Agron. Sustain. Dev.

30 (2010) 649–656 Available online at:


c INRA, EDP Sciences, 2010 www.agronomy-journal.org
DOI: 10.1051/agro/2010002 for Sustainable Development

Review article

Improving nitrogen fertilization in rice by site-specific N management.

A review*

Shaobing Peng1 **, Roland J. Buresh1 , Jianliang Huang2 , Xuhua Zhong3 , Yingbin Zou4 , Jianchang Yang5 ,
Guanghuo Wang6 , Yuanying Liu7 , Ruifa Hu8 , Qiyuan Tang4 , Kehui Cui2 , Fusuo Zhang9 , Achim Dobermann1

1
Crop and Environmental Sciences Division, International Rice Research Institute, DAPO Box 7777, Metro Manila, Philippines
2
Crop Physiology and Production Center, MOA Key Laboratory of Huazhong Crop Physiology, Ecology and Production, Huazhong Agricultural University,
Wuhan, Hubei 430070, China
3 Rice Research Institute, Guangdong Academy of Agricultural Science, Guangzhou, Guangdong 510640, China
4 Crop Physiology, Ecology, and Production Center, Hunan Agricultural University, Changsha, Hunan 410128, China
5 Agronomy Department, Agricultural College, Yangzhou University, Yangzhou, Jiangsu 225009, China
6 College of Environmental and Natural Resources Sciences, Zhejiang University, Hangzhou, Zhejiang 310029, China
7 College of Resources and Environmental Sciences, Northeast Agricultural University, Harbin, Heilongjiang 150030, China
8 Centre for Chinese Agricultural Policy, Institute of Geographical Sciences and Natural Resource Research, Chinese Academy of Sciences, Beijing 100101,

China
9 College of Resources and Environmental Sciences, China Agricultural University, Beijing 100094, China

(Accepted 21 December 2009)

Abstract – Excessive nitrogen (N) application to rice (Oryza sativa L.) crop in China causes environmental pollution, increases the cost of
rice farming, reduces grain yield and contributes to global warming. Scientists from the International Rice Research Institute have collaborated
with partners in China to improve rice N fertilization through site-specific N management (SSNM) in China since 1997. Field experiments
and demonstration trials were conducted initially in Zhejiang province and gradually expanded to Guangdong, Hunan, Jiangsu, Hubei and
Heilongjiang provinces. On average, SSNM reduced N fertilizer by 32% and increased grain yield by 5% compared with farmers’ N practices.
The yield increase was associated with the reduction in insect and disease damage and improved lodging resistance of rice crop under the
optimal N inputs. The main reason for poor fertilizer N use efficiency of rice crop in China is that most rice farmers apply too much N fertilizer,
especially at the early vegetative stage. We observed about 50% higher indigenous N supply capacity in irrigated rice fields in China than
in other major rice-growing countries. Furthermore, yield response of rice crop to N fertilizer application is low in China, around 1.5 t ha−1
on average. However, these factors were not considered by rice researchers and extension technicians in determining the N fertilizer rate for
recommendation to rice farmers in China. After a decade of research on SSNM in China and other Asian rice-growing countries, we believe
SSNM is a matured technology for improving both fertilizer N use efficiency and grain yield of rice crop. Our challenges are to further simplify
the procedure of SSNM and to convince policy-makers of the effectiveness of this technology in order to facilitate a wider adoption of SSNM
among rice farmers in China.

site-specific nitrogen management / nitrogen use efficiency / grain yield / nitrogen response / rice / China

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
2 SSNM principles and procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 650
3 IRRI-China collaboration on SSNM research . . . . . . . . . . . . . . . . . . . . . . . . . 651
4 Key research findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
5 Remarks on site-specific N management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 655

* Article from the International Workshop on Sustainable Nutrient Management: Technology and Policy organized by Prof. Fusuo Zhang
and Dr. Yuxin Miao, May 28–21, 2008, Shijiazhuang, Hebei Province, China.
** Corresponding author: s.peng@cgiar.org
650 S. Peng et al.

1. INTRODUCTION The high rates of N fertilizer input and improper timing of

N application in China have led to low recovery efficiency and
Increase in fertilizer nutrient input, especially N fertilizer, agronomic N use efficiency. Zhu (1985) reported that recovery
has contributed significantly to the improvement of crop yields efficiency in China was less than 30% for ammonium bicar-
in the world (Cassman et al., 2003). Development of semi- bonate and 30–40% for urea. Li (1997) estimated that recovery
dwarf rice (Oryza sativa L.) varieties in the 1960s assured efficiency for rice in China was around 30–35%. However, Li
the achievement of high grain yield under increased N fertil- (2000) observed that the average recovery efficiency of rice in
izer rates because of their lodging resistance at high N inputs Jiangsu Province was only 20%. This was further confirmed
(Yoshida, 1972). To maximize grain yield, farmers often ap- by Wang et al. (2001), who reported that recovery efficiency
ply a higher amount of N fertilizer than the minimum required of farmers’ N fertilizer practice was 18% in an on-farm ex-
for maximum crop growth (Lemaire and Gastal, 1997). Over- periment conducted in Zhejiang. Peng et al. (2006) found that
application of N fertilizer may actually decrease grain yield recovery efficiency of farmers’ N fertilizer practice was 20 to
by increasing susceptibility to lodging (Pham et al., 2004) and 30% in four provinces in China.
damage from pests and diseases (Cu et al., 1996). Yoshida (1981) estimated agronomic N use efficiency to be
China is currently the world’s largest consumer of N fer- 15–25 kg rough rice per kg applied N in the tropics. Cassman
tilizer. In 2006, annual N fertilizer consumption in China was et al. (1996b) reported that agronomic N use efficiency was
31 million metric tons or 31.7% of the global N consumption 15 to 18 kg kg−1 N in the dry season in the farmers’ fields
(Heffer, 2008). About 18% of this N was used for rice produc- in the Philippines. In China, agronomic N use efficiency was
tion in China (Heffer, 2008). China accounts for about 19% of 15–20 kg kg−1 N from 1958 to 1963 and declined to only
the world’s rice planting area and produced 29% of global rice 9.1 kg kg−1 N between 1981 and 1983 (Lin, 1991). Since then,
production in 2006 (FAOSTAT, 2008). Rice crops in China use agronomic N use efficiency has further decreased in China be-
about 36% of the total N fertilizer used for rice production in cause of the increase in N rate (Peng et al., 2002). Wang et al.
the world (Heffer, 2008). (2001) reported that agronomic N use efficiency of farmers’
There are several sources of information on the average N fertilizer practice was 6.4 kg kg−1 in Zhejiang. Peng et al.
N rate of rice production in China. The International Fertil- (2006) reported that rice yield increases by only 5 to 10 kg
izer Industry Association (IFA, 2002) reported that China’s for every kg of N fertilizer input by using farmers’ N fertilizer
national average N rate for rice was 145 kg ha−1 in 1997. practice in China.
Based on data from 1995 to 1997, the average rate of N ap-
plication for rice production in China was 180 kg ha−1 (FAO,
unpublished data, 2001). If we use the data for the rice plant- 2. SSNM PRINCIPLES AND PROCEDURES
ing area from FAOSTAT and N consumption for rice from
Heffer (2008), China’s national average N rate for rice was Site-specific N management (SSNM) was developed to in-
193 kg ha−1 in 2006, about 90% higher than the world aver- crease fertilizer N use efficiency of irrigated rice (Dobermann
age (FAOSTAT, 2008; Heffer, 2008). Nitrogen rates of 150 to et al., 2002). In SSNM, N application is based on the crop de-
250 kg ha−1 are common (Wang et al., 2001; Peng et al., 2006). mand for N. Climatic factors (solar radiation and temperature)
In Jiangsu Province, China, the average N rate reached 300 kg and indigenous N supply largely affect crop N demand. In-
ha−1 in some counties (Q. Zhu, pers. commun., 2001). Average digenous N supply includes N from soil mineralization, irriga-
rice grain yield in China was 6.27 t ha−1 in 2006, about 53% tion water and crop residues. Soil chemical analysis is not reli-
higher than the world average (FAOSTAT, 2008). Rice crop in able to quantify indigenous N supply for paddy soil (Cassman
China received twice as much N fertilizer as in Japan but the et al., 1996a). Therefore, the measurement of grain yield in
grain yield was similar in the two countries. N-omission plots is used to obtain field-specific estimates of
Fertilizer N use efficiency can be measured in different the indigenous N supply (Cassman et al., 1996b). During the
ways using N-omission plots (Novoa and Loomis, 1981). The growing season, leaf N status measured with a chlorophyll me-
yield increase that results from N application in comparison ter (SPAD) or leaf color chart is a good indicator of crop N de-
with no N application is defined as the agronomic N use ef- mand (Peng et al., 1996). Both the SPAD and leaf color chart
ficiency (kg grain yield increase per kg N applied). The ap- provide a good estimation of leaf N content on a leaf-area ba-
parent recovery efficiency of fertilizer N is used to express sis (Peng et al., 1993; Yang et al., 2003). In-season upward
the percentage of fertilizer N recovered in aboveground plant or downward adjustments of predetermined N topdressings at
biomass at the end of the cropping season. Yield response of critical growth stages are made based on SPAD or leaf color
rice crop to N fertilizer application is defined as N response, chart readings at these stages. In this approach, the timing and
which is calculated as the difference between yields with and number of N applications are fixed while the rate of N top-
without N fertilizer (IRRI, 2006). The partial factor produc- dressing varies across seasons and locations.
tivity from applied N is the ratio of grain yield to N applied In current SSNM practice, four steps are needed to estimate
and it provides an integrative index that quantifies total eco- total N rate based on indigenous N supply capacity and target
nomic output relative to utilization of all N resources in the yield: (1) set an attainable yield target based on 85% of yield
system, including indigenous soil N and fertilizer N (Cassman potential, (2) estimate indigenous N supply – yield without N
et al., 1996b). This parameter for quantifying fertilizer N use fertilizer, (3) estimate N response – the difference between tar-
efficiency does not need N-omission plots. get yield and yield without N fertilizer, and (4) estimate N rate
 Improving nitrogen fertilization in rice by site-specific N management. A review 651

based on N response and agronomic N use efficiency. A realis- Collaborating organizations in China are Zhejiang University,
tic target agronomic N use efficiency is chosen by considering Guangdong Academy of Agricultural Sciences, Hunan Agri-
the yield and agronomic N use efficiency of the location in pre- cultural University, Yangzhou University, Huazhong Agri-
vious seasons (IRRI, 2006). The total N is distributed at one cultural University, Northeast Agricultural University, China
day before transplanting, midtillering, panicle initiation and Agricultural University and the Chinese Academy of Sciences.
heading with approximate proportions of 35, 20, 30 and 15%, From 1997 to 2000, on-farm demonstration of SSNM was
respectively (Peng et al., 2006). The pre-determined N rate is conducted in Zhejiang province in farmers’ fields by compar-
30 kg ha−1 at midtillering and 40 kg ha−1 at panicle initiation. ing SSNM with farmers’ N fertilizer practice (Wang et al.,
The actual rates of N topdressing at midtillering and panicle 2001, 2004). From 2001 to 2003, SSNM was compared
initiation are adjusted by ±10 kg N ha−1 according to leaf N with other N treatments including real-time N management,
status measured with a SPAD or leaf color chart. When leaf N farmers’ N fertilizer practice, modified farmers’ N fertilizer
is below the low level of the SPAD or leaf color chart thresh- practice, and three fixed-N split treatments in the replicated
olds, N rate will be increased by 10 kg ha−1 ; when leaf N is on-farm trials in Zhejiang, Guangdong, Hunan and Jiangsu
greater than the upper level of the SPAD or leaf color chart provinces (Peng et al., 2006). In real-time N management, N
thresholds, it will be decreased by 10 kg ha−1 . No adjustment was applied only when the leaf N content was below a criti-
in N rate is needed when leaf N is within the thresholds. At cal level. The farmers’ N fertilizer practice was modified by
heading, N is applied only if leaf N is below the SPAD or leaf reducing the total N rate by 30% and this reduction was re-
color chart thresholds. We used the same SPAD and leaf color stricted to within 10 days after transplanting. The three fixed-
chart thresholds across different growth stages for a given vari- N split treatments had total N rates of 60, 120 and 180 kg ha−1
ety (Peng et al., 1996). The SPAD and leaf color chart thresh- with 35% applied at basal, 20% at midtillering, 30% at pani-
olds vary with varieties, which needs to be determined exper- cle initiation, and 15% at heading. From 2005 to 2007, SSNM
imentally. Japonica generally requires higher thresholds than was compared with the real-time N management using four
indica. For example, SPAD thresholds of 35–37 were used for SPAD thresholds in the replicated on-farm trials in Guang-
indica varieties and they were 2 units higher for japonica vari- dong, Hunan, Hubei and Jiangsu provinces. Overall, SSNM
eties (Huang et al., 2008). The SPAD value of 35 corresponded performed better than the real-time N management and the
to a leaf color chart score of 3.2, 4.8 and 5.2 for the leaf color modified farmers’ N fertilizer practice because the total N rate
charts developed by the International Rice Research Institute of SSNM was closer to the optimal level. In on-farm demon-
(IRRI), Zhejiang Agricultural University, China, and Univer- stration and farmer participatory research, we focused mainly
sity of California-Davis, USA, respectively (Yang et al., 2003). on SSNM technology for improving fertilizer N use efficiency
The procedures of SSNM can be simplified in two ways. in irrigated rice in China.
Firstly, if there is no N-omission plot and its grain yield is un-
known, one can estimate the N response based on the informa- On-farm demonstration of SSNM was done in the six
tion of climatic yield potential and soil fertility (i.e. skipping provinces from 2003 to 2007. Only one location was chosen
the first two steps in estimating total N rate). Secondly, if farm- in each province. About 10 farmers from each location par-
ers do not want to measure leaf N status using the SPAD or leaf ticipated in the demonstration experiment. Each farmer’s field
color chart, upward or downward adjustments of N topdress- had three treatments: farmers’ N fertilizer practice, SSNM and
ings can be made based on visual leaf N status (IRRI, 2006). N omission. The field was divided into two equal parts with
Since 1997, SSNM has been evaluated in farmers’ fields a levee, and farmers’ N fertilizer practice and SSNM treat-
in eight major irrigated rice domains in Asia (Dobermann ments were randomly assigned to each half of the field. The
et al., 2002), including rice farms in Zhejiang Province, China N-omission plot (5 m × 5 m) was installed in the half of the
(Wang et al., 2001, 2004). Across all sites in Asia, average field with SSNM treatment. In the farmers’ N fertilizer prac-
grain yield increased by 11% and average recovery efficiency tice, farmers apply N fertilizer (form, rate and timing) based
increased from 31% to 40%, with 20% of all farmers achieving on their own decision. Other crop management practices in-
more than 50% recovery efficiency (Dobermann et al., 2002). cluding P and K application were identical among the three
treatments.
We used farmer participatory research for on-farm test-
3. IRRI-CHINA COLLABORATION ON SSNM ing of N fertilization by standard and farmer-modified SSNM
RESEARCH for irrigated rice in Guangdong, Hunan, Hubei and Jiangsu
provinces from 2003 to 2005 (Hu et al., 2007). Twelve to
15 farmers were randomly selected in each study village in
IRRI scientists in collaboration with partners in China each year for a dialogue with the research team and for a
have started research work on improving rice N fertilization rapid rural technology assessment. Based on the information
through SSNM in China since 1997 (Wang et al., 2001). Field obtained from the rapid rural technology assessment, mod-
experiments were conducted initially in Zhejiang province ified SSNM schemes were developed through dialogue be-
in 1997 and gradually expanded to Guangdong, Hunan and tween a research team and farmers at a workshop in each vil-
Jiangsu provinces in 2001, to Hubei province in 2003, and to lage. Modification mainly involved decreasing the number of
Heilongjiang province in 2005 (Peng et al., 2006). These six N fertilizer topdressings and increasing the rate of basal N
provinces occupy about 45% of the rice planting area in China. application. Based on the farmers’ willingness, 144 farmers
652 S. Peng et al.

Grain yield (t ha-1) Grain yield with N (t ha-1)

12 11

10
10
9

8
8

6
6

5
4

2 3
0 100 200 300 400 500 2 3 4 5 6 7 8 9
-1 -1
N rate (kg ha ) Grain yield of zero-N (t ha )

Figure 1. Grain yield and N rate of site-specific N management Figure 2. Relationship between grain yield with and without N fertil-
(solid symbols) and other N treatments (open symbols). Data are from izer. Data are from on-farm field trials, on-farm demonstration and
on-farm field trials, on-farm demonstration and farmer participatory farmer participatory research conducted in six provinces in China
research conducted in six provinces in China from 2001 to 2007. from 2001 to 2007. y = 2.16 + 0.886x (r = 0.70; n = 544).

were selected to conduct an experiment to compare SSNM or

major rice-growing countries. Crop N uptake at maturity
modified SSNM with farmers’ N fertilizer practice.
when N fertilizer was not applied was in the range of 90–
Data from on-farm field trials, on-farm demonstration and
100 kg ha−1 , compared with 60–70 kg ha−1 under tropical con-
farmer participatory research in the six provinces in China
ditions (Dobermann et al., 2003). Grain yield of zero-N control
from 2001 to 2007 were pooled to compare SSNM with farm-
ranged from 3 to 8 t ha−1 and averaged at 5.77 t ha−1 (Fig. 2).
ers’ N fertilizer practice and other N treatments. The same
There was a positive relationship between grain yields with
dataset was used to determine the relationships among yield
and without N fertilizer.
with N fertilizer, zero-N yield, N response and agronomic N
use efficiency. There were 544 observations for these analyses. Averaged across 107 farmers who conducted on-farm
Data from the on-farm demonstration of SSNM in Zhejiang demonstration, SSNM produced 5% higher grain yield than
province from 1997 to 2000 were published elsewhere (Wang farmers’ N fertilizer practice (Tab. I). The yield increase was
et al., 2001, 2004) and not included in the analyses of this associated with the reduction in insect and disease damage
paper. and improved lodging resistance of rice crop under the op-
timal N inputs (Zhong et al., 2006a, b). This small yield in-
crease was achieved when total N rate was reduced from 195 to
4. KEY RESEARCH FINDINGS 133 kg ha−1 , a 32% saving of N fertilizer over farmers’ N fer-
tilizer practice. The N response of SSNM was 28% higher than
There were wide ranges in both grain yield and total N rate that of farmers’ N fertilizer practice. Consequently, SSNM al-
across all experiments (Fig. 1). Grain yield varied from 4 to most doubled farmers’ N fertilizer practice in agronomic N
10 t ha−1 with an average of 7.28 t ha−1 , while total N rate use efficiency, and had 55% higher partial factor productivity
ranged from 50 to 400 kg ha−1 with a mean of 157 kg ha−1 . of applied N than farmers’ N fertilizer practice. Similar results
SSNM was able to narrow the range of total N rate to 80– were obtained when SSNM was compared with other N treat-
200 kg ha−1 with a mean of 120 kg ha−1 . There was no corre- ments such as real-time N management, farmers’ N fertilizer
lation between grain yield and total N rate. This poor relation- practice, modified farmers’ N fertilizer practice, and fixed-N
ship was caused by many factors including location, season, split treatments in 25 replicated on-farm trials in five provinces
variety, pest damage, other crop management practices, etc., between 2001 and 2007 (Tab. II). Yield increase was insignif-
which affected grain yield aside from total N rate. icant but SSNM saved 25% N fertilizer over other N treat-
We observed a relatively high indigenous N supply ca- ments. SSNM increased N response by 15%, agronomic N use
pacity in irrigated rice fields in China compared with other efficiency by 55%, and partial factor productivity of applied N
 Improving nitrogen fertilization in rice by site-specific N management. A review 653

Table I. Grain yield, total N rate, yield response to N application, agronomic N use efficiency, and partial factor productivity of applied N (PFP)
of farmers’ fertilizer practice and site-specific N management (SSNM). Data were from on-farm demonstrations conducted by 107 farmers from
six provinces in China between 2003 and 2007. Average grain yield of zero-N control was 5.69 t ha−1 across the 107 farmers.

Parameters Farmers’ practice SSNM Difference (%)

Grain yield (t ha−1 ) 7.08 b 7.47 a 5
N rate (kg ha−1 ) 195 a 133 b 38
N response (t ha−1 ) 1.39 b 1.78 a 25
Agronomic N use efficiency (kg kg−1 ) 7.1 b 13.4 a 61
PFP (kg kg−1 ) 36.3 b 56.2 a 43
Within a row, means followed by different letters are significantly different at the 0.05 probability level according to the least significant
difference (LSD) test.

Table II. Grain yield, total N rate, yield response to N application, agronomic N use efficiency, and partial factor productivity of applied N
(PFP) of fixed N-rate treatments and site-specific N management (SSNM). Data were from 25 on-farm field experiments conducted in five
provinces in China between 2001 and 2007. Average grain yield of zero-N control was 5.87 t ha−1 across the 25 experiments.

Parameters Other N treatments SSNM Difference (%)

Grain yield (t ha−1 ) 7.45 a 7.69 a 3
N rate (kg ha−1 ) 161 a 120 b 29
N response (t ha−1 ) 1.58 a 1.82 a 14
Agronomic N use efficiency (kg kg−1 ) 9.8 b 15.2 a 43
PFP (kg kg−1 ) 46.3 b 64.1 a 32
Within a row, means followed by different letters are significantly different at the 0.05 probability level according to the least significant
difference (LSD) test.

by 38% compared with other N treatments. In farmer partici- N response (t ha-1)

patory research, the rate and distribution of N fertilizer during 5
the growing season of modified SSNM were in between those
of SSNM and farmers’ N fertilizer practice (Hu et al., 2007). 4
Both SSNM and modified SSNM, compared with farmers’ N
fertilizer practice, maintained rice yields with significantly less
N fertilizer and no significant increase in total labor input. The 3
reduction in N fertilizer input averaged 48 kg ha−1 for SSNM
and 23 kg ha−1 for modified SSNM. 2
Large variation in N response was observed across 544
observations (Fig. 3). The 1.5 t ha−1 average N response
across all experiments is very low compared with the average 1
N response of 3.0 t ha−1 observed under tropical conditions
(Dobermann et al., 2003). Negative N response occurred when 0
the application of N fertilizer actually reduced grain yield.
Yield reduction was often observed under excessive N input
due to greater pest damage and lodging. There was no corre- -1
lation between N response and total N rate. High total N rates
did not result in high N response. The N response was corre- -2
lated positively with grain yield when N was applied (Fig. 4). 0 100 200 300 400 500
There was a negative but weak correlation between N response -1
N rate (kg ha )
and the grain yield of zero-N control.
There was a negative relationship between agronomic N Figure 3. Relationship between yield response to N application (N
use efficiency and total N rate (Fig. 5). Large variation in response) and N rate. Data are from on-farm field trials, on-farm
agronomic N use efficiency was observed when total N rates demonstration and farmer participatory research conducted in six
were between 80 and 120 kg ha−1 . When total N rates were provinces in China from 2001 to 2007.
greater than 300 kg ha−1 , agronomic N use efficiency was be-
low 10 kg kg−1 . Negative agronomic N use efficiency was ob- correlated positively with the grain yield of N-applied treat-
served when grain yield was reduced by the application of N ments but negatively with the grain yield of zero-N control
fertilizer. Like N response, agronomic N use efficiency was (Fig. 6).
654 S. Peng et al.

N response (t ha-1) N response (t ha-1)

5 5
y = -1.75 + 0.448x y = 2.16 - 0.114x
4 (r = 0.62; n = 544) 4 (r = -0.13; n = 544)

3 3

2 2

1 1

0 0

-1 -1
a b
-2 -2
3 4 5 6 7 8 9 10 11 3 4 5 6 7 8 9 10 11
-1 -1
Grain yield with N (t ha ) Grain yield of zero-N (t ha )

Figure 4. Relationship between yield response to N application (N response) and grain yield with N fertilizer (a) and grain yield without N
fertilizer (b). Data are from on-farm field trials, on-farm demonstration and farmer participatory research conducted in six provinces in China
from 2001 to 2007.

Agronomic N use efficiency (kg/kg)

the provinces where total N rate is high and agronomic N use
50 efficiency is low.
The main reason for poor fertilizer N use efficiency of rice
crop in China is that most rice farmers apply excess N fertil-
40 izer, especially at the early vegetative stage (Peng et al., 2006).
High indigenous N supply capacity and low N response were
not considered by rice researchers and extension technicians
30 when N fertilizer rate was determined for recommendation to
rice farmers in China. This explains why SSNM usually have a
lower total N rate than farmers’ N fertilizer practice in China.
20 A twelve-season long-term field experiment demonstrated that
a decrease in total N rate may reduce indigenous N supply ca-
pacity, but will not cause yield reduction in subsequent rice
10 crops as long as optimal N management is practiced (data not
shown).

0
5. REMARKS ON SITE-SPECIFIC
N MANAGEMENT
-10
Improved N management such as SSNM increases both
0 100 200 300 400 500
grain yield and fertilizer N use efficiency compared with farm-
N rate (kg ha-1) ers’ N fertilizer practice in China. This is achieved by reducing
total N rate and by reducing N rate during the early vegetative
Figure 5. Relationship between agronomic N use efficiency and N
rate. Data are from on-farm field trials, on-farm demonstration and
stage. The total N rate is reduced because of small N response
farmer participatory research conducted in six provinces in China and high target agronomic N use efficiency. The reduction in
from 2001 to 2007. N rate during the early vegetative stage is considered because
irrigated rice soils in China have a high indigenous N supply
capacity, which provides sufficient N for the early vegetative
Grain yield and total N rate were generally higher in the growth. In implementing the SSNM procedure, we have two
North than in the South. Grain yield of zero-N control sug- steps to determine the right N rate for rice crop. The first step
gests that indigenous N supply was the lowest in Guangdong is done before crop establishment. The total N rate is estimated
compared with the other five provinces. The N response and based on N response and target agronomic N use efficiency.
agronomic N use efficiency were the highest in Heilongjiang The percentage of the first N application is decided based on
compared with the other five provinces. Therefore, there is a previous experimental results and indigenous N supply capac-
great potential to improve N management through SSNM in ity. The second step is the upward or downward adjustments
 Improving nitrogen fertilization in rice by site-specific N management. A review 655

Agronomic N use efficiency (kg/kg) Agronomic N use efficiency (kg/kg)

50 50
y = -7.74 + 2.595x y = 19.37 - 1.426x
40 (r = 0.43; n = 544) 40 (r = -0.18; n = 544)

30 30

20 20

10 10

0 0
a b
-10 -10
3 4 5 6 7 8 9 10 11 3 4 5 6 7 8 9 10 11
Grain yield with N (t ha-1) Grain yield of zero-N (t ha-1)

Figure 6. Relationship between agronomic N use efficiency and grain yield with N fertilizer (a) and grain yield without N fertilizer (b). Data
are from on-farm field trials, on-farm demonstration and farmer participatory research conducted in six provinces in China from 2001 to 2007.

of topdressing N rate by 10 kg N ha−1 at critical growth stages ing SSNM could be complicated for many farmers. These
based on leaf N status. problems may be overcome by developing remote sensing
Several limitations are associated with SSNM. First of all, technology such as a canopy reflectance sensor or satellite re-
grain yield of zero-N control varies with cropping history, va- mote sensing for determining the timing and rate of fertilizer
riety, climatic conditions and crop management practices. As N topdressing during the rice-growing period (Xue and Yang,
a consequence, N response is not very stable for a given lo- 2008).
cation across seasons. Secondly, agronomic N use efficiency Despite the above limitations, a decade of collaborative re-
is affected by many factors such as N response and total N search work on SSNM between IRRI and Chinese scientists
rate. Therefore, total N rate determined by N response and has changed the perception of researchers and farmers about N
agronomic N use efficiency provides only an approximate fertilizer management in irrigated rice in China. The following
range. Thirdly, varietal differences in tillering capacity and concepts are accepted now by many rice researchers in China:
early growth vigor are not considered when the percentage of (1) target yield and indigenous N supply capacity should be
the first N application is decided. Finally, the in-season adjust- considered when total N rate is determined; (2) determination
ment of N rate by ±10 kg ha−1 may be inadequate because of N application rate at basal and during the early vegetative
more adjustment could be needed when yield level and the stage should also be based on indigenous N supply; (3) the rate
magnitude of N response are high. Because of these limita- of N topdressing depends on leaf N status and crop N demand,
tions, only 3–5% increase in grain yield can be achieved by and (4) regional blanket recommendation for N management
SSNM in China. Furthermore, SSNM improved agronomic N will not work well. Nowadays, many researchers have devel-
use efficiency only up to an average of 15 kg kg−1 in China. oped local N management practices based on SSNM principles
Future research should focus on accurate determination of N and by integration with other management practices. Many
response and agronomic N use efficiency using weather data farmers who have been exposed to SSNM understand that it
such as solar radiation and air temperature, and information is not true that “the higher the N fertilizer input, the higher
on variety and soil characteristics. Experiments are needed to the grain yield” and that “the greener the leaves, the better the
test if in-season adjustment of N rate should be linked with the rice crop”. They know that it is necessary to reduce N input
difference between actual SPAD readings and SPAD threshold at basal and the early vegetative stage and increase N input
values. at late stages. The dynamic change in leaf N status has been
Several problems exist in the adoption of SSNM technol- embedded in some farmers’ minds. They can judge visually
ogy. First of all, information on the yield of N-omission plots when and how much N should be applied without using the
and target yield are not available to many farmers. Secondly, leaf color chart.
N response is variable and largely affected by variety, crop
management practices, season and location. Thirdly, some
farmers are reluctant to invest time in monitoring leaf N sta- 6. CONCLUSIONS
tus using the leaf color chart. Many farmers have difficulty
in determining leaf color chart readings accurately. Fourthly, After one decade of research on SSNM in China and other
frequent changes in varieties require adjustments in leaf color Asian rice-growing countries, we believe SSNM is a matured
chart threshold values. Finally, the procedures for implement- technology for improving both fertilizer N use efficiency and
656 S. Peng et al.

grain yield of irrigated rice crop. Our study suggests that there Lemaire G., Gastal F. (1997) Nitrogen uptake and distribution in plant
is potential for large-scale dissemination of SSNM technology canopies, in: Lemaire G. (Ed.), Diagnosis of the Nitrogen Status in
Crops, Springer-Verlag, Berlin, pp. 3–43.
in China. Accurate estimation of N response and selection of
the right agronomic N use efficiency are needed for a better Li Q. (1997) Fertilizer issues in the sustainable development of China
performance of SSNM. In-season adjustment in N rate should agriculture, Jiangxi Science and Technology Press, China.
also consider the differences in location, season and variety. Li R. (2000) Efficiency and regulation of fertilizer nitrogen in high-yield
SSNM principles are useful for researchers to develop N man- farmland: A case study on rice and wheat double maturing system
agriculture area of Tai lake for deducing to Jiangsu Province, Ph.D.
agement practices for their local conditions. Our challenges Dissertation, China Agricultural University, Beijing, China.
are to further simplify the procedure of SSNM and to convince
Lin B. (1991) Make the most efficient use of fertilizers in increasing
policy-makers of the effectiveness of this technology in order crop production, in: Soil Science in China: Present and Future, Soil
to facilitate a wider adoption of SSNM among rice farmers in Science Society of China, Jiangsu Science and Technology Press,
China. Future research is needed to develop SSNM based on China, pp. 29–36.
remote sensing technology so that SSNM can be practiced on Novoa R., Loomis R.S. (1981) Nitrogen and plant production, Plant Soil
a large scale. 58, 177–204.
Peng S., Buresh R.J., Huang J., Yang J., Zou Y., Zhong X., Wang G.,
Zhang F. (2006) Strategies for overcoming low agronomic nitrogen
REFERENCES use efficiency in irrigated rice systems in China, Field Crops Res.
96, 37–47.
Cassman K.G., Dobermann A., Sta Cruz P.C., Gines G.C., Samson M.I., Peng S., Garcia F.V., Laza R.C., Cassman K.G. (1993) Adjustment for
Descalsota J.P., Alcantara J.M., Dizon M.A., Olk D.C. (1996a) Soil specific leaf weight improves chlorophyll meter’s estimate of rice
organic matter and the indigenous nitrogen supply of intensive irri- leaf nitrogen concentration, Agron. J. 85, 987–990.
gated rice systems in the tropics, Plant Soil 182, 267–278.
Peng S., Garcia F.V., Laza R.C., Sanico A.L., Visperas R.M., Cassman
Cassman K.G., Dobermann A., Walters D.T., Yang H.S. (2003) Meeting K.G. (1996) Increased N-use efficiency using a chlorophyll meter
cereal demand while protecting natural resources and improving on high-yielding irrigated rice, Field Crops Res. 47, 243–252.
environmental quality, Ann. Rev. Environ. Resour. 28, 315–358.
Peng S., Huang J., Zhong X., Yang J., Wang G., Zou Y., Zhang F., Zhu Q.,
Cassman K.G., Gines G.C., Dizon M.A., Samson M.I., Alcantara J.M. Buresh R., Witt C. (2002) Challenge and opportunity in improving
(1996b) Nitrogen-use efficiency in tropical lowland rice systems: fertilizer-nitrogen use efficiency of irrigated rice in China, Agric.
contributions from indigenous and applied nitrogen, Field Crops Sci. China 1, 776–785.
Res. 47, 1–12.
Pham Q.D., Abe A., Hirano M., Sagawa S., Kuroda E. (2004) Analysis of
Cu R.M., Mew T.W., Cassman K.G., Teng P.S. (1996) Effect of sheath lodging-resistant characteristics of different rice genotypes grown
blight on yield in tropical, intensive rice production system, Plant under the standard and nitrogen-free basal dressing accompanied
Disease 80, 1103–1108. with sparse planting density practices, Plant Prod. Sci. 7, 243–251.
Dobermann A., Witt C., Abdulrachman S., Gines H.C., Nagarajan R., Son Wang G.H., Dobermann A., Witt C., Sun Q.Z., Fu R.X. (2001)
T.T., Tan P.S., Wang G.H., Chien N.V., Thoa V.T.K., Phung C.V., Performance of site-specific nutrient management for irrigated rice
Stalin P., Muthukrishnan P., Ravi V., Babu M., Simbahan G.C., in southeast China, Agron. J. 93, 869–878.
Adviento M.A.A. (2003) Soil fertility and indigenous nutrient sup-
ply in irrigated rice domains of Asia, Agron. J. 95, 913–923. Wang G.H., Sun Q., Fu R., Huang X.H., Ding X.H., Wu J., He Y.F.,
Dobermann A., Witt C. (2004) Site-specific nutrient management
Dobermann A., Witt C., Dawe D., Gines H.C., Nagarajan R.,
Satawathananont S., Son T.T., Tan P.S., Wang G.H., Chien N.V., in intensive irrigated rice systems of Zhejiang province, China, in:
Thoa V.T.K., Phung C.V., Stalin P., Muthukrishnan P., Ravi V., Babu Dobermann A. et al. (Eds.), Increasing Productivity of Intensive
M., Chatuporn S., Kongchum M., Sun Q., Fu R., Simbahan G.C., Rice Systems through Site-specific Nutrient Management. Science
Adviento M.A.A. (2002) Site-specific nutrient management for in- Publishers, Inc., Enfield, NH (USA), and International Rice
tensive rice cropping systems in Asia, Field Crops Res. 74, 37–66. Research Institute, Los Baños, Philippines, pp. 243–263.

FAOSTAT (2008) FAO Statistical databases, http://www.fao.org, Food Xue L., Yang L. (2008) Recommendations for nitrogen fertilizer top-
and Agriculture Organization (FAO) of the United Nations, Rome. dressing rates in rice using canopy reflectance spectra, Biosyst. Eng.
100, 524–534.
Heffer P. (2008) Assessment of fertilizer use by crop at the global level,
International Fertilizer Industry Association (IFA), Paris, France, Yang W.-H., Peng S., Huang J., Sanico A.L., Buresh A.L., Witt C. (2003)
pp. 1–5. Using leaf color charts to estimate leaf nitrogen status of rice,
Agron. J. 95, 212–217.
Hu R., Cao J., Huang J., Peng S., Huang J., Zhong X., Zou Y., Yang
J., Buresh R.J. (2007) Farmer participatory testing of standard and Yoshida S. (1972) Physiological aspects of grain yield, Annu. Rev. Plant
modified site-specific nitrogen management for irrigated rice in Physiol. 23, 437–464.
China, Agric. Syst. 94, 331–340. Yoshida S. (1981) Fundamentals of rice crop science. International Rice
Huang J., He F., Cui K., Buresh R.J., Xu B., Gong W., Peng S. (2008) Research Institute, Los Baños, Philippines, 269 p.
Determination of optimal nitrogen rate for rice varieties using a
chlorophyll meter, Field Crops Res. 105, 70–80. Zhong X., Peng S., Buresh R.J., Huang N., Zheng H. (2006a) Some
canopy indices influencing sheath blight development in hybrid
IFA (2002) Fertilizer use by crop, 5th ed. http://www.fertilizer.org/ifa/ rice, Chinese J. Rice Sci. 20, 535–542.
statistics.asp, International Fertilizer Industry Association (IFA),
International Fertilizer Development Center (IFDC), International Zhong X., Zheng H., Huang N., Ou J. (2006b) Performance of site-
Potash Institute (IPI), Potash and Phosphate Institute (PPI), and specific nutrient management in double-season rice in South China,
Food and Agriculture Organization (FAO). China Rice 3, 30–32.
International Rice Research Institute (IRRI) (2006) Site-specific nutrient Zhu Z. (1985) Research progresses on the fate of soil N supply and ap-
management, www.irrri.org/irrc/ssnm/, accessed 20 March 2006. plied fertilizer N in China, Soil 17, 2–9.