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


     

Published online 6 February 2007
Published in Agron J 99:405-414 (2007)
DOI: 10.2134/agronj2006.0027
© 2007 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
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 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 Google Scholar
Google Scholar
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Agricola
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Related Collections
Right arrow Nutrient Management
Right arrow Nitrogen
Right arrow Site-Specific Analysis

Site-Specific Analysis & Management

Comparison of Nitrogen Management Zone Delineation Methods for Corn Grain Yield

Nathan E. Derby*, Francis X. M. Casey and David W. Franzen

Dep. of Soil Sci., North Dakota State Univ., Fargo, ND 58105-5638

* Corresponding author (nathan.derby{at}ndsu.edu)

Received for publication January 31, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
It is important for today's farmer to manage his fields efficiently to maximize economic return and minimize environmental impacts. Managing agricultural inputs based on spatial variability in a field may help to achieve these goals. The objective of this study was to develop and compare different N management zone delineation methods and investigate the effect that variable N fertilization directed by these zones had on corn (Zea mays L.) grain yield over 4 yr. The zone delineation methods were based on: (i) supervised classification of bare soil color; (ii) cluster analysis of soil N, apparent electrical conductivity (ECa), and 1 yr of yield data; or (iii) a summation of standardized ECa, elevation, and 5 yr of yield data. The different delineation methods resulted in relatively similar zone patterns. Applied N rates were mostly affected by varying yield goals due to low variability of soil test N between zones. The effect of zone method on yield was inconsistent over the study period, with the soil color zone yields corresponding best to hypothesized productivity potential. The uniform treatment yields were not significantly different than yields from most of the N management zones. Increases in applied N for management zone treatments resulted in small yet not statistically significant yield increases in most years, with economic benefits over the uniform treatment observed in 2 yr. Yield was affected by adverse weather conditions in 2 yr, resulting in significant increases in subsurface soil N and reduced economic benefits of zone management versus conventional management.

Abbreviations: DGPS, differentially corrected global positioning system • ECa, apparent electrical conductivity • GDD, growing degree days • GIS, geographical information system • RCB, randomized complete block • U, uniform zone delineation method • V1, variable rate 1 zone delineation method • V2, variable rate 2 zone delineation method


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
DELINEATION of nutrient management zones for site-specific farming is rapidly becoming an important aspect of agricultural practices. The necessity for precision farming stems from the fact that an agricultural field is not a homogeneous medium on which uniform inputs will result in uniform outputs. The impetus to delineate zones varies and includes: (i) reducing or describing soil test variability (Flowers et al., 2005), (ii) improving nutrient management with variable rate fertilizer application (Fleming et al., 2000), (iii) understanding the spatial variability of crop growth and yield (Machado et al., 2002), and (iv) protecting groundwater quality (Powers et al., 2000) as surface-applied chemicals have been shown to move rapidly to shallow groundwater under small depressions (Derby and Knighton, 2001).

In the past decade, many researchers have developed different methods to delineate semihomogenous zones within fields. Fleming et al. (2004) utilized a farmers experience in combination with aerial photos to define areas of similar soil properties and Carr et al. (1991) used aerial photos to modify soil survey boundaries. Aerial-infrared photos were used by Tomer et al. (1997) to predict spatially variable corn yield. Grid soil sampling for soil chemical and physical properties is a widely used approach for characterizing spatial variability (Franzen et al., 2000) and is often used as a comparison to other methods of zone delineation (Flowers et al., 2005). Soil zones based on topography have also been used for management zones (Franzen et al., 2000). Kravchenko and Bullock (2000) found topography to be a very important yield-limiting factor. Others have used elevation or other topographic attributes in conjunction with additional spatial data to delineate zones (Kitchen et al., 2003; Schepers et al., 2004). With the recent availability of real time kinematic (RTK) GPS for detailed topographic maps, elevation will likely become more attractive than ever to precision farming (Clark and Lee, 1998). Yield maps created from GPS-equipped grain yield monitors also provide much information about the spatial variability of a field. Taylor et al. (2001) discussed several methods for using yield monitor data to develop yield goal maps. However, their yield prediction results were inconsistent across site–years and they concluded that multiple years of data would be needed to develop appropriate yield goal maps for a given location. Similarly, Machado et al. (2002) found that cluster zones of yield monitor data related differently to elevation and soil texture from year to year due to weather factors. Also, Jaynes and Colvin (1997) found spatial yield data to be unstable from year to year, even though some similarities were noted.

Soil apparent electrical conductivity (ECa) has recently become one of the most used soil properties to describe the spatial variability of a field (Corwin and Lesch, 2003). Lund et al. (1999) have shown that bulk ECa measured with a Veris 3100 sensor cart (Veris Technologies, Salina, KS) is a cost-effective way to delineate zones for precision agriculture. They also found these zones to be repeatable and consistent with other soil properties such as clay content. Electrical conductivity has been used to develop zones for precision agriculture because it is correlated, either positively or negatively, with many soil factors related to yield potential (Johnson et al., 2003). The use of ECa in conjunction with other georeferenced data such as elevation (Kitchen et al., 2003), aerial imagery (Schepers et al., 2004) and grain yield (Franzen and Nanna, 2002) has aided in the delineation of management zones.

Cluster analysis has been used in many studies to group points or cells that have similar attributes into zones. These data points may be similar to each other, as related to precision agriculture, based on any or all of the attributes described previously. Two commonly used cluster analysis methods are hierarchical and k-means (SAS Institute, 1994). It is beyond the scope of this paper to expand on the differences in the hierarchical and k-means algorithms. Either method produces as many or as few clusters as the user requires. Fridgen et al. (2004) developed software to calculate statistics and perform cluster analysis for a range of cluster numbers so the user can decide how many zones are best for a given field. Diker et al. (2002) utilized frequency analysis of yield monitor data as another method to delineate management zones.

It is clear that delineation of semihomogeneous zones within a field has become an important aspect of today's agriculture and existing methods of zone delineation vary in their degree of complexity. It is also clear that no single method of zone delineation is correct for every situation. There is agreement however, that use of multiple layers of data is necessary to adequately describe the spatial variability of a field. The objective of this study was to develop and evaluate a new method for N management zone delineation, on the basis of maintaining or increasing corn grain yield and N use efficiency to maximize economic return and minimize environmental impacts, compared to another zone delineation method and conventional uniform N application in a southeastern North Dakota field. The results presented are from one site that was part of a multistate project to evaluate the effectiveness of nutrient management zone determination methods.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Site Description
The research site was located in southeastern North Dakota, USA (46.05°N lat, 98.11°W long) and comprised 17 ha of a 64-ha center pivot sprinkler irrigated field with a surface elevation range of 396.2 to 398.2 m. The soil types on the location were Hecla loamy fine sand (sandy, mixed, frigid Oxyaquic Hapludoll), Wyndmere fine sandy loam (coarse-loamy, mixed, superactive, frigid Aeric Calciaquoll), Stirum fine sandy loam (coarse-loamy, mixed, superactive, frigid Typic Natraquolls), and Ulen fine sandy loam (sandy, mixed, frigid Aeric Calciaquolls), which had average shallow apparent electrical conductivities (ECa) of 10, 18, 24, and 14 mS m–1, respectively. The shallow ECa range of the entire site was 4 to 40 mS m–1. Cumulative growing degree days (GDD) from planting to September 30th for 2001 to 2004 were 1358, 1379, 1317, and 1121°C, respectively. Total precipitation plus irrigation for the same time period each year was 637, 586, 561, and 645 mm. In the 2 yr before the present study, potato (Solanum tuberosum L.) and soybean [Glycine max (L.) Merr.] had been grown at this location, respectively. Corn was grown each of the 4 yr reported for this study. Continuous corn is a common rotation in this area and had been practiced at this site from 1989 to 1998 as well, with a corn yield range of 2.4 to 12.7 Mg ha–1 measured in 1994. A previous study conducted in 1990–1995 (Derby et al., 2005) indicated that the site responds well to N fertilizer.

The area was divided into 60 plots (Fig. 1 ). Each plot was 21.3 m (28 corn rows) wide and 128 m long. The horizontal gap near the north edge of the study area was a grass trail used to gain access to sampling equipment. Within groups of three plots, treatments were assigned randomly for a randomized complete block (RCB) design. Treatments were the zone delineation methods (U) uniform, no zones; (V1) variable rate 1, zones were based on bare soil image; and (V2) variable rate 2, zones were based on multiple layers of data and evolved throughout the study. The zones to be used for variable rate fertilizer application within a given plot were then dependent on the treatment assigned to that plot. Square 21.3-m cells were identified within each plot to serve as boundaries for the zones and subsequently, the variable rate N application. In 2001, the treatments were 18.3 m wide (24 corn rows). However, the variable rate applicator used was 21.3 m wide. Hence, variable rate N application was actually done on 21.3 m treatments and treatments in subsequent years were 21.3 m wide as discussed above. In 2002, six plots to the north of the access trail were destroyed before harvest when the farmer cooperator mistakenly chopped that area for silage.


Figure 1
View larger version (32K):
[in this window]
[in a new window]

  		Fig. 1. Plot layout with treatment designations and Order 1 soil survey.

 
Surface elevation measurements were taken manually at 100-m intervals (with additional elevation measurements within the 100-m grid framework taken in obvious small depressions, etc., to more completely assess the spatial variability) with a transit and rod to create a digital elevation model of the site. Soil test nitrate N was also measured on a 100-m grid in October 2000. Apparent soil electrical conductivity (ECa) was measured with a Veris 3100 Sensor Cart in 2001 after harvest on untilled soil. Yield maps were created from differentially corrected global positioning system (DGPS) equipped on-the-go grain yield monitor data for corn in 1994, 1995, 1997, and 1998 and soybeans in 2000.

Management Zone Delineation
A summary of the zone delineation methods used for each treatment and the yield goals used for fertilizer N recommendations within the zones is presented in Table 1. Management zone treatment U was simply a uniform fertility treatment based on average soil test nitrate and an average yield goal for the entire area. The first method of determining different zones (V1) was based on supervised classification of a bare soil image of the site. Boundaries between soils of different reflectance were drawn to create four zones of similar soil, i.e., dark soil, light soil, and two intermediate soil shades were identified. Our extensive knowledge of the field also aided us in delineating zones from the bare soil image. The resulting zones corresponded well to the Order 1 soil survey (Fig. 1). The bare soil image was imported into ArcMap GIS 8.2 (ESRI, 2000; where GIS is geographical information system) and a 21.3- by 21.3-m grid was imposed on it. Each 21.3-m square was given a value of 1, 2, 3, or 4 corresponding to the zone on the image (Fig. 2 ). In the 1st year of the study (2001) the V2 treatment was the same as V1, that is, the bare soil image was used to create the zones for V1 as well as V2.


View this table:
[in this window]
[in a new window]

  		Table 1. Zone delineation methods and yield goals used for each treatment.

 

Figure 2
View larger version (43K):
[in this window]
[in a new window]

  		Fig. 2. Zones based on bare soil color in 21.3-m grid format. Thin lines are original zone lines drawn on aerial image. Thick lines are the arbitrary boundaries between areas of varying N credit for soybeans in 2000. The N credits for 2001 fertilizer recommendation were 34, 22, and 45 kg N ha–1 for the west, center, and east portions of the field, respectively. The extracted plots in the lower two maps are Treatments V1 and V2, which were both based on bare soil color in 2001.

 
For the 2nd year, the V2 zones were developed by Ward's hierarchical cluster analysis (SAS Institute, 1994) of three layers of data. The data layers were 0- to 60-cm soil test nitrate N (Franzen, 1999), Veris deep ECa, and 2001 corn yield. Deep ECa was used because it was correlated more with yield and soil N than was shallow ECa (Ralston 2003). Before cluster analysis, uniform data grids with 21.3- by 21.3-m cells were created from the raw data using Transform, a part of Noesys (Noesys, 1999). The uniform data grid, the dimensions of which were specified during data import into Transform, was constructed by nearest neighbor interpolation to fill in missing data. If more than one data point occupied any given cell, the points were averaged. Four clusters were chosen to be consistent with the four zones from the bare soil image. Kitchen et al. (2002) also found that four zones were optimal. The cluster analysis assigned a cluster number to each set of data points. These numbers were reordered to reflect yield potential in the field based on previous yield maps, with Zone 1 having the lowest yield potential and Zone 4 the highest. The four zones resulting from the cluster analysis and the V2 treatment plots are shown in Fig. 3 . This method was abandoned for the remainder of the study due to the low variability and transient nature of soil test N and the need for a better method based on more stable zonal indicators.


Figure 3
View larger version (29K):
[in this window]
[in a new window]

  		Fig. 3. Zones developed by Ward's hierarchical cluster analysis for 2002 season in 21.3-m grid format. The extracted plots in the lower map are the V2 treatments plots.

 
A simple summation method was developed to delineate the zones for Treatment V2 in 2003 and 2004. A schematic summary of this method is presented in Fig. 4 . This method involved combining three layers of data: elevation, Veris deep ECa, and yield rank. Elevation and ECa were chosen as components of the zone delineation method due to their temporally stable nature (Fraisse et al., 2001). Initially, the uniform data grid was created from the data as discussed previously. Yield rank was calculated for each of 5 yr of preprecision agriculture yield monitor data by assigning a value of either 1, 0, or –1, if a given yield value was higher, equal to, or lower than the average of the entire dataset, respectively. Ranking the yield in this fashion allowed for averaging across years as well as different crops (Kitchen et al., 2003). The 5 yr of ranked yields were then averaged for each grid cell to create the yield rank dataset used, with values ranging from –1 to 1. Averaging multiple years of yield data in this fashion created a more robust layer of data that accounted for many factors not described by data such as topography and apparent EC, making yield a more intrinsic property of the field. The values for elevation were standardized to values between 0 and 1 by subtracting the minimum value of the dataset from each value and then dividing by the range of the elevation values. Deep ECa was also standardized in the same fashion. Additionally, the standardized ECa value was then inverted by subtracting it from 1. This step was necessary because historically at this location, areas of high EC typically had lower yields and low elevation, and areas of low EC had high yields and high elevation. Linear regression indicated positive correlation of yield and elevation and negative correlation of yield and ECa for the 5 yr of yield data collected before the present study. The standardized-inverted EC, standardized elevation, and average yield rank values were then added together for each 21.3-m cell. In the absence of any evidence to give more weight to a given data layer, each layer was given equal weight. Further research could be done to determine which data layer, if any, is most important at this site and warrants more weight than the others. The summed data was divided into four groups in ArcMap GIS 8.2 using the Natural Breaks (Jenks) classification to create the zones for Treatment V2 (Fig. 4). Zone numbers did not have to be reordered when using the summation method since areas of low elevation and high EC (low inverted EC) had the lowest yield potential, etc.


Figure 4
View larger version (68K):
[in this window]
[in a new window]

  		Fig. 4. Maps of standardized inverted deep ECa, elevation, and average of 5 yr ranked yield, summed to create the summation method zones by for 2003 and 2004 seasons. The V2 treatment plots are the extracted plots in the lower map.

 
Determining Fertility Rates
Urea was applied before spring tillage with a DGPS equipped variable rate broadcast applicator with a Falcon controller (AGCO Global Technologies, Duluth, GA), which was operated by a local fertilizer dealer. The applicator utilized a georeferenced prescription map to direct the variable application to each zone. The operation of the application equipment was monitored by research personnel so that any problems with the application would be noted. For example, the speed of the applicator was reduced to minimize any lag that may occur during rate changes. Also, the center of each pass was flagged to allow the operator to follow the correct row during application, eliminating overlap and skips. The application equipment was calibrated by the fertilizer dealer to assure that the correct rates were being applied, and the amount of material that was prescribed was verified with the amount actually applied to the field. The prescription maps were converted from ArcGIS shape files with SGIS software (AGCO Global Technologies, 2003). The urea was incorporated with a coulter disk/chisel plow implement as soon as possible after application to reduce volatilization losses.

In 2001, variable N fertility rates were based on supplying 17.9 kg N ha–1 for each megagram per hectare of corn expected (yield goal), reduced by the amount of soil test nitrate N and other fertilizer N sources (the entire area received 11 kg N ha–1 with starter fertilizer and 56 kg N ha–1 from fertigation through the sprinkler irrigator each year). In subsequent years, the recommendation was increased to 21.4 kg N ha–1 for each megagram per hectare of corn expected (Franzen, 2003). Soil test nitrate N was determined on two composited 0- to 60-cm samples taken postharvest of the previous year from the center of each plot.

For the U treatment, N rate was determined using a yield goal of 11.80 Mg ha–1 and the average soil test N values from all U plots. Plots with the V1 treatment were fertilized based on the same uniform yield goal of 11.80 Mg ha–1; however, the soil test values used in the recommendation calculations were zone specific, i.e., average soil test of the samples taken from within the respective zones. Treatment V2 N rates were calculated using soil test averages from each respective V2 zone as well as variable yield goals for each V2 zone. The average yield goal was based on past knowledge of the yield potential of the specific areas of the field and the farmer cooperators desired yield. The variable yield goals for the V2 treatment were set to bracket the average yield goal and were based on the spatial yield history. In 2001, a N credit for the previous soybean crop was used to adjust the fertility recommendation since N mineralization from organic matter increases after soybeans (Franzen, 2003). Based on the soybean yield in 2000, the western third of the field was given a 34 kg N ha–1 credit, the center third was given a 22 kg N ha–1 credit, and the eastern third was given a 45 kg N ha–1 credit (Fig. 2).

Yield Monitoring
Corn grain yield was measured after physiological maturity with a Case IH Axial Flow combine (Case Corporation, Racine, WI). The combine was equipped with a Case IH Advanced Farming System (AFS) yield monitor and a Trimble AgGPS 132 DGPS system (Trimble Navigation Ltd., Sunnyvale, CA) using a base station for differential correction. The yield monitor was maintained in good working condition by the farmer cooperator and calibrated with a weigh wagon immediately before harvest of the study area. Lag time was accounted for within the AFS software and the yield data was cleaned by removing the high and low outliers. The combine had an 8-row header, which meant that every fourth pass through the field, the harvester would overlap two treatments. Care was taken to omit these overlap data from the calculations of mean yield for each 21.3-m cell using ArcMap GIS software. The yields reported for zones are the averages of all of the 21.3-m cells for the given zone, across all plots assigned the given treatment. Treatment averages are the average of all 21.3-m cells within a given treatment. Statistical analysis was performed with JMP (SAS Institute, 1994) using the Fit Y by X procedure.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil test nitrate N was very similar between zones for each year of this study. Consequently, the rates of fertilizer N applied to the zones were mostly affected by the variable yield goals of the V2 treatment (Table 2). The lack of soil N variability was somewhat surprising considering the spatial variability in soil type, soil ECa, and topography observed, but others have reported similar lack of soil N variation and still saw some benefits of variable N application (Ferguson et al., 1996). On this site however, the zonal variation of applied N did not seem to have a direct impact on the zone averaged yields as discussed below.


View this table:
[in this window]
[in a new window]

  		Table 2. Yield goals, average soil test N and fertilizer N rate applied with variable rate applicator. Soil test N values are averages of samples taken from each zone within a given treatment.

 
Zone Yield Comparisons
The effect of delineation method and zones on corn grain yield was inconsistent across years. Mean corn grain yield for treatments zones for each year are presented in Table 3. Only a few differences in yields between treatment zones were statistically significant at {alpha} = 0.05.


View this table:
[in this window]
[in a new window]

  		Table 3. Mean corn grain yields by treatment and zone for each year. Ordered by decreasing yield within each year.

 
For 2001, all variable rate zones had higher average yields than the U treatment plots. In 2001, both the V1 and V2 delineation methods were based on supervised classification of bare soil color, which is reflected in the mean yields being not significantly different between treatments for each zone. Also of note is that the mean yields decreased with decreasing zone class number, indicating that the hypothesis of Zone 1 having the lowest productivity potential and Zone 4 the highest, was correct. These results indicated that nutrient management based on bare soil color resulted in better productivity and N use efficiency over a uniform N application, albeit very slightly. As there was almost no difference in the N rates between zones for 2001 (Table 2) the natural productivity for each zone is apparent.

When cluster zones were used for Treatment V2 in 2002, the yield for Zone 2 of V1 was significantly lower than that of Zone 2 of the V2 treatment. This indicated that the cluster zoning method delineated a less productive area of the field for Zone 2 than the bare soil color method did. The arrangement of yield by zone for V2 was not as ordered as in 2001, although Zone 4 for both zone delineation methods was at or near the top of the yield range. This indicated that both zoning methods in 2002 correctly delineated areas with high productivity. The N application rates were more variable for V2 in 2002, however the yields do not reflect that fact. If the V2 zones had adequately described the field variability in terms of production potential, the yields should have been highest for Zone 4 and lowest for Zone 1. Apparently the yield was influence by factors not considered in the cluster analysis. On the other hand, the V1 zone yields were ordered similarly to 2001, which further substantiates that the zones were ordered correctly in terms of production potential. A number of the V1 and V2 treatment zones had higher average yields than the U treatment indicating a marginal benefit of zoning over a uniform application of N.

The summation zone method was used for V2 in 2003. There seemed to be a general reversal of which zones had the highest yields in 2003, and the range of yields between zones was less as compared to the previous years, with the lower zone numbers being toward the higher end of the yield range and the higher zone numbers being near the lower end of the yield range. This was true for both the V1 and V2 zones and was possibly due to significant wind damage that occurred in the 1st week of July, which broke off as many as 50% of the corn plants in some areas. It is possible that there was more damage to the plants in historically high-yielding Zones 3 and 4 of Treatment V2 and Zone 3 of V1, resulting in lower yields compared to the other zones. These zones were at topographically higher locations on the field, so they could have been impacted more by the wind. Also, although no plant height measurements were taken, the plants in V2 Zones 3 and 4 may have been taller and hence, more affected by the wind. Conversely, yield in Zone 4 of Treatment V1 was still comparatively high. This was a topographic depressional area and would have been somewhat protected from the wind. The wind damage seemed to have an averaging effect on yields, as is evident by the lower standard deviations and the relationship of the U treatment yield to the zone yields. The wind damage was definitely the cause for the lower than average yields observed in this year. As a result of the lower than expected yields, average residual soil nitrate N at a depth of 60 to 120 cm increased from 17 kg ha–1 in the fall of 2001, and 29 kg ha–1 in the fall of 2002, to 64 kg ha–1 in the fall of 2003.

The summation zone delineation method was used again in 2004 for V2. As in 2001 and 2002, the V1 zone yields were ordered correctly based on our hypothesis of yield potential. The yields for the V2 zones were again not consistent with the amount of N applied (Table 2) to each zone. Zones 4 and 3 of V1 and V2, respectively had significantly higher yields than the U treatment, demonstrating a slight advantage of zone management in 2004. However, yields were lower than normal over all treatments in 2004 due to below normal growing season temperatures. The cumulative growing degree days (GDD) for the 2004 growing season were only 1121°C. The corn varieties planted on the site were 100-d relative maturity hybrids, which require greater than 1300°C (GDD) to reach maturity (Sutton and Stucker, 1974). A previous study at this location also observed low yields and reduced N response in years with low cumulative GDD (Derby et al., 2004). Similar to 2003, residual soil nitrate N at 60 to 120 cm was higher than in 2001 and 2002, indicating lower N use efficiency. Although weather played a major role in the yield potential of this field in 2003 and 2004, the management zone techniques that were developed were compared to the uniform treatment within each year, and the weather affected each treatment similarly.

Zone Pattern Comparisons
The different zone delineation methods resulted in similar zone patterns with some minor differences (Fig. 2Go–4). This is common when investigating different methods of zone delineation (Fleming et al., 2004). For example, all three methods generally delineated the northern portion just west of center as a zone of lower productivity potential. In all years except 2003, these are the lowest yield zones and are located in the Stirum and Wyndmere soils (Zones 1 and 2). Similarly, all methods resulted in higher zone numbers on either end of the area, with the highest zone number (higher yield potential) on the east end. This indicated that the zoning methods delineated areas with different production potential, even though the yield variability was not large. By comparison, Zone 4 delineated by bare soil color (V1) corresponded to medium zone numbers (2 and 3) for both the cluster method of 2002 and the summation method of 2003 and 2004. As shown in Fig. 2 and 4, V1 Zone 4 is a relatively small, isolated area of Ulen soil with lower elevation than the immediately surrounding area. The higher yields in this area are probably due to increased water availability caused by runoff without the detrimental effects of salinity and sodicity seen on the Stirum and Wyndmere soils on the other high ECa–low elevation areas of the field. In all years, yields for Zone 4 of Treatment V1 are at or near the maximum of the yield range for all treatment–zone combinations. This is probably due to the fact that Zone 4 of V1 was only present in one small area of the field, while V2 Zones 2 and 3 were more spatially prevalent.

Supervised classification of bare soil color resulted in the most consistent arrangement of average zone yields from year to year compared to the other zone delineation methods. The spatial patterns of the zones appeared to be similar between treatments with the exception of the small area of V1, Zone 4. In this one small area, the higher EC and lower elevation that resulted in lower productivity in other areas of the field did not negatively affect yield. Generally, in all years of the study, there was only significant yield difference between the highest yielding zone and the lowest yielding zone. This suggests that possibly only two zones should have been considered for management on this field. Furthermore, the low yield variability between zones with different yield goals suggests that the use of yield goals may not be appropriate on this field.

Treatment Comparisons and Economics
The V1 and V2 management zone treatments resulted in slightly higher average N fertilizer application rates compared to the U treatment (Table 4). The V2 treatment consistently resulted in a higher average N rate over the V1 and U treatments. Average yields for the V1 and V2 treatments however were only slightly higher and rarely significantly higher than the U treatment. Obviously, the additional fertilizer N applied to the zoned treatments did not translate into higher yields as would have been expected.


View this table:
[in this window]
[in a new window]

  		Table 4. Treatment averages of N fertilizer rate applied with variable rate applicator, corn grain yield, costs and benefits as compared to the U treatment, and postharvest soil test nitrate N for each year.

 
A simple cost/benefit comparison was also included in Table 4 to further investigate the feasibility of using zone management. For these comparisons, a hypothetical field size of 65 ha and a cost of $0.66 kg–1 of N for urea were used. For the V1 and V2 treatments, the local fertilizer dealer would have charged $11.00 ha–1 to create four zones and $360.00 for soil sampling ($90.00 per zone, $5.54 ha–1). The charge to spread a single product such as urea was $13.50 ha–1. Comparatively, if the entire field were managed conventionally (U), the costs would have been $90.00 for soil sampling ($1.38 ha–1 for 65 ha) and $13.50 ha–1 for N application. The gross income (benefit) was calculated based on a hypothetical corn grain price of $78.57 Mg–1 ($2.00 bu–1). The zone management did not pay in 2 of the 4 yr of this study, as the additional cost associated with the V1 and V2 treatments over the U treatment was greater than the additional benefits realized in 2003 and 2004. For example, in 2004 the additional cost to zone manage for V1 was $21.29 ha–1 greater than the cost of conventional management (U), but the additional benefit from the slightly increased yield was only $9.37 greater than U management. This was a net decease in income of $11.92 ha–1 from using V1 zone management instead of conventional N management. However, in 2001 and 2002, the additional economic benefits of zone management outweighed the additional costs when compared to the conventionally managed U treatment. For example, in 2002 the additional cost to utilize zone management scheme V2 over U was $18.56 ha–1. The additional income compared to the U treatment was $25.65 ha–1, a net increase in income of over $7.00 ha–1.

Besides the economic considerations, the environmental costs of N management need to be weighed since in areas with shallow groundwater such as this, there is a risk of groundwater contamination from excess nitrate N in the subsurface soil. In 2003 and 2004, the years when the additional fertilizer N applied to the V1 and V2 treatment areas did not translate into additional yield, the deep (60–120 cm) soil test nitrate N increased (not significant at {alpha} = 0.05) over that of the U treatment (Table 4). This indicated that the crop did not fully utilize the fertilizer N and some of it leached to the subsoil, perhaps making it unavailable to the crop in subsequent years. Although differences in treatment soil test nitrate N within years were observed, none of these were significant at the {alpha} = 0.05 level. However, deep soil test nitrate N yearly averages for 2003 and 2004 were significantly higher than for 2001 and 2002. Note that the subsurface nitrate N for the U treatment was also higher in 2003 and 2004 than in previous years, indicating that the higher N rates on the V1 and V2 treatments were not solely responsible for the higher residual soil nitrate N. A reduction in crop N uptake as a result of the adverse weather conditions discussed previously played a major role in increasing the deep soil nitrate N.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Three different methods for delineation of management zones and N application based on: (i) bare soil color; (ii) clustering of soil test nitrate, deep ECa, and yield; and (iii) summation of standardized deep ECa, elevation, and 5 yr of ranked yield, were compared with each other and to a uniform N application treatment as related to corn grain yield over 4 yr. Variable rate N application was based on a constant yield goal and zone averaged soil test N for the zones delineated based on soil color. Nitrogen was variable rate applied to the cluster and summation zones based on zone averaged soil test N and variable yield goals. The zone delineation methods resulted in similar zones, with some minor differences in the patterns and order of the zone rank. Yield results were inconsistent from year to year and the range of yields between treatments and zones was small, indicating no clear benefit to one zone delineation method over another. However, yields within zones from supervised classification of bare soil color corresponded to hypothesized productivity potential in most years. The higher N rates applied to the zones with higher yield goals did not necessarily result in higher yields, indicating that additional N for variable yield goals wasn't justified in at least 2 yr of the study and that the use of yield goals to direct N rates needs to be re-evaluated. However, crop damage from wind in 1 yr and insufficient heat units in another year may have affected yield response to N. The uniform N application treatment resulted in yields that were not statistically different from the other treatments in most years. Even though this site had significant variation in soil ECa, topography, and soil color, zone management seemed to result in uniformity of yield across the site. It appeared that weather had a more significant impact on yield than did other factors such as soil variability and topography in at least 2 yr of this study. For that reason, it was not cost effective to use variable rate technology at this location in years when weather limited the crops growth potential. Subsurface soil test nitrate N also increased significantly after the 2 yr of adverse weather. In the 2 yr when crop growth was not limited by weather factors, there was economic benefit in using either of the zone management methods over the uniform treatment. Considering the inconsistent yield results from year to year, the uniformity in yield between treatments and zones, and the inconsistent economic benefits of the zone management schemes, it is difficult to recommend one of the developed zone delineation methods over the other, and that possibly a two-zone management scheme may have been better suited to the field. It is also difficult to recommend either zoning method/nutrient management scheme over the conventional uniform N application at this location since neither are economically feasible when weather is a limiting factor.


   ACKNOWLEDGMENTS
 
This research was funded by a CSREES Initiative for Future Agriculture & Food Systems (IFAFS) grant.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





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 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 Google Scholar
Google Scholar
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Agricola
Right arrow Articles by Derby, N. E.
Right arrow Articles by Franzen, D. W.
Related Collections
Right arrow Nutrient Management
Right arrow Nitrogen
Right arrow Site-Specific Analysis

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