Soybean Seed Composition, Aboveground Growth, and Nutrient Accumulation with Phosphorus and Potassium Fertilization in No-Till and Strip-Till Soil Fertility & Crop Nutrition
Bhupinder S. Farmaha, Fabián G. Fernández,* and Emerson D. Nafziger ABSTRACT
Strip-till can improve soybean [Glycine max (L.) Merr.] yield and increase P and K availability relative to no-till. Our objectives were to evaluate the effect of P and K rate and placement in no-till and strip-till on soybean-seed oil and protein content, aboveground growth and P and K accumulation, and soil water. A 3-yr field experiment was conducted near Urbana, IL, with tillage/fertilizer placement treatment as the main plots: no-till/broadcast (NTBC), no-till/deep band (NTDB), and strip-till/deep band (STDB) with banded fertilizer at 15 cm beneath the planted row. Phosphorus (0, 12, 24, and 36 kg ha–1 yr–1) was the subplot and K (0, 42, 84, and 168 kg ha–1 yr–1) was the sub-subplot. Higher protein and oil yields were produced with STDB than the no-till treatments. Increase protein yield with P fertilization occurred only for the no-till treatments, but STDB maintained higher protein yield than the no-till treatments when no P was applied, indicating that STDB was more effective at making P available to the crop. Phosphorus and K placement made no difference in protein and oil concentration or yield. Leaf area index (LAI) was greater for STDB than NTBC and NTDB and greater for NTDB than NTBC. Phosphorus and K fertilization increased LAI relative to the check by V2 stage illustrating the importance of fertilization for early growth. The advantage for soybean production with STDB over the no-till treatments was the result of greater soil water content in STDB during the reproductive stages.
S
oybean is typically planted in rotation with corn (Zea mays L.); and second to corn, is the most widely grown crop in the Midwest with approximately 26.8 million hectares planted in 2010 (USDA/NASS, 2010). In 2008, 62% of full-season, and 76% of double-cropped (planted after wheat [Triticum aestivum L.] harvest), soybean were planted under no-till or other conservation tillage systems in the United States (CTIC, 2008). While no-till provides soil conservation benefits relative to other tillage systems (Rhoton, 2000) and economic efficiency compared with conventional tillage (Smart, 1999), no-till can also pose challenges for crop production. Greater crop residue cover in no-till compared with conventional tillage systems lowers soil water evaporation and reduces the amount of heat radiation from the sun that reaches the soil. For these reasons, no-till generally results in wetter and cooler soils early in the growing season relative to conventionally-tilled soils (Fortin, 1993). Since corn is usually planted earlier than soybean, these early-season conditions constitute a greater concern for corn production and have been studied to a greater length than for soybean (Hoeft et al., 2000). However, wetter and cooler conditions can affect soybean production as well. Studies have indicated that cool and wet soil conditions can reduce availability of slow-mobile nutrients such as P and K (Barber, 1971) and decrease seed yield (Philbrook et al., 1991; Dep. of Crop Sciences, Univ. of Illinois, N-315 Turner Hall, MC-046, 1102 South Goodwin Ave., Urbana, IL 61801. Received 9 Jan. 2012. *Corresponding author (
[email protected]). Published in Agron. J. 104:1006–1015 (2012) Posted online 11 May 2012 doi:10.2134/agronj2012.0010 Copyright © 2012 by the American Society of Agronomy, 5585 Guilford Road, Madison, WI 53711. All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.
1006
West et al., 1996; Yin and Al-Kaisi, 2004). Also, in recent years, several reports have indicated a yield increase of up to 9% with early-planted soybean (late April) compared with late May (De Bruin and Pedersen, 2008). Potentially greater interest in early soybean planting could result in similar concerns for no-till soybean as those observed for no-till corn. Still, even if soybean planting practices do not drastically change toward the early season, currently, little is known about soybean cultivation with strip-till. Given some of the limitations of no-till for crop production, strip-till has been proposed as an option that combines the improved seedbed conditions of conventional tillage with the conservation benefits of no-till (Jones et al., 1994; Morrison, 2002). The narrow tilled, residue-free planting row created with strip-till can increase early-season soil temperature and aeration and reduce excess soil water in the seedbed similar to conventionally-tilled soils. Conversely, the undisturbed residue left on the soil surface outside the tilled strip protects soil from water and wind erosion and helps conserve soil moisture similar to no-till treatments (Hares and Novak, 1992). The enhanced seedbed conditions in strip-till have been reported to improve early planting, uniform emergence, and, ultimately, seed yield compared with no-till (Bolton and Booster, 1981; Kaspar et al., 1991; Vyn and Raimbault, 1992; Opoku et al., 1997; Vyn et al., 1998; Vetsch and Randall, 2002; Licht and Al-Kaisi, 2005; Perez-Bidegain et al., 2007). Farmaha et al. (2011) reported increased soybean yield with strip-till compared with no-till even without early planting, and attributed the benefit to improved soil conditions for nutrient uptake. Strip-till allows simultaneous subsurface application of fertilizer during tillage operation. Repeated broadcast P and K fertilizer applications in no-till often result in vertical Abbreviations: LAI, leaf area index; NTBC, no-till/broadcast; NTDB, notill/deep band; STDB, strip-till/deep band.
A g r o n o my J o u r n a l
•
Vo l u m e 10 4 , I s s u e 4
•
2 012
MATERIALS AND METHODS Site Description Soybean were grown for 3 yr (2007–2009) in rotation with corn at the Crop Sciences Research & Education Center near Urbana, IL, on a Flanagan silt loam soil (fine, smectitic, mesic Aquic Argiudolls) with small sections of Drummer silty clay loam soil (fine-silty, mixed, superactive, mesic Typic Endoaquolls). These soils are classified as somewhat poorly drained and poorly drained, respectively, but the site has artificial subsurface tile drainage. Pre-treatment soil test values in the top 18 cm of soil were: cation exchange capacity, 17 cmolc kg–1; organic matter, 3.7%; pH (1:1 soil/water ratio) 5.7; Bray P1 (Bray and Kurtz, 1945) 20 mg kg–1; and 1 M NH4OAc extractable K (Warncke and Brown, 1998) 167 mg kg–1. The study was conducted in a randomized complete-block design with a split-split-plot arrangement and three replications. The main (whole) plot included three tillage/fertilizer placement treatments: NTBC, NTDB, and STDB. The splitplot treatments were four P application rates (0, 12, 24, and 36 kg P ha–1 yr–1), and the split-split plot treatments were four K application rates (0, 42, 84, and 168 kg K ha–1 yr–1). For the unfertilized plots (0 kg P ha–1 yr–1 and 0 kg K ha–1 yr–1) each tillage/fertilizer placement treatment received the corresponding soil disturbance created by the application equipment. Each split-split plot was 6 by 23 m with 76-cm row spacing. Broadcast fertilizer treatments were applied using a hand-held spin-spreader. Band application of fertilizers was at 15-cm depth directly underneath the planting row using a Gandy Orbit Air applicator (Model 6212C, Gandy, Owatonna, MN). The NTDB treatments were applied using 2-cm-wide, low-disturbance, NH3 thin profile knives (minimum tillage knife Model 003-0000018, Fertilizer Dealer Supply, Philo, IL). The STDB operation was made with a wavy cutting coulter and row cleaners (residue managers) in front of modified NH3 knives (original mole knife, Model 003-0100411, Fertilizer Dealer Supply) with closing discs (berm shapers) behind the mole knife. Cultivar and planting date were confounded with year in our experiment. The cultivar Hi-Soy2846 (Maturity Group 2.8) was planted on 25 May 2007 and 13 June 2008 and cultivar Pioneer 93M42 (Maturity Group 3.4) was planted on 24 June 2009 at a seeding rate of 376,000 seeds ha–1, 5-cm deep in the soil.
stratification of these nutrients with greater concentrations in the soil surface layer (Holanda et al., 1998; Houx et al., 2011). This stratification may reduce nutrient availability and nutrient accumulation in soybean shoots when the nutrient-rich surface dries out, while subsurface applications may reduce the potential negative effect of soil surface drying (Eckert and Johnson, 1985). Hairston et al. (1990) reported that deep placement of K produced significantly greater no-till soybean seed yield than broadcast surface applications. Similarly, applying P and K at 15- to 20-cm depth in ridge-till increased soybean uptake of both nutrients as early as the V5–V6 development stage compared with broadcast treatments (Borges and Mallarino, 2003). By contrast, several studies have observed no additional increase in seed yield over broadcast applications with subsurface banding (Borges and Mallarino, 2000; Yin and Vyn, 2002; Rehm and Lamb, 2004). While it has been shown that soybean yields benefit with strip-till over no-till (Vyn et al., 1998; Farmaha et al., 2011), it is still poorly understood how or when during soybean development strip-till influences crop growth, nutrient accumulation, and, ultimately, seed yield. Similarly, the inconsistency in results from nutrient placement in the above-mentioned studies suggests the need for further investigation. In-season growth analysis has long been recognized as a way to quantify crop response relative to environmental factors (Radford, 1967; Hunt, 1982). For example, Yusuf et al. (1999) observed that compensatory growth can help early season, slow-growing notill soybean attain similar yields to conventionally-tilled soybean. Although soybean plants take up about 75% of the total K during pod development stages (starting at R3) (Hanway and Weber, 1971), Fernández et al. (2008, 2009) found that having adequate K supply early in development was important to improve no-till soybean growth and seed yield. Finally, not many studies have been designed to evaluate the effect of striptill and nutrient placement on seed quality parameters such as protein and oil. Although soybean protein and oil concentrations are largely affected by genetic factors and environmental conditions during the seed-fi lling period, management practices can also have a significant impact (Wilcox, 1985; Brummer et al., 1997; Temperly and Borges, 2006). Yin and Vyn (2003) observed that compared with surface broadcast, 10-cm deep placement of K increased plant K uptake, seed K concentration, and oil concentration of soybean in low-K testing soils. Others have observed K deficiency reduces protein concentration (Koch and Mengel, 1977) and oil concentration (Sale and Campbell, 1986) likely due to a decrease in photosynthesis and phloem translocation (Wallingford, 1980). Concurrent work (Farmaha et al., 2011, 2012) examined the response of P and K fertilization, no-till, and strip-till on belowground parameters including: root development and water and nutrient distribution in the soil, and how those parameters influenced seed yield and seed nutrient concentration. The objectives of this study were to evaluate the effect of P and K rate and placement in no-till and strip-till on soybeanseed oil and protein content, aboveground growth and P and K accumulation, and soil water. Agronomy Journal
•
Volume 104, Issue 4
•
Sample Collection and Analysis Aboveground soybean tissue samples were collected only during the last 2 yr of the study (2008 and 2009). These samples were collected from 0.5 m2 (0.66 by 0.76 m) starting at V1 development stage and continued at 9- to 13-d intervals through R6 for a total of eight sampling times each year. Development stages were determined as described by Fehr and Caviness (1977). In 2008, samples were collected at V1, V2, V4, R1, R2, R4, R5, and R6 development stages and V1, V2, R1, R2, R4, R5, R5, and R6 in 2009. While the eight sampling events occurred during slightly different development stages in the 2 yr, the seasonal patterns were similar across years. Possibly due to late planting in 2009, soybean developed faster in 2009 than 2008. In 2009 the R1 sample was done when soybean began flowering, but the plants were only at the V4/V5 stage in terms of vegetative growth. Thus, while some of the development stages for a particular sampling time were different between 2012
1007
years, in terms of growth, the sampling periods were similar, thus data were grouped across years by sequential sampling event. Aboveground plant samples were collected from plots receiving factorial P–K combinations of 0 or 36 kg P ha–1 yr–1 and 0 or 168 kg K ha–1 yr–1, representing extreme low and high P and K application rates. Canopy height was measured from the soil surface to the uppermost trifoliate. Plants were cut at the soil surface and after leaf area was measured with a LI-COR 3100 leaf area meter (LICOR, Lincoln, NE), the aboveground tissues were oven dried at 60°C for at least 72 h to constant weight, and dry matter was recorded. Plant materials were combined, ground to pass a 1-mm mesh screen on a Wiley mill (Standard Model 3; Arthur H. Thomas Co., Philadelphia, PA), and analyzed for total P and K concentrations by A&L Great Lakes Laboratories, Inc. (Fort Wayne, IN) by nitric acid-perchloric acid mixture (HNO3– HClO4) digestion following the official methods of analysis of AOAC International (Horwutz, 2000). Nutrient accumulation was calculated as the product of dry matter accumulation by nutrient concentration. Accumulation rates of P and K were calculated by dividing the difference in aboveground tissue P and K accumulation between consecutive sampling times by the number of days between sampling times. Leaf area index was calculated by dividing leaf area by soil surface area. Continuous (hourly) measurement of soil water content throughout the 2008 and 2009 growing seasons was accomplished with ECH2O EC-5 and EC-20 moisture probes and Em-50 digital data loggers (Decagon Devices Inc., Pullman, WA). Soil water content was measured at the 0- to 5-, 5- to 10-, 10- to 15-, 15- to 20-, and 20- to 40-cm soil depth increments at in-row and between-rows positions from the three tillage/fertilizer placement treatments receiving the 0–0 and 36–168 kg P–K ha–1 yr–1 rates. These hourly soil water measurements were averaged across a 7-d period preceding each aboveground tissue sampling time. Since concurrent work (Farmaha et al., 2012) showed most nutrient uptake and changes in soil water occurred within the top 10 cm of the soil, we only present soil water data for that portion of the soil. At the end of each of the three growing seasons, soybean seed yield was recorded and seed samples were collected and measured for protein and oil concentration using near-infrared reflectance spectroscopy (Infratec Model 1229 Grain Analyzer, Foss Tecator Hoganas, Sweden). Seed protein and oil yield was calculated as the product of seed yield and seed protein and oil concentration, respectively. Statistical Analysis Seed protein and oil data were analyzed using the GLIMMIX procedure of SAS (SAS Institute, 2011). Tillage/fertilizer placement and fertilizer rate were considered fi xed effects while years, blocks and their interactions with treatments were considered random effects. The normality assumption of the residuals was tested using the Shapiro–Wilk test with the UNIVARIATE procedure of SAS (SAS Institute, 2011). The homogeneity of variances assumption was tested visually from the residual plots (plot the residuals vs. fitted values). Mean comparisons among treatments with balanced data were done using Tukey’s studentized range honestly significant difference (HSD) test to control experiment-wise error and with the 1008
Tukey-Kramer test for the unbalanced data. Significant fi xed effects and their interactions were further analyzed using the SLICE and SLICEFDIFF options available in the GLIMMIX procedure. Single degree of freedom contrasts were constructed for all the response variables between years to check the necessity of ANCOVA analysis. The REG procedure of SAS (SAS Institute, 2011) was used to regress protein and oil concentration and yield against P and K fertilization rates. Regression analysis was also used to regress protein and oil yield against seed yield and seed yield against canopy height. Due to large variability between years for seed protein and oil parameters, the hypothesis on equal slopes among years was tested. The response of LAI, dry biomass accumulation, and canopy height to tillage/fertilizer placement treatment, P and K fertilization rate, and development stage was described with the following three-parameter exponential function (Ratkowsky, 1990) using the NLMIXED procedure of SAS (SAS Institute, 2011): Y exp(a bX cX 2 )
[1]
where Y is the predicted value of the response function, a, b, and c are unknown constants, and X is the development stage. The response of P and K concentration and accumulation to tillage/fertilizer placement treatment, P and K fertilization rate, and development stage was described with the following type 1, sigmoidal logistic function (Seber and Wild, 2003): Y a / {1 exp[ k ( X Xc)]}
[2]
where, Y is the predicted value of the response function, a is the amplitude, Xc is the value of development stage when the value response variable reaches 50%, k is a coefficient, and X is the development stage. The likelihood of the model was maximized using quasi-Newton optimization and approximated using the adaptive Gaussian quadrature approach. The multiplicative additive effect of tillage/fertilizer placement, P and K fertilization rate, and random plot effects were also added to the above described functions. The variance of the response variables: LAI, dry biomass accumulation, and canopy height, increased with increasing size of means and violated equal variance assumption. Therefore these response variables were log-transformed to achieve the constant variance assumption. When the response variables were log-transformed then the nonlinear function, multiplicative additive effect of treatments, and random plot effects were also subjected to log transformation. However, the results were reported back to original scale for simplicity using anti-log transformation. Difference of least square means were then adjusted using Holm (1979) technique included in PROC MULTTEST of SAS (SAS Institute, 2011) to control the familywise error rate. Treatment effects were declared significant at an α ≤ 0.1. RESULTS AND DISCUSSION Seed Protein and Oil Concentration and Yield A significant tillage/fertilizer placement treatment × P fertilization rate interaction for protein yield (Table 1) was explained by a linear increase in protein yield with increased P fertilization for NTBC and NTDB treatments and no response for the STDB treatment (Fig. 1). The interaction also Agronomy Journal
•
Volume 104, Issue 4
•
2012
Table 1. Analysis of variance for soybean seed protein and oil concentration (g kg –1) and yield (Mg ha –1) over 3 yr (2007–2009). Seed protein Source of variation
df†
Tillage/fertilizer placement (T) Phosphorus fertilization (P) T×P Potassium fertilization (K) T×K P×K T×P×K
2 3 6 3 6 9 18
2007–2008 2007–2009 2008–2009
1 1 1
Concentration F P>F 2.92 0.17 12.31
