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


     

Published in Agron J 99:1564-1578 (2007)
DOI: 10.2134/agronj2006.0151
© 2007 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
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 HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Agricola
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Related Collections
Right arrow Nutrient Management
Right arrow Site-Specific Analysis
Right arrow Soil Fertility and Productivity

Site-Specific Analysis & Management

Evaluation of Zone Soil Sampling Approaches for Phosphorus and Potassium Based on Corn and Soybean Response to Fertilization

Jorge Sawchikb and Antonio P. Mallarinoa,*

a Dep. of Agronomy, Iowa State Univ., Ames, IA 50011
b Inst. Nacional de Investigaciones Agropecuarias (INIA), La Estanzuela, Colonia, Uruguay

* Corresponding author (apmallar{at}iastate.edu)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Soil sampling approaches have been compared based on soil-test variation. This study evaluated sampling approaches for P and K based on yield response to fertilization. Strip trials were established on four fields for P and three fields for K managed with corn (Zea mays L.) and soybean (Glycine max L. Merr.) rotations and evaluated 3 or 4 yr (27 site-years). Treatments replicated three to four times were fertilizer and no fertilizer application. Soil test results from a dense grid-point sampling (DG) approach (0.08 to 0.27 ha) were used to simulate six approaches: (i) 1.0-ha grid cells (GC), and zones delineated based on (ii) soil series from digitized survey maps (SMZ); (iii) elevation (EZ); (iv) apparent soil electrical conductivity, ECa (ECZ); (v) EZ and ECZ (EECZ); and (vi) EZ, ECZ, and slope (EECSZ). Grain yield monitors, global positioning systems (GPS), and geographical information systems (GIS) were used to describe crop responses. Estimates of soil-test variation were largest for DG, intermediate for GC, and less for other approaches. Crops responded (P ≤ 0.05) to fertilization in 20 site-years. Sampling approaches DG, GC, EZ, EECZ or EECSZ, ECZ, and SMZ identified a differential within-field yield response in 16, 8, 5, 3, 2, and 2 site-years, respectively. Differential yield responses seldom were explained by zone-mean soil-test values. Zone approaches often identified areas with different yield levels but were less effective than DG or GC at describing within-field variation of soil tests and yield response to fertilization. Zone approaches may be more effective in fields with shorter fertilization histories or soils with more contrast in properties.

Abbreviations: DG, dense grid-point sampling • ECa, apparent soil electrical conductivity • ECZ, apparent soil electrical conductivity zone • EECSZ, combined EZ, ECa, and slope zones • EECZ, combined EZ and ECa zones • EZ, elevation zone sampling method • GC, grid-cell sampling • GIS, geographical information systems • GPS, global positioning systems • SMZ, soil series zone • STK, soil-test K • STP, soil-test P

Evaluation of Zone Soil Sampling Approaches for Phosphorus and Potassium Based on Corn and Soybean Response to Fertilization

Jorge Sawchikb and Antonio P. Mallarinoa,*

a Dep. of Agronomy, Iowa State Univ., Ames, IA 50011
b Inst. Nacional de Investigaciones Agropecuarias (INIA), La Estanzuela, Colonia, Uruguay

* Corresponding author (apmallar{at}iastate.edu)

Received for publication May 13, 2006.
Soil sampling approaches have been compared based on soil-test variation. This study evaluated sampling approaches for P and K based on yield response to fertilization. Strip trials were established on four fields for P and three fields for K managed with corn (Zea mays L.) and soybean (Glycine max L. Merr.) rotations and evaluated 3 or 4 yr (27 site-years). Treatments replicated three to four times were fertilizer and no fertilizer application. Soil test results from a dense grid-point sampling (DG) approach (0.08 to 0.27 ha) were used to simulate six approaches: (i) 1.0-ha grid cells (GC), and zones delineated based on (ii) soil series from digitized survey maps (SMZ); (iii) elevation (EZ); (iv) apparent soil electrical conductivity, ECa (ECZ); (v) EZ and ECZ (EECZ); and (vi) EZ, ECZ, and slope (EECSZ). Grain yield monitors, global positioning systems (GPS), and geographical information systems (GIS) were used to describe crop responses. Estimates of soil-test variation were largest for DG, intermediate for GC, and less for other approaches. Crops responded (P ≤ 0.05) to fertilization in 20 site-years. Sampling approaches DG, GC, EZ, EECZ or EECSZ, ECZ, and SMZ identified a differential within-field yield response in 16, 8, 5, 3, 2, and 2 site-years, respectively. Differential yield responses seldom were explained by zone-mean soil-test values. Zone approaches often identified areas with different yield levels but were less effective than DG or GC at describing within-field variation of soil tests and yield response to fertilization. Zone approaches may be more effective in fields with shorter fertilization histories or soils with more contrast in properties.

Abbreviations: DG, dense grid-point sampling • ECa, apparent soil electrical conductivity • ECZ, apparent soil electrical conductivity zone • EECSZ, combined EZ, ECa, and slope zones • EECZ, combined EZ and ECa zones • EZ, elevation zone sampling method • GC, grid-cell sampling • GIS, geographical information systems • GPS, global positioning systems • SMZ, soil series zone • STK, soil-test K • STP, soil-test P


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SITE-SPECIFIC MANAGEMENT recognizes that within-field variability in crop yield, nutrient availability, and other soil properties should be assessed to improve crop management. Precision farming technologies such as yield monitors, differential global positioning systems (DGPS), geographical information systems (GIS), various forms of remote sensing, and variable-rate applicators can be used for site-specific management. Many studies have shown that soil-test P (STP) and soil-test K (STK) levels vary considerably within fields. The variation patterns sometimes are related to soil series or soil map units, but fertilization, manure application, and other management practices often create new and large variability patterns at various scales (Franzen and Peck, 1995; Mallarino, 1996; Mallarino and Wittry, 2004). In these instances, use of a single P or K fertilizer rate throughout a field or soil map unit may result in excessive fertilization in some areas and suboptimal fertilization in others. Variable-rate technology allows for P and K application to specific field areas, may improve nutrient use efficiency and farm profitability, but requires reliable and cost-effective assessments of soil-test values.

Soil survey maps at scales ranging from 1:12,000 to 1:24,000 have been used to delineate soil sampling zones for many years. However, studies (Jaynes, 1996; Brevik et al., 2001; Mallarino and Wittry, 2004) have shown that because of the insufficient detail of these soil survey maps, zone delineation using this approach may not be sufficient for effective site-specific management. Grid sampling subdivides a field into a systematic arrangement of small areas or cells. Soil-test results from composite soil samples collected from these cells usually are interpolated by a variety of techniques for estimating soil-test values at unknown positions (Wollenhaupt et al., 1994; Franzen and Peck, 1995). Several studies have indicated that the sampling density required for effective use of variable-rate technology differs across nutrients, fields, and geographic regions, but sampling cells smaller than about 1 ha seem needed for P and K in many fields. For example, Wollenhaupt et al. (1994) recommended 0.36-ha cells to guide variable-rate application of P and K in Wisconsin. Working in fields with long histories of P and K fertilization, Franzen and Peck (1995) determined that using 0.44-ha cells was superior to 1-ha cells. Mallarino and Wittry (2004) reported that cells larger than 0.8 ha in size did not represent P and K levels appropriately in various Iowa soils. These and other studies showed that a large portion of the nutrient variation can be missed if the grid size is too large and, for that reason, interpolations may perform poorly (Mulla et al., 2001; Mueller et al., 2001). Sampling 1-ha cells is the most frequently used grid sampling method in Iowa and row-crop production areas of the U.S. Midwest. One composite sample is collected from each cell either by collecting the cores from a small portion of the cell or over the entire cell area.

Targeted or zone sampling can be used to reduce the number of samples and sampling costs while gathering acceptable information about within-field nutrient variation. Criteria and attributes suggested to delineate sampling zones vary greatly. Soil survey maps and landscape position have been used to delineate sampling zones for a long time. Yield maps can aid zone delineation because yield can be related to nutrient availability and nutrient removal. However, research has indicated that long-term yield data are required to reliably establish patterns of crop yield variation (Lark and Stafford, 1998; Jaynes et al., 2003). Also, several studies have shown that soil-test P or K in samples collected using dense grid sampling methods are related to topography or elevation in some fields but not in others (Franzen and Peck, 1995; Franzen et al., 1998; Mallarino and Wittry, 2004).

Apparent soil electrical conductivity maps based on on-the-go measurements using electro-magnetic induction or direct-contact methods have been used to delineate sampling zones and study variation in soil properties (Jaynes, 1996; Kitchen et al., 2003a; Heiniger et al., 2003; Johnson et al., 2003; Sudduth et al., 2003). Studies often have shown a significant association between ECa and terrain attributes or crop yield (Sudduth et al., 1996; Corwin et al., 2003; Kitchen et al., 2003b). However, studies that used ECa to delineate sampling zones for P and K testing have yielded inconsistent results. Chang et al. (2001) used soil-test values from a dense grid sampling approach to simulate approaches based on larger grid cells, elevation, and ECa. Dividing the field into four large cells or blocks reduced the average within-block variance of STP compared with the whole field variance in three fields, whereas use of ECa reduced the variance only in one field. Fridgen et al. (2001) delineated sampling zones for two fields based on ECa, elevation, and slope using clustering methods. The within-zone variance of STP and STK compared with the whole-field variance was reduced only in one field, where zones with higher ECa also had higher STK and clay content.

Soil or crop canopy images have also been suggested as aids for zone sampling delineation for various nutrients, often in combination with other approaches (Fleming et al., 2000; Luchiari et al., 2003; Mulla et al., 2001; Fleming and Buchlieter, 2003). In Iowa, Mallarino and Wittry (2004) used a combination of field images, soil map units, elevation, and yield to compare the efficacy of various zone sampling approaches and grid sampling (1.2- to 1.6-ha cells) for P and K in eight fields. They compared and ranked approaches based on differences between within-zone and across-zone STP and STK variability. The results showed that grid sampling was always more efficient for STP, while grid sampling and an integrated zone approach based on field imagery, soil map units, elevation, and yield were similarly efficient for STK.

While several approaches have been proposed to delineate sampling zones and some have been evaluated based on comparisons of soil-test means and variance within and among zones, these approaches have not been evaluated based on their efficacy to identify field areas with different yield response to P and K fertilization. Therefore, the objective of this study was to assess the efficacy of various sampling approaches based on crop response to P and K fertilization in seven Iowa fields.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Sites, Soil Sampling, and Treatments
Soil-test and yield data used for this study were derived from replicated strip trials conducted on seven Iowa farmer's fields located in Boone, Guthrie, Linn, and Tama counties. At each field, approximately 6 to 12 ha located at least 40 m away from field borders were selected to establish the trials. Table 1 shows information about the experimental areas and the dominant soil series according to digitized, 1:12000 scale soil survey maps (Iowa Cooperative Soil Survey, 2001). The fields are referred to as Sites 1 though 7. All fields were managed with corn-soybean rotations and chisel-plow/disk tillage (Sites 1, 2, 3, 4, and 7) or no-till (Sites 5 and 6). Other management practices were those used by each farmer and, therefore, corn hybrids, soybean varieties, seeding rates, and planting dates varied among fields. Yield data from two treatments applied at each P trial site (Sites 1–4) and each K trial site (Sites 5–7) were appropriate for the objectives of this study: a nonfertilized control and a uniform fertilizer rate applied to the entire length of the strip. Yield data from a third treatment (variable-rate application) were not used because high-testing areas of strips corresponding to this treatment were not fertilized. A RCBD was used at all sites, with three replications at Sites 1 and 2 and four replications at the others. The treatment strips had a width of 18.3 m in all sites and their lengths (exclusive of at least 40-m borders on each side) varied from 360 to 550 m among the sites.


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

  		Table 1. Location, size, and predominant soil series for experimental areas in fields with P or K trials.

 
Table 2 shows the uniform P and K application rates used, the year of treatment application, crops evaluated, and cropping sequences. Suffixes "a" and "b" in the code used for each site-year indicate the first and second crop of the corn-soybean rotation, and suffixes "a2" and "b2" denote crops of the second rotation cycle. A fourth site-year was not evaluated for the K trial Site 5. The fertilizer rates applied before the first crop of each rotation cycle (twice at each location) were those suggested by the Iowa State University Soil Testing Laboratory for a single fertilizer application for 2-yr corn-soybean rotations, which is the most common fertilization practice in Iowa for this rotation. The fertilizers used were granular monoammonium phosphate (MAP, NH4H2PO4, 11–23–0 N–P–K) or potassium chloride (KCl, 0–0–52 N–P–K), which were spread with commercial spreaders and were incorporated into the soil by chisel plowing and disking except for sites managed with no-tillage (Sites 5 and 6). We used MAP because this fertilizer and diammonium phosphate [(NH4)2HPO4] are the P sources being sold and used by farmers in Iowa and the western Corn Belt. No corrective N rate was used to offset the small amount of N applied with MAP but at least 168 kg N ha–1 (as anhydrous ammonia) were uniformly applied for corn, which is the highest N rate recommended in Iowa for corn after soybean (Blackmer et al., 1997). A yield response to higher N rates for corn after soybean is unlikely in Iowa (Sawyer, 2006; Mallarino and Ortiz-Torres, 2006). A yield response from soybean to the N applied with MAP also is very unlikely. Iowa research (Mallarino, unpublished data, 1999; Sawyer, 2001) showed no soybean yield response to N fertilizer application at various rates. Fertilizer P (MAP) was applied across all strips of the K trials and fertilizer K was applied across all strips of the P trials at recommended rates.


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

  		Table 2. Crops and fertilizer application rates for fields with P or K trials.

 
Soil samples (15-cm depth) were collected from each site before applying treatments (twice) using a dense grid-point sampling (DG) approach adapted to the experimental layout. Before applying treatments for the first time, the separation of the grid lines across strips of P trial sites coincided with the width of each strip (18.3 m) and the separation distance along strips was 45 m (grid cell size was 0.082 ha). For K trial sites, the separation of the grid lines across strips coincided with the width of each replication (54.9 m), and the separation distance along strips was 45 m at Sites 5 and 6 and 50 m at Site 7 (grid cell size was 0.25 or 0.27 ha). Soil cores (8–12) for each composite sample were collected from the entire area (following no specific pattern) of a circle approximately 100 m2 in size at the center of each cell, and the center of the circle was georeferenced as a sample point. Before applying treatments for a second time (i.e., before the first crop of the second rotation cycle), soil samples were collected from the nonfertilized strips to be able to relate yield response of crops of the second rotation cycle to soil-test values of these nonfertilized strips. Sampling methods and cell length were similar to those used for the first sampling date and were assumed to represent the same areas, although fewer samples were taken for the P trial sites because fewer strips were sampled compared with the initial sampling date.

Soil samples from P Sites 1 and 2 were analyzed with the Mehlich-3 P test because soil survey maps indicated the presence of small areas of Harps soil (Typic Calciaquolls) with a high-pH surface layer due to CaCO3, whereas samples from Sites 3 and 4 were analyzed with the Bray-P1 test to accommodate to the farmers' commonly used test. The Bray-P1 and Mehlich-3 tests measure approximately similar amounts of P in noncalcareous soils when extracted P is measured with a colorimetric method (Mallarino, 2003), and Iowa interpretations are similar (Sawyer et al., 2002). However, only the Mehlich-3 and Olsen tests are recommended for calcareous Iowa soils because the Bray-P1 test often underestimates plant-available P in calcareous soils. Procedures used followed methods suggested for the North-Central Region (Frank et al., 1998) using a colorimetric determination of extracted P. Soil samples from the K trials (Sites 5–7) were analyzed for STK with the ammonium-acetate test following methods suggested for the North-Central Region (Warncke and Brown, 1998). Iowa STP interpretation classes (Sawyer et al., 2002) were used in this study. The classes for Bray-P1 and Mehlich-3 P tests are (in mg P kg–1): ≤8 for very low, 9 to 15 for low, 16 to 20 for optimum, 21 to 30 for high, and ≥31 for very high. The classes for STK are (in mg K kg–1): ≤90 for very low, 91 to 130 for low, 131 to 170 for optimum, 171 to 200 for high and ≥201 for very high.

Grain Yield Measurements
Grain yield was harvested with farm combines equipped with impact flow-rate yield monitors and DGPS receivers using differential correction from the U.S. Coast Guard AM beacon transmitter. The monitors were calibrated by weighing grain harvested along combine passes outside the experimental areas. A sensor located in the grain augers measured grain moisture, and yield was adjusted to moisture contents of 155 g kg–1 for corn and 130 g kg–1 for soybean. Yield data used for the study were unaffected by borders because experimental areas were at least 40 m away from field borders and data from combine passes that included border rows between strips were not used. Two 4.57-m or one 7.62-m-wide combine passes were used from each soybean strip, and two to four 4.57-m-wide combine passes were used from each corn strip. Yield monitor data were imported into ArcView GIS (Environmental Systems Research Inst. Inc., Redlands, CA), analyzed for any common yield monitor problem (Mallarino et al., 2001) such as effects of waterways or unplanned combine stops, and affected data deleted.

Delineation of Large Grid Cells and Field Zones
Soil-test values from samples collected with the DG approach were used to simulate several sampling approaches using a procedure developed and used by others (Franzen and Peck, 1995; Mulla et al., 2001; Bianchini and Mallarino, 2002; Mallarino and Wittry, 2004). The simulated sampling approaches were (i) 1.0-ha grid cells (GC); (ii) soil series zones (SMZ); (iii) elevation zones (EZ); (iv) apparent soil electrical conductivity zones (ECZ); (v) zones delineated using elevation and apparent soil electrical conductivity (EECZ); and (vi) zones delineated using elevation, apparent soil electrical conductivity, and slope (EECSZ). Maps with associated information for each approach were generated using ArcView GIS by creating appropriate polygons to represent grid cells or zones. All soil-test values from the DG sampling approach within each zone's borders were averaged to obtain a zone average. Therefore, the simulated zone approaches imply collecting more soil cores per zone than would normally be used in production agriculture, and the results of this study would indicate the highest potential efficacy of zone approaches.

For the 1.0-ha GC approach at P trial sites, STP data from 12 adjacent sampling points of the DG approach were averaged (six points across strips and two points along strips). For the K trial sites, where the DG approach was sparser, STK data from four adjacent sampling points were averaged (two points across strips and two points along strips). Soil series zones for the SMZ approach were imported to ArcView GIS from digitized, 1:12000 scale Iowa soil survey maps (Iowa Cooperative Soil Survey, 2001). Elevation and soil ECa were measured once for all sites in fall 2001 after harvesting crops using a real-time kinematic DGPS receiver and electromagnetic induction sensors mounted in an all-terrain vehicle and a density of one observation every 70 to 100 m2. A stationary base-station GPS receiver located at one side of each field was used to differentially correct the roving GPS receiver. Soil ECa measurements in P Sites 1 and 2 were made with an EM-38 electromagnetic induction sensor (Geonics Ltd., Mississauga, ON, Canada), while in other sites they were made with a Veris 3100 direct-contact sensor (Veris Technologies, Salina, KS). The ECa is an attribute with significant temporal variability primarily affected by soil moisture (Kitchen et al., 2003b) and, therefore, ECa results for these fields should not be directly compared. Elevation and ECa surface models based on a 10- by 10-m regular grid were created from the georeferenced point maps using ArcView GIS and the inverse-distance interpolation method with a distance-weighing exponent of 2.0. Also, a slope map was created with ArcView from the interpolated surfaces by calculating the maximum elevation change from each grid cell compared with each surrounding cell.

The EZ, ECZ, EECZ, and EECSZ zones were delineated based on elevation and ECa with the Management Zone Analyst 1.0 software (Fridgen et al., 2004), which uses fuzzy c-means and an unsupervised clustering algorithm (Bezdek et al., 1984; Fridgen et al., 2001; Kitchen et al., 2003a). This algorithm minimizes an objective function defined as the sum of squared distances from all data points in the cluster domain to the cluster center. Clustering algorithms require a quantitative measure of similarity (or distance) between observations (Sharma, 1996; Johnson and Wichern, 2002). We used Euclidean distance for the EZ and ECZ zoning approaches and the Mahalanobis distance for the combined approaches EECZ and EECSZ as suggested by Fridgen et al. (2001) because descriptive statistics of the clustering variables revealed unequal variances and nonzero covariances. Because multiple outcomes are inherent to the MZA clustering algorithm, we ran MZA three times and used the Fuzziness Performance Index (FPI) and the Normalized Classification Entropy (NCE) clustering performance indices as suggested by Fridgen et al. (2004) to assess the outcomes and select the number of zones. The FPI represents the degree of membership sharing between clusters, and values near zero represent distinct clusters with little membership sharing. The NCE index represents the amount of disorganization created by dividing a data set into classes. When a comparison of these two indices did not identify a single best outcome, we selected the outcome with fewer clusters (zones) or the one with the smaller within-zone pooled variance of the attribute compared with the total variance.

Descriptive statistics were calculated for STP and STK values from sampling points of the DG sampling approach (i.e., the center of 100-m2 areas) included within borders of zones of each simulated sampling approach for each site. Differences between mean soil-test values of zones of each sampling approach were assessed by ANOVA with PROC MIXED of SAS (SAS Institute, Cary, NC), in which zone was considered a fixed effect and the error term was the pooled within-zone variances of the soil-test values.

Evaluation of Yield Response to Fertilization
The grain yield response to fertilization over the experimental area of each site was assessed by ANOVA for a RCBD using PROC MIXED of SAS, in which fertilization was considered a fixed effect and replication (blocks) was considered a random effect. Yield inputs were means of all yield monitor points recorded at 1-s intervals within each treatment strip. Yield responses also were assessed for areas of each field that initially tested within different soil-test interpretation classes as identified by the DG and GC grid sampling approaches using a procedure used previously by Oyarzabal et al. (1996) and Bianchini and Mallarino (2002). The input data were prepared using ArcView GIS by overlaying yield, soil-test values, and field design maps. Yield inputs were means for areas defined by the width of each strip and the separation distance of the soil sampling grid lines along the strips. The soil-test inputs for analyses of crops for the first rotation cycle represented areas defined by the width of a replication and the separation distance of the grid lines along the strips (means of three composite samples for P trials and the result of a single composite soil sample for K trials). Therefore, two yield means (one for each treatment) corresponded to one STP or STK value. The F test from a separate one-way ANOVA for each STP or STK class with replications (blocks) and fertilization treatments as sources of variation assessed fertilization effects on yield. These analyses were performed for STP or STK classes in which both treatments were represented in at least two replications of the experimental design (blocks). Yield data for cells along strips of one replication that corresponded to the same soil-test interpretation class were considered as samples within experimental units (the analysis used the replication mean). Similar procedures were followed for the second rotation cycle, although soil-test inputs were those from samples collected from the nonfertilized strips before planting the first crop of the second rotation cycle.

An ANOVA procedure (PROC MIXED of SAS) was used to assess differences in yield response to fertilization among zones for each field, year, and zoning approach (SMZ, EZ, ECZ, EECZ, and EECSZ) individually. Fertilization, zone, and the fertilization by zone interaction were considered fixed effects while replication (blocks) was considered a random effect. Small zones of any approach that were not represented in at least two contiguous field replications were not considered. Only probabilities of F tests for the two main effects and the interaction are shown in tables. The zone main effect assessed average yield differences across zones. The fertilization main effect assessed average yield response to fertilization across zones and was similar to the whole-field analysis (or was approximately similar for sites in which data for small zones were not included because of insufficient replication). A statistically significant interaction of fertilization by zone indicated that the yield response to fertilization differed across the zones of a zoning approach.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Grain Yield Response to Fertilization
Phosphorus and K fertilization increased (P ≤ 0.05) grain yield of corn and soybean as evaluated by strip averages (Table 3 ) with the exception of four site-years for P trials and three site-years for K trials. Interpretations of yield responses from crops of the first and second crop rotation cycles differ slightly because the fertilizer treatment applied before the first crop of the second rotation cycle was reapplied to the same strips and the control strips were not fertilized since the beginning of the trials. For example, a response to fertilization from crops of the second rotation cycle probably resulted from cumulative effects of the first and second fertilizer applications. However, this study did not emphasize evaluations of the magnitude of a response and fertilization strategies.


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

  		Table 3. Mean corn and soybean grain yield response to P and K fertilization across the entire strip length of each treatment.

 
Comparisons of yield responses and initial STP or STK values (Table 4 ) indicated that results for the responsive site-years were reasonable because median soil-test values were very low, low, or optimum. The probability of yield response to P and K fertilization for these classes is 80%, 60%, and < 25%, respectively (Sawyer et al., 2002). A large proportion of the areas of the P sites tested very low or low, with site proportions ranging from 36% (Site 4) to 75% (Site 1). A lack of yield responses at Sites 1b and 2b was not expected because a large proportion of the experimental areas tested very low or low. Both were second-year crops, but the fact the P treatment was not reapplied should not explain the result because of the large amount of P applied before the first-year crop. A small and inconsistent yield response to P at Site 4 is reasonable because initial mean and median STP was in the upper part of the optimum class, and 46% of the site area tested high or very high. The K sites had very small or no areas testing very low, but 30 to 61% tested low and 27 to 65% tested optimum.


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

  		Table 4. Descriptive statistics and classification into interpretation classes of soil-test P and K values from soil samples collected using a dense grid-point (DG) sampling approach before applying treatments for the first time.

 
The yield responses to P and K were also assessed for field areas testing within different soil-test interpretation classes as classified by the dense, base DG sampling approach. Results for the P trials (Table 5 ) showed a yield response where STP was very low, at 8 of 16 areas where STP was low, and at 1 of 11 areas where STP was optimum. No yield responses were observed in high-testing areas. The lack of yield response in low-testing areas of Sites 1b, 4b, and 4a2 coincided with lack of response for the strip-average analyses in these sites. However, yield responses detected by the strip-average analysis for Sites 3b, 4a, and 4b2 were not detected by the analysis by STP class. Analyses for the K trials (Table 6 ) showed a response to K in 8 of the 11 areas testing low in STK. There was no yield response in areas testing optimum or higher in STK. The lack of response to K in low-testing areas of Site 6a coincided with results for the strip-average analysis. However, yield responses detected by the strip-average analyses for Sites 5b and 7b2 were not detected by the analysis by STK class. A failure of the analysis by soil-test class to detect a yield response when the strip-average analysis did (three site-years for P and two site-years for K) may be explained by yield responses not associated as expected with the classification by soil-test class. However, sometimes similar yield differences with both analyses suggest that large variability or lesser power of the statistical analysis by soil-test class contributed to the failure to confirm yield differences.


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

  		Table 5. Corn and soybean yield response to P fertilization for field areas testing in different soil-test P (STP) interpretation classes as identified by a dense grid-point (DG) sampling approach.

 

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

  		Table 6. Corn and soybean yield response to K fertilization for field areas testing in different soil-test K (STK) interpretation classes as identified by a dense grid-point (DG) sampling approach.

 
Soil-Test Values for Large Grid Cells and Field Zones
Phosphorus Trial Sites
The GC approach (1.0-ha cells) identified field areas encompassing three STP interpretation classes at Site 1 and two classes in the other sites (Table 7 ). However, GC described a narrower range of STP values and fewer STP classes than the base DG approach (Table 4) because each GC value was a mean of several point values. One or both extreme classes that GC identified at each field were not identified by DG, and field areas testing in the central optimum class always were larger for DG than for GC (at least twice as large). Results may have been different if GC had been simulated by randomly selecting GC point values. Our simulated methods for GC approximated a dense grid-cell sampling method wherein cores for a composite sample are collected from several points of the cell area.


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

  		Table 7. Descriptive statistics and classification into interpretation classes of soil-test P and K for a simulated 1.0 ha grid-cell (GC) sampling approach based on soil-test values from samples collected before applying treatments for the first time using a dense grid-point (DG) approach.

 
Soil-test P differences (P ≤ 0.05) among SMZ, EZ, or ECZ zones were infrequent (Tables 8 and 9) . Soil-test P differed among SMZ zones only at Site 4 (Table 8), where areas of Canisteo soil had higher STP than areas of Clarion or Webster soils. This site had STP ranging across the five STP interpretation classes according to the DG approach (Table 4). Webster and Canisteo soils occupy the low-lying landscape positions and have approximately similar topsoil texture and profile, although the Canisteo soil tends to be in lower in the landscape and topsoil often is calcareous. The higher STP for Canisteo could be explained by accumulation of sediment and runoff P fertilizer relative to topographically higher areas of the field. Data in Table 9 show that STP differed among zones of all approaches at Site 2 and among EECZ zones at Site 3. The differences at Site 2 are explained by higher STP for a zone on a low-lying and concave topographic position with high ECa. This area was poorly drained, had consistently lower yield across years (and possibly lower P removal), and could have sediment and runoff fertilizer P deposition from higher elevations. At Site 3, the reasons for the differences among the EECZ zones are not clear because STP increased with increasing ECa but there was no clear trend with elevation, although the lowest STP was observed for the zone with lowest ECa and lowest elevation.


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

  		Table 8. Soil-test values for the dominant soil series (survey map zone, SMZ) within each site from soil samples collected using a dense grid-point (DG) sampling approach before applying treatments for the first time.

 

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

  		Table 9. Summary of attributes and soil-test P values in zones delineated using different approaches for the P trials.

 
Potassium Trial Sites
The GC approach (1.0-ha cells) identified field areas encompassing two STK interpretation classes at Sites 5 and 6 and three classes at Site 7 (Table 7). As expected, this approach described a narrower range of STK values and fewer STK classes than the DG approach (Table 4). Although the within-field STK range was comparatively narrower than for STP (smaller or no field areas tested very low in STK), the results also showed that a grid-cell sampling approach based on composite soil samples collected over larger area may result in less extreme STK values when compared with a grid-point approach.

Similar to the results for STP, differences in STK among zones of zones approaches tested were infrequent. Soil-test K differed among SMZ zones at Sites 5 and 7, and among EZ zones at Site 7. At Site 5, the Klinger soil tested highest in STK, Donnan tested intermediate, and Kenyon tested lowest (Table 8). Although the Kenyon soil tested low, STK for the Klinger and Donnan soils were in the optimum class, which would result in the same K fertilizer rate. In contrast, DG revealed STK variation across the very low to high classes in this field (Table 4), and GC identified areas testing low and optimum (Table 7). At Site 7, the Colo-Ely soil complex tested optimum in STK and the Tama soil tested low (Table 8), and STK of the three EZ zones increased as elevation decreased (Table 10 ). The results for SMZ and EZ at this site are in agreement because the Tama soil is found at higher elevations and its topsoil has coarser texture than the Colo and Ely soils. Therefore, perhaps finer texture and runoff K fertilizer from higher elevations can explain high STK for low topographic positions with Colo and Ely soils.


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

  		Table 10. Summary of attributes and soil-test K values in zones delineated using different approaches for the K trials.

 
Summary of Soil-Test Results for Sampling Approaches
A lack of frequent and large soil-test differences among zones for approaches based on soil series, ECa or topography attributes could be explained by several reasons. First, differences in soil parent materials and topography in these fields probably were not large enough to result in large differences in nutrient concentrations. Soil-test values differed among zones in very few fields, but when this occurred, the higher levels were found in low-lying areas having soils with finer texture and higher organic matter than the upland soils (Iowa Cooperative Soil Survey, 2001). This finding could be explained by differences in parent materials and/or long-term or recent transport and deposition of sediment, P, and K through erosion. Kravchenko and Bullock (2000) also reported a negative relationship between elevation and STK in Illinois fields with approximately similar ranges in elevation and slope. Because low-lying areas sometimes yield less than higher elevations (mainly with soybean and with higher than average rainfall) and there were histories of uniform P and K application, less P and K removal with harvest may also explain greater STP and STK in topsoil of low-lying field areas. Second, long fertilization histories may have significantly reduced the influence of soil forming factors on STP and STK of topsoil. A lack of relationship between topography and soil-test levels in fields with long fertilization histories has been observed previously, even with larger differences in elevation and slope (Franzen et al., 1998; Kravchenko and Bullock, 2000; Mallarino and Wittry, 2004). Zoning approaches based on ECa were not more useful at identifying areas with contrasting soil-test values compared with approaches based on soil series or elevation. Previous research in North Carolina (Heiniger et al., 2003) showed no large nor consistent relationship between ECa and STP and a weak relationship between ECa and STK. Perhaps ECa would be a better indicator of STK variation in fields with greater contrast in soil texture and cation exchange capacity (Sudduth et al., 2003).

A result of practical interest for grid sampling was the large soil-test variation observed among the point samples within each 1.0-ha grid cell (Table 7). Such large small-scale soil-test variability has been demonstrated in previous studies (Franzen and Peck, 1995; Mallarino, 1996; Mallarino and Wittry, 2004; Flowers et al., 2005), and confirms that the grid-point sampling method often used in the Midwest for a 1.0-ha grid size may not appropriately represent nutrient availability of those cells. Our experience in Iowa indicates that few producers and consultants collect cores from the entire cell area; most collect cores from a single central area about 100 to 400 m2 in size because it is faster and less costly.

Yield Response to Fertilization for Large Grid Sampling Cells
Table 11 shows the yield response to P fertilization for field areas with STP in different interpretation classes as identified by the GC approach. There was a yield response (P ≤ 0.05) in four (Sites 2a, 2a2, 3a, 3a2) of seven corn sites where more than one class was represented with enough area to conduct a statistical test, and the responses were observed in areas testing optimum or less. At two sites testing low (1b and 4b2), corn did not respond to P. For soybean, however, P fertilization increased yield in three (Sites 1a, 2b, 2b2) of seven sites where more than one STP class was represented, and the responses were observed only in low-testing areas. In contrast, as discussed previously, when the base DG approach was used there was a differential within-field yield response to P in 8 of 16 sites (Table 5). Table 12 shows similar information for the K trial sites. Potassium fertilization increased corn or soybean yield in three (Sites 5a2, 7a, and 7a2) of nine sites where more than one STK class was represented with enough area to conduct a statistical test. At Sites 5a2 and 7a2, K fertilization increased yield only in areas testing low. At Site 7a, K fertilization increased corn yield in areas testing low and optimum but not in areas testing high. In contrast, when the base DG approach was used, there was a differential within-field yield response to K in 8 of 11 sites (Table 6).


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

  		Table 11. Corn and soybean yield response to P fertilization for field areas testing in different soil-test P (STP) interpretation classes according to a simulated 1.0-ha grid-cell (GC) sampling approach.

 

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

  		Table 12. Corn and soybean yield response to K fertilization for field areas testing in different soil-test K (STK) interpretation classes according to a simulated 1.0-ha grid-cell (GC) sampling approach.

 
Results from both the P and the K trial sites indicated that the GC approach was less successful at identifying field areas with different yield response than the DG approach probably because the larger cell size encompassed small areas with various degrees of yield response. The frequency of significant yield responses to P or K in field areas testing optimum was slightly higher for GC than for the base DG approach. This result may be explained by the presence of low-testing areas in the larger cells of the GC approach.

Yield Response to Fertilization for Zones of Zone Sampling Approaches
All zone approaches sometimes identified field areas with different yield levels. Because this study emphasized analysis of yield response to fertilization, statistics for yield differences among zones are shown in tables but are discussed only when the results were relevant to interpret differences in yield response among zones.

Phosphorus Trial Sites
Soil Survey Map Approach
The yield response to P did not differ (P ≤ 0.05) among soil series at any site as indicated by the lack of significant interactions between P fertilization and soil series effects (Table 13 ). There was only a weak trend at Site 2b2 (at P = 0.09), where soybean tended to respond more for the Clarion soil than for the Webster soil probably because of much lower STP for Clarion (Table 8).


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

  		Table 13. Corn and soybean yield responses to P fertilization for field areas with different soil series (survey map zone, SMZ).

 
Elevation Approach
Corn yield response to P differed between EZ zones only at Sites 1b and 3a as indicated by statistics shown in Table 14 . At Site 1b, a higher yield response to P (not shown) was observed at higher elevations (EZ-2), where STP was lower by only 2 mg P kg–1, and values for the two zones were both low (Table 9). At Site 3a, a yield response to P (not shown) was higher at lower elevations (EZ-1), where STP was also lower by 2 mg P kg–1 than for EZ-2 but was classified low compared with optimum. The soybean response to P (Table 15 ) differed among EZ zones only at Site 2b2, where it was larger (not shown) for the highest elevation zone. However, this differential response cannot be explained with certainty because the difference in elevation between zones was small (Table 9), and although STP was lowest (11 mg P kg–1) for the highest elevation and very high (28 mg P kg–1) for the lowest elevation (EZ-2), it was also low (14 mg P kg–1) for the intermediate elevation. Excess rainfall could explain lower soybean yield at low elevations in this and other sites. Other research (Kaspar et al., 2003; Jaynes et al., 2003) also showed lower soybean yield in poorly drained soils of lower topographic positions in this region, which is explained by complex interactions between soil moisture, pH, and soybean cyst nematode (Heterodera glycines Ichinohe) infestation.


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

  		Table 14. Statistical significance of the corn yield response to P fertilization for zones delineated using different sampling approaches.

 

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

  		Table 15. Statistical significance of the soybean yield response to P fertilization for zones delineated using different sampling approaches.

 
Electrical Conductivity-based Approach
The corn yield response to P fertilization differed (P ≤ 0.05) among ECZ zones only at Site 3a (Table 14), where the response to P (not shown) was higher for the two zones (ECZ-1 and ECZ-2) with higher ECa (Table 9), which also had lower elevation. This inverse relationship between ECa and elevation can be explained by the usually finer texture and moisture retention capacity of soils located at lower landscape positions in this region. The differential response to P across ECZ zones at this site cannot be explained by STP because it was optimum for the two responsive zones and low for the nonresponsive zone, although values were borderline and the difference was only 3 mg P kg–1 (Table 9). Perhaps excess soil moisture mainly early in the season at this site-year (rainfall was 245 mm higher than average) can explain the higher response to P at the zones with lowest elevation and higher ECa. Soybean yield response to P (Table 15) differed among ECZ zones only at Site 1a, where P increased yield in all zones but more for the ECZ-1 zone (not shown). This zone had intermediate ECa and the lowest STP (10 mg P kg–1), although all zones tested low (Table 9).

Approaches based on Combinations of ECa and Topography Attributes
The corn yield response to P fertilization differed among EECZ zones in one site (Site 3a, Table 14) and did not differ among EECSZ zones at any site. At Site 3a, the yield response to P was higher (not shown) for the EECZ-2 zone, which had the highest ECa and intermediate elevation but also had optimum STP that was higher than for two other zones (Table 9). The analysis for EZ had shown larger yield response and lower STP at lower elevation (EZ-1) than at higher elevation, and the analysis for ECZ showed larger yield response and STP for the two zones with higher ECa (ECZ-1 and ECZ-2). Therefore, a larger yield response for the EECZ-2 zone might be explained by excessive rainfall (not shown) and the wet, low-lying topographic position. It is possible that in the lower-testing and better-drained parts of the field, root development and P uptake were not affected, while in the EECZ-2 zone the crop became more dependent on P application.

Potassium Trial Sites
Soil Survey Map Approach
Yield response to K differed (P ≤ 0.05) between SMZ zones only for corn at Sites 7a and 7a2 (Table 16 ). In both years the yield response was larger for the Colo-Ely soil complex than for the Tama soil. We cannot explain this result with certainty because mean STK was higher for the Colo-Ely complex and corn yield levels did not differ consistently between soils. Perhaps high yield response in small low-testing areas explains the larger response for the Colo-Ely soil because the STK variability for this complex was the highest of all soil series (Table 8).


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

  		Table 16. Corn and soybean yield responses to K fertilization for field areas with different soil series (survey map zones, SMZ).

 
Elevation Approach
The yield response to K differed among EZ zones only for corn at Sites 6b2 and 7a2 (Table 17 ). At Site 6b2, the yield response to K was higher (not shown) for the zone with the lowest elevation (EZ-1, Table 10), but we cannot explain this result with certainty because mean STK of the two zones was similar. The responsive zone had STK with lower SD and a higher overall yield level (not shown). Perhaps the larger response is explained by a combination of more uniform STK in a responsive class and higher yield potential. At Site 7a2, a larger corn yield response to K (not shown) was observed for the intermediate and high EZs (EZ-2 and EZ-3, Table 10), which is reasonable because mean STK for these zones was low while it was optimum for EZ-1.


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

  		Table 17. Statistical significance of the corn yield response to K fertilization for zones delineated using different sampling approaches.

 
Approaches based on ECa and Topography Attributes
The approach based on ECa alone never identified a within-field differential yield response to K. However, ECa in combination with topography attributes identified a differential corn yield response to K at one corn site-year and one soybean site-year. The corn yield response differed among EECZ and EECSZ zones at Site 7a2 (Table 17). The response was greater for EECZ-2 (not shown), which had intermediate elevation and STK (a value borderline with the low and optimum classes) and the highest ECa compared with other zones (Table 10). The response was greater for the EECSZ-1 and EECSZ-2 zones. These zones had intermediate elevation, but EECSZ-1 was the least steep and had the highest ECa and STK, while EECSZ-2 was the steepest and had the second-highest ECa and the lowest STK (Table 10). The soybean yield response to K differed among EECSZ zones at Site 5a2 (Table 18 ). The yield response was similar for EECSZ-1 and EECSZ-2 and there was no response for EECSZ-3 (not shown). Elevation and slope did not explain the differential responses among these zones (Table 10), but the two responsive zones had the highest ECa and lowest STK (values in the low class compared with the optimum class for EECSZ-3). We cannot explain with certainty the differential corn yield responses among zones, although the responsive zones had the highest ECa and/or lowest STK. In this region, high ECa suggests finer soil texture, exchangeable Ca, and moisture retention capacity, which might explain a higher crop response to K fertilization even when STK was not the lowest but still within a responsive interpretation class. However, we cannot explain the reason ECa alone did not identify a differential crop response to K in any field.


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

  		Table 18. Statistical significance of the soybean yield response to K fertilization for zones delineated using different sampling approaches.

 
Summary of Crop Response to Fertilization Across Zones
The grain yield response to P or K fertilization seldom differed across zones of SMZ, EZ, ECZ, EECZ, or EECSZ. In the P trial sites (16 site-years), a differential crop response was identified by EZ at three sites (Sites 1a, 2b2, and 3a), by ECZ at two sites (Sites 1a and 3a), and by EECZ at one site (Site 3a). In the K trial sites (11 site-years), a differential response was identified by SMZ at two sites (Sites 7a and 7a2), by EZ at two sites (Sites 6b2 and 7a2), and by either EECZ or EECSZ at two sites (Sites 5a2 and 7a2). In very few instances could a within-field differential response be explained by the mean or median STP or STK of the zones. The soil-test values usually were quite similar across zones, were within the same interpretation class, or the differential response did not follow the expected relationship with soil-test values. In contrast, GC identified a within-field differential yield response directly explained by soil-test values in five P trial sites and three K trial sites. The zone approaches were useful to identify field areas with different yield levels, which could be useful for nutrient management because P and K removal is considered for determining maintenance fertilization rates. However, yield maps from already commonly used yield monitors are more useful in this regard because they directly describe within-field yield variation.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Study of yield response to fertilization for the strip averages of each site indicated a response in 20 of 27 site-years and only when median STP or STK was <20 mg P kg–1 (Bray-P1 or Mehlich-3 tests) or <171 mg K kg–1 (ammonium-acetate test). Analysis of yield responses for field areas testing within different soil-test interpretation classes as described by a dense (0.08 to 0.25-ha) grid-point sampling approach showed differential responses in eight of 13 responsive P site-years and in all eight responsive K site-years. Responses were large and frequent for low-testing areas (<16 mg P kg–1 or <131 mg K kg–1) and infrequent for areas testing up to 20 mg P kg–1 or 170 mg K kg–1.

The GC approach, which was based on a cell size similar to that commonly used in corn and soybean production areas of the Midwest, was 50% as efficient at identifying within-field response variation as the denser grid-point approach; GC identified within-field response variation at 8 of 27 site-years, while DG identified it at 16 of 27 site-years. The zone sampling approaches were less successful in this regard, and a differential crop response was identified by EZ, EECZ or EECSZ, ECZ, and SMZ in five, three, two, and two site-years, respectively. The differences in response among zones seldom were explained by differences in soil-test values because usually these were numerically similar, within the same interpretation class, or the relationship was contrary to what might be predicted (i.e., greater response with higher soil-test values). Zoning approaches often identified areas with different yield levels, which may be useful when yield maps are not available to consider possible differences in nutrient removal for soil-test maintenance recommendations.

The efficacy of the approaches evaluated in this study may differ for this and other regions depending on the method used to collect composite soil samples, the number of cores per sample, the soil series within a field, and the field management history. The DG approach showed very large soil-test variability in most fields. For the simulated sampling approaches, we used more composite samples and more soil cores per cell or zone than those commonly used in production agriculture, so our results likely indicate the highest potential efficacy of these approaches. Long histories of fertilization for originally low-testing Iowa soils probably reduced the impact of topography and soil properties on soil-test variation. Overall, the results for the fields and methods in this study indicated that zone sampling approaches were less efficient than grid sampling at predicting yield response to P and K fertilization, although zone sampling may be more efficient in fields with greater contrast in soil series within fields and shorter histories of fertilization.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Research supported in part by the Foundation for Agronomic Research (FAR) and the Iowa Soybean Association.


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




This article has been cited by other articles:


Home page
 Agron. J.Home page
J. Sawchik and A. P. Mallarino
Variability of Soil Properties, Early Phosphorus and Potassium Uptake, and Incidence of Pests and Weeds in Relation to Soybean Grain Yield
Agron. J., September 8, 2008; 100(5): 1450 - 1462.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
D. L. Shaner, R. Khosla, M. K. Brodahl, G. W. Buchleiter, and H. J. Farahani
How Well Does Zone Sampling Based on Soil Electrical Conductivity Maps Represent Soil Variability?
Agron. J., September 8, 2008; 100(5): 1472 - 1480.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
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 HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Agricola
Right arrow Articles by Sawchik, J.
Right arrow Articles by Mallarino, A. P.
Related Collections
Right arrow Nutrient Management
Right arrow Site-Specific Analysis
Right arrow Soil Fertility and Productivity

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