.000
NTA (g m^−2)	0.490**	0.187 ns	0.507**	0.235*	0.301*	0.678**	0.597**	1.000
NUEg (gGW gN^−1)	0.148 ns	0.216*	0.310*	0.472**	−0.106 ns	0.108 ns	−0.165 ns	−0.278*	1.000
Grain Yield (g m^−2)	0.568**	0.331**	0.704**	0.548**	0.193*	0.515**	0.401**	0.722**	0.447**	1.000
2001
No. panicles m^−2	1.000
No. spikelets panicle^−1	−0.359**	1.000
No. spikelets m^−2	0.411**	0.691**	1.000
Filled grain percentage (%)	0.137 ns	0.051 ns	0.145 ns	1.000
1000-grain weight (g)	−0.005 ns	−0.241**	−0.285**	−0.096 ns	1.000
Pre-NTA (g m^−2)	0.219**	0.107 ns	0.260**	0.044 ns	0.071 ns	1.000
Post-NTA (g m^−2)	0.313**	0.126 ns	0.388**	0.035 ns	−0.080 ns	−0.211**	1.000
NTA (g m^−2)	0.423**	0.186 ns	0.516**	0.063 ns	−0.007 ns	0.630**	0.626**	1.000
NUEg (gGW gN^−1)	0.011 ns	0.327**	0.315**	0.597**	0.169*	−0.242**	−0.211**	−0.361*	1.000
Grain yield (g m^−2)	0.364**	0.468**	0.735**	0.625**	0.148 ns	0.265**	0.291**	0.443**	0.646**	1.000
NTA, total nitrogen absorption; NUEg, nitrogen use efficiency for grain yield; Pre-NA, pre-heading nitrogen absorption; Post-NA, post-heading nitrogen absorption. Significant difference at *P = 0.05 and
**P = 0.01; ns, not significant at P = 0.05.

Table 4 Results of multiple regression analysis and contribution ratios

Resultant trait	Component trait	R^2		Contribution ratio (%)
		2000	2001	2000	2001
Y1:	X1: No. panicles per hill			31.8	32.2
Grain yield (g m^−2)	X2: No. spikelets per panicle	0.803	0.808	26.0	25.5
	X3: Filled grain percentage			24.4	24.1
Y2:	X4: Pre-NA	1.000	1.000	76.7	78.7
NTA (g m^−2)	X5: Post-NA			23.3	21.3
Y3:	X6: NUEg	0.978	0.943	56.3	56.6
Grain yield (g m^−2)	X7: NTA			43.7	43.4
NTA, total nitrogen absorption; NUEg, nitrogen use efficiency for grain yield; Pre-NA, pre-heading nitrogen absorption; Post-NA, post-heading nitrogen absorption.

Figure 1  Relationship between total nitrogen absorption (NTA) and nitrogen use efficiency for grain yield (NUEg) among recombinant inbred lines (RILs) in (A) 2000 and (B) 2001. The curve lines in the figure indicate the iso-grain yield. Grain yield = NTA × NUEg. Numerals on the right side in both (A) and (Bb) indicate the grain yield (g m−2), **P = 0.01. Clusters A–F in the figure correspond to those in Table 2.

Relationships among growth duration, nitrogen absorption at different rice growth stages and NUEg

Growth duration from transplanting to heading showed a significant positive correlation with pre-NA (r ₀₀  = 0.635, P = 0.01; r ₀₁  = 0.363, P = 0.01, respectively) and NTA (r ₀₀  = 0.447, P = 0.01, r ₀₁  = 0.409, P = 0.01, respectively) in both years. Similar results were recorded between the growth duration from transplanting to maturity and pre-NA (r ₀₀  = 0.606, P = 0.01; r ₀₁  = 0.419, P = 0.01, respectively) and NTA (r ₀₀  = 0.516, P = 0.01; r ₀₁  = 0.402, P = 0.01, respectively). Growth duration from transplanting to heading showed a significant negative correlation with NUEg in both years (r ₀₀  = −0.257, P = 0.01, r ₀₁  = −0.404, P = 0.01, respectively), and similar results were obtained between the growth duration from transplanting to maturity and NUEg (r ₀₀  = −0.348, P = 0.01; r ₀₁  = −0.375, P = 0.01, respectively).

DISCUSSION

Experimental conditions and materials used

As it is not possible to avoid interference from soil fertility when using soil culture, the current experiment was carried out under hydroponic culture. Ponds were flooded with a nutrient solution continuously from transplanting to maturity to guarantee that all rice plants were grown under the same nutrition conditions, ensuring that any differences in nitrogen absorption and utilization for grain yield were caused by genotype differences only. In the current study, all the rice lines were cultured at a density of 62 plants (hills) m⁻² because of limited space; this was higher than the Chinese conventional density of approximately 44 plants (hill) m⁻². Grain yield and yield components and related functional components were determined at this density.

F₁ (SY 63) plants crossed between ZX 97 and MH 63 showed larger values for grain yield, NTA and NUEg because of hybrid vigor. The results were in agreement with those from previous reports (Ichii and Nakamura 1990; Yao et al. 2000b). To produce RILs, the F₁ (SY 63) plants were selfed, and then the generations were advanced using the single-seed descent method to the F₁₂/F₁₃ generations. With advancement of the generations up to F₁₂ or F₁₃, the genotypes of these rice lines became comparatively stable. Therefore, the abundant agronomically desirable traits of these RIL lines enabled the selection of superior genotypes according to the needs of rice production.

Grain yield and yield components of the RILs

The mean grain yield was lower in 2001 than in 2000, possibly because of differences in the weather conditions. The shorter sunshine hours before, during and/or after the heading period of most RILs in 2001 resulted in a lower number of spikelets per panicle and/or a lower filled grain percentage compared to those in 2000 (Table 2) (Mastsushima 1976). Large genotypic variations in grain yield among the RILs were revealed under hydroponic culture. RILs with a higher grain yield tended to display a larger number of panicles m⁻² and of spikelets per panicle, and a higher filled grain percentage. The number of spikelets m⁻² exhibited the highest correlation coefficient with grain yield (Table 3). This was supported by the fact that further increases in rice yield depended on the presence of a higher number of spikelets per unit area (Mastsushima 1976; Yamamoto et al. 1991; Yao et al. 2000c). Compared to the other yield components, the number of panicles m⁻² contributed most to grain yield in both years (Table 4). A similar finding was reported by Yoshida et al. (1972) for transplanted rice and by Miller et al. (1991) for water-seeded rice.

Differences in NTA and NUEg among RILs

Genotypic variations in NTA and NUEg were significant among the RILs under hydroponic culture. Many authors have found significant genotypic variations in NTA and NUEg (Broadbent et al. 1987; Datta De and Broadbent 1990; Hasegawa 2003; Higuchi and Yoshino 1986; Inthapanya et al. 2000; Jiang and Dai 2003; Motomatsu et al. 1988; Singh et al. 1998; Tirol-Padre et al. 1996; Yao et al. 2000 a,b,c; Yoshida 1981). However, the wide genotypic differences in NTA (CV was 24.0% and 22.1% in 2000 and 2001, respectively) and NUEg (CV was 19% and 26.5% in 2000 and 2001, respectively) observed in the current study had not been recorded in these previous reports. Therefore, the current results implied that selection of cultivars or lines with high NTA and NUEg values could be achieved, although differences in the NTA and NUEg values among the RILs should be clarified under different levels of nitrogen application in further studies (Seino 1975).

NTA and NUEg in relation to grain yield and yield components

The functional components of grain yield were found to be NTA and NUEg. Genetic variations in grain yield among the RILs at the same N level were attributed to variability in NTA at different rice growth stages and NUEg. Increasing total pre-NA and post-NA or increased NUEg resulted in an increase in grain yield. The contribution of NUEg to the total variation in grain yield was highest among all the functional components, accounting for 56% of the variation in both years (Table 4). Tirol-Padre et al. (1996), using 180 lines, reported that NUEg may be more consistent across environments than the total nutrient content and may be a more useful character for developing new cultivars adapted to low soil fertility conditions. The current results suggested that more attention should be paid to NUEg for improving grain yield, because improving NUEg could reduce not only environment pollution but also agriculture costs. In the current study, some high-yielding lines showed a high NUEg value and comparatively low nitrogen absorption (Fig. 1). Along the iso-grain yield line at 800 g m⁻², the highest nitrogen use efficiency showed a minimum NTA (14.9 g m⁻²) and maximum NUEg (51.3 gGW gN⁻¹) in 2000, the corresponding values were 13.2 g m⁻² and 61.6 gGW gN⁻¹, respectively, in 2001. A similar phenomenon was also observed in winter wheat, which might be caused by the higher translocation of nitrogen to the panicles (Li 2000).

Based on the present study, it can be concluded that there were significant differences in grain yield and yield components, and N absorption and use efficiency among RILs under hydroponic culture. In general, the higher-yielding rice plants displayed a larger number of panicles and spikelets per panicle, a higher filled grain percentage, higher total pre-NA and post-NA and NTA values, and higher NUEg values. At the same high-yielding level, there were large differences in the NTA and NUEg values. Therefore, it might be possible to breed higher-yielding cultivars with higher N use efficiency and provide a simple criterion for selecting and breeding such cultivars.

ACKNOWLEDGMENTS

The authors thank Dr Zhang QF, Huazhong Agricultural University, China, for providing the rice materials used in this study. The authors also thank Dr Kenji Nagata, National Agricultural Research. Center for the Tohoku Region and Dr Kenji Kato, Faculty of Agriculture, Okayama University for their valuable suggestions.