HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Weisz, R. Articles by Farrer, D. C. Search for Related Content PubMed Articles by Weisz, R. Articles by Farrer, D. C. Agricola Articles by Weisz, R. Articles by Farrer, D. C. Related Collections Other Crop Management Best Management Practices Wheat

Wheat

In-Season Tissue Testing to Optimize Soft Red Winter Wheat Nitrogen Fertilizer Rates

Influence of Wheat Biomass

^a Dep. of Crop Science, North Carolina State Univ., Raleigh, NC 27695
^b Dep. of Soil Science, North Carolina State Univ., Raleigh, NC 27695

^* Corresponding author (Randy_Weisz{at}ncsu.edu)

Received for publication April 12, 2006.

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
In the southeastern USA, soft red winter wheat (Triticum aestivum L.) N fertilizer recommendations are based on growth stage (GS) 30 tissue testing and models that assume that the relationship between tissue N concentration (N_con) and optimum N fertilizer rates (MaxN₃₀) is stable across fields differing in GS-30 biomass. However, previous research has indicated this may not be the case. Consequently, it was critical to re-evaluate these models. Using a split-split plot design, six experiments were conducted in North Carolina between 2002 and 2004. Main plots were planting date–seeding rate combinations that produced wheat with different GS-30 biomass. Subplots and sub-subplots were five N rates applied at GS-25 and GS-30, respectively. Wheat yield was responsive to fertilizer N at all site-years. The overall relationship between MaxN₃₀ and N_con was weak (r² = 0.43). The relationship between MaxN₃₀ and N uptake (N_con x biomass) was weaker (r² = 0.27). However, when the data were divided into different biomass classes, the overall model improved (R² = 0.75). For biomass < 340 kg ha^–1, the N_con at which no additional N fertilizer was required (N_critical) was 70.0 g N kg^–1. As biomass increased, N_critical decreased to 33.2 g N kg^–1. Intermediate classes had slopes of MaxN₃₀ versus N_con and N_critical values that were similar to those previously reported. This study indicates that to use tissue testing to determine N fertilizer recommendations across a range of GS-30 biomass conditions requires information regarding dry matter biomass.

Abbreviations: GS, Zadoks winter wheat growth stage • LP_OptN₃₀, economically optimum growth stage-30 nitrogen fertilizer rates determined using linear-plateau functions • MaxN₃₀, the growth stage-30 nitrogen fertilizer rate that resulted in the maximum yield • N₂₅, nitrogen rate applied at growth stage 25 • N₃₀, nitrogen rate applied at growth stage-30 • N_con, whole plant tissue nitrogen concentration • N_critical, whole plant tissue nitrogen concentration at which no additional nitrogen fertilizer was required to optimize yield • NUE, nitrogen-use efficiency • OptN₃₀, economically optimum growth stage-30 nitrogen fertilizer rate • PDSR, planting date-seeding rate combination • Q_OptN₃₀, economically optimum growth stage-30 nitrogen fertilizer rates determined using quadratic functions • W_biomass, the wheat biomass at growth stage-30

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
THE N-use efficiency (NUE) of winter wheat (Triticum aestivum L.) is generally low with, on average, only about 33% of applied N fertilizer utilized by the crop (Raun and Johnson, 1999). In-season tissue testing has been used to optimize wheat N fertilizer rates and increase NUE (Adams and Chapman, 1984; Alley et al., 1994; Weisz and Heiniger, 2004a). Tissue testing has been an especially valuable tool in warm humid regions of the world where soil mineral N content is too variable to be a good indicator of N availability to crops (Fox and Piekielek, 1978). For example, in the humid southeastern USA, soil mineral N measured in the spring when winter wheat is usually fertilized can range from 27 to 96% of the value measured in the fall before planting (Scharf and Alley, 1994).

In many winter wheat production areas, at least half of the total N fertilizer is applied pre-plant. An exception to this is the southeastern USA where most N fertilizer is applied at growth stage (GS) (Zadoks et al., 1974) 25 or 30. This is because warm temperatures and high precipitation typical of the southeastern USA result in denitrification and leaching of pre-plant N fertilizer (Counce et al., 1984; Scharf et al., 1993). Under these conditions, an N application may be recommended at GS-25 if tiller densities are low (Scharf and Alley, 1993; Weisz et al., 2001), but the majority of N fertilizer is usually applied at GS-30 based on a tissue test (Scharf et al., 1993; Flowers et al., 2004). In on-farm tests in Virginia, in-season N fertilizer rate optimization based on GS-25 tiller density and GS-30 tissue testing increased producer income by an average of $73 ha^–1 (Scharf and Alley, 1993) and increased apparent fertilizer use efficiency (Scharf et al., 1993) compared with applying a single fixed N rate. In North Carolina, a similar in-season N fertilizer recommendation system (Weisz and Heiniger, 2004a) was compared with single fixed N-rate applications and was found to increase wheat yields up to 2.27 Mg ha^–1 and to reduce N inputs by up to 48.6% (Flowers et al., 2004). The North Carolina and Virginia recommendation systems are based on a relationship between optimum GS-30 N fertilizer rates and GS-30 whole-plant tissue N concentration (N_con) developed by Scharf et al. (1993).

Given the success in-season wheat N fertilizer rate recommendation systems have had, Flowers et al. (2004) tested the system developed by Weisz and Heiniger (2004a) on a within-field, site-specific basis. In these fields there was a large variance in N_con. Using the field average N_con and applying the recommended optimum N rate across the whole field resulted in over-application of N in five fields and under-application in two. Correcting for this by tissue sampling, determining optimum N rates, and applying GS-30 N on a site-specific basis improved fertilizer use efficiency but did not result in higher yields compared with the whole-field optimization. The results demonstrated an improvement in using tissue-based N fertilizer optimization compared with typical growers' practices, but the failure to improve yield on a site-specific basis suggested that the recommended fertilizer rates might not be accurate under conditions where crop growth varied (Flowers et al., 2004).

The question of how reliable tissue-based systems are for N fertilizer rate recommendations is not new. Early research in relating N_con to N fertilizer requirements focused on finding "critical values" (N_critical) or "sufficiency levels" for N_con (i.e., the value of N_con that is high enough that no further N fertilizer is required). Adams and Chapman (1984) reported an N_critical of 27.5 g kg^–1 in Arkansas. In North Dakota, Engel and Zubriski (1982) found N_critical ranging from 52 to 56 g kg^–1. Vaughan et al. (1990) reported N_critical for dryland hard winter wheat grown in the west-central Great Plains (USA) of 32 g kg^–1. Beringer and Hess (1979) concluded that N_critical was too variable to be useful in Germany. Focusing specifically on soft red winter wheat in the eastern USA, N_critical values of 38 to 41, 40, and 35 g N kg^–1 were reported by Donohue and Brann (1984), Baethgen and Alley (1989), and Roth et al. (1989), respectively. The consistency of these results led to the conclusion that N_con could be used to determine N fertilizer rate recommendations in Pennsylvania (Roth et al., 1989) and Virginia (Baethgen and Alley, 1989; Scharf et al., 1993). In the subsequent development of a relationship between optimum GS-30 N fertilizer rates and N_con, Baethgen and Alley (1989) and Scharf et al. (1993) limited their sampling to experimental plots or to farm fields that were uniform and had high yield potential. Even so, the relationships they found between optimum N rate and N_con were weak (r² = 0.59 and 0.51).

We hypothesized that the relationship between soft red winter wheat N_con and GS-30 N_critical or optimum N fertilizer rates may not be as stable across the wide range of conditions typically found in wheat fields. If this is true, then the model developed by Scharf et al. (1993) may not be well adapted to low-density wheat with low yield potential, and this might explain the finding of Flowers et al. (2004). We wanted to determine if the relationship between optimum GS-30 N fertilizer rate and N_con might be more sensitive to the density of the wheat crop (e.g., its biomass) than previously thought. If so, this might explain some of the variation in the model developed by Scharf et al. (1993). This is important because this recommendation system is being used by southeastern USA producers to determine N fertilizer rates across a wide range of field conditions, not only in high-yield potential fields. In addition, such information would be valuable for site-specific N management.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Research Locations and Experimental Design
Experiments were conducted at four sites throughout North Carolina during 2002, 2003, and 2004. Specific locations were the Piedmont Research Station near Salisbury, NC, in 2003 (P2003); the Cunningham Research Station near Kinston, NC, in 2002, 2003, and 2004 (C2002, C2003, and C2004); the Lower Coastal Plain Tobacco Research Station near Kinston, NC, in 2002 (L2002); and the Tidewater Research Station near Plymouth, NC, in 2004 (T2004). Taxonomic classification of the soils at these sites is shown in Table 1.

Lime and fertilizer rates other than N followed standard recommendations for North Carolina based on annual soil tests (Hardy et al., 2002; Crozier et al., 2004). All study sites followed corn and were conventionally tilled. Pre- and postemergence herbicides were applied as recommended (York, 2004), and weed management was excellent at all site-years except L2002, where weed cover between GS-25 and GS-30 was rated at approximately 10% in PDSR-1, 14% in PDSR-2, and 33% in PDSR-3.

Data Collection
The number of tillers with a minimum of three leaves in a 1-m section of row was determined at two random locations in each subplot before N₂₅ application. This resulted in 10 samples per main plot, and main plot tiller density was estimated as the average of these samples. Plant samples for biomass and N_con were taken at GS-30 before N₃₀ applications. These samples were taken randomly within each sub-subplot from four 0.46-m sections of row by cutting whole plants just above the soil level. Subplot biomass was estimated as the mean of these 20 samples (i.e., four samples within each sub-subplot multiplied by five sub-subplots per subplot). At site-year L2002, where weed populations were high, weed tissue was removed from the samples before biomass and N_con measurements were made. Plant samples were dried at 60°C for 48 h for dry matter determination. Tissue N concentration of the dried samples was determined by Waters Agriculture Laboratories (Camilla, GA) using a CHN analyzer (McGeehan and Naylor, 1988). At harvest, the center 2 m of each sub-subplot was cut with a Massey Ferguson MF-8 or Gleaner K2 plot combine (AGCO Corp., Duluth, GA), and yields were measured with a HarvestMaster grain gauge (Juniper Systems, Logan, UT). Yields were adjusted to a moisture content of 135 g kg^–1.

Statistical Analysis
The PDSR combinations were designed to produce different GS-30 biomass levels at each site-year and depended on the soils and actual dates when wheat could be planted at each site-year. Consequently, PDSR combinations were not uniform across site-years, and ANOVA for GS-25 tiller density, GS-30 biomass, and grain yield were done individually by site-year. At GS-25, ANOVA for a RCBD was used to determine the effects of PDSR on tiller density. At GS-30, ANOVA for a split-plot design was used to determine the effects of PDSR and N₂₅ on wheat biomass. After harvest, ANOVA for a split-split plot design was used to determine the effects of PDSR, N₂₅, and N₃₀ on grain yield at each site-year. For these analyses, PROC GLM in SAS Version 6 (SAS Institute, Cary, NC) was used.

Determination of Maximum Growth Stage-30 Nitrogen Rates
For each PDSR-N₂₅ combination at each site-year, the N₃₀ fertilizer rate that resulted in the maximum yield (MaxN₃₀) was calculated. Initially, the yield responses to N₃₀ were modeled using linear-plateau, quadratic-plateau, the exponential Mitscherlich, and second-order polynomial functions. In the majority of cases the quadratic-plateau, Mitscherlich, and second-order polynomial functions failed to fit the data (i.e., they were unable to predict a statistically significant MaxN₃₀ or they predicted a value that was higher than what is agronomically reasonable for soft red winter wheat). The yield data were, however, well modeled with the linear-plateau function (for example see Fig. 1 ), and consequently MaxN₃₀ values were estimated with the following rules:

1. The yield response to N₃₀ was determined by ANOVA in PROC GLM. If the response to N₃₀ was not significant, MaxN₃₀ was assumed to be zero.
   
2. For PDSR-N₂₅ combinations for which the response was significant, Fishers Protected LSD ( = 0.05) was used for means separation. If the highest mean yield was for N₃₀ = 0, then MaxN₃₀ was determined to be zero.
   
3. If data sets had significant and positive yield responses to N₃₀, they were analyzed as follows:
   

   a. A linear-plateau function was fit to the data using PROC NLIN (SAS Institute, Cary, NC). If the 95% confidence interval for the estimated slope of the linear portion of the model, and for the breakpoint (the N₃₀ value beyond which the yield response was flat) did not contain zero, MaxN₃₀ was determined to be equal to the breakpoint (Fig. 1). In four cases, the yield response to N₃₀ was initially positive, reached a plateau, and then decreased at higher N rates. When this occurred, the N₃₀ treatment levels that had significantly decreased yield compared with the N₃₀ rate with the highest yield as determined by the Fisher's protected LSD ( = 0.05) mean separation were deleted.
   

   b. If the linear-plateau function did not result in a meaningful fit to the data (i.e., if the 95% confidence interval for the linear slope or for the breakpoint included zero), then treatment means were compared in PROC GLM using a series of linear contrasts to determine MaxN₃₀.
   

   
   

View larger version (15K): [in this window] [in a new window]   Fig. 1. Response of wheat yield (mg ha–1, solid squares) and partial profit ($ ha–1, open circles) to N fertilizer applied at growth stage (GS) 30. Yield was modeled with a linear-plateau function. The N rate resulting in the highest yield (MaxN30) was estimated as the breakpoint. Partial profit was computed assuming a N/grain price ratio of 10, modeled with a quadratic function, and economic optimal N rate (OptN30) estimated as the inflection point. These example data were from the Lower Coastal Plains Research Station 2002, planting date–seeding rate combination = 3, and GS-25 N application = 67 kg ha–1.

Determination of Optimum Economic Growth Stage-30 Nitrogen Rates
The economically optimum N₃₀ rate (OptN₃₀) is not necessarily equal to MaxN₃₀. Baethgen and Alley (1989) showed that because wheat yield response to N is sufficiently large, the impact of changing grain and/or N prices on OptN₃₀ is small. In fact, they showed that for N/grain price ratios between 2 and 6, OptN₃₀ was essentially equal to MaxN₃₀. However, since 1989, the price of N has increased significantly, and price ratios between 6 and 10 are more representative of current conditions. Consequently, we evaluated the impact a price ratio of 10 would have on OptN₃₀ calculations.

When the yield response to N₃₀ is modeled with a linear-plateau function, MaxN₃₀ (estimated at the breakpoint) is equal to OptN₃₀ if the linear slope is greater than or equal to the N/grain price ratio. If the slope is lower than the N/grain price ratio, then the linear-plateau model predicts that OptN₃₀ will be equal to 0. This approach was used for data sets that conformed to rule 3a. For data sets that followed rules 1 or 2, OptN₃₀ was determined to be zero. Data sets that followed rule 3b (n = 13) were eliminated from this analysis.

A second and potentially more robust way to estimate OptN₃₀ is to convert the yield data directly to partial profit and model the partial profit response to N₃₀. For each PDSR-N₂₅ combination at each site-year, partial profit was estimated by subtracting the cost of N₃₀ from the income received from the grain yield assuming a N/grain price ratio of 10. These data were well fit with a second-order polynomial function (see Fig. 1). If the inflection point was below zero, OptN₃₀ was set to zero; otherwise, OptN₃₀ was determined to equal the N₃₀ value at the inflection point.

	RESULTS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Tiller Density, Biomass, and Grain Yield
Planting date–seeding rate combinations successfully produced different tiller densities at all site-years (Table 3), with mean GS-25 tiller density being consistently highest for PDSR-1 and lowest for PDSR-3. Biomass levels at GS-30 could have been affected by PDSR and N₂₅. There was a significant PDSR x N₂₅ interaction only at C2003. The main effects of PDSR and N₂₅ were significant at all site-years except C2004, where only N₂₅ had an effect (Table 3). At all site-years except C2004, GS-30 biomass was higher with earlier planting and/or higher seeding rates (Fig. 2A ). At all site-years, GS-30 biomass increased as N₂₅ increased to 67 kg ha^–1, with no further increase at higher N rates (Fig. 2B). Despite the PDSR x N₂₅ interaction at C2003, these trends were still apparent at that site-year (data not shown). These data indicate that the PDSR treatments were successful in establishing a wide range of GS-30 biomass.

View larger version (28K): [in this window] [in a new window]   Fig. 2. Mean growth stage (GS) 30 biomass for (A) wheat planted with planting date–seeding rate combinations (PDSR) 1, 2, and 3. (B) Plots receiving five different rates of N at GS-25. These data are from five site-years (Cunningham Research Station 2002, and 2004 indicated as C2002 and C2004, respectively; Lower Coastal Plains Research Station 2002, L2002; Piedmont Research Station 2003, P2003; and Tidewater Research Station 2004, T2004) where the PDSR by GS-25 N rate interaction was not statistically significant. Means within site-years with different letters were significantly different at the 0.05 probability level.

Growth Stage-30 Nitrogen Rates for Maximum Yield
At each site-year, grain yield response to N₃₀ for each PDSR–N₂₅ combination was analyzed using ANOVA (data not shown). For 23 PDSR–N₂₅ combinations, N₃₀ was not significant, or the highest yield was attained at N₃₀ = 0 kg ha^–1. In these cases, MaxN₃₀ was defined as 0 kg N ha^–1. For the remaining PDSR–N₂₅ combinations, response to N₃₀ was modeled using a linear-plateau function (e.g., Fig. 1). Of these, 52 resulted in meaningful fits to the data, and MaxN₃₀ was estimated as the resultant model breakpoint. For 13 PDSR–N₂₅ combinations, MaxN₃₀ was defined using a series of linear contrasts.

The overall relationship between MaxN₃₀ and N_con (MaxN₃₀ = 156.5 – 2.73N_con) was weak (r² = 0.43) (Fig. 3A ). The linear relationships between MaxN₃₀ and N_con previously reported by Scharf et al. (1993) (MaxN₃₀ = 235 – 4.8 N_con; n = 136; r² = 0.51) and by Baethgen and Alley (1989) (MaxN₃₀ = 314 – 7.90 N_con; n = 16; r² = 0.59) are shown in Fig. 3A. The relationship reported by Baethgen and Alley (1989) had the steepest slope and lowest N_critical (x axis intercept) of 39.7 g N kg^–1. Scharf et al. (1993) used substantially more data in their analysis, which resulted in a somewhat lower slope and N_critical close to 49.0 g N kg^–1. The slope we found for the overall relationship between MaxN₃₀ and N_con was substantially lower than those previously reported and resulted in an N_critical level of 57.3 g N kg^–1.

View larger version (27K): [in this window] [in a new window]   Fig. 3. The growth stage (GS)-30 N fertilizer rate that resulted in the maximum yield (MaxN30) versus GS-30 whole plant tissue N concentration (Ncon) for the entire data set (A, open circles) and for five GS-30 biomass classes (B–F). The entire data set is shown in each graph, but for B through F, individual biomass classes are highlighted using larger solid symbols, and the linear regression model for that biomass class only (solid line and r2) is indicated. The linear models previously reported by Scharf et al. (1993) (dotted line) and by Baethgen and Alley (1989) (dashed line) are shown for comparison in each graph.

[1]

[2]

where W_biomass (kg ha^–1) was the GS-30 wheat biomass. Both models had an R² = 0.69. Consequently, biomass was an important parameter in determining MaxN₃₀. To visualize this biomass effect on the relationship between MaxN₃₀ and N_con, we divided the complete data set into five biomass classes. The relationships between MaxN₃₀ and N_con for each of these five classes are illustrated graphically in Fig. 3B–F.

Biomass Less than 340 kg ha^–1
Maximum N rates for wheat with biomass values lower than 340 kg ha^–1 fell along the upper right edge of the total data cloud (Fig. 3B). The linear regression model for this biomass class was only weakly significant (r² = 0.49; p = 0.08). Although the data seemed to be more quadratic than linear, the quadratic term in a second-order polynomial fit to these data was not statistically significant. Maximum N₃₀ rates for this biomass class were generally higher than those reported by Scharf et al. (1993) (dotted line in Fig. 3B) and had a much higher N_critical (x axis intercept) of 70.0 g N kg^–1 (Table 4). To our knowledge, this is higher than any N_critical value for soft red winter wheat reported in the literature.

Biomass from 1000 to 1500 kg ha^–1
The next highest biomass class (1000–1500 kg ha^–1) fell in the center of the full data cloud (Fig. 3D). The linear regression model of this data subset had a N_critical (x axis intercept) of 49.2 g N kg^–1 (Table 4). This was nearly equal to the N_critical found by Scharf et al. (1993) (dotted line in Fig. 3D). Based on the overlap of the SEs of the slopes of this regression line and either of those from the two previous biomass classes (Table 4), the slopes of these three regressions (Fig. 3B–D) were not statistically different. Apparently, as the GS-30 biomass increased from below 340 up to 1500 kg ha^–1 the major effect was to shift the relationship between MaxN₃₀ and N_con progressively to the left resulting in decreasing N_critical values.

Biomass from 1500 to 2000 kg ha^–1
This biomass class fell in the left half of the full data cloud (Fig. 3E) and had a steeper slope compared with the lower biomass classes based on the non-overlapping SEs of the slope estimates (Table 4). This biomass class had a linear regression model that was similar to that previously reported by Baethgen and Alley (1989) (dashed line in Fig. 3E). The value of N_critical for this biomass class (41.4 g N kg^–1) continued the trend of lower values with increasing GS-30 biomass.

Biomass Greater than 2000 kg ha^–1
The highest biomass class (>2000 kg ha^–1 (Fig. 3F) was on the left-most edge of the data cloud. The linear regression model for this biomass class had a slope similar to that reported by Baethgen and Alley (1989) (dashed line in Fig. 3F) and the lowest N_critical (33.2 g N kg^–1) of any biomass class.

Combined Model of Maximum Growth Stage-30 Nitrogen Fertilizer Rates
The stepwise regression models in Eq. [1] and [2] imply that as biomass increases, the slope of the linear relationship between MaxN₃₀ and N_con continually decreases (becomes more negative). Figures 3B–F demonstrate that this was not the case. The slope was relatively stable across the three lower biomass classes (Fig. 3B–D) and decreased in the two higher biomass classes (Fig. 3E and 3F and Table 4). To more accurately represent these data, MaxN₃₀ values from the entire data set were modeled using N_con as a continuous variable and biomass as a class variable (Table 4). Tissue N concentration and the interaction of N_con and biomass class were significant (p = 0.0001), with an overall model R² = 0.75. The full model fit for each biomass class is represented by the regression lines shown in Fig. 3B–F.

Economically Optimum Growth Stage-30 Nitrogen Rates
Because the relationship between MaxN₃₀ and N_con was not statistically significant (at the 0.05 probability level) for wheat with biomass <340 kg ha^–1, this biomass class was not included in the economic analysis. Following the results of the statistical analysis of MaxN₃₀, the economically optimum N₃₀ rates that were determined using quadratic functions fitting partial profit (Q_OptN₃₀) and from linear-plateau functions fitting yield (LP_OptN₃₀) were modeled using N_con as a continuous variable and biomass as a class variable (Table 4). For Q_OptN₃₀ and LP_OptN₃₀, tissue N concentration and the interaction of N_con and biomass class were significant (p = 0.0001), and the overall model R² values were 0.79 and 0.77, respectively (Table 4). The relationship between MaxN₃₀, Q_OptN₃₀, and LP_OptN₃₀ for each biomass class is illustrated in Fig. 4 . In general, estimated economically optimal N rates at any N_con within a given biomass class, whether estimated using the linear-plateau models of yield or the quadratic models of partial profit, were nearly identical (Fig. 4). In fact, based on the non-overlap of the SEs of the slopes and intercepts (Table 4), the regression models of MaxN₃₀, Q_OptN₃₀, and LP_OptN₃₀ versus N_con for a given biomass class were not statistically different.

View larger version (13K): [in this window] [in a new window]   Fig. 4. Linear regression of optimum growth stage (GS) 30 N fertilizer rates computed using three different methods (LPOptN30, economically optimum N fertilizer rates determined using linear-plateau functions; MaxN30, N fertilizer rates that resulted in the maximum yield; and QOptN30, economically optimum N fertilizer rates determined using quadratic functions) versus GS-30 whole plant tissue N concentration (Ncon). The data are divided into five GS-30 biomass classes.

	DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Using all the data, the relationship we found between MaxN₃₀ and N_con was the weakest reported in the literature for soft red winter wheat. However, when the entire data set was subdivided into GS-30 biomass classes, the model relating MaxN₃₀ and N_con resulted in an R² = 0.75, which is the highest value we are aware of in the published literature. Wheat in the lowest biomass class had the highest N_critical and required the highest N₃₀ to optimize yield for any given N_con. Conversely, wheat in the highest biomass class had an extremely low N_critical and the lowest N₃₀ requirement to maximize yield for any given N_con. The GS-30 biomass classes representing 340 to 1000 kg ha^–1 (Fig. 3C) and 1500 to 2000 kg ha^–1 (Fig. 3E) had a regression models that were similar to those reported by Scharf et al. (1993) and Baethgen and Alley (1989), respectively. The amount of biomass at GS-30 had a direct impact on the prediction of N_critical and MaxN₃₀. This might explain some of the differences among the relationships previously reported in the literature and might explain some of the unaccounted-for variability in those models. It also indicates that tissue testing, when combined with an estimate of GS-30 biomass, should lead to more accurate GS-30 N rate recommendations.

Baethgen and Alley (1989) suggested that MaxN₃₀ would be best modeled using NUP instead of N_con. However, their conclusion was based on a data set of 16 observations, and Roth et al. (1989) found that the relationship between MaxN₃₀ and NUP differed across years. Flowers et al. (2003) also concluded that predicted N fertilizer rates based on NUP were highly unreliable. The relationship reported here between MaxN₃₀ and NUP was statistically significant but very weak (r² = 0.27). For example, consider two wheat fields: the first with biomass of 700 kg ha^–1 and N_con of 55 g kg^–1 (lower right corner of Fig. 3C) and the second with biomass of 2500 kg ha^–1 and N_con of 15 g kg^–1 (upper left corner of Fig. 3F). Both these hypothetical fields have the same NUP, but MaxN₃₀ would be 0.0 and 122 kg ha^–1 for the first and second field, respectively. These results demonstrated that the relationship between MaxN₃₀ and biomass was more complex than can be explained by NUP, which is the simple product of N_con and biomass. Equation 1 can be rearranged such that:

[3]

or

[4]

where

[5]

One way to interpret Eq. [4] is that MaxN₃₀ can be described as a family of linear models with each using N_con as the independent variable but where the slope of any given line is a function of the biomass (Eq. [5]). This is consistent with the trend shown in Fig. 3 and demonstrates that there is a more complex relationship between MaxN₃₀, W_biomass, and N_con than can be explained by NUP alone.

Models predicting MaxN₃₀ and economically optimal N₃₀ rates (assuming a N/grain price ratio of 10) were not statistically different (Fig. 4). This is consistent with the findings of Baethgen and Alley (1989), who examined lower price ratios than we did. Apparently, the wheat yield response to GS-30 N fertilizer is large compared with current economic considerations.

Figures 3 and 4 indicate that a field with a thick, well developed wheat stand needs a different calibration curve for determining optimum GS-30 N fertilizer rates than one with a thin stand or low GS-30 biomass. This is a critically important consideration in using GS-30 tissue N concentration to optimize N fertilizer rates in farmers' fields. These results also suggest that if a site-specific N fertilizer recommendation system is to be based on GS-30 tissue N concentration, then an estimate of site-specific biomass would be needed to develop an optimal GS-30 N prescription.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
This research was sponsored in part by Initiative for Future Agricultural and Food Systems Grant no. 00-52103-9644 from the USDA Coop. State Research, Education, and Extension Service.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

• Adams, D., and S.L. Chapman. 1984. Monitoring wheat for nutritional and disease levels as a management decision aid. p. 20. In Agronomy abstracts. ASA, Madison, WI.
• Alley, M.M., P. Scharf, D.E. Brann, W.E. Baethgen, and J.L. Hammons. 1994. Nitrogen management for winter wheat: Principles and recommendations. Ext. Circ. 424-026. Virginia Coop. Ext. Serv., Blacksburg.
• Baethgen, W.E., and M.M. Alley. 1989. Optimizing soil and fertilizer nitrogen use by intensively managed winter wheat: II. Critical levels and optimum rates of nitrogen fertilizer. Agron. J. 81:120–125.
• Beringer, V.H., and G. Hess. 1979. Brauchbarkeit der pflanzenanalze zur bemessung spatter N-gaben zu winterweizen. Landwirtsch. Forsch. 32:384–394.
• Counce, P.A., B.R. Wells, J.R. Davis, and A.M. Simpson. 1984. Response of wheat to N-source, N-rate and nitrification inhibitors. Arkansas Farm Res. 33:2.
• Crozier, C., R. Heiniger, and R. Weisz. 2004. Lime, phosphorus, potassium, sulfur, manganese, copper, magnesium, and calcium management for small grains. In Small grain production guide 2004–2005. Ext. Circ. AG-580. North Carolina Coop. Ext. Serv., Raleigh.
• Donohue, S.J., and D.E. Brann. 1984. Optimum N concentration in winter wheat grown in the coastal plain region of Virginia. Commun. Soil Sci. Plant Anal. 15:651–661.
• Engel, R.E., and J.C. Zubriski. 1982. Nitrogen concentrations in spring wheat at several growth stages. Commun. Soil Sci. Plant Anal. 13:531–534.
• Flowers, M., R. Weisz, and R. Heiniger. 2003. Quantitative approaches for using color infrared photography for assessing in-season nitrogen status in winter wheat. Agron. J. 95:1189–1200.[Abstract/Free Full Text]
• Flowers, M., R. Weisz, R. Heiniger, D. Osmond, and C. Crozier. 2004. In-season nitrogen optimization and site-specific management in soft red winter wheat. Agron. J. 96:124–134.[Abstract/Free Full Text]
• Fox, R.H., and W.P. Piekielek. 1978. Field testing of several nitrogen availability indices. Soil Sci. Soc. Am. J. 42:747–750.
• Hardy, D.H., D.L. Osmond, and A. Wossink. 2002. An overview of nutrient management with economic considerations. Publ. AG-565. North Carolina Coop. Ext. Serv., North Carolina State Univ., Raleigh.
• McGeehan, S.L., and D.V. Naylor. 1988. Automated instrumental analysis of carbon and nitrogen in plant and soil samples. Commun. Soil Sci. Plant Anal. 19:493–505.
• Raun, W.R., and G.V. Johnson. 1999. Improving nitrogen use efficiency for cereal production. Agron. J. 91:357–363.[Abstract/Free Full Text]
• Roth, G.W., R.H. Fox, and H.G. Marshall. 1989. Plant tissue tests for predicting nitrogen fertilizer requirements of winter wheat. Agron. J. 81:502–507.
• Scharf, P.C., and M.M. Alley. 1993. Spring nitrogen on winter wheat: II. A flexible multicomponent rate recommendation system. Agron. J. 85:1186–1192.
• Scharf, P.C., M.M. Alley, and Y.Z. Lei. 1993. Spring nitrogen on winter wheat: I. Farmer-field validation of tissue test-based rate recommendations. Agron. J. 85:1181–1186.
• Scharf, P.C., and M.M. Alley. 1994. Residual soil nitrogen in humid region wheat production. J. Prod. Agric. 7:81–85.
• Vaughan, B., K.A. Barbarick, D.G. Westfall, and P.L. Chapman. 1990. Tissue nitrogen levels for dryland hard red winter wheat. Agron. J. 82:561–565.
• Weisz, R., C.R. Crozier, and R.W. Heiniger. 2001. Optimizing nitrogen application timing in no-till soft red winter wheat. Agron. J. 93:435–442.[Abstract/Free Full Text]
• Weisz, R., and R. Heiniger. 2004a. Nitrogen management for small grains. In Small grain production guide 2004–2005. Ext. Circ. AG-580. North Carolina Coop. Ext. Serv., Raleigh.
• Weisz, R., and R. Heiniger. 2004b. Small grain seeding rates for North Carolina. In Small grain production guide 2004–2005. Ext. Circ. AG-580. North Carolina Coop. Ext. Serv., Raleigh.
• York, A. 2004. Small grain weed control. In Small grain production guide 2004–2005. Ext. Circ. AG-580. North Carolina Coop. Ext. Serv., Raleigh.
• Zadoks, J.C., T.T. Chang, and C.F. Zonzak. 1974. A decimal code for the growth stages of cereals. Weed Res. 14:415–421.

This article has been cited by other articles:

				R. P. Sripada, D. C. Farrer, R. Weisz, R. W. Heiniger, and J. G. White Aerial Color Infrared Photography to Optimize In-Season Nitrogen Fertilizer Recommendations in Winter Wheat Agron. J., October 15, 2007; 99(6): 1424 - 1435. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Weisz, R. Articles by Farrer, D. C. Search for Related Content PubMed Articles by Weisz, R. Articles by Farrer, D. C. Agricola Articles by Weisz, R. Articles by Farrer, D. C. Related Collections Other Crop Management Best Management Practices Wheat