| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Agron., Iowa State Univ., Ames, IA 50011
* Corresponding author (apmallar{at}iastate.edu)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: DGPS, differential global positioning systems • GIS, geographical information systems • RCBD, randomized complete block design
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
In contrast to major research efforts with foliar fertilization at soybean reproductive stages, little research has been conducted on foliar fertilization at early vegetative stages. Foliar fertilization at early stages could increase P and K supplies at the time when the root system is not well developed. Also, a small amount of foliar-applied N is not likely to inhibit N2 fixation and could increase plant development and grain yield. Rosolem et al. (1982) showed no yield differences when two N–P–K formulations (5–15–15 and 14–4–7) were sprayed at 45 and 60 d after seedling emergence. The soil tested lower than optimum for both P and K according to local interpretations and received 37 and 19 kg ha-1 P and K, respectively, at planting. Research in Iowa (Haq and Mallarino, 1998) showed that foliar fertilization with various rates of a commercial 3–8–15 fertilizer applied at the V5 growth stage (Fehr et al., 1971) increased soybean grain yield in 7 of 48 trials and reduced yields slightly in two trials, with a mean yield increase of 54 kg ha-1 across all trials. The results of this study showed that responses as high as 700 kg ha-1 are possible, even in soils that test optimum or higher in P and K. More recently, Haq and Mallarino (2000) showed that N–P–K foliar fertilization with 28 to 56 L ha-1 of 3–8–15, 10– 4–8, or 8–0–7 fertilizers sprayed at the V5 stage affected yields significantly at 6 of 27 sites. Some or all treatments either increased yield up to approximately 400 kg ha-1 in some sites or decreased yield by a similar amount at other sites. The 3–8–15 fertilizer caused no leaf damage, and other fertilizers caused little damage; however, leaf damage was not related to yield decreases. Although no simple relationship between yield response and site variables occurred in the two studies, multivariate factor analyses showed that the responsive sites had lower plant P concentration in soybean tissue at the V5 or R2 growth stages; lower rainfall in late spring to midsummer; and lower N, P, and K uptake at the R2 growth stage (nutrient uptake at the V5 growth stage was not measured) compared with nonresponsive sites.
Most fields used by Haq and Mallarino (1998)(2000) had soils that tested optimum or above in P and K according to current soil test interpretations for soybean in Iowa (Voss et al., 1999). Furthermore, only N–P–K mixtures were used. Although soybean usually has not responded to soil-applied S in Iowa and other midwestern states (Webb, 1990; Sweeney and Granade, 1993; Sexton et al., 1998), the response to foliar fertilization observed by Garcia and Hanway (1976) occurred only when S was added to N–P–K mixtures. Others have reported conflicting information concerning soybean response to foliar-applied S or micronutrients at various reproductive stages (Boote et al., 1978; Varsa, 1978; Parker and Boswell, 1980; Poole et al., 1983; Reinbott and Blevins, 1995). Leaf damage due to higher concentrations of N and S was suggested as one possible cause for either yield decreases or the lack of yield increases.
On-farm research using 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 relatively narrow and long strips, which usually are as long as the field length, and the grain is harvested with farm combines and weighed using large-capacity balances. Precision farming technologies that include yield monitors, differential global positioning receivers (DGPS), and geographical information systems (GIS) have been successfully adapted to these types of field trials (Oyarzabal et al., 1996; Reetz, 1996; Mallarino et al., 1998). Furthermore, these technologies can be used to assess within-field variation in yield response to fertilization resulting from variation in soil nutrients, soil types, topography, and other factors.
This study was conducted with two main objectives. One objective was to evaluate soybean yield response to foliar fertilization at early vegetative stages with commercial fertilizers varying in N, P, K, S, and micronutrient composition. The second objective was to study the within-field variation in soybean yield response to one of the fertilizers used in the small-plot trials using replicated strip trials and precision agriculture technologies.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Two treatments were evaluated in the strip trials. One was a control with only glyphosate herbicide, and the other was 28 L ha-1 3–8–15 fertilizer (without S or micronutrients) mixed with the glyphosate. This fertilizer was selected for these trials because it produced the most consistent responses in previous studies (Haq and Mallarino, 1998, 2000). The fertilizer treatment was applied with farm-size equipment when soybean was at the V4 to V5 growth stage. Glyphosate was applied following label specifications at rates of 0.84 to 1.05 kg ha-1 acid equivalent using flat-fan spray tips with approximately 100 L ha-1 water. The fertilizer was mixed into the applicator tank immediately before the application. The experimental design was a RCBD with either three or four replications. These two treatments were applied to strips 18 or 27 m in width and as long as the fields. Although treatments were sprayed along the entire field length, the length used varied from 300 to 780 m because borders (40–50 m) were not evaluated.
Measurements
Composited soil samples (12 cores per sample) were collected each spring from the 0- to 15-cm soil depth for all trials. In the small-plot trials, samples were collected from each replication. In strip trials, composite soil samples were collected from grid cells that were 0.12 to 0.30 ha in size (depending on the field). Grid lines were spaced 60 to 90 m in the direction along the length of the strips and were spaced twice the width of a strip across the strips so that the width of each cell coincided with the width of a replication (there were two strips in each replication). Soil samples were analyzed for pH, organic matter, P, K, Ca, Mg, Fe, and Zn by procedures described for the soils of the North-Central region (Brown, 1998). Briefly, pH was analyzed in a 1:1 soil/water ratio; organic matter by loss of weight on ignition; P by the Bray P-1 method; K, Ca, and Mg by the ammonium acetate [NH4(C2H3O2)] method; and Fe and Zn by the 0.1 M HCl method. Cation exchange capacity was estimated by the sum of K, Ca, and Mg extracted by the ammonium acetate extractant and exchangeable H+ as estimated from measurements of pH and buffer pH by the Shoemaker, McLean, and Pratt method (Brown, 1998). Soil test values are shown for small-plot trials in Table 2 and for strip trials in Table 3. Soil test interpretation classes from Iowa State University were used for soil test P and K (Voss et al., 1999). Soil test P (0–15 cm depth, Bray P-1 extractant) boundaries for the classes very low, low, optimum, high, and very high are 0 to 8, 9 to 15, 16 to 20, 21 to 30, and >30 mg kg-1, respectively. Corresponding classes for K are 0 to 60, 61 to 90, 91 to 130, 131 to 170, and >170 mg kg-1, respectively.
|
|
Data Management and Analysis
For data from small-plot trials, analyses of variance were conducted for yield from each site, across the nine sites within each year, and across the 18 sites according to RCBD and using the procedure GLM of SAS (SAS Inst., 1996). The treatment sum of squares was partitioned into an orthogonal contrast of the control vs. the mean of the fertilized treatments. Means were further compared by LSD (P 
0.1) when either the treatment main effect or the orthogonal contrast of the control vs. the mean of all fertilized treatments was significant (P 0.1).
For the strip trials, yield and soil test data were imported into ArcView (Environ. Syst. Res. Inst., Redlands, CA) for GIS management and to SAS for statistical analyses following two procedures. In one procedure, the design assumed was a RCBD that accounted for the spatial correlation of yield. This procedure, its applications, and minor modifications were described in detail by others (Hinz, 1987; Bhatti et al., 1991; Hinz and Lagus, 1991; Stroup et al., 1994; Mallarino et al., 1998). In this study, we followed the procedure adapted for strip trials harvested with yield monitors described by Mallarino et al. (1998). Briefly, the procedure used had three main steps. The first step involved preparing the yield input data for the analysis. The yield values used were means of values recorded by the yield monitor (at 1-s intervals in this study) for small areas delineated by the width of the combine head and the length of a soil sampling cell along the crop rows. Individual yield monitor data were not directly considered because of the known lack of accuracy of yield monitors over distances <30 to 40 m (Colvin et al., 1995; Lark et al., 1997). The second step involved using nearest-neighbor analysis to calculate values of a yield covariate (one value for each yield input value) that were later included in a RCBD analysis of covariance. The covariate values were calculated from yield residuals after removing treatment and block effects using a conventional RCBD and subtracting each yield residual from the mean value of its four neighbors.
The other procedure assessed treatment effects for areas of each field having contrasting soil test values and different soil types. The procedure used was first described by Oyarzabal et al. (1996) and was later used by Mallarino et al. (1998). The yield input data were yield means for small areas defined by the width of each strip (not of each combine trip as in the first procedure) and the length of a soil-sampling cell along the crop rows. The soil test input data were values for each area defined by the width of each replication (i.e., two strips) and the separation distance of the sampling grid lines in the direction along crop rows. Thus, two yield values (one for each treatment) matched one soil test value, both representing areas of similar size. Each soil test value for P and K was assigned to interpretation classes defined by Iowa State University. The pH values were assigned to acidic (less than pH 6.0, for which lime is recommended), slightly acid or near neutral (pH 6.0 to 7.0), and alkaline [pH higher than 7.0, due to calcium carbonate (CaCO3) in all cases] classes. As expected, not all classes were represented in each field. Other soil properties (organic matter, cation exchange capacity, Ca, and Mg) were arbitrarily assigned to classes low, medium, and high by splitting the values found in each field. A particular class was included in the analysis only if at least two replications (blocks) had at least one cell with a soil test value corresponding to that class. This restriction assured that there were at least two true replications for the analysis. Thus, the analysis included estimates for effects of blocks, treatment, soil test class, and the interaction of the treatments with the soil test classes. The soil test classes were considered as repeated measures within the experimental units. A significant treatment x soil test class interaction suggests that treatment effects differed for areas of the field with different soil test levels.
Simple correlation and regression analysis (CORR and GLM procedures of SAS) were used to study relationships between relative yield increases due to foliar fertilization. Relative yield increases (one per site) were calculated for a combined analysis across all trials (26) and for a within-field analysis for each strip trial. For the analysis between trials, the mean yield of the control was subtracted from the mean of the statistically highest-yielding treatment (or treatments) or from the mean of all treatments for nonresponsive sites. The result was then divided by the control mean and multiplied by 100. The matching soil test values used were means for each experimental area. For a within-field analysis of each strip trial, one relative yield value was calculated for each area defined by a replication (across strips) and the length of a soil-sampling cell (along strips). Thus, the number of data pairs was similar to the number of cells in each field (32–59, depending on the field).
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
0.1) below the yield of the control in this study.
|
The results showed no evidence of yield response to the micronutrients mixed with the 10–4–8–1 fertilizer. Iowa State University does not provide micronutrient soil test interpretations for soybean. Interpretations for three midwestern states derived mostly from Michigan soils (Vitosh et al., 1995) for the extractant used in this study suggest that Fe was low only in Site 6 (but no treatment increased yield at this site), and Zn was adequate in all soils. Addition of S slightly improved the effect of the 10–4–8 fertilizer on soybean yield. The 10–4–8–1 fertilizer produced higher yields than the control and the 10–4–8 fertilizer across all sites. However, yields for the 10–4–8–1 and 3–8–15 fertilizers did not differ, and the 3–8–15–1 mixture did not increase yields over the control. The lack of yield response to the 3–8–15–1 fertilizer compared with the 3–8–15 and 10–4–8–1 fertilizers cannot be fully explained with the methods used in this study. It is possible that a 3–8–15–1 ratio was unfavorable for plant growth. The nutrient ratio in the 10–4–8–1 fertilizer is more similar to the best N–P–K–S ratio found by Garcia and Hanway (1976) for foliar fertilization of soybean at reproductive stages. These authors reported that the most successful mixture was a 10–1–3–0.5 (N–P–K–S) ratio, and they noted that this ratio closely resembles the nutrient ratio in soybean seeds. The advantage of the 10–4–8–1 mixture compared with the 3–8–15–1 mixture occurred in spite of small leaf damage in some sites. Visual estimates showed that only the 10–4–8–1 and 10–4–8–1–M fertilizers produced a barely detectable leaf damage (<5% of the leaf area) in Sites 9, 13a, 16, 17, and 18. It is likely that the higher proportion of N in the mixture with S caused this light damage.
Yield Responses in Eight Strip Trials
The results of the eight strip trials (Table 5) showed a yield response (P < 0.1) to the 3–8–15 fertilizer only at Site 2. The response was very small (35 kg ha-1) and was not observed in the adjacent small-plot trial at this site (Table 4). Also, the small responses observed in other small-plot trials were not observed in the strip trials conducted in the same fields. The general lack of response could be partly explained by predominantly high soil-test P and K values. The mean values across each field were optimum or higher for soybean in all fields although there was large variation within each field (Table 3). Only Sites 10, 11, and 13b had cells with soil test P values testing low, and the low-testing areas were 25% in Site 10, 16% in Site 11, and 5% in Site 13b. Only Sites 3 and 10 had cells with low soil K, and the areas testing low were 24% in Site 3 and 14% in Site 10. Moreover, areas testing low in P and K seldom coincided.
|
Analyses of variance and regression analyses showed no statistically significant yield response for field areas that tested below optimum in soil test P (such as in Sites 10 and 11) or soil test K (such as in Sites 3 and 10). Various reasons could explain this result. One is that other unknown growth factors could have been more limiting than soil P or K availability. Another possible reason is that low-testing areas identified by grid-point sampling were seldom contiguous across each field, and areas testing low in P and K seldom coincided. Other research has shown very high small-scale variability in soil P and K (Mallarino et al., 1998) and failures of yield monitors to appropriately measure yield of small areas (Colvin et al., 1995; Lark et al., 1997), which may explain a lack of relationship between estimates of yield response and soil test values from field-scale trials.
Comparison of the yield responses observed in this study with prevailing prices for the fluid fertilizers and soybean grain prices suggest that foliar fertilization of soybean across all production conditions will not result in an average yield response large enough to offset fertilizer and application costs. An informal telephone survey among several Iowa distributors conducted by the authors in spring 1999 showed retail prices ranging from $0.53 to $0.92 L-1 for the fluid fertilizers used in this study. Mean soybean grain price for 1999 in the United States was approximately $0.20 kg-1. The mean yield response to 3–8–15 fertilizer across all small-plot trials was 93 kg ha-1 and was almost zero across all strip trials. Thus, if mean prices are assumed, the mean yield response would not offset the cost of the fertilizer. The probability of economic benefits could be increased, however, if available information before fertilization can be used to restrict applications to conditions that increase the probability of response. The methods used do not allow for recommendations or speculations concerning soil, management, or environmental factors that could increase the probability of response. Previous research (Haq and Mallarino, 1998, 2000) suggested, however, that responses are more likely with low rainfall in late spring to midsummer and when growing conditions result in low N, P, and K uptake. Fertilization costs could be decreased by mixing the fertilizer with a postemergence herbicide. A variety of reasons led us to use glyphosate herbicide for glyphosate-resistant soybean in this study. Such reasons include its widespread use in the Midwest, information from the manufacturers of the fluid fertilizer and glyphosate suggesting unlikely negative interactions from the mixture, and that glyphosate does not produce leaf damage. This study does not imply that fluid fertilizers should be mixed with glyphosate or other herbicides, and further research should address possible interactions between fluid fertilizers and postemergence herbicides.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. U. Haq and A. P. Mallarino Response of Soybean Grain Oil and Protein Concentrations to Foliar and Soil Fertilization Agron. J., May 13, 2005; 97(3): 910 - 918. [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]
|
|
|
|
|
|
A. A. Bianchini and A. P. Mallarino Soil-Sampling Alternatives and Variable-Rate Liming for a Soybean-Corn Rotation Agron. J., November 1, 2002; 94(6): 1355 - 1366. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Bermudez and A. P. Mallarino Yield and Early Growth Responses to Starter Fertilizer in No-Till Corn Assessed with Precision Agriculture Technologies Agron. J., September 1, 2002; 94(5): 1024 - 1033. [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 | |||
