| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Agron., Iowa State Univ., Ames, IA 50011
* Corresponding author (apmallar{at}iastate.edu)
Received for publication October 3, 2001.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: ANOVA, analysis of variance • GIS, geographical information system • GPS, global positioning system • ISU, Iowa State University • NNA, nearest-neighbor analysis • RCBD, randomized complete block design • SD, standard deviation • STK, soil-test potassium • and STP, soil-test phosphorus
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Starter fertilization is a common fertilizer practice in some areas of the USA to increase early growth and grain yield. Small fertilizer amounts are applied at planting time in the furrow with the seeds or in a band beside and below the seeds. Mascagni and Boquet (1998) showed the potential harm of direct contact between high rates of fertilizer and corn seed but also that application of N and P fertilizers in the furrow at low rates can advance silking date, increase grain yield, and decrease grain moisture at harvest. In Iowa, for example, an N–P–K starter mixture is recommended for corn when soil conditions are expected to be cooler and wetter than normal and with high residue cover (Voss et al., 1999). Mengel et al. (1992) found that starter fertilization increased corn yield in only one site under conventional tillage but in eight sites under no-till management in Indiana. The response to starter fertilizer usually is attributed to the P in the mixture (Randall and Hoeft, 1988), which is consistent with known high-P requirements for early plant growth and development. In some situations, however, responses to N also occur (Ritchie et al., 1995). Scharf (1999) found larger responses to N–P starter fertilizers compared with N-only starter in sites where STP was low but found no differences when STP was above optimum. Rehm et al. (1988) suggested that the magnitudes of increased growth and yield due to starter fertilization increase when the starter is applied to soils with low STP but also found significant responses to P–K starter fertilization in high-STP soil during a cool and wet spring season. Starter fertilization often increases early growth to a greater extent than grain yield (Randall and Hoeft, 1988; Bullock et al., 1993). Mengel et al. (1988) observed that seed-applied N–P starter fertilizer in southern Indiana soils enhanced early growth and increased corn yield in both plow and no-till systems at low rates of P and K preplant fertilization. At higher preplant fertilization rates, however, starter fertilization enhanced early vegetative growth but had no significant effect on grain yield. Furthermore, the starter fertilizer increased grain yield in soils with high STP only when no fertilizer or low fertilizer rates were applied before planting. The effects of starter in enhancing corn growth and development probably explains results of research in Wisconsin (Bundy and Andraski, 1999), which showed that the response to starter fertilization was more frequent for hybrids of long relative maturity that were planted late.
The advent of yield monitors, GPS, and GIS provides researchers and producers useful tools to describe and study soil nutrient variability over the landscape. Major factors producing soil-test variability are soil types, topography, previous crops, and soil management practices such as tillage. This variability likely results in different responses to fertilization within fields. The ability to accurately record geographical coordinates and data for specific locations within a field can improve field-scale research. On-farm research on the basis of strip plots is an accepted methodology for complementing traditional small-plot research, generating local recommendations, and demonstrating management practices (Rzewnicki et al., 1988; Shapiro et al., 1989). Treatments are applied to narrow and long strips (usually the length of the fields), and the grain is harvested with common combines and weighed using large-capacity balances. Precision agriculture technologies were successfully adapted to these types of field trials (Oyarzabal et al., 1996; Mallarino and Wittry, 1997; Mallarino et al., 2001). Evaluation of treatments using classical ANOVA may not adequately account for the spatial variability encountered in long and heterogeneous strips. Procedures that account for spatial correlation, such as NNA, provide better alternatives. Accounting for spatial correlation with NNA or other techniques can reduce the experimental error and make the analysis more sensitive in discerning treatment differences (Hinz, 1987; Bhatti et al., 1991; Hinz and Lagus, 1991; Stroup et al., 1994; Mallarino et al., 1998; Helms et al., 1999).
Research is needed to study the relationships between early growth and grain yield responses of no-till corn to starter fertilization across varied production conditions. Furthermore, research is needed to assess within-field variation of responses to starter over landscapes. The objective of this study was to evaluate yield and early growth responses to starter fertilization for no-till corn in farmers' fields using precision agriculture technologies and statistical analysis that account for spatial correlation.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Soil samples collected at planting followed a systematic grid-point sampling scheme. The width of the cells in the direction perpendicular to the future corn rows coincided with the width of each replication, and their length (parallel to the corn rows) ranged from 24 to 36 m, depending on the field. Composite samples (10 to 12 cores from a 15-cm depth) were collected from an area approximately 20 m2 in size located at the center of each cell. Soil samples were analyzed for P (Bray-P1 method), K (ammonium acetate method), organic matter (Walkey–Black method), and pH (1:1 soil/water) following standard soil-test procedures recommended for the North-Central region (Brown, 1998). Iowa State University (ISU) soil-test interpretation classes for P and K in corn grain production (Voss et al., 1999) were used to classify soil-test ranges. Five STP classes were (i) Very Low (
8 mg kg-1), (ii) Low (9–15 mg kg-1), (iii) Optimum (16–20 mg kg-1), (iv) High (21–30 mg kg-1), and (v) Very High (
31 mg kg-1). Five soil-test K (STK) classes were (i) Very Low (60 mg kg-1), (ii) Low (61–90 mg kg-1), (iii) Optimum (91–130 mg kg-1), (iv) High (131–170 mg kg-1), and (v) Very High (170 mg kg-1). The aboveground part of corn plants was sampled when corn height to the center of the whorl averaged 15 to 25 cm across treatments and field areas (V5 to V6 growth stage). Ten plants were cut at ground level from the center of each of the two treatment strips within each replication and each soil sampling cell along the crop rows. Thus, two plant samples (one that received starter and one without starter) correspond to one soil sample for all fields. Plant samples were dried at 60°C and weighed.
Grain yields were measured and recorded using combines equipped with yield monitors and differential GPS receivers. The yield monitors used were impact flow-rate sensors Ag Leader 2000 (Ag Leader Technol., Ames, IA), Green Star (John Deere, Moline, IL), or Micro-Trak (Micro-Trak Syst., Eagle Lake, MN). Differential corrections were obtained through the U.S. Coast Guard AM signal. The spatial accuracy was checked by georeferencing several positions in the field (border rows and natural markers such as waterways) with a hand-held differential GPS receiver. Yield data were unaffected by field borders because at least 40 m from any border was harvested but not used. While harvesting, each combine trip (a 4.5-m swath) was identified with a unique number that was recorded with the georeferenced yield data. The raw yield data recorded by the yield monitors were carefully analyzed for common errors such as incorrect geographic coordinates due to partial loss of good differential correction, the effects of waterways, and incorrect settings in the time lag for the grain path through the combine. Affected data were corrected (such as grain path lags) or deleted (for example, yield points near waterways and when a combine stopped within the trial area). The data were imported into spreadsheets and then exported to ArcView (Environ. Syst. Res. Inst., Redlands, CA) for GIS work and later to the SAS statistical package (SAS Inst., 1996) for statistical analyses. The maps in Fig. 1 and 2 are ArcView layouts that show an example (for Field 3) of the strip trial methodology used and the type of maps generated using ArcView GIS. Figure 1 shows soil series and various soil-test values. Figure 2 shows treatments, yield points, means of grain yield and early growth by strip, and grain yield differences by strip and soil sample cell.
|
|
Procedure 3 assessed treatment effects separately for parts of the experimental areas with different soil-test values or soil map units following a procedure described by Oyarzabal et al. (1996) and later by Mallarino et al. (2001). Five analyses were performed to consider separately STP, STK, pH, organic matter, and the soil series at each field. ArcView GIS was used to produce the input data from different areas of each field for these analyses. The yield data were means for areas defined by the width of each strip (12–24 m) and the separation distance along crop rows of the soil sampling grid lines (24–36 m). Each line of data in the resulting data set for each field consisted of a yield value and a set of five codes that corresponded to an interpretation class of the four soil tests and to one soil series. The soil-test data were values for areas defined by the width of each replication and the separation distance of the sampling grid lines in the direction along crop rows. Values were classified into the five ISU interpretation classes for STP and STK and into arbitrary interpretation classes for pH (pH < 5.5, 5.5–6.2, 6.3–7.0, and >7.0) and organic matter (<30, 30–40, and >40 g kg-1, respectively) because ISU and most other universities do not have specific interpretation classes for these soil tests. Field areas were coded for soil series using digitized soil survey maps based on a 1:12000 mapping scale (Iowa Coop. Soil Survey, 2001). The number of usable soil series for each field varied from three to seven. The F test from a one-way ANOVA was used to estimate the consistency of starter effects for each interpretation class of each soil test and for each soil series. The numerator mean square (between groups) represented variation introduced by the treatments (starter and no starter), and the denominator means square represented variation within groups (cells with a similar classification). Values were not used for these analyses when there were less than three yield cells for any soil-test class or soil series within a field.
Treatment effects on corn early growth (dry weights at V5 to V6 growth stages) were analyzed with the three procedures in the same manner as described using the yield data. The growth input data were derived from one sampling point from each small cell defined by the treatment strip and the separation distance of grid sampling lines along crop rows.
The Correlation (CORR) and General Linear Models (GLM) procedures of SAS were used to study relationships between relative yield increases due to starter fertilization and soil-test values for areas defined by each strip and soil sampling cell. Relative yield increases were used in an attempt to minimize differences in absolute yields between fields and areas within a field. The relative increases were calculated from treatment means (without starter and with starter for the area defined by a soil sampling cell) by subtracting the yield without starter from the yield with starter, dividing by the yield without starter, and multiplying by 100.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
0.1) yield increases at Fields 2, 5, and 11 and markedly increased the significance of the response at Field 8. The responses at Fields 5 and 11 were small, however (least-square means differences were 148 and 179 kg ha-1, respectively). The starter treatment did not influence yield variability consistently (Table 4). Trends were the opposite for two fields (Fields 1 and 3), and there were no significant differences in yield variability for other fields. Starter fertilization did not influence grain moisture recorded with the yield monitors significantly (P 0.1) at any field (data are not shown). The yield responses to starter fertilization at Fields 3 and 7 could be explained by deficient STP because these were the only fields in which the mean STP was below optimum for corn according to ISU soil-test interpretations (<16 mg kg-1 by the Bray-P1 test; Voss et al., 1999). The larger response at Field 3 (compared with Field 7) could be attributed to lower mean STP and proportionally larger areas testing Very Low or Low. However, other measured variables differed between these two fields. Spring rainfall and air temperature in April and June for Field 3 were the lowest among all fields (Table 2), whereas Field 7 received the highest spring rainfall among all fields, and air temperature was among the highest. Also, the hybrids used at these fields were different, and plant population was markedly lower at Field 7 than at Field 3. Neither the large response at Field 8 nor a smaller response at Field 11 can be obviously explained by STP (because the mean STP was in the Very High class in both fields), rainfall or temperature (because both were among the highest and better distributed during the April to June period among all fields), or other soil tests. These were the only two fields where the starter was applied beside and below the seeds, however, and the rates applied were the highest (163 and 170 kg ha-1) among all fields. Although many factors may have determined the response to starter in these two trials (which cannot be fully identified with the methods used), one likely factor was N availability. Nitrogen could also explain smaller responses in other soils with high STP and where the starter was applied in the furrow (Fields 1, 2, and 5). Only Fields 10 and 11 received preplant N fertilization. Previous research has shown that responses to N–P–K starter usually are due to P, but often are also explained by the N in the starter when preplant or sidedressed N rates are not high enough, and seldom are explained by K (Randall and Hoeft, 1988; Ritchie et al., 1995; Scharf, 1999). The N rate applied beside and below the seeds at Field 8 (16.3 kg N ha-1) could explain the high response (465 kg ha-1) at this field because no N was applied before planting (N was sidedressed at the V6 growth stage at 145 kg ha-1). A smaller response (165 kg ha-1) in Field 11, which received starter (27.2 kg N ha-1) beside and below the seeds, could be explained by insufficient preplant N rate or N losses after application of N approximately 5 mo before planting. We cannot ignore the possibility of a response to a small amount of S applied with the starter at this field (Table 2). Small responses at other fields with STP optimum or higher for corn could be explained by the small amount of N (3.9–9.1 kg N ha-1) applied in the furrow or by no-till corn yield response to starter even at levels considered optimum or high.
Table 5 shows the influence of starter fertilization on early growth of corn for the nine trials where plants were sampled at the V5 to V6 growth stage (plant samples were not collected for trials conducted in 1995). Plant weights were higher for the starter treatments in all fields. Both methods of analysis showed significant (P < 0.1) responses in Fields 3, 6, 7, 9, 10, and 11. Differences in standard errors between statistical procedures were less important than for grain yield. Procedure differences were meaningful only in Field 3, in which the starter effect achieved statistical significance at the P < 0.05 level only with the RCBD-NNA procedure. The largest early growth differences were found in Fields 3, 4, 6, 7, and 9 (32–61% relative increase) where mean STP ranged from Very Low to Very High. Smaller differences were found at other fields (5–19%) where mean STP ranged from Optimum to Very High. Other studies found that early growth response to starter fertilization often does not relate consistently to soil-test values or weather conditions (Mengel et al., 1988; Randall and Hoeft, 1988; Bullock et al., 1993). Starter fertilization increased early growth variability (P 0.10) in five fields (Fields 4, 6, 9, 10, and 11), which is in contrast with results for yield.
|
0.1) yield responses ranged from -2.7 to 9% (4.1% mean) for areas testing Very Low or Low in STP, -2.7 to 3.8% for areas testing Optimum (0.6% mean), and -4.2 to 5.4% for areas testing High or Very High (2.0% mean). However, responses were not observed in all low-testing areas and sometimes were observed in high-testing areas. For example, starter fertilization increased yield for all STP classes present at Field 1 (Low to Very High), and there was no response at any area with similar STP classes at Field 5. It is noteworthy, however, that the RCBD-NNA analysis detected an overall small response at Field 5 (Table 4). One possible reason for these discrepancies is that soil tests are not perfect estimates of soil P availability (due to analytical or sampling error). Other possible reasons for responses to starter fertilization in some high-testing soils include response to nutrients other than P (most likely N) and the influence of other growth factors that could not be identified with the methods used in this study.
|
|
0.1) only in Fields 1, 2, and 7. Responses were higher for the highest organic matter class (>40 g kg-1) in Fields 1 and 2 but were higher for the lowest class in Field 7 (<30 g kg-1). Similar inconsistent relationships were observed across fields when absolute or relative yield responses were regressed on organic matter values (trends were not significant or yield responses increased with increasing organic matter in some fields but decreased in others). In these landscapes, higher organic matter values usually are associated with higher late-spring soil moisture. The within-field variation in soil pH did not allow for a meaningful study of relationships between response and pH. Although most fields had areas of acidic pH where lime would be recommended according to ISU interpretations (pH < 6.0 or < 6.3, depending on the soil series; Voss et al., 1999), these areas comprised more than 10% of the experimental areas in only four fields (Fields 1, 8, 10, and 11). Responses were not related to pH in Fields 8 and 10 and were higher for the more acidic areas in Fields 1 and 11.
Yield responses to starter were inconsistent for field areas with contrasting soil series although most Iowa fields (and those in our study) do not have as highly contrasting soil properties as those of other regions. (Table 1 shows the two dominant soil series for each field.) We expected higher responses to starter fertilizer in field areas with map units representing low-laying, wet, and poorly drained soils (possibly colder and with low nutrient availability in spring) even though soil-test values could differ. Responses differed (P 0.1) only for soil series in five fields. Four fields (Fields 1, 2, 8, and 11, all in eastern Iowa) shared similar soil series although not all series were represented in all fields. In Fields 1 and 2, there was response in areas with Atterberry, Klinger, or Olin soil series and no response in areas with Dinsdale or Kenyon series. Dinsdale and Kenyon series are well-drained, loam or silty clay loam soils of upland and moderately sloping topographic positions; Olin is an excessively well-drained, sandy loam soil found in upland positions; and Atterberry and Klinger series are somewhat poorly drained, silty loam or silty clay loam soils found in upland positions with little slope. In Field 8, there was response in areas with Dickinson series and no response in areas with Klinger series. Dickinson series is an excessively well-drained soil found in upland positions with moderate slopes. In Field 11, there was response only in field areas with Clyde soil series and no response in areas with Dickinson, Dinsdale, or Klinger series. Clyde is a poorly drained soil found in concave areas of upland positions. Yield responses also differed for soil series of Field 7, which was in western Iowa. At this field, responses were significant for areas with Marshall series with 2 to 5% and 5 to 9% slope but not significant for areas of Marshall with steeper slopes or for areas mapped as Colo or Colo-Judson complex. The Marshall series is a well-drained, silty clay loam soil found in ridges or slopes of upland positions, and the Colo and Colo-Judson complex include poorly to moderately well-drained, silty clay loam soils found in valleys (often subject to flooding) and upland drainage ways.
Although much speculation is possible concerning reasons for these inconsistent results, the methods used in this study allow for few relevant comments. The lack of relationship between response to starter and soil map units or soil organic matter, which should be related to soil conditions during seedling emergence and early growth stages, could be explained by various reasons. Perhaps the most obvious and likely reason is that many factors interact. For example, soils of higher landscape positions usually had lower soil organic matter but likely were warmer early in the season (measurements were not collected) because these soils are better drained. Another possible reason is that the soil survey maps used and the 0.14- to 0.32-ha grid soil sampling method were not precise enough to describe variation in soil properties.
Correlation and regression analyses within and across fields showed a poor relationship between responses in early growth and grain yield. These analyses were not performed for fields where early growth was not measured (Fields 1 and 2) and where records of yield monitor data points within a strip were not recorded (Fields 4 and 6). The mean grain yield response to starter fertilization across all fields was very small (2.4%) compared with a much larger mean response of early growth (32%). Across all fields, relative yield response increased linearly as relative early growth response increased, but the correlation was poor (r = 0.44). Responses in early growth and grain yield were linearly correlated (P 0.05) in Field 3 (r = 0.41), Field 5 (r = 0.21), Field 7 (r = 0.24), and Field 8 (r = 0.56), and were not correlated at Fields 9, 10, and 11. These results suggest that starter fertilization increases early growth more and more frequently than grain yield and that large responses of no-till corn early growth are not necessarily reflected in grain yield. Obviously, growth conditions during the rest of the season also influence yield and may offset any starter effect in increasing early growth.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Early growth responses to starter were large and occurred in most fields and in most areas within fields. Growth responses were larger when STP was low but also were significant when STP was high. Across all fields, starter fertilization increased early growth by 32% and grain yield only by 2.4%. The grain yield response was positively but poorly correlated with early growth response within and across fields. Starter fertilization had no consistent effects on yield variability but often increased early growth variability. Overall, the results suggested that large yield responses of no-till corn to starter are more likely when STP is below optimum and (or) when preplant or sidedress N rates are deficient and that precision agriculture tools are effective for on-farm research.
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. M. Kyveryga, A. M. Blackmer, and T. F. Morris Disaggregating Model Bias and Variability when Calculating Economic Optimum Rates of Nitrogen Fertilization for Corn Agron. J., June 5, 2007; 99(4): 1048 - 1056. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Bermudez and A. P. Mallarino Impacts of Variable-Rate Phosphorus Fertilization Based on Dense Grid Soil Sampling on Soil-Test Phosphorus and Grain Yield of Corn and Soybean Agron. J., May 11, 2007; 99(3): 822 - 832. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. W. Roth, D. B. Beegle, S. M. Heinbaugh, and M. E. Antle Starter Fertilizers for Corn on Soils Testing High in Phosphorus in the Northeastern USA Agron. J., June 27, 2006; 98(4): 1121 - 1127. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. S. Wortmann, S. A. Xerinda, M. Mamo, and C. A. Shapiro No-Till Row Crop Response to Starter Fertilizer in Eastern Nebraska: I. Irrigated and Rainfed Corn Agron. J., January 5, 2006; 98(1): 156 - 162. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. E. Kaiser, A. P. Mallarino, and M. Bermudez Corn Grain Yield, Early Growth, and Early Nutrient Uptake as Affected by Broadcast and In-Furrow Starter Fertilization Agron. J., March 1, 2005; 97(2): 620 - 626. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. J. Niehues, R. E. Lamond, C. B. Godsey, and C. J. Olsen Starter Nitrogen Fertilizer Management for Continuous No-Till Corn Production Agron. J., September 1, 2004; 96(5): 1412 - 1418. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Bermudez and A. P. Mallarino Corn Response to Starter Fertilizer and Tillage across and within Fields Having No-Till Management Histories Agron. J., May 1, 2004; 96(3): 776 - 785. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. J. Wittry and A. P. Mallarino Comparison of Uniform- and Variable-Rate Phosphorus Fertilization for Corn-Soybean Rotations Agron. J., January 1, 2004; 96(1): 26 - 33. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
