Agronomy Journal Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (24)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Agricola
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Related Collections
Right arrow Economics
Right arrow Rice
Right arrow Nutrient Management
Right arrow Site-Specific Analysis
Right arrow Irrigation
Agronomy Journal 93:869-878 (2001)
© 2001 American Society of Agronomy

SOIL MANAGEMENT

Performance of Site-Specific Nutrient Management for Irrigated Rice in Southeast China

Guanghuo Wanga, Achim Dobermann*,b, Christian Wittc, Quingzhu Sund and Rongxing Fue

a College of Environ. and Nat. Resources Sci., Zhejiang Univ., Huajiachi, Hangzhou 310029, P.R. China
b Dep. of Agron. and Hortic., Univ. of Nebraska, P.O. Box 830915, Lincoln, NE 68583-0915
c Int. Rice Res. Inst. (IRRI), DAPO, Box 7777, Metro Manila, Philippines
d Agric. Res. Stn., Jinhua, Zhejiang, P.R. China
e Agric. Technol. Ext. Stn., Jinhua, Zhejiang, P.R. China

* Corresponding author (adobermann2{at}unl.edu)

Received for publication October 2, 2000.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND...
 REFERENCES
 
Rice (Oryza sativa L.) yield increases in Zhejiang, China have slowed since 1985 despite the increasing use of hybrids and fertilizers. On-farm experiments at 21 sites were conducted to evaluate a new approach for site-specific nutrient management (SSNM). Field- and season-specific N–P–K applications were calculated by accounting for the indigenous nutrient supply, yield targets, and nutrient demand as a function of the interactions between N, P, and K. Nitrogen applications were fine-tuned based on season-specific rules and field-specific monitoring of crop N status. The performance of SSNM was tested for four successive rice crops. Compared with the current farmers' fertilizer practice (FFP), average grain yield increased from 5.9 to 6.4 Mg ha-1 while plant N, P, and K uptake increased by 8 to 14%. The gross return over fertilizer cost was about 10% greater with SSNM than with FFP. Yields were about 20% greater in late rice (hybrid cultivars) than in early rice (inbred cultivars), but SSNM performed equally better than FFP in both seasons. Improved timing and splitting of fertilizer N increased N recovery efficiency from 0.18 kg kg-1 in FFP plots to 0.29 kg kg-1 in SSNM plots. The agronomic N use efficiency (grain yield increase per kilogram fertilizer applied) was 80% greater with SSNM than with FFP. As defined in our study, SSNM has potential for improving yields and nutrient efficiency in irrigated rice. Future research needs to develop a practical approach for achieving similar benefits across large areas without field-specific modeling and with minimum crop monitoring.

Abbreviations: AEN, agronomic efficiency of applied fertilizer N • DAT, days after transplanting • ER, early rice • FFP, farmers' fertilizer practice • GRF, gross return above fertilizer cost • IKS, indigenous K supply • INS, indigenous N supply • IPS, indigenous P supply • LR, late rice • PFPN, partial factor productivity of applied N • PI, panicle initiation • REN, recovery efficiency of applied N • SSNM, site-specific nutrient management


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND...
 REFERENCES
 
ZHEJIANG PROVINCE IN SOUTHEAST CHINA accounts for 7% of the national rice production. About 1.3 million ha of arable land is used for growing irrigated rice, mainly in double-cropping systems that were adopted in the 1960s. Rice production increased rapidly from 1970 to 1985 but started to decline thereafter (Fig. 1). The initial increase (3.5% yr-1 from 1970–1985) was mainly due to the widespread adoption of hybrid rice and an increase in fertilizer use (3% yr-1). However, since 1985, industrialization and urbanization have caused a decline in the rice area at a rate of about 2% yr-1, and yield growth rates have slowed down to 0.3% yr-1, despite further rising fertilizer consumption. Current average rice yields in Zhejiang are only about 50 to 60% of the estimated genetic and climatic yield potential (Zheng et al., 1997b). Fertilizer prescriptions are not based on soil testing or more site-specific knowledge of soil nutrient status. Environmental pollution by nutrient leaching or runoff from rice fields has become another concern (Zhang and Wang, 1999). On-farm studies have been conducted in China, but they indicate low N use efficiency in irrigated rice (Natl. Soil and Fert. Stn., 1993; Wang et al., 1994).



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 1. Trends of rice production, area, yield, and total fertilizer consumption in Zhejiang Province, China. Yield trends shown are averages of two rice crops per year. Annual rates of yield and production changes were estimated from a piece-wise linear regression. (Data source: annual statistical yearbooks of Zhejiang Province, Zhejiang Statistical Bureau, Beijing, China.)

 
Trends similar to those in Zhejiang have been observed in other key irrigated rice regions of Asia where early adoption of intensive rice cropping took place (Cassman and Dobermann, 2001). At issue then is how soil and crop management must change to reaccelerate yield and productivity growth in the future. We hypothesized that rice yields, profit, plant nutrient uptake, and N use efficiencies can be significantly increased by applying fertilizers on a field- and season-specific basis, i.e., through more knowledge-intensive, site-specific approaches of nutrient management. On-farm research conducted in different countries has demonstrated large and potentially manageable variability in soil nutrient supply and crop response to nutrients among rice farms or small rice fields (Angus et al., 1990; Cassman et al., 1996; Olk et al., 1999; Adhikari et al., 1999). Therefore, we defined site-specific nutrient management (SSNM) as the dynamic, field-specific management of nutrients in a particular cropping season to optimize the supply and demand of nutrients according to their differences in cycling through soil–plant systems (Dobermann and White, 1999).

The specific objectives of our research were to (i) quantify the variation in soil nutrient supply in irrigated rice fields within a typical rice domain of Zhejiang and (ii) develop a new approach for SSNM and compare its agronomic and economic performance to the current practice of fertilizer use.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND...
 REFERENCES
 
Site Characteristics
The experimental domain is located in the Jinhua-Quzhou Basin in central Zhejiang, near Jinhua City (29°5'N, 119°47'E), and represents a rice-growing area of about 145000 ha in the Jinhua district. Soils at the study site mainly include alluvial soils with high fertility (Aquents, Aquepts, Aqualfs) and various red soils (Aquults) with lower soil fertility status (Soil Survey Office of Zhejiang, 1994). The study site has a subtropical climate (annual average temperature of 17°C and 1300–1500 mm rainfall; Fig. 2). Double-rice cropping, which started about 30 yr ago, is the main cropping system. Early rice (ER) is grown from April to mid- or late July using both hybrids and inbred rice varieties. Late rice (LR) is 90% hybrid rice and is grown from mid-July to late October. Land preparation is done on wet soils with tractors. Rivers and reservoirs irrigate all paddy fields. Average nutrient concentrations in irrigation water measured in 1999 were N, 1.2; K, 1.7; Mg, 1.4; and Ca, 13 mg L-1, indicating that total N and K inputs from this source are typically small.



View larger version (46K):
[in this window]
[in a new window]
  		Fig. 2. Climatic conditions at Jinhua, Zhejiang Province, China during 1997 to 1999. Sunshine hours and temperature data are 9-d moving averages; rainfall is monthly total.

 
On-Farm Experiments
On-farm experiments were conducted from 1997 to 1999 to develop and test a new SSNM approach. Within a radius of 20 km around Jinhua City, 21 rice farms belonging to seven different villages were selected to represent the range of socioeconomic conditions and soil types in the whole domain. All farmers grew rice in a ER–LR system, but farm sizes ranged from small farms (0.2–0.5 ha rice area) to larger farms (2–5 ha). Soils included alluvial (19 sites) and red soils (2 sites). Treatments included:

FFP—Farmers' Fertilizer Practice (1997–1999)
All crop and fertilizer management was done by the farmer on a single rice field (0.1–0.4 ha) with no interference by the researcher. Ranges of farmers' fertilizer rates in 1998 and 1999 were 120 to 220, 0 to 45, and 0 to 180 kg ha-1 N, P, and K, respectively, applied to ER or 95 to 250, 0 to 43, and 0 to 120 kg ha-1 N, P, and K, respectively, applied to LR. Most farmers applied all N, P, and K fertilizer within the first 2 wk after planting.

0-N—Nitrogen Omission Plot (1997–1999)
Only P (30 kg P ha-1) and K (50 kg K ha-1) were applied to a strip plot (80–100 m2) embedded in the FFP to ensure that macronutrients other than N did not limit plant N uptake from indigenous sources.

0-P—Phosphorus Omission Plot (1997–1998)
Only N (160 kg N ha-1 in ER and 180 kg N ha-1 in LR) and K (150 kg K ha-1) were applied to a strip plot (80–100 m2) embedded in the FFP to ensure that macronutrients other than P did not limit plant P uptake from indigenous sources.

0-K—Potassium Omission Plot (1997–1998)
Only N (160 kg N ha-1 in ER and 180 kg N ha-1 in LR) and P (35 kg P ha-1) were applied to a strip plot (80–100 m2) embedded in the FFP to ensure that macronutrients other than K did not limit plant K uptake from indigenous sources.

SSNM—Site-Specific Nutrient Management (1998–2000)
Nutrient applications were prescribed to a larger plot (300–1000 m2) located within the farmer's field (FFP) on a field- and crop-specific basis following the SSNM approach described below. Ranges of fertilizer rates in 1998 and 1999 were 90 to 170, 10 to 35, and 30 to 120 kg ha-1 N, P, and K applied to ER or 90 to 180, 10 to 28, and 30 to 140 kg ha-1 N, P, and K applied to LR.

In 1997, only FFP, 0-N, 0-P, and 0-K treatments were established to measure baseline data of productivity, nutrient use efficiency, and indigenous nutrient supply in ER and LR crops grown on each farm. Nutrient omission plots were rotated within the field after each crop to avoid residual effects. The SSNM plot was established with the 1998 ER crop and remained at the same location for four consecutive rice crops grown in 1998 and 1999.

Rice varieties were chosen by the farmers and were the same in all treatments. Farmers planted conventional modern varieties in ER (about 10 different ones, such as ‘Jinzhao22’, ‘Zhefu802’, ‘Zhong903’, or ‘Zhe733’) but mostly hybrid rice in LR (e.g., "Xieyou64', ‘Ilyou88’, ‘Ilyou92’, or ‘Xieyou963’). Average plant density in FFP and SSNM plots was 23 hills m-2 for ER in both 1998 and 1999. In 1998, LR was planted at an average density of 19 hills m-2 in FFP and SSNM, but the density differed by about 15% in 1999 (FFP: 19 hills m-2; SSNM: 22 hills m-2). Farmers did all water management and pest control in both FFP and SSNM plots following the commonly adopted methods. After an initial flooding period, the floodwater was drained off at about 25 d after transplanting (DAT), and the soil was kept drained for 7 to 10 d to halt tillering. Fields were then alternately shallow-flooded and drained until booting stage (about 50 DAT) when they were flooded again until past flowering (65 DAT) to avoid flowering-stage water stress. The soil was then drained again and kept saturated by occasional irrigation until maturity. No severe incidence of pests was observed during the experimental period.

Agronomic Measurements
Initial soil samples were collected in spring of 1997 to determine general soil properties in the 0.0- to 0.15-m depth. Composite samples of 15 soil cores per field were analyzed following standard procedures (van Reeuwijk, 1992). Plant sampling procedures followed a standard procedure at all experimental sites (Witt et al., 1999; Dobermann and Fairhurst, 2000). Two sampling areas (6 by 6 m in FFP and SSNM; 4 by 4 m 0-N, 0-P, and 0-K plots) were randomly selected in each treatment for replicated plant sampling. A 12-hill plant sample was collected at physiological maturity to determine yield components and nutrient concentrations in plant tissue. Grain yields were obtained from a central 5-m2 harvest area in each sampling plot at harvestable maturity and are reported at a standard moisture content of 0.14 kg kg-1 H2O fresh wt. Grain and straw subsamples from the 12-hill sample were dried to constant weight at 70°C. Straw yields were estimated from the oven-dry grain yield of the 5-m2 harvest area and the grain/straw ratio of the 12-hill sample. Nitrogen concentrations in grain and straw were measured by micro-Kjeldahl digestion, distillation, and titration (Bremner and Mulvaney, 1982). Tissue P was measured by the molybdenum-blue colorimetric method and tissue-K by atomic adsorption spectrophotometer after wet digestion (Walinga et al., 1995). Other measurements included chlorophyll meter (SPAD 502, Minolta, Ramsey, NJ) readings (SPAD values) of the uppermost fully expanded leaf in the SSNM and FFP treatments. Beginning at 20 DAT, 20 leaf readings per plot were averaged, and measurements continued in 7- to 10-d intervals until about 10 d after flowering.

Site-Specific Nutrient Management Approach
Details of the SSNM approach used are provided elsewhere (Dobermann and White, 1999; Witt et al., 1999; Dobermann et al., 2001). The SSNM approach applied in this case study focused on managing the spatial variation in the indigenous N, P, and K supply among individual rice fields and the temporal variability in crop N demand occurring within a field during a growing season. Soil and plant nutrient analysis indicated that nutrients other than N, P, and K did not limit rice growth. Therefore, SSNM mainly involved prediction of field-specific optimal fertilizer rates and development and implementation of a site-specific N management scheme that accounted for real-time variation in crop N demand at major growth stages of rice.

We used a modification of the QUEFTS (QUantitative Evaluation of the Fertility of Tropical Soils) model (Janssen et al., 1990; Smaling and Janssen, 1993; Janssen, 1998; Witt et al., 1999) to work out field-specific fertilizer recommendations for each farm at the beginning of each season. Information needed to estimate the total amount of N, P, and K to be applied included (i) climatic yield potential; (ii) yield goal; (iii) definition of the relationship between grain yield and nutrient uptake; (iv) recovery efficiencies of fertilizer N, P, and K; (v) field-specific estimates of the indigenous N, P, and K supply; and (vi) potential constraints to fertilizer use.

Based on previous crop simulation analysis for the Jinhua area, the climatic yield potential was set to 9 Mg ha-1 for ER and 10 Mg ha-1 for LR (Zheng et al., 1997b). Yield goals were constrained to a range of 70 to 80% of the climatic yield potential because beyond that level, internal nutrient efficiencies in the plant decline (Witt et al., 1999). Moreover, practical experience indicates that yields of about 80% of the climatic yield potential appear to represent a ceiling for what can be achieved by most farmers under field conditions (Cassman and Harwood, 1995). The generic empirical models proposed for rice by Witt et al. (1999) were used to model the relationship between grain yield and plant accumulation of N, P, and K. Average first crop recovery fractions of 0.4, 0.2, and 0.5 kg kg-1 were assumed for fertilizer N, P, and K, respectively. These numbers were based on values measured in good farms during the two crops grown in 1997 as well as data obtained from an accompanying field experiment. The potential supply of N, P, and K from soil and other indigenous sources was estimated as plant nutrient uptake in the nutrient omission plots. The indigenous N supply (INS) was defined as total plant N accumulation at maturity in 0-N plots, indigenous P supply (IPS) was defined as total plant P accumulation at maturity in 0-P plots, and indigenous K supply (IKS) was defined as total plant K accumulation at maturity in 0-K plots (Janssen et al., 1990). In 1998, INS, IPS, and IKS values measured during the 1997 LR crop were used as model inputs, and the target yield was set to 7.5 to 8.0 Mg ha-1 in ER and 7.5 to 8.5 Mg ha-1 in LR. In 1999, average INS, IPS, and IKS values measured in the 1998 ER and LR crops were used as model inputs, and the target yield was set to 7.2 to 8.0 Mg ha-1 in ER and 8 Mg ha-1 in LR. A linear optimization procedure was used to find the best combination of N, P, and K fertilizer rates to achieve the yield goal under the constraint of optimizing the internal N, P, and K efficiencies in the plant. The model was constrained to arrive at a solution close to the situation of most balanced nutrition, i.e., where the ratio between uptake and potential supply of each macronutrient was close to 0.95 (Janssen et al., 1990). Upper limits of 180 and 35 kg ha-1 N and P, respectively, were set to avoid predicting unrealistically high yields and fertilizer rates on soils with low fertility, assuming that application of N, P, and K alone cannot completely substitute for low inherent soil fertility. Excessive N rates would also increase the risk of pest damage or lodging. Lower limits of 10 to 15 kg P ha-1 and 30 kg K ha-1 were set as the minimum amount to be applied to replenish net removal from the field and minimize risk.

Field-specific, a priori N recommendations calculated using QUEFTS assume average climatic conditions and no or minimal stresses. Under field conditions, the actual climate varies, and some stresses such as water, pests, or mineral disorders cannot fully be excluded. Our goal was to gradually develop a N management strategy that allows adjusting N rates seasonally within each field. In the FFP, farmers typically applied all N in two splits of about 40% preplant and 60% within 7 to 10 DAT, and only few applied a third dose at later growth stages. The first N management scheme tested in the SSNM in 1998 ER and LR was one with fixed split applications at preset growth stages. Urea [(NH2)2CO] was applied in three splits—40% shortly before transplanting (incorporated), 20% topdressed at 7 to 14 DAT, and 40% topdressed at panicle initiation (PI). In 1999 ER and LR, we monitored plant N status with a chlorophyll meter (SPAD 502, Minolta, Ramsey, NJ) at preset critical growth stages at which N must be applied, but we varied the amount based on the actual plant N status:


N1
Preplant
40% of model-predicted fertilizer N rate
N2 7 to 14 DAT 20% of model-predicted fertilizer N rate
N3 35 to 45 DAT (PI) if SPAD > 36 30 kg N ha-1
if SPAD 33–36 40 kg N ha-1
if SPAD < 33 50 kg N ha-1
N4 50 to 55 DAT if SPAD > 36 0 kg N ha-1
if SPAD 33–36 20 kg N ha-1

This strategy accounted for variation in INS during early growth and variation in late-season N demand depending on the actual growth conditions. Late N at 55 DAT was only applied in cases with good crop stand to support the extra yield potential by adding more N for grain filling (Perez et al., 1996). Fertilizer sources used were urea (46% N), single superphosphate (6.1% P), and muriate of potash [KCl] (50% K). All P fertilizer was incorporated into the soil before transplanting (100% basal). Potassium fertilizer was split into 50% basal plus 50% at PI stage.

Calculations and Statistical Analysis
Nitrogen use efficiencies were estimated using the differences between N-fertilized treatments and the 0-N plots, as described by Cassman et al. (1998). Terms used are agronomic efficiency of applied N (AEN; kilogram grain yield increase per kilogram N applied), apparent recovery efficiency of applied N (REN; kilogram N taken up per kilogram N applied), and partial factor productivity of applied N (PFPN; kilogram grain per kilogram N applied). Economic calculations were made using U.S. dollars as standard currency:

(1)

(2)
where

TFC = total fertilizer cost ($ ha-1)

PN = price of N fertilizer ($0.40 kg-1 N)

FN = amount of N applied (kg N ha-1)

PP = price of P fertilizer ($1.11 kg-1 P)

FP = amount of P applied (kg P ha-1)

PK = price of K fertilizer ($0.36 kg-1 K)

FK = amount of K applied (kg N ha-1)

GRF = gross return above fertilizer cost ($ ha-1)

PR = price of rice ($0.16 kg-1 paddy)

YR = rice yield (kg ha-1) All prices used were average retail prices surveyed at Jinhua from 1997 to 1999. Because of the difficulties in imputing costs to family labor, it was not possible to calculate the absolute level of profit with and without SSNM. However, the incremental profitability of SSNM ({Delta}GRF, $ ha-1) was measured as the difference in gross returns due to different grain yields for SSNM and FFP minus the change in total fertilizer costs due to different fertilizer usage in the two treatments:

(3)

PROC GLM of SAS (SAS Inst., 1988) was used to perform ANOVA on the differences between SSNM and FFP ({Delta} = SSNM - FFP) measured at each farm for four consecutive rice crops using the following model:


Village
df = 6
Farm within village df = 16
Crop
Year 1 vs. Year 2 df = 1
ER vs. LR df = 1
Year x season df = 1
Village x crop df = 18
Residual df = 42

Crops in Year 1 included the 1998 ER and LR crop; crops in Year 2 included the 1999 ER and LR crops. Crop was partitioned into three orthogonal components (i.e., year, season, and year x season interaction), which allows for the specific testing of year or season main effects and the year x season interaction effect. A fixed-effects model was used to analyze the on-farm data because the sampling locations were not selected truly randomly. All effects, except village, were tested against the residual. Village effect was tested against farm within village as error term.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
GENERAL DISCUSSION AND...
 REFERENCES
 
Variation in Indigenous Nutrient Supply
Most farms had a clay loam to silty clay soil texture and relatively high soil organic matter content but an acidic pH and low to moderate cation exchange capacity (Table 1). Plant-available levels of soil K, Mg, Ca, P, and Zn were highly variable among the fields sampled (CVs 42–66%). Olsen-P content ranged from 8 to 61 mg P kg-1, but 80% of all farms had Olsen-P contents above the commonly proposed critical level of >10 mg P kg-1. Potassium extracted by 1 M ammonium acetate ranged from 0.12 to 0.85 cmolc kg-1, but one-third of all fields had <0.2 cmolc K kg-1, a critical level often used for rice soils with little K fixation (Dobermann and Fairhurst, 2000). However, plant-based indicators of the IKS indicated greater available K reserves than suggested by extraction with 1 M ammonium acetate (see below). In all fields, extractable Zn was above the commonly used critical level of 1 mg kg-1 (Ponnamperuma et al., 1981).


View this table:
[in this window]
[in a new window]
  		Table 1. General soil properties in 21 rice farms at Jinhua, Zhejiang, China.

 
The averages of INS, IPS, and IKS were similar in all crops sampled, but two-fold ranges were measured among the 21 rice fields in each season (Table 2). The INS ranged from about 50 to 115 kg N ha-1, IPS ranged from about 13 to 31 kg P ha-1, and IKS ranged from about 70 to 180 kg K ha-1. There was no consistent difference between indigenous nutrient supplies measured in ER or LR crops. Over the short term, the present average levels of indigenous N, P, and K supply would be sufficient for achieving rice yields of about 5 Mg ha-1 without applying N and 7.5 to 8 Mg ha-1 without applying P and K, assuming nutrient requirements of 14.7, 2.6, and 14.5 kg of N, P, and K, respectively, per 1000 kg grain yield (Witt et al., 1999). Among farms, CVs of INS, IPS, and IKS ranged from 14 to 30% in each cropping season. Among the four crops sampled in the same rice field, average standard deviations of indigenous nutrient supplies were 10, 3, and 17 kg ha-1 N, P, and K, respectively, which is equivalent to an average CV of about 15% for all three nutrients. This average estimation error (or crop-to-crop variability) of a field-specific measurement includes spatial variability due to rotating omission plots, climatic and crop management factors, different varieties grown in the same field in different seasons, and the errors associated with plant sampling and chemical analysis.


View this table:
[in this window]
[in a new window]
  		Table 2. Variation of the indigenous nutrient supply among 21 rice farms at Jinhua, China.

 
Grain Yield and Nutrient Uptake
Compared with FFP, SSNM significantly increased grain yields and N, P, and K uptake in the four crops grown in 1998 and 1999 (Table 3). The average yield difference between SSNM and FFP for the four crops grown was 0.5 Mg ha-1 (8%, P = 0.006) and was similar in the ER and LR crops. Seasonal differences in the performance were not statistically significant (crop season, P = 0.269). Yield increases over FFP were similar in 1998 and 1999 although both SSNM and FFP yields in 1999 (Year 2) were about 1 Mg ha-1 lower than in 1998 due to unfavorable climatic conditions. Heavy rains in mid-June during the flowering period of both ER crops caused low grain-filling percentages of 73% compared with about 87% in LR crops. Rice yields in 1999 were further lowered because of lower-than-normal solar radiation (ER) or late planting (LR). Yields were highest in the 1998 LR crop when the average yield for SSNM was 7.4 Mg ha-1. In five farms, yields in the SSNM exceeded 8 Mg ha-1, with a maximum of 8.7 Mg ha-1. In three farms, yield increases compared with FFP were >1 Mg ha-1, showing the potential of this approach.


View this table:
[in this window]
[in a new window]
  		Table 3. Effect of site-specific nutrient management (SSNM) on grain yield, plant nutrient accumulation, and fertilizer use in 21 rice farms at Jinhua, Zhejiang Province, China (1998–1999).

 
There were significant increases in plant N, P, and K accumulation in SSNM compared with FFP treatments. On average, plant N accumulation increased by 8 kg ha-1 (8%, P = 0.003), P accumulation by 2.4 kg ha-1 (14%, P = 0.002), and K accumulation by 12 kg ha-1 (10%, P = 0.014). Crop-season effects were not significant, i.e., similar increases in nutrient uptake were achieved in ER and LR crops (Table 3). However, increases in N uptake only became statistically significant in 1999 (Year 2) when N management in the SSNM was further improved by using the chlorophyll meter (Table 3).

Fertilizer Use
Compared with other regions in Asia, fertilizer use in the FFP at our study site was relatively high. Farmers applied average rates of 165 to 170 kg N ha-1 and 19 kg P ha-1 to both ER and LR crops, despite a 1 Mg ha-1 yield difference between these two climatic seasons (Table 3). However, K use in the FFP was 44 kg K ha-1 in ER compared with 63 kg K ha-1 in LR. In the LR season, most farmers grew hybrids and seemed to follow the recommendation that rice hybrids require a narrower N/K ratio for optimal performance (Yang et al., 1997). However, most farmers had no means of adjusting their fertilizer rates according to the actual soil fertility status. Correlations between N rate and INS were -0.34 in 1998 ER and -0.31 in 1998 LR, confirming similar observations made in other parts of Asia (Olk et al., 1999). Similarly, P rates were not significantly correlated with IPS (r = 0.06 in 1998 ER; r = 0.34 in 1998 LR), and K rates were not correlated with IKS (r = -0.33 in 1998 ER; r = 0.30 in 1998 LR). Moreover, there was little consistency in farmers' fertilizer N and P use. For the whole period of six rice crops sampled from 1997 to 1999, correlations of fertilizer rates between two successive cropping seasons ranged from -0.05 to 0.45 for N and 0.06 to 0.36 for P but were 0.16 to 0.62 for K. The latter may indicate that at least some farmers regularly followed recommendations for applying K.

On average, 34 kg ha-1 less fertilizer N was used in SSNM treatments than in FFP (-26%, P = 0.000), particularly in Year 2 (-61 kg N ha-1, Table 3). Crop year effects were all significant for N–P–K fertilizer applications, and the general trend was that rates in Year 2 were much smaller than in Year 1. For the 1999 ER crop, 56, 11, and 4 kg ha-1 less N, P, and K, respectively, were applied in the SSNM treatment compared with FFP. For the LR crop of the same year, 63, 7, and 21 kg ha-1 less N, P, and K, respectively, were applied in the SSNM treatment than in FFP. Lower fertilizer rates in the SSNM treatment resulted from model-based predictions that accurately accounted for the high native soil fertility status measured as plant nutrient uptake in 1997 and 1998 omission plots.

Nitrogen Use Efficiency
Significant increases in N use efficiency were achieved through the field- and season-specific N management practiced in the SSNM treatment (Table 4). In general, compared with the FFP, less N fertilizer was applied and AEN, REN, and PFPN were significantly increased. Across all four crops grown, AEN increased by 5 kg kg-1 (78%, P = 0.000), REN by 0.11 kg kg-1 (61%, P = 0.000), and PFPN by 12 kg kg-1 (33%, P = 0.000). Differences in the impact of SSNM on N use efficiency between ER and LR were not significant. However, the differences in AEN and REN between SSNM and FFP treatments increased with time and were significantly greater in 1999 than in 1998 (Crop year effect, Table 4). This indicates that the gradual fine-tuning of N management to the local conditions, including the use of a chlorophyll meter, was successful.


View this table:
[in this window]
[in a new window]
  		Table 4. Effect of site-specific nutrient management (SSNM) on fertilizer N use efficiency, fertilizer cost, and gross return above fertilizer (GRF) cost in 21 rice farms at Jinhua, Zhejiang Province, China (1998–1999).

 
Compared with the FFP, N applications in the SSNM treatment were more uniform among farms, spread more evenly through the growing season, and avoided heavy single applications at early growth stages (Fig. 3). For the 1999 ER and LR crops, almost all farmers applied two large doses of about 40 to 160 kg N ha-1 each during the first 2 wk after planting but hardly any N thereafter. Many farmers used N fertilizer to stimulate tillering, particularly in LR crops planted at low density. In addition, midseason drainage from about 25 to 35 DAT (maximum tillering to PI stage) is difficult to control and probably was a major cause for large N losses in the FFP. In the SSNM treatment, preplant N application was much smaller than in the FFP treatment. The topdressed N application at 7 to 14 DAT was also small (about 30 kg N ha-1), but a third dose of 30 to 50 kg N ha-1 was applied between 20 and 55 DAT, with the date and amount depending on plant N status. For example, the ER crop in 1999 had poor weather so that the climatic yield potential was low and N accumulated in the rice plants. Nevertheless, the farmers applied their usual high amounts of N (average 170 kg N ha-1), all at early stages when they could not know the weather conditions and crop yield potential in advance. In contrast, only 115 kg N ha-1 was applied in the SSNM treatment based on knowledge of the INS and SPAD readings at critical growth stages, resulting in AEN and REN almost twice that of FFP (data not shown).



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 3. Fertilizer N applications in the farmers' fertilizer practice (FFP) and site-specific nutrient management (SSNM) plots during 1999 early (ER) and late (LR) rice crops.

 
Our results provide on-farm evidence that current N management practices in southeast China are inconsistent with the physiological N requirements of the rice crop and lead to large N losses. Nitrogen supply appears to be excessive during early vegetative growth but deficient during grain filling. During early growth, chlorophyll meter readings in the FFP plots were mostly larger than those in SSNM plots, but the reverse was true during reproductive growth stages (Fig. 4). The importance of sufficient late-season N supply for achieving high rice yields has been highlighted in other studies (Perez et al., 1996; Peng et al., 1996b; Peng and Cassman, 1998). Hybrid rice, grown in the LR season at our site, takes up 15 to 20% of the total plant N accumulation after heading, and this capacity to utilize late-season N is seen as one of the reasons for the yield advantage over inbred varieties (Yang et al., 1999). Other studies in China have shown that applying more N fertilizer during middle growth stages improved N use efficiency and increased N uptake and grain yields (Zheng et al., 1997a). At the same site in Jinhua, in an experiment conducted in 1994 for LR, skipping the basal N application and applying N in four splits during 8 to 38 DAT increased the grain yield from 6.0 to 6.4 Mg ha-1 and AEN from 7.4 to 11.3 kg kg-1 compared with the current recommended practice (ten Berge et al., 1997).



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 4. Difference between chlorophyll meter readings ({Delta}SPAD) in site-specific nutrient management (SSNM) and farmers' fertilizer practice (FFP) during early vegetative growth and late reproductive growth. Values shown are for the 1999 early (ER) and late (LR) rice crops.

 
Profitability of Site-Specific Nutrient Management
Site-specific nutrient management led to a reduction of the average fertilizer cost by $15 ha-1 crop-1 and an increase in GRF by $88 ha-1 crop-1 (10%, P = 0.001) (Table 4) compared with FFP. Increases in GRF were significantly larger in 1999 ($108 ha-1 crop-1, 14%) than in 1998 ($69 ha-1 crop-1, 8%) and larger for LR ($103 ha-1 crop-1) than for ER ($74 ha-1 crop-1), partly because hybrid rice varieties with longer growth periods and higher yield potential were grown in the LR season. Profit increases continuously rose over time, suggesting a gradual improvement of the SSNM strategy (Fig. 5). Attributing meaning to the calculation of {Delta}GRF (Eq. [3]) implicitly assumes that the only difference in crop management between SSNM and FFP is different quantities of nutrients and different timing of a certain constant number of applications so that all other management practices and quantities of input use are held constant.



View larger version (25K):
[in this window]
[in a new window]
  		Fig. 5. Difference in total fertilizer cost ({Delta}TFC) and gross return above fertilizer cost ({Delta}GRF) between the farmers' fertilizer practice (FFP) and site-specific nutrient management (SSNM) in 21 rice farms at Jinhua, China during early rice (ER) and late rice (LR) crops in 1998 and 1999 (means ± standard error). The P values shown indicate the probability of a significant mean difference in GRF between FFP and SSNM in each season.

 
At issue is to what extent slight differences in crop management occurred and how they would affect the profitability of SSNM. First, a cost is associated with obtaining field-specific estimates of INS, IPS, and IKS. The combined size of these plots for an individual field was <0.03 ha, and yield losses would mainly occur in 0-N plots. Assuming that a broader use of the SSNM approach will probably not be based on placing such an omission plot into each rice field (see below), the cost per hectare appears to be negligible. Secondly, in the 1999 LR crop, rice in the SSNM plot was planted at about 15% greater density than in the FFP plot, mainly to increase N use efficiency during early growth. However, considering the high tillering ability of hybrid rice, the effect on yield was probably not significant. If adoption of SSNM would require planting LR at a density of 22 hills m-2 rather than 19 hills m-2, the extra seed cost would amount to about $4 ha-1. Planting at the greater density would require about four 8-h labor days ha-1 more than under FFP, which is equivalent to an extra cost of about $21 ha-1. Thus, the total additional crop establishment cost in the 1999 LR season was $25 ha-1, which compares to an average {Delta}GRF of $132 ha-1 measured in the same crop. Thirdly, the real-time N management approach used for SSNM is associated with an extra cost. Using a chlorophyll meter or a simple leaf color chart (Tao et al., 1990) to gather information about crop N status requires about 0.5 h per field. However, if this can be done by paid crop consultants or local extension aides, the cost per hectare becomes small, probably well below $5 ha-1 per crop cycle.

Another issue is labor for applying N fertilizer because SSNM was often associated with an extra topdressing of N. For example, in 1999, the average number of N applications was 2.4 for FFP vs. 3 for SSNM (Fig. 4). Assuming that it takes one person about 3 h to apply N on 1 ha, the additional cost compared with FFP is <$2 ha-1. In summary, although SSNM was associated with an additional cost, those expenses were far below the increase in GRF measured for all four crops.


   GENERAL DISCUSSION AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND...
REFERENCES
 
Because of the opportunity cost of farmers' time, widespread adoption of new technologies will be greatly facilitated if they either have very large financial advantages or are relatively simple to implement. The SSNM approach significantly increased grain yield and nutrient uptake with less fertilizer applied, resulting in large increases in N use efficiency and profit. What determines the performance of SSNM at the farm level and how can this approach be implemented in larger areas?

Increases in N use efficiency and profit by SSNM were larger in 1999 than in 1998. Compared with 1998, the main improvements in 1999 included availability of better estimates of indigenous nutrient supply and an improved N management algorithm that included a chlorophyll meter for decision making. However, despite an overall 60 to 80% increase, N use efficiencies obtained with SSNM remained below AEN of >20 kg kg-1 and REN of >0.5 kg kg-1 typically achieved in irrigated rice with good crop management (Peng et al., 1996a; Peng and Cassman, 1998). Moreover, average yields under SSNM were 82% of the yield goal in the 1998 ER, 92% in 1998 LR, 69% in 1999 ER, and 80% in 1999 LR. This unattained yield gap of 8 to 31% was mainly caused by (i) climatic factors, (ii) poor water management, and (iii) insufficient planting density due to labor shortages. Little can be done about climatic variability, except for optimizing planting dates and adopting a real-time N management approach as was attempted in our study. Future research should focus on further fine-tuning of the N application regime so that it becomes fully synchronized with local water and pest management practices and vice versa. Furthermore, as cropping systems and management operations change over time, knowledge-intensive forms of nutrient management must be updated, particularly with regard to the timing and amount of N applications. For example, beginning in 2000, many farmers at Jinhua started using direct seeding of ER to overcome the labor problems causing nonoptimal plant densities, which in turn will require a modification of the optimal N management strategy to account for principal differences between transplanted and direct-seeded rice (Dobermann and Fairhurst, 2000).

It remains an open question whether results similar to those presented here can be achieved over large areas without field-specific modeling because such modeling cannot be done for more than a tiny fraction of rice fields in Zhejiang or elsewhere. Presumably, it will not be necessary to measure indigenous nutrient supplies on a field-specific basis because the cost associated with this would be too high and our data indicate that the achievable precision is only about ± 15%. Instead, using a limited number of categories—each representing a range of INS, IPS, and IKS values—will probably be sufficient to work out site-specific fertilizer regimes using simple decisions aids such as a fertilizer chart derived from a model such as QUEFTS. Understanding the natural and anthropogenic causes of the spatial and temporal variation in indigenous nutrient supplies among rice farms and across larger regions will be required to establish such categories for specific rice-growing domains. Combined with locally adopted forms of real-time N management, the performance of a simplified approach is likely to be similar to that obtained in our study as long as major principles of yield potential and nutrient interactions are honored as driving factors for crop nutrient demand. Further studies in larger areas must be conducted to verify this assumption.


   ACKNOWLEDGMENTS
 
We acknowledge contributions made by other scientists involved in this research, including Professor Huang Changyong and Dr. He Yunfeng of Zhejiang University and Mr. Xianghai Ding, Ms. Xueping Huang, and Mr. Jiangxiang Wu of the Agricultural Bureau of Jinhua. We are grateful to the farmers in Jinhua County for their patience and excellent cooperation in conducting the on-farm experiments since 1997. We wish to thank David Dawe for comments on the economic analysis, Ma. Arlene Adviento for help with the soil and plant analysis work, and Gregorio C. Simbahan (all IRRI) for help with data management and statistical analysis. Funding for this research was provided by the International Fertilizer Industry Association (IFA), the Potash and Phosphate Institute (PPI/PPIC), the International Potash Institute (IPI), and the Swiss Agency for Development and Cooperation (SDC).


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 GENERAL DISCUSSION AND...
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
P. He, S. Li, J. Jin, H. Wang, C. Li, Y. Wang, and R. Cui
Performance of an Optimized Nutrient Management System for Double-Cropped Wheat-Maize Rotations in North-Central China
Agron. J., November 1, 2009; 101(6): 1489 - 1496.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
X. Q. Liang, H. Li, M. M. He, Y. X. Chen, G. M. Tian, and S. Y. Xu
The Ecologically Optimum Application of Nitrogen in Wheat Season of Rice Wheat Cropping System
Agron. J., January 11, 2008; 100(1): 67 - 72.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
H. S. Khurana, S. B. Phillips, Bijay-Singh, A. Dobermann, A. S. Sidhu, Yadvinder-Singh, and S. Peng
Performance of Site-Specific Nutrient Management for Irrigated, Transplanted Rice in Northwest India
Agron. J., October 15, 2007; 99(6): 1436 - 1447.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
A. Dobermann, C. Witt, S. Abdulrachman, H. C. Gines, R. Nagarajan, T. T. Son, P. S. Tan, G. H. Wang, N. V. Chien, V. T. K. Thoa, et al.
Soil Fertility and Indigenous Nutrient Supply in Irrigated Rice Domains of Asia
Agron. J., July 1, 2003; 95(4): 913 - 923.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
A. Dobermann, C. Witt, S. Abdulrachman, H. C. Gines, R. Nagarajan, T. T. Son, P. S. Tan, G. H. Wang, N. V. Chien, V. T. K. Thoa, et al.
Estimating Indigenous Nutrient Supplies for Site-Specific Nutrient Management in Irrigated Rice
Agron. J., July 1, 2003; 95(4): 924 - 935.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (24)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Agricola
Right arrow Articles by Wang, G.
Right arrow Articles by Fu, R.
Related Collections
Right arrow Economics
Right arrow Rice
Right arrow Nutrient Management
Right arrow Site-Specific Analysis
Right arrow Irrigation

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Crop Science Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome