Soybean Sowing Date: The Vegetative, Reproductive, and Agronomic Impacts

The sensitivity of soybean [Glycine max (L.) Merr.] main stem node accrual to ambient temperature has been documented in greenhousegrown plants but not with fi eld-grown plants in the north-central United States. Biweekly V-node and R-stage, stem node number, internode length, and other traits were quantifi ed in an irrigated split-plot, four-replicate, randomized complete block experiment conducted in Lincoln, NE, in 2003–2004. Main plots were early-, mid-, lateMay, and mid-June sowing dates. Subplots were 14 cultivars of maturity groups 3.0 to 3.9. Node appearance was surprisingly linear from V1 to R5, despite the large increase in daily temperature from early May (10–15°C) to July (20–25°C). The 2003 and 2004 May planting date regressions exhibited near-identical slopes of 0.27 node d–1 (i.e., one node every 3.7 d). Cold-induced delays in germination and emergence did delay the V1 date (relative to planting date), so the primary effect of temperature was the V1 start date of linearity in node appearance. With one exception, earlier sowings led to more nodes (earlier V1 start dates) but also resulted in shorter internodes at nodes 3 to 9 (cooler coincident temperatures), thereby generating a curved response of plant height to delayed plantings. Delaying planting after 1 May led to signifi cant linear seed yield declines of 17 kg ha–1 d–1 in 2003 and 43 kg ha–1 d–1 in 2004, denoting the importance of early planting for capturing the yield potential available in soybean production, when moisture supply is not limiting. A.M. Bastidas, T.D. Setiyono, A. Dobermann, K.G. Cassman, G.L. Graef, and J.E. Specht, Dep. of Agronomy and Horticulture, Univ. of Nebraska, Lincoln, NE 68583-0915; R.W. Elmore, Dep. of Agronomy, Iowa State Univ., Ames, IA 50011. Contribution of the Dep. of Agronomy and Horticulture, Univ. of Nebraska, Lincoln, NE 68583. Published as Paper no. 15201, Journal Series Nebraska Agric. Exp. Stn. Project No. 12-194. Funding for this research was received from the Nebraska Agricultural Research Division, Nebraska Soybean Development, Utilization, and Marketing Board, United Soybean Board, and Fluid Fertilizer Foundation. Received 4 May 2006. *Corresponding author ( [email protected]). Abbreviations: CI, confi dence interval; DAP, days after planting; Ddf, denominator degrees of freedom; DOY, day of year; ET, evapotranspiration; MG, maturity group; Ndf, numerator degrees of freedom; R, reproductive; V, vegetative. Published in Crop Sci. 48:727–740 (2008). doi: 10.2135/cropsci2006.05.0292 © Crop Science Society of America 677 S. Segoe Rd., Madison, WI 53711 USA 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. Permission for printing and for reprinting the material contained herein has been obtained by the publisher. R e p ro d u c e d fr o m C ro p S c ie n c e . P u b lis h e d b y C ro p S c ie n c e S o c ie ty o f A m e ri c a . A ll c o p y ri g h ts re s e rv e d . 728 WWW.CROPS.ORG CROP SCIENCE, VOL. 48, MARCH–APRIL 2008 with mid-May plantings, with yield declining signifi cantly if sowing is delayed into late May or early June—even more with delays beyond mid-June. Despite this wealth of information, reports diff er as to whether it is better to plant in early May (or late April) or in mid-May, primarily because yields in the former have not been consistent. While variation in soybean planting date is expected to impact the pattern of soybean growth and development, very few reports have examined this issue in detail. A method for evaluating soybean vegetative (V) and reproductive (R) development was fi rst documented by Fehr et al. (1971). The key V-node and R-stage parameters were graphically illustrated in a later bulletin (Fehr and Caviness, 1977) and with pictures by Pedersen (2004). This staging system is now the standard method used to document phenological development in soybean. Pedersen and Lauer (2003, 2004a, 2004b) conducted one of the few detailed studies on the eff ects of early (3–6 May) vs. late (23–27 May) planting dates by examining soybean growth, development, and yield in a 4-yr experiment located in Wisconsin. They observed that the start of each reproductive stage—from R1 (begin fl ower) to R5 (begin seed)—was delayed by the 3-wk delay in planting date, except for stage R6 (full seed), which occurred coincidently in both planting dates at 105 d after emergence. Seed number and pod number were greater, but seed per pod was lower, in the early May planting date. However, these yield component diff erences were small, off ering little explanation for the diff erence in 4-yr seed yield means between 4.23 Mg ha–1 recorded in the early May plantings and 3.85 Mg ha–1 in the late May plantings. Pedersen and Lauer (2004a) also used data they collected at 20-d intervals to examine seasonal patterns in plant height and node appearance. At 64 d after emergence, plants in the late May planting were 35 cm shorter than plants in the early May planting, but at R6, plants in both planting dates were nearly equal in height. The authors concluded that planting date did not have an eff ect on plant height at harvest. Many soybean producers in Nebraska and elsewhere have drawn a similar conclusion based on their own experience with planting dates, but unfortunately this has led some to believe that plants in late plantings can “catch up” with the plants in early May plantings in traits other than just plant height. Pedersen and Lauer (2004a) did note that plants in the early May planting averaged 16.3 main stem nodes at maturity, compared to an average of 15.5 mature nodes in the late May planting. Our examination of their data indicated that stem node 8, which was attained about 59 d after emergence in their early May sowing date, was attained about 43 d after emergence in their later May sowing date. Thus, node appearance in the latter was about 5 d behind that in the former. Node production did cease at the beginning of seed-fi ll (reproductive stage R5) in both planting dates. Fehr and Caviness (1977) stated that from emergence to the fi fth node, a new node appeared on the main stem about every 5 d, but also noted that this could vary from 3 to 8 d. They also noted that after node 5, a new node appeared on the main stem about every 3 d, but again noted that this could range from 2 to 5 d, depending on the temperature. The foregoing numbers have been restated many times since then (Monks et al., 1988; Pedersen, 2004), although these generalizations are often misinterpreted by producers. Zhang et al. (2004) reported on a 5-yr fi eld study in Mississippi in which the calendar dates of successive main stem nodes were recorded every other day for cultivars ranging from maturity group (MG) early or mid-3 to late 5 grown in early March to late June planting dates. The diff erence in days between planting and emergence (VE) ranged from a high of 14 d for all cultivars in the early (cooler) March planting date to a low of 5 d in the (warmer) mid-May to late June plantings. The respective diff erences in development at other stages ranged from 16 to 5 d from VE to VC (= V0), 6 to 5 d from V0 to V1, 6 to 4 d from V1 to V2, and 5 to 3 d for each successive node thereafter. Aside from the foregoing Mississippi study, studies where the measurement period for V-node and R-stage assessment was as short or shorter than the period between node appearance, as recommended by McMaster and Hunt (2003), are not available in the literature. This is also true for western Corn Belt locations where irrigation is practiced. The objective of our study was to quantify the impact of planting date on the vegetative, reproductive, and agronomic performance of 14 MG 3.0 to 3.9 cultivars planted at about 16-d intervals over a 7-wk span during which Nebraska producers typically plant such cultivars. These data and those for yield and other measured agronomic variables were collected in an east-central Nebraska irrigated production system that was optimally managed to allow expression of the available seed yield potential (Specht et al., 1999, 2006). MATERIALS AND METHODS A 2-yr fi eld experiment was conducted in 2003 and 2004 on the Agronomy Farm at the East Campus of the University of Nebraska, Lincoln (40°51′ N, 96°45′ W) on a deep Kennebec silt loam soil (fi ne-silty, mixed, superactive, mesic Cumulic Hapludoll). The previous crop each year was maize (Zea mays L.). The fi eld was fall-plowed after maize harvest, then fi eldcultivated twice in the spring of each year. The experimental design each year was a split-plot randomized complete block with four replicates (i.e., blocks). Main plots were four planting dates scheduled each year at about 16-d intervals. In 2003, those dates were 2 May (day of year [DOY] 122), 17 May (137), 30 May (150), and 16 June (167), but in 2004 were 28 April (119), 16 May (137), 2 June (154), and 17 June (169). Subplots were 14 soybean cultivars of MG R e p ro d u c e d fr o m C ro p S c ie n c e . P u b lis h e d b y C ro p S c ie n c e S o c ie ty o f A m e ri c a . A ll c o p y ri g h ts re s e rv e d . CROP SCIENCE, VOL. 48, MARCH–APRIL 2008 WWW.CROPS.ORG 729 node (Vn) on the calendar date when leafl ets at the next node above (Vn+1) have just unrolled so that their edges are no longer touching. Sinclair (1984b) noted that leafl ets meeting this “ just unrolled” criterion were 21 mm in length. Reproductive development in each planting date of each year was tracked using the R1 (begin fl ower), R2 (full fl ower), R3 (begin pod), R4 (full pod), R5 (begin seed), R6 (full seed), R7 (begin maturity), and R8 (full maturity) stages defi ned by Fehr et al. (1971). Plant development was scored biweekly to match an anticipated 3to 5-d node appearance rate, as recommended by McMaster and Hunt (2003). On each biweekly scoring date, 10 adjacent plants were individually scored to obtain 10-plant mean values for Vn and Rn development, as recommended by Fehr and Caviness (1977). The time interval between the appearance of two successive leaves on the main stem is known as a phyllochron (Wilhelm and McMaster, 1995; McMaster and Hunt, 2003; Hunt et al., 2003), not a plastochron (Hofstra et al., 1977). By botanic defi nition, the latter term is restricted to the time period between successively initiated leaf primordia (Erickson and Michelini, 1957), which can only be observed via daily microscopic dissection and observation of a main stem apical meristem (Miksche, 1961). All agronomic data and the fi nal main stem node and internode length data were subjected to an analysis of variance, using the PROC MIXED (Littell et al., 1996) procedure of SAS (SAS Institute, 1999). Only planting date was considered a fi xed eff ect, so the RANDOM statement included the terms year, block (year), year × planting date, block × planting date (year), cultivar, year × cultivar, planting date × cultivar, and year × planting date × cultivar. The MODEL statement included a DDFM = KR option to specify a Kenward–Roger adjustment in the computation of a Satterthwaite-type denominator degrees of freedom (Ddf ) for the evaluation of mixed model eff ects. ESTIMATE statements were used to compute the means and standard errors of main eff ects and interactions. CONTRAST statements were used to partition the planting date mean square into preplanned single-degree-of-freedom mean squares attributable to the linear, quadratic, and cubic eff ects. A Type 1 error value of α = 0.05 was chosen as the F-test signifi cance criterion. The observably triphasic seasonal pattern of seasonal V-node appearance in each of the eight planting dates was fi t to a three-segment linear regression model. This model was chosen because fi ve of its six estimable parameters (i.e., the B1, B2, and B3 regression coeffi cients for the three successive linear phases, with the X0 and X1 breakpoints separating the linear phases) have biological meaning. The sixth parameter (I1) is the y-intercept when X = 0, and it was set to an arbitrary V-node value of −2.0 to represent the day of planting. The Fehr and Caviness (1977) system uses the term VE to denote seedling emergence, but in this study, we assigned this stage an arbitrary V-node value of −1.0. These assignments were necessary to retain monotonicity in V-node stage from planting onward, thus avoiding the awkward “hidden stage” parameter that Pachepsky et al. (2002) devised to deal with pre-V0 vegetative development. Model-fi tting was implemented with GraphPad Prism 4 software (Motulski and Christopoulos, 2003) using this (componentized) GraphPad equation: Y1 = I1 + B1 × X, Y at 3.0 to 3.9, which is the MG range recommended for the latitude of this test location. The 14 cultivars were selected from the highest yielding entries in 2or 3-yr performance trials conducted before 2003 in Nebraska and Iowa. These cultivars were Asgrow AG3401 (relative maturity 3.4); Dekalb DKB 31-52 (3.1); Kaup 335 (3.3); Kruger K323+RR (3.2); Latham 1067RR (3.1); Nebraska strains NE3001 (3.0), NE3201 (3.1), and NEX8903 (3.1); Nebraska experimental lines U98-307162 (3.4), U98-307917 (3.4), and U98-311442 (3.9); Pioneer 93B36 (3.3) and 93B47 (3.4); and Stine 3632-4 (3.6). All were indeterminate, except for the semideterminate cultivar NE3001. High quality, fungicide-treated seed was obtained from the Nebraska Seed Foundation or from the indicated companies. The fourrow subplot row length was 4.3 m, with an interrow spacing of 0.76 m. The viable seeding rate was 390,000 seeds ha–1 and the sowing depth was 2.5 cm. In 2003, weed control was accomplished with a preplant application of 0.03 kg ha–1 of fl umetsulam {N-(2,6-difl uorophenyl)5-methyl-[1,2,4]-triazolo-[1,5a]-pyrimidine-2-sulfonamide} and 1.07 kg ha–1 of s-metolachlor {2-chloro-N-(2-ethyl6-methylphenyl)-N-[(1S )-2-methoxy-1-methylethyl]acetamide}. In 2004, the preplant application was 0.23 kg ha–1 of fl ufenacet {N-(4-fl uorophenyl)-N-(1-methylethyl)-2-[[5(trifl uoromethyl)-1,3,4-thiadiazol-2-yl]oxy]acetamide} and 0.06 kg ha–1 of metribuzin [4-amino-6-(1,1-dimethylethyl)-3(methylthio)-1,2,4-triazin-5(4H)-one]. Irrigation was applied with a solid set sprinkler system. Except for a light irrigation immediately after the fi rst planting date in 2004 (due to dry soil conditions), irrigation was not needed until late June, and was scheduled as needed to routinely replenish soil moisture with adjustments made for local rainfall and crop daily evapotranspiration (ET). The Penman–Monteith equation (Allen et al., 1998) was used to estimate daily ET from air temperature, radiation, humidity, and wind speed data collected from a nearby automated weather station operated by the High Plains Regional Climate Center (http://www.hprcc.unl.edu). The data collected from the central two rows of each four-row subplot included plant maturity (days from planting to maturity; i.e., when 95% of the pods are mature); mature (standing) plant height, measured from the ground surface to the tip of the main stem; plant population, based on a plant count in a 4.3-m section of a one subplot row (at maturity in both years and just after emergence in 2004); seed yield, based on the weight and moisture of the seed harvested with a 2-row plot combine with the fi nal yield adjusted to 13% seed moisture; and 100-seed weight, based on the weight of a sample of 100 random seeds (also adjusted to 13% moisture). A 75-g sample of the harvested seed of each subplot was subjected to a nearinfrared analysis to estimate seed protein and oil content at 13% seed moisture. Two representative plants per subplot were gathered at harvest to obtain a two-plant mean measure of the fi nal length of each internode, starting with internode one located between the cotyledonary node (V0) and unifoliolar node (V1), and ending with the last visible internode between the last two nodes at the tip of the stem. The soybean Vand R-staging system described by Fehr et al. (1971) was used to track plant development in each of the 14 cultivars in each of the planting dates. For readers unfamiliar with this system, a V-number is assigned to a given stem R e p ro d u c e d fr o m C ro p S c ie n c e . P u b lis h e d b y C ro p S c ie n c e S o c ie ty o f A m e ri c a . A ll c o p y ri g h ts re s e rv e d . 730 WWW.CROPS.ORG CROP SCIENCE, VOL. 48, MARCH–APRIL 2008 X0 = I1 + B1 × X0, Y2 = Y at X0 + B2 × (X − X0), Y at X1 = Y at X0 + B2 × (X1 − X0), Y3 = Y at X1 + B3 × (X − X1), and Y = IF(X < X0, Y1, IF(X < X1, Y2, Y3)). Model-fi tting was robust when suitable initial values were supplied for the fi ve estimable parameters, and R2 values of 0.996 or better were achieved for the fi ts in each of the eight data sets. Attempts to globalize the model parameter fi ts to all planting dates are described in the Results section. The length of each successive internode on a mature main stem exhibited a parabolic-like pattern when plotted as a function of nodal position (after node 3, since internode length declined from node 0 to 3). This led us to model the data for each planting date with a three-parameter Lorentzian function: Y = A/{1 + [(X − C)/(W)]2}, where C is the centering X value at which the Lorentzian peak attains its greatest amplitude (A), and W is the peak half-width when measured at peak halfamplitude. These parameters had biological meaning relative to hypotheses as to whether planting date delays (and thus the accompanying warmer temperatures) would shift the peak of internode length to a lower node, or increase peak amplitude, or lessen its width, or any combination of these. The global fi tting eff orts are described in the Results section. Soybean reproductive development is characterized by changes in organ morphology that are qualitative, not quantitative, despite the assignment of consecutive integers (i.e., 1–8) to the successive R stages. The R-stage trend in each planting date had an irregular pattern that could not be fi t to a simple mathematical model whose parameters had intrinsic biological meaning. However, the focus in the present paper was determining the coincidence of key R-stages with the dates of key phases in the modeled patterns of node accrual and internode length. With that in mind, a nonlinear “standard curve” was generated for each planting date by fi tting the R-stage data to the best-fi tting polynomial equation (as described by Motulski and Christopoulos, 2003). An F-test was used to determine if the gain in model R2 generated when a polynomial of a given order was incremented to its next higher order was due to chance. A quintic (i.e., fi fth order) polynomial was judged by such F-tests to suffi ciently account for most of the nonlinearity present in each R-stage data set. Figure 1. Temperature and phenological data for (A) 2003; (B) 2004. Top: Daily (thin line) and 15-d (thick line) mean temperatures from Day 91 to 294 on a day of year (DOY) scale. Middle: Progression of biweekly vegetative node number (Vn) in each planting date. Vn values of −2 and −1 were arbitrarily used to denote the sowing and emergence (VE) stages. Bottom: Progression of biweekly reproductive stage (Rn) number in each planting date. The staging system of Fehr and Caviness (1977) was used. Each Vand R-stage data symbol represents a mean of 140 plants. R e p ro d u c e d fr o m C ro p S c ie n c e . P u b lis h e d b y C ro p S c ie n c e S o c ie ty o f A m e ri c a . A ll c o p y ri g h ts re s e rv e d . CROP SCIENCE, VOL. 48, MARCH–APRIL 2008 WWW.CROPS.ORG 731 RESULTS Seasonal temperature and rainfall patterns in the two experimental seasons were quite diff erent (Table 1). In 2003, July and August were 3°C warmer, May and September were 3°C cooler, and June was 1.5°C cooler than those same months in 2004. Seasonal temperatures in 2003 followed the historical pattern, rising steadily from May to June to July at a rate of 5°C per month, holding at 26°C in July and August, and then declining sharply from August to September (Fig. 1). In contrast, the 2004 seasonal temperature pattern had less in-season variation, with May, June, and September warmer, but July and August cooler, than their historical means. Rainfall during the growing season (May– September) totaled to 378 mm in 2003 and 359 mm in 2004 (Table 1). However, after a rainy June of 2003, rainfall was limited during July and August. Daily crop ET was also greater in these two abnormally warm months, making more frequent irrigation necessary. In fact, all irrigation needed in 2003 (250 mm) was applied during these 2 mo. Less irrigation was needed in 2004 (161 mm), due to the cooler temperatures that year that reduced daily crop ET during the critical July–August reproductive growth period. Vegetative and Reproductive Development Vegetative and reproductive development data for individual cultivars within each planting date were utilized and presented in Setiyono et al. (2007), so only the planting Table 1. Precipitation, irrigation, and temperature on a monthly and seasonal basis during 2003 and 2004 at the Agronomy Farm on the East Campus of the University of Nebraska at Lincoln. Variable and year May June July August September Season