| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,*,a
a Int. Rice Res. Inst. (IRRI), DAPO Box 7777, Manila, Philippines
b Res. Inst. for Rice (RIR), Sukamandi, Indonesia
c Philippine Rice Res. Inst. (PhilRice), Maligaya, Nueva Ecija, Philippines
d Natl. Inst. for Soils and Fert. (NISF), Hanoi, Vietnam
e Cuu Long Delta Rice Res. Inst. (CLRRI), Omon, Cantho, Vietnam
f Zhejiang Univ. (ZU), Hangzhou, P.R. China
g Tamil Nadu Rice Res. Inst. (TNRRI), Aduthurai, Tamil Nadu, India
h Soil and Water Manage. Res. Inst. (SWMRI), Thanjavar, Tamil Nadu, India
* Corresponding author (adobermann2{at}unl.edu)
Received for publication July 2, 2002.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: AD, Aduthurai • HA, Hanoi • HYS, high-yielding season • IKS, indigenous potassium supply • INS, indigenous nitrogen supply • IPS, indigenous phosphorus supply • IS, indigenous supply of a nutrient • JI, Jinhua • LYS, low-yielding season • MA, Maligaya • OM, Omon • SSNM, site-specific nutrient management • SU, Sukamandi • TH, Thanjavur • 0-K, potassium omission (plot) • 0-N, nitrogen omission (plot) • 0-P, phosphorus omission (plot)
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
At issue is whether more quantitative yet simple approaches for estimating fertilizer needs can be adopted for practical use. The potential supply of a nutrient can be defined as the cumulative amount of that nutrient originating from all indigenous and exogenous sources that circulates through the soil solution surrounding the entire root system during one complete crop cycle (Janssen et al., 1990). Likewise, the proportion of potential supply of a nutrient derived from all indigenous (nonfertilizer) nutrient resources is then defined as the effective IS. Knowledge of IS would allow the calculation of fertilizer rates based on the general equations (Dobermann and Cassman, 2002)
 | [1] |
 | [2] |
In an irrigated system, IS includes plant available nutrients derived from sources such as (i) chemical and biological transformations of soil solids, (ii) biological N2 fixation in the floodwater–soil system, (iii) atmospheric deposition, and (iv) solutes and sediments deposited by irrigation and/or naturally occurring flooding (Cassman et al., 1998; Dobermann et al., 1998). Assuming that the rate of nutrient uptake is not limited by the plant itself, but mainly by the rate of nutrient delivery to the root surface, the INS can be measured as plant N accumulation in a 0-N plot, which receives P, K, and other nutrients but no fertilizer N. Likewise, IPS can be measured as plant P accumulation in a 0-P plot, which receives N, K, and other nutrients, and IKS can be measured as plant K accumulation in a 0-K plot, which only receives N, P, and other nutrients. These assumptions are only fully valid for growing seasons with high yield potential in which climatic factors are less limiting the crop's potential to act as a sink for nutrients.
Because dry matter and nutrient uptake by roots are related to the aboveground biomass and are difficult to measure, IS can also be approximated by only measuring plant nutrient accumulation with the aboveground biomass in a nutrient omission plot. This approach was successfully used in recent studies on site-specific nutrient management (SSNM) in irrigated rice (Wang et al., 2001; Dobermann et al., 2002). However, measurement of nutrient uptake in omission plots is not feasible on a routine basis because it involves destructive plant sampling and plant tissue analysis. Cost is high and errors are associated with sampling, sample processing, and analysis. Therefore, alternative methods for estimating IS in SSNM must be found and calibrated against measured IS as the key reference.
A first option is to estimate INS, IPS, and IKS from information such as soil tests and climate. Rapid soil tests do not extract the exact amount of nutrient that is available for crop uptake, but they provide a relative measure of nutrient supply. Traditionally, soil tests have been used to empirically estimate the yield response to fertilizer application (Dahnke and Olson, 1990). Alternatively, empirical calibration of soil test values against nutrient uptake measured in omission plots would allow estimating IS, and this approach has been used for upland crops such as maize (Zea mays L.) (Janssen et al., 1990; Smaling and Janssen, 1993).
A second alternative is to only measure grain yield in 0-N, 0-P, and 0-K plots. Grain yield measurements are conducted using larger harvest areas, and they are easy to teach to farmers or extension staff. For developing fertilizer recommendations, grain yield in an omission plot can be directly used as intercept (Y0) in a yield–fertilizer application response curve, and fertilizer needs can be estimated from the difference between target yield and intercept (
Y) by assuming a certain amount of plant nutrient uptake per unit yield increase (Witt et al., 2002). If the quantitative approach described in Eq. [1] and [2] is used, grain yields must be converted into total plant accumulation of the nutrient omitted (IS) using empirical models. The latter is feasible if a generic relationship between grain yield and total plant accumulation of the nutrient omitted can be established (Witt et al., 1999).
Magnitudes and causes of spatial and temporal variability in INS, IPS, and IKS in irrigated rice environments were discussed previously (Dobermann et al., 2003). Using the same data, the objectives of this paper are to (i) evaluate relationships among grain yield, plant nutrient accumulation, and soil tests in nutrient omission plots and (ii) develop guidelines for use of nutrient omission plots in SSNM. Practical utilization of simple crop-based estimates of IS such as grain yield in a nutrient omission plot requires clarification of several key issues. What is the achievable precision of estimating IS? How many years (crops) and replicated omission plots are needed to estimate the average IS for a field? If the goal is to develop fertilizer recommendations for larger domains with similar biophysical and socioeconomic characteristics, how many years, fields (or farms) within the domain, and omission plots within a field are needed to estimate the average IS for such a domain? Can an unfertilized plot (-F plot) be used to estimate INS instead of using a true 0-N plot in which P, K, and other nutrients are applied? What is the best strategy for using omission plots in the management of N, P, and K?
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
At each farm, a single rice field served as the principal experimental unit. Omission plots were established as duplicate or triplicate sets in each field and separated from the surrounding field by bunds. Omission plots were moved to a different location within the same field after each crop. Nitrogen omission plots, to which either no fertilizer (-F plot) or P and K (+PK) were applied, were sampled for eight consecutive rice crops grown from 1997 to 2001 at each site to estimate INS. From 1995 to 1996, paired -F and +PK plots were sampled for three to four rice crops in all farms at AD, MA, OM, and SU to compare estimates of INS using both types of 0-N plots. Phosphorus omission plots, to which N and K were applied, were sampled for three to four consecutive rice crops grown from 1997 to 1999 to estimate IPS. Potassium omission plots, to which N and P were applied, were sampled for three to four consecutive rice crops grown from 1997 to 1999 to estimate IKS.
In two to four cropping seasons at each site, soil samples were collected at 20 to 30 d after planting from the nutrient omission plots and analyzed for organic C (Walkley, 1947) and total N (Bremner, 1996) in 0-N plots, 0.5 M NaHCO3 extractable P (Olsen et al., 1954) and Bray-I P (Bray and Kurtz, 1945) in 0-P plots, and 1 M ammonium acetate extractable K (van Reeuwijk, 1992) in 0-K plots. Initial soil data and those from different cropping seasons were pooled to obtain averages of extractable nutrients for each field. Plant sampling included measurement of grain yields (expressed at 0.14 g H2O g-1 fresh weight), plant N accumulation in 0-N plots, P accumulation in 0-P plots, and K accumulation in 0-K plots, all at physiological maturity of rice.
Data Analysis
For statistical analysis, the two main rice seasons at each site were classified as either high-yielding seasons (HYS) or low-yielding seasons (LYS). At the tropical sites, the HYS is usually a dry season with low rainfall and high solar radiation, whereas the LYS is a wet (monsoon) season with lower yield potential due to cloudy conditions and high rainfall. Sukamandi was the only exception to this rule because dry-season yields at this site are lower than wet-season yields so that the latter was defined as HYS. For the two subtropical sites, the HYS is the late rice crop in Zhejiang and the early (spring) rice crop in northern Vietnam. Linear regression analysis was used to study relationships between various variables measured.
The PROC MIXED procedure of the SAS/STAT software (SAS Inst., 1999) was used to perform analysis of variance (ANOVA) for plant nutrient uptake and grain yield measured in omission plots across all sites and seasons. For 0-N plots, this analysis included measurements of four HYS crops sampled from 1997 to 2001. For 0-P and 0-K plots, the ANOVA was based on two HYS crops sampled at each site from 1997 to 1999. A mixed-model ANOVA (Littell et al., 1996) was conducted separately for each site to estimate the precision of the domain mean as a function of the number of farms (fields) sampled, years, and replicate omission plots within a field. All effects in this ANOVA model were treated as random in PROC MIXED. If Yijk is the nutrient uptake or grain yield measured for the kth omission plot in crop (or year) j in field i, then a general ANOVA model can be written as
 | [3] |
, Eq. [3] can be rewritten as:
 | [4] |
The variance of the domain mean 
can then be expressed as a function of the various variance components obtained from the ANOVA:
 | [5] |
The relative standard error (or precision) of the domain mean is then estimated as:
 | [6] |
Equations [5] and [6] were used to (i) calculate the relative standard error SE% as a function of f and p with y fixed at two HYS crops (years) and (ii) calculate SE% as a function of f and y with p fixed at one omission plot per field. The number of fields needed to estimate the domain mean with a 5 and 10% precision using one or two replicate nutrient omission plots per field and two HYS crops were also calculated based on (i).
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Soil K extracted by 1 M ammonium acetate showed no common relationship with plant K uptake or grain yield in 0-K plots (Fig. 1c and 1f). Within domains, correlation coefficients ranged from -0.23 to 0.69 for K uptake or -0.37 to 0.66 for grain yield in 0-K plots. Statistically significant correlations were only found for grain yield at HA and SU (r = 0.53–0.66, P < 0.01) or K uptake at JI and MA (r = 0.61–0.69, P < 0.01). Only up to 23% of the variation in IKS could be explained from measured soil properties. Multiple regression including other soil properties such as clay content, organic matter, cation exchange capacity, or pH did not improve the predictability of INS, IPS, or IKS significantly.
Relationships between Plant Nutrient Accumulation and Grain Yield
Plant N accumulation in 0-N plots was closely correlated with grain yield (R2 = 0.62, Fig. 2) using an average regression coefficient of 13.4 kg plant N Mg-1 grain yield. This compares to 14.7 kg N Mg-1 proposed for a situation of balanced plant nutrition or current farm averages of about 17 kg N Mg-1 (Witt et al., 1999) and suggests N limitation in the 0-N plots. The regression coefficients differed slightly between HYS and LYS, and differences occurred among sites (Table 1). At AD and TH, for example, regression coefficients were 10.6 to 11.4 kg N Mg-1, and standard errors of prediction were less than 7 kg N ha-1, probably because crop growth and yield in 0-N plots were primarily limited by N, resulting in strong dilution of N in the plant. In contrast, at SU, average grain yield in 0-N plots was low (3.12 Mg ha-1) even though 17.5 kg N Mg-1 was taken up. Other factors such as climate; soil constraints other than N, P, or K supply; insufficient plant density; and pests decreased the internal efficiency of N in the plant at this site.
|
|
|
The relationship between plant K accumulation and grain yield in 0-K plots was most scattered (R2 = 0.24, SE = 26.3 kg K ha-1), and large site differences occurred (Table 1). At sites with low soil K status and K deficiency in the 0-K plots (HA and TH), only 11.3 to 13.1 kg K Mg-1 was taken up by the plant. At other sites, K supply in 0-K plots did not limit yield at present levels, resulting in average K accumulation in the plant of more than 16 to 17 kg K Mg-1 and more scattered relationships between K uptake and yield. Standard errors for predicting IKS were lowest at AD and TH (9.8–10.8 kg K ha-1). Using total dry matter (grain plus straw) instead of grain yield did not significantly improve prediction of INS, IPS, or IKS (data not shown). About 80% of the aboveground plant K is contained in vegetative plant parts of rice (Witt et al., 1999) where it is mainly stored in the central vacuoles of cells (Marschner, 1995). Unless there is a deficiency, vacuolar K+ or straw K concentrations continue to respond to variations in external K supply but have little effect on K concentrations in the grain or grain yield itself (Dobermann et al., 1996c). Therefore, estimates of IKS based on grain yield measurement tend to be less accurate than those for nutrients that are mostly stored in the grain (N and P).
Attainable Precision and Sampling Requirements
Considering the greater uncertainties and potential underestimation of IS associated with LYS crops (Dobermann et al., 2003), nutrient omission plots should be primarily established in the HYS to obtain an estimate of the potentially available nutrient supply. At issue then is how many HYS crops (years), fields within a domain, or plots within a field need to be sampled to obtain a representative estimate of the mean IS for a specific field or for a larger domain. Variance components and sampling requirements were similar for nutrient uptake measurements and grain yields in omission plots so that only grain yield data will be discussed below. It should also be noted that statistical analysis was conducted for the rice domains as they were sampled at each site, including different within-domain uniformities and a considerable amount of stratified spatial variation within some domains (e.g., at HA, SU, and MA). Therefore, the sampling requirements described below should be considered as preliminary results. They varied from site to site, depending on how domains were delineated and sampled.
The largest proportions of variance were generally due to variation among fields, years, and their interaction, but site differences occurred (Table 2). At HA and SU, for example, more than 50% of the variation in HYS grain yield in 0-N was due to farm effects, which reflects the distinctively different soil types included in these domains. In contrast, farms at AD and OM contributed only 3 to 4% to the overall variation in 0-N plots because less variation in soil types and cropping practices occurred within these domains. At many sites, most of the variation in grain yield in 0-P and 0-K was caused by year and farm x year effects, probably because the data set included 0-P and 0-K plots for only two HYS crops (compared with four HYS measured for 0-N) causing greater relative variability due to climate.
|
|
|
At many sites, sampling of just one HYS crop in 5 to 10 farms would allow estimating the domain mean IS with about ±10% precision, which may be sufficient for the purpose of working out a domain-specific fertilizer recommendation. However, significant site differences in sampling requirements occurred, which were related to the uniformity of the domains sampled in our study. At JI, only five farms would need to be sampled for one HYS crop to estimate the domain IS means of all three nutrients at a precision of at least ±11% (Table 2) because farms differed less in soil types, cropping systems, and crop management practices. At MA, however, sampling five farms for 1 yr would only result in a precision of 17 to 20% for grain yield in 0-N, 0-P, and 0-K because the domain sampled was more heterogeneous and clustered into three villages with different soil types, planting dates (yield potential), and crop establishment methods. Similarly, at HA, precision was low for estimating IKS (Table 2) because the domain as sampled here included 12 farms on degraded soils and 12 farms on more fertile alluvial soils.
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The soil tests evaluated in this study were poor predictors of IS or rice grain yield measured in nutrient omission plots. The achievable precision of estimating INS, IPS, or IKS was mostly less than that of measuring grain yield in an omission plot (Fig. 1 and 2). Key factors that cause poor predictability of IS from soil tests in irrigated rice environments include (i) inherent soil variation; (ii) variations in fallow periods, tillage method and depth, water and crop residue management that affect microbial activity, organic matter quality and turnover, chemical soil processes, and root growth (Willett, 1989; Buresh et al., 1993; Olk et al., 1996; Kundu et al., 1996; Witt et al., 2000; Bucher, 2001); (iii) nonsymbiotic biological N2 fixation at rates of about 30 to 50 kg N ha-1 per crop (Kundu and Ladha, 1995; Roger, 1996); (iv) variable nutrient input through irrigation and sediments (Dobermann et al., 1998); and (v) fixation and release of NH4+ and K+ (Keerthisinghe et al., 1985), including effects of redox conditions on those processes (Schneiders and Scherer, 1998). Considering this, it is no surprise that commonly used rapid soil tests often fail to predict crop nutrient uptake from indigenous sources under field conditions even though correlations with soil nutrient release and plant growth may be high in laboratory or greenhouse studies with irrigated rice.
There is little evidence for a consistent association between soil organic C or total soil N and the amount of plant N acquired from indigenous sources under field conditions in tropical lowland rice systems (Cassman et al., 1998). Other soil tests based on chemical extractions or soil N mineralization assays have been studied widely (Sahrawat, 1983; Wilson et al., 1994; Bronson et al., 2001), but soil N tests typically explain less than 20 to 40% of the total variation in INS or grain yield in 0-N plots under field conditions (Dolmat et al., 1980; Bajaj, 1984; Zhu et al., 1984; Manguiat et al., 1994; Stalin et al., 1996). Laboratory measures of soil N release also have limited potential for practical use because temperature effects on in-season N mineralization, inputs of N from sources other than mineralization of soil organic matter, and sensitivity of soil N dynamics to soil drying, length of fallow period, crop rotation, and residue management are major factors affecting INS under field conditions (Shiga and Ventura, 1976; Cassman et al., 1996). Predictions of N mineralization and INS can potentially be improved by including other factors such as soil properties or temperature into empirical models (Sahrawat, 1983; Inubushi et al., 1985), but such data are rarely available for practical nutrient management.
Relationships between soil test P and K and plant nutrient uptake in lowland rice soils are strongly affected by clay content, clay mineralogy, base saturation, and redox conditions, causing wide ranges of critical soil test levels reported in the literature (Dobermann et al., 1998). Correlations between soil tests and plant nutrient uptake in long-term field experiments with rice across a similar range of environments were better than those obtained in our on-farm studies. Olsen P explained 43% of P uptake but up to 65% when other variables were added to the regression (Dobermann et al., 1996b). Ammonium acetate extractable K explained 49% of K uptake but up to 88% when other soil properties such cation exchange capacity of the clay fraction, K saturation percentage, and exchangeable Ca and Mg were added to the regression (Dobermann et al., 1996a). However, most of these additional soil properties are not routinely measured, and the large treatment contrasts built up in long-term experiments are likely not representative of on-farm conditions with varying input use over time.
In summary, rapid laboratory soil tests or field test kits may allow placing rice soils into different categories of extractable nutrients, but these categories are not necessarily related to the actual IS under field conditions. This questions the usefulness of such soil tests as a basis for making fertilizer decisions in lowland rice. Some improvements may be possible through more location-specific soil test calibration or dynamic soil tests that attempt to predict patterns of nutrient release over time (Wanasuria et al., 1981; Sahrawat, 1983; Manguiat et al., 1994; Dobermann et al., 1996a, 1997). These techniques are, however, laborious and currently not used in routine analysis. They also do not account for all nutrient sources that determine IS during the whole growing season, and the existing infrastructure for soil analysis is insufficient for a wider-scale adoption of such techniques in south and southeast Asia.
Crop-Based Estimates of Indigenous Nutrient Supply in Rice
The key advantage of any crop-based IS measurement is that it integrates the supply of truly plant-available nutrient forms across the effective rooting depth under field conditions. This includes readily available nutrient pools that are typically extracted with soil tests but also more slowly available nutrient pools and nutrients supplied from other indigenous sources in flooded rice fields. Crop-based estimates of IS result in a measurement unit that is directly usable for calculating fertilizer needs in absolute terms rather than mass-based soil test units, such as milligrams per kilograms of soil, that require empirical calibration for conversion. They also account for plant-mediated processes to acquire nutrients through chemical and microbial processes in the rhizosphere, which play an important role for nutrient uptake by rice (Kirk and Saleque, 1995; Reichardt et al., 2000; Kirk, 2001).
A potential uncertainty of crop-based estimates of IS is that the nutrient requirements of a crop could be underestimated when measuring grain yield or nutrient uptake in aboveground biomass because the storage of nutrients in the root system are neglected (Appel, 1994). The IS measured as crop uptake in an omission plot may also be larger than that in a fertilized plot due to enhanced root growth under nutrient deficiency (Morita and Yamazaki, 1993). However, nutrient accumulation by the root system can be ignored if the difference in root nutrient accumulation between unfertilized and fertilized plots is negligible. There is some evidence that this is true for shallow-rooting crops such as irrigated lowland rice grown on fine-textured soils in the tropics. In field studies, root N accumulation at physiological maturity of rice accounted for only 2 to 15% of the total above- and belowground plant N (Wopereis et al., 1994). Root biomass was relatively insensitive to increases in aboveground biomass due to fertilizer N application (Witt et al., 2000), and indirect effects of applied N fertilizer on uptake of indigenous N were small (Schnier, 1994; Cassman et al., 1998). Genotypic differences in root growth and nutrient uptake may also exist among modern rice cultivars, but they tend to be small on fertile soils and under irrigated conditions (Teo et al., 1995).
The general relationships shown in Fig. 2 and Table 1 can be used to obtain estimates of INS, IPS, or IKS based on grain yield measurements in nutrient omission plots. Linear regressions were mainly used for practical reasons. It should be noted, however, that the internal nutrient utilization decreases as yields approach the genetic–climatic yield potential, causing a more curvilinear relationship between grain yield and plant nutrient accumulation at high yield levels (Witt et al., 1999). With good local calibration and conduct of omission plots, INS, IPS, and IKS can be estimated from grain yields with a precision of about ±5 to 10 kg N ha-1, ±2 to 3 kg P ha-1, and ±10 to 20 kg K ha-1, respectively. Site differences may represent factors that are inherent to a site (climate and soil) but also crop management and genotypic variation in plant nutrient utilization or harvest index.
Using Nutrient Omission Plots for Site-Specific Nutrient Management
Different crop-based concepts for nutrient management have recently been proposed for rice. Some include measurement of IS in omission plots for making field-specific preplant decisions on amounts of N, P, and K to apply (Dobermann et al., 2002), whereas others are restricted to postemergence management of N using leaf N diagnostics, without preseason determination of INS (Peng et al., 1996; Balasubramaniam et al., 1999). Work is ongoing to combine these approaches into a simple, modular framework for SSNM (Witt et al., 2002). Depending on the local conditions and the nutrient of interest, SSNM can be of field-specific nature (e.g., N) or include domain-specific fertilizer recommendations (e.g., P and K). The following practical guidelines can be given for the use of nutrient omission plots:
1. Because estimates of IS are inherently variable under field conditions, simplified SSNM approaches should include only a few broad categories of IS as the basis for determining fertilizer rates. Their local interpretation should include a minimum amount of soil and other data, including indigenous farmer knowledge.
2. Nutrient omission plots should be part of an extension strategy because they offer possibilities for estimating fertilizer needs and visualizing nutrient deficiencies and their effects on crop growth and pest infestation.
3. Measurements of grain yield in nutrient omission plots may provide a better index of the expected yield response to fertilizer application than measurements of IS (nutrient uptake) because site-specific factors affecting yield other than nutrients are accounted for and the measurement error is small. Grain yields in different nutrient omission plots were highly correlated with each other and also with the yields obtained by farmers in their fertilized fields (Dobermann et al., 2003).
4. Nitrogen omission plots allow quantifying the intercept (yield at zero fertilizer application) of a crop yield–N application curve. This is of less importance if crop-based postemergence N management is conducted although knowing INS is also useful for deciding on the N need during early growth when crop diagnostics cannot be reliably measured. Where crop-based N management is not practiced, knowing the intercept is important for developing general recommendations for N use and timing of applications. Where soil P and K levels are not extremely low, an unfertilized total omission plot (-F) can be used to estimate INS instead of 0-N plots to which P, K, and other nutrients are applied.
5. In double- or triple-rice monoculture systems, crop-based measurements of IS should be conducted in HYS crops in which growth is least affected by abiotic and biotic constraints to yield. Estimates of IS obtained from HYS omission plots are applicable to LYS crops, except at sites where crop management and fallow periods differ significantly between two cropping seasons. Where such seasonal differences are consistent and affect the dynamics of soil nutrients such as N and P, estimates of IS should be obtained separately for HYS and LYS crops. Typical examples for this are rice systems that include an upland crop, a long aerated fallow period before one crop, or season-specific nutrient inputs, for instance, from sedimentation during naturally occurring seasonal floods.
6. Grain yield can be measured in a single omission plot per field and nutrient because differences between replicate plots embedded in small rice fields are typically less than 5%. Omission plots should be relocated within a field after each crop. Management of omission plots must be done with great care, particularly with regard to proper seed quality, seedling age, planting density, water supply, and pest control.
7. Measurement of grain yield in at least one HYS rice crop allows obtaining an initial IS estimate for fertilizer decisions. However, two or more HYS crops should be sampled over time using a single omission plot for each nutrient to obtain a more accurate field- or domain-specific IS estimate. Despite large seasonal fluctuations, changes in IS over time tend to be slow so that measured IS values are likely to be representative for many years. This is particularly true for INS (Dobermann et al., 2003), whereas medium-term trends in IPS and IKS may fluctuate more, depending on the overall input–output budgets of these nutrients.
8. Sampling requirements for estimating domain-specific values of INS, IPS, or IKS depend on the homogeneity of the domain of interest. Preliminary indications are that for irrigated rice domains of about 100 to 200 km2, grain yield in omission plots should be measured in at least one HYS crop in about 10 farms (range 5 to 
15, depending on size and variability of domains), with one omission plot per nutrient and farm to estimate the domain means of INS, IPS, and IKS. Resampling in a second HYS crop is recommended for subsequent verification and fine-tuning.
These suggestions require further validation, particularly for estimating IPS and IKS, because (i) the domain delineation and sampling strategies used were not aimed at minimizing the within-domain variance in IS, (ii) the variance component analysis for IPS and IKS was based on only two HYS crops sampled compared with four HYS crops for INS, and (iii) the statistical approach ignored spatial autocorrelation among omission plots within a field or among farms within a domain. Future research should focus on (i) obtaining better understanding of the precision required for estimating IS within the context of developing domain-specific fertilizer recommendations and (ii) developing geospatial techniques for delineating potential recommendation domains. The latter must be based on a minimum set of biophysical and socioeconomic information that allows assessing yield potential, IS, and the expected response to fertilizer. This may include information about climate, parent material, soil types, cropping systems, major crop management practices, and historical yield records. Following the guidelines described above, such initial spatial classification should form the basis for stratified sampling to estimate domain-specific IS using omission plots. This may then be followed by redrawing the final recommendation domain boundaries to minimize spatial variation in measured IS within domains and maximize the variation between different domains.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Crop-based approaches to nutrient management may be both more accurate and easier to implement in irrigated rice systems than soil test–based concepts. Measurements of grain yield in nutrient omission plots provide reliable, though not highly accurate, estimates of IS. This information appears sufficiently robust for calculating location-specific fertilizer rates by placing a field or a domain into broader categories of INS, IPS, or IKS. Time and costs are associated with measuring IS through nutrient omission plots, but sampling requirements are not excessive, and advantages exist because of the educational value of omission plots and the potential for participatory development of fertilizer recommendations. Changes in IS tend to be slow so that measured IS values are likely to be representative for many years.
Measurements of grain yield in nutrient omission plots may play a role in field-specific fertilizer management or the development of fertilizer recommendations for larger areas (domains) with similar characteristics. Within such domains, management of the spatio-temporal variability in INS, IPS, and IKS can include domain- or field-specific approaches. Future success of crop-based SSNM approaches will depend on how well recommendation domains can be delineated and how tools such as nutrient omission plots can be efficiently used in obtaining information about IS.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
A. Dobermann, current address: Dep. of Agron. and Hortic., Univ. of Nebraska, P.O. Box 830915, Lincoln, NE 68583-0915. 
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  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]
|
|
|
|
|
|
Z. Segda, S. M. Haefele, M. C. S. Wopereis, M. P. Sedogo, and S. Guink |
