Introduction
Early-season management of Palmer amaranth is critical in cotton production to maximize yield potential (Fast et al. Reference Fast, Murdock, Farris, Willis and Murray2009; MacRae et al. Reference MacRae, Culpepper, Webster, Sosnoskie and Kichler2013; Norsworthy et al. Reference Norsworthy, Schrage, Barber and Schwartz2016). Palmer amaranth continues to be one of the most challenging weed species to manage in cotton and other agronomic crops (Webster Reference Webster2013). Implications of Palmer amaranth interference in cotton production have been well documented (Fast et al. Reference Fast, Murdock, Farris, Willis and Murray2009; MacRae et al. Reference MacRae, Culpepper, Webster, Sosnoskie and Kichler2013; Morgan et al. Reference Morgan, Baumann and Chandler2001; Norsworthy et al. Reference Norsworthy, Smith, Steckel and Koger2009; Rowland et al. Reference Rowland, Murray and Verhalen1999; Smith et al. Reference Smith, Baker and Steele2000; Webster and Grey Reference Webster and Grey2015). Vann et al. (Reference Vann, York, Cahoon, Buck, Askew and Seagroves2017a) reported cotton height reduction ranged from 7% to 57% and lint yield reductions from 8% to 42% when the first POST herbicide applications were delayed 7 to 28 d, respectively. Furthermore, prolonged insufficient control of Palmer amaranth can lead to a rapid increase in Palmer amaranth populations and contribution of seed to the soil seedbank (Inman et al. Reference Inman, Jordan, York, Jennings, Monks, Everman, Bollman, Fowler, Cole and Soteres2016).
Resistance management has been at the forefront of weed management programs in cotton production since glyphosate-resistant (GR) Palmer amaranth was first confirmed in 2005 (Culpepper et al. Reference Culpepper, Grey, Vencill, Kichler, Webster, Brown, York, Davis and Hanna2006). The use of soil-applied residual herbicides combined with timely POST applications and integrated management strategies have become requirements to effectively manage GR Palmer amaranth and other herbicide-resistant weeds (Culpepper et al. Reference Culpepper, Webster, Sosnoskie, York and Nandula2010; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012; Sosnoskie and Culpepper Reference Sosnoskie and Culpepper2014). POST herbicide options are limited for cotton growers. Since the transition from GR cotton to glyphosate- and glufosinate-resistant cotton, growers have intensively relied on glufosinate (Barnett et al. Reference Barnett, Culpepper, York and Steckel2013; Sosnoskie and Culpepper Reference Sosnoskie and Culpepper2014). Glufosinate can be effective in controlling GR Palmer amaranth when timely applications are made (Barnett et al. Reference Barnett, Culpepper, York and Steckel2013; Cahoon et al. Reference Cahoon, York, Jordan, Seagroves, Everman and Jennings2015a; Corbett et al. Reference Corbett, Askew, Thomas and Wilcut2004). However, control is generally reduced when glufosinate is applied to Palmer amaranth taller than 8 cm (Coetzer et al. Reference Coetzer, Al-Khatib and Peterson2002; Culpepper et al. Reference Culpepper, Webster, Sosnoskie, York and Nandula2010). The rapid growth and competitive ability of Palmer amaranth (Ward et al. Reference Ward, Webster and Steckel2013) creates challenges for growers in making well-timed herbicide applications. In addition, at-plant residual herbicides may not control weeds adequately and may further decrease a grower’s flexibility in making timely POST applications (Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012).
The commercialization of dicamba-resistant cotton has provided growers with an additional POST option for managing GR Palmer amaranth (Cahoon et al. Reference Cahoon, York, Jordan, Everman, Seagroves, Culpepper and Eure2015b). Dicamba has proved to be an effective tank-mix partner in combinations with glyphosate or glufosinate for control of glyphosate-susceptible (GS) and GR Palmer amaranth (Cahoon et al. Reference Cahoon, York, Jordan, Everman, Seagroves, Culpepper and Eure2015b; Johnson et al. Reference Johnson, Young, Matthews, Marquardt, Slack, Bradley, York, Culpepper, Hager, Al-Khatib, Steckel, Moechnig, Loux, Bernards and Smeda2010; Merchant et al. Reference Merchant, Sosnoskie, Culpepper, Steckel, York, Braxton and Ford2013; York et al. Reference York, Culpepper, Sosnoskie and Bollman2012). Depending on POST application timing and location, late-season Palmer amaranth control was increased at least 14% and 41% when dicamba was tank-mixed with glufosinate and glyphosate, respectively, compared to when they were applied without dicamba (Cahoon et al. Reference Cahoon, York, Jordan, Everman, Seagroves, Culpepper and Eure2015b). Regardless of rate, POST mixtures of glufosinate and dicamba were more effective at controlling 16- to 23-cm Palmer amaranth 12 d after application compared to glufosinate or dicamba alone (Vann et al. Reference Vann, York, Cahoon, Buck, Askew and Seagroves2017b). Similar findings were reported by Merchant et al. (Reference Merchant, Sosnoskie, Culpepper, Steckel, York, Braxton and Ford2013), where a 17% to 22% increase in control was observed with 20-cm-tall Palmer amaranth when dicamba was included with glufosinate compared to glufosinate alone.
The performance of a POST herbicide application is related to herbicide rate, herbicide coverage, and susceptible weed size at application. Adequate weed control could be compromised when one or more of these components are not observed, thus increasing selection pressure of the herbicides applied and potentially contributing to herbicide resistance. With unpredictable weather conditions, executing timely herbicide applications can be challenging, as weed control may be reduced because of an increase in weed size (Stewart et al. Reference Stewart, Nurse, Hamill and Sikkema2010). The potential increase in cost, due to the possibility of applying an increased number of different herbicides at separate application timings, may lead to growers’ reluctance to follow a well-timed spraying schedule. However, short-term and long-term benefits can be attained through weed management programs that offer greater diversity of herbicides and cultural practices (Inman et al. Reference Inman, Jordan, York, Jennings and Monks2017; Jordan et al. Reference Jordan, York, Seagroves, Everman, Clewis, Wilcut, Shaw, Owen, Wilson, Young and Weller2014).
Although the importance of weed control timing has been well reported, only limited research in North Carolina has evaluated specific POST herbicide frequency and application timings in dicamba-resistant cotton. Furthermore, the peer-reviewed literature contains few data that address the economics of various weed management herbicide intensities in dicamba-resistant cotton. It is also important to understand how PRE herbicides can influence the dynamics of POST herbicide frequency. Therefore, two separate experiments were conducted to determine the most effective POST herbicide application timings in dicamba-resistant cotton. The objective for the first experiment was to compare different timings of POST herbicide applications of glufosinate and glyphosate plus dicamba on Palmer amaranth and annual grass control, cotton yield, and economic net returns without PRE herbicides. The objective for the second experiment was to follow the same POST timings and herbicides as experiment 1, comparing Palmer amaranth control, cotton yield, and economic net returns with and without PRE herbicides.
Materials and Methods
Experiment 1
Experiments were established in North Carolina across six environments during 2015 and 2016 near Clayton (35.67oN, 78.51oW) and Rocky Mount (35.89oN, 77.64oW). Soils in Clayton were a Dothan loamy sand (fine-loamy, kaolinitic, thermic Plinthic Kandiudults) with 0.27% humic matter. Soils in Rocky Mount were a Norfolk loamy sand (fine-loamy, kaolinitic, thermic Typic Kandiudults) with 0.5% humic matter. Dyna-Gro® cotton ‘3385 B2XF’ (Crop Production Services, Loveland, CO) was planted in 2015. Cotton ‘DP 1522 B2XF’ (Monsanto, St Louis, MO) was planted in 2016. Cotton was planted in conventionally tilled, raised beds at a seeding rate of 14 seeds m–1 of row. Plot sizes ranged from three to four rows (91-cm spacing) by 9 to 12 m, depending on location. Other than treatments imposed for the experiment, cotton was managed according to North Carolina Cooperative Extension Service recommendations (Edmisten et al., Reference Edmisten, Yelverton, Bachelor, Koenning, Crozier, Meijer, Cleveland, York, Hardy and Bullen2015).
The experimental design was a randomized complete block with four replicates. Treatments consisted of POST herbicides applied 2, 3, 4, and 5 wk after planting (WAP); 3, 4, and 5 WAP; 4 and 5 WAP; and 5 WAP only. Additional treatments included herbicides applied 2 WAP only, 2 and 3 WAP, and 2, 3, and 4 WAP. Glufosinate was applied 2 and 3 WAP. At 4 and 5 WAP, glyphosate plus dicamba was applied. A nontreated control was also included. No PRE herbicides were applied at planting. Rates for all herbicides are provided in Table 1. All herbicides were applied using CO2-pressurized backpack sprayers equipped with Turbo TeeJet® Induction 110025 nozzles (TeeJet Technologies, Wheaton, IL) calibrated to deliver 140 L ha–1 at 165 kPa.
Table 1. Herbicide common names, trade names, application rates, and manufacturers.
All environments had natural infestations of Palmer amaranth with a mixture of GS and GR Palmer amaranth; density was 75 plants m–2 or greater. Palmer amaranth height and crop growth stage at each application timing can be found in Table 2. Most environments had dense populations of annual grass species ranging from 25 to 100 plants m–2. Broadleaf signalgrass [Urochloa platyphylla (Munro ex C. Wright) R.D. Webster], goosegrass [Eleusine indica (L.) Gaertn.], large crabgrass [Digitaria sanguinalis (L.) Scop.], and Texas millet [Urochloa texana (Buckley) R.D. Webster] were the dominant grass species. Palmer amaranth control was estimated visually using a 0 to 100 scale (Frans et al. Reference Frans, Talbert, Marx, Crowley and Camper1986) and Palmer amaranth density recorded in each plot by counting the number of plants from a randomly determined 1 m2 in each plot biweekly from 2 to 8 WAP. Palmer amaranth aboveground fresh biomass was collected from row middles in treated plots (17 to 23 m2) within 3 wk prior to harvest and from 1 m2 in the nontreated plots. Data collected for estimated visible control, densities, and aboveground biomass for annual grass were recorded in the same manner as Palmer amaranth. All treated plots were mechanically harvested in mid-October to mid-November with a spindle picker modified for small-plot harvesting.
Table 2. Cotton growth stage and Palmer amaranth height at first POST application, averaged over all environments, for experiments 1 and 2
Experiment 2
Experiments were established in North Carolina across four environments during 2016 and 2017 near Clayton (35.67oN, 78.51oW) and Rocky Mount (35.89oN, 77.64oW). Soils in Clayton were a Goldsboro sandy loam (fine-loamy, siliceous, subactive, thermic Aquic Paleudults) with 0.41% humic matter. Soils in Rocky Mount were an Aycock very fine sandy loam (fine-silty, siliceous, subactive, thermic Typic Paleudults) with 0.5% humic matter. Cotton ‘DP 1522 B2XF’ (Monsanto, St Louis, MO) was planted in 2016. Cotton ‘DP 1538 B2XF’ (Monsanto, St Louis, MO) was planted in 2017. Cotton was planted in conventionally tilled, raised beds at a seeding rate of 14 seed m–1 of row. Plot sizes were four rows (91-cm spacing) by 9 to 12 m, depending on location. Other than treatments imposed for the experiment, cotton was managed according to North Carolina Cooperative Extension Service recommendations (Edmisten et al., Reference Edmisten, Yelverton, Bachelor, Koenning, Crozier, Meijer, Cleveland, York, Hardy and Bullen2015).
The experimental design was a randomized complete block with four replicates. Treatments consisted of the same POST herbicide treatments as experiment 1 with and without PRE herbicides applied immediately after planting. The PRE herbicide program consisted of acetochlor plus diuron plus fomesafen. A nontreated control was also included. Rates for all herbicides are provided in Table 1. Each site received a minimum of 10 mm of rainfall within 2 wk of PRE herbicides being applied.
All environments had natural infestations of Palmer amaranth with a mixture of GS and GR Palmer amaranth; density was 100 plants m–2 or greater. Average Palmer amaranth height and crop growth stage at each application timing can be found in Table 2. Palmer amaranth control was estimated visually using a 0 to 100 scale (Frans et al. Reference Frans, Talbert, Marx, Crowley and Camper1986) and Palmer amaranth density recorded in each plot by counting the number of plants from a randomly determined 1 m2 in each plot biweekly from 2 to 8 WAP. Palmer amaranth aboveground fresh biomass was collected from row middles in treated plots (17 to 23 m2) within 3 wk prior to harvest and from 1 m2 in the nontreated plots. All treated plots were mechanically harvested in mid-October to mid-November with a spindle picker modified for small-plot harvesting.
For both experiments, estimated economic net return was calculated based on the North Carolina Cooperative Extension Service budget for cotton (Edmisten et al., Reference Edmisten, Yelverton, Bachelor, Koenning, Crozier, Meijer, Cleveland, York, Hardy and Bullen2015), with a total production cost of $1,268.12 ha–1, excluding herbicide cost. Herbicide cost was based on pricing from local chemical retailers and factored into total production cost. Ginning cost was based on seed cotton yield for each plot at a price of $0.27 ha–1. Economic return was calculated for three lint prices as the difference between the product of yield (45% lint at $1.54 ha–1, $1.76 ha–1, and $1.98 ha–1 with 55% seed at $0.30 ha–1) and total production cost.
Data for both experiments were analyzed using the PROC Mixed procedure in SAS (v. 9.4; SAS Institute, Cary, NC). Treatments were considered a fixed factor, and replication and environment were considered random factors, as this allows inferences over a broad range of environments (Blouin et al. Reference Blouin, Webster and Bond2011; Carmer et al. Reference Carmer, Nyquist and Walker1989). Significant treatment-by-environment interactions for Palmer amaranth control, lint yield, and economic returns were observed for both experiments. Similar trends were found when environments were analyzed individually; therefore, analyses were combined across environments. Furthermore, the treatment mean square was at least 3-fold greater than the treatment-by-environment interaction mean square, providing justification to combine results over environments. Type III statistics were used to test all fixed effects, and least square means were calculated based on P ≤ 0.05 (Moore and Dixon Reference Moore and Dixon2015). Treatment means were separated using Fisher’s Protected LSD at P ≤ 0.05.
Results and Discussion
Experiment 1. POST-Only Herbicide Treatments
Regardless of herbicide sequence, Palmer amaranth was controlled 98% or greater 8 WAP when herbicides were applied three times. When glyphosate and dicamba were applied twice (4 and 5 WAP), no difference in Palmer amaranth control was observed 8 WAP compared to treatments with at least three herbicide applications (Table 3). Control of Palmer amaranth declined to 71% when a single glyphosate plus dicamba application was delayed to 5 WAP. Vann et al. (Reference Vann, York, Cahoon, Buck, Askew and Seagroves2017a) reported 75% control of Palmer amaranth when the first POST application of glufosinate plus dicamba was delayed to 5 WAP. Glufosinate alone applied 2 WAP and 2 and 3 WAP controlled Palmer amaranth 58% and 82% 8 WAP, respectively. Although effective control was observed 14 d after treatment, longevity of herbicide applications at only 2 WAP and 2 and 3 WAP are not sufficient for season-long weed control. Trends among treatments for annual grass control were like that of Palmer amaranth control (Table 3). The greatest control was observed when at least three herbicide applications were delivered or when glyphosate and dicamba was applied 4 and 5 WAP. A single application of glyphosate plus dicamba 5 WAP controlled annual grasses 85%, compared to glufosinate alone 2 WAP or 2 and 3 WAP resulting in 60% and 89% control, respectively. Glyphosate is generally more effective than glufosinate on grass species, especially goosegrass (Corbett et al. Reference Corbett, Askew, Thomas and Wilcut2004; Culpepper et al. Reference Culpepper, York, Batts and Jennings2000). Average aboveground biomass for the nontreated check was 21,500 kg ha–1 and 11,200 kg ha–1 for Palmer amaranth and annual grasses, respectively (data not shown). Treatments comprising three or more herbicide applications reduced Palmer amaranth biomass at least 96% (data not shown). Annual-grass biomass was reduced at least 93% with three or more herbicide applications or sequential applications of glyphosate plus dicamba at 4 and 5 WAP.
Table 3. Palmer amaranth and annual grass control 8 wk after planting (WAP) as affected by POST only application timing in experiment 1, conducted in North Carolina in 2015 and 2016.a
a Means within a column followed by the same letter are not different according to Fisher’s Protected LSD test at P ≤ 0.05. Data are pooled over six environments. Nontreated control was not included in data analysis.
b Annual grass consisted of broadleaf signalgrass, goosegrass, large crabgrass, and Texas millet.
When at least three herbicide applications were made, no difference in cotton lint yield was observed regardless of timing sequence (Table 4). However, yields following three applications at 2, 3, and 4 WAP were lower compared to yields following herbicides applied at 3, 4, and 5 WAP and 2, 3, 4, and 5 WAP. Buchanan and Burns (Reference Buchanan and Burns1970) suggested that cotton should be maintained weed-free for approximately 8 WAP to protect maximum yields. Although weed control 8 WAP was similar between these application timings, the lack of herbicide beyond 4 WAP proved to be critical by the end of the season. Despite sequential applications of glyphosate plus dicamba applied 4 and 5 WAP providing similar weed control compared to three or more herbicide applications, early-season weed interference reduced lint yield 23% to 30% (Table 3). Vann et al. (Reference Vann, York, Cahoon, Buck, Askew and Seagroves2017a) reported a 23% reduction in lint yield when the first POST application was delayed 14 d. Furthermore, Everman et al. (Reference Everman, Burke, Allen, Collins and Wilcut2007) reported a 72% reduction in lint yield when no early-POST herbicide was used compared to glufosinate alone early POST. Glufosinate alone at 2 WAP provided the lowest yields. Similar yields were observed with glufosinate at 2 and 3 WAP and a single application of glyphosate plus dicamba 5 WAP (Table 4). Comparable to yield trends, economic net returns were highest following three or four herbicide applications. Although cotton lint yield was lower for sequential applications of glyphosate plus dicamba (4 and 5 WAP), reductions in herbicide cost allowed for similar returns compared to three and four applications (Table 4).
Table 4. Lint yield and economic net return as affected by POST application timing, in experiment 1, conducted in North Carolina in 2015 and 2016.a
a Means within each column followed by the same letter are not different according to Fisher’s Protected LSD test at P ≤ 0.05. Data are pooled over six environments.
b Abbreviations: WAP, weeks after planting.
c Cotton price based on 45% lint and 55% cottonseed.
Experiment 2. PRE and POST Herbicide Treatments
All POST herbicide sequences provided similar or greater control of Palmer amaranth when following PRE herbicides (Table 5). Palmer amaranth was controlled 99% or greater 8 WAP when three POST herbicide applications were made, with or without the use of PRE herbicides. Palmer amaranth control increased by 30%, 27%, 8%, and 8% with POST timings 2 WAP only, 5 WAP only, 2 and 3 WAP, 4 and 5 WAP with PRE herbicides compared to no PRE, respectively. The PRE herbicide program alone provided 79% control of Palmer amaranth, like sequential POST herbicide applications at 2 and 3 WAP. Previous studies have shown excellent Palmer amaranth control by fomesafen applied PRE alone or in combination with acetochlor and/or diuron (Cahoon et al. Reference Cahoon, York, Jordan, Everman, Seagroves, Braswell and Jennings2015c; Whitaker et al. Reference Whitaker, York, Jordan, Culpepper and Sosnoskie2011). The importance of PRE herbicides has been well documented in combating herbicide-resistant Palmer amaranth and reducing early-season weed interference (Everman et al. Reference Everman, Clewis, York and Wilcut2009; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012; Whitaker et al. Reference Whitaker, York, Jordan, Culpepper and Sosnoskie2011).
Table 5. Palmer amaranth control 8 wk after planting (WAP) and lint yield in response to POST application timing with and without PRE herbicides, in experiment 2 conducted in North Carolina in 2016 and 2017.a
a Means within columns (Palmer amaranth control or lint yield) followed by the same letter are not different according to Fisher’s Protected LSD test at P ≤ 0.05. Data are pooled over four environments. Nontreated for Palmer amaranth control was not included in data analysis.
Although not always significant, greater cotton lint yields were observed when the PRE herbicide program was used compared with no PRE herbicides (Table 5). There were no differences in lint yields when at least three POST applications were applied, regardless of timing sequence or PRE herbicides. The PRE herbicide program alone resulted in yields similar to that of POST-only applications at 2 WAP, 2 and 3 WAP, and 5 WAP. The inclusion of PRE herbicides allowed for greater flexibility in the number of POST application timings. This was more evident with later POST application timings, 4 and 5 WAP, and 5 WAP only, compared to early application timings, 2 WAP and 2 and 3 WAP. Everman et al. (Reference Everman, Burke, Allen, Collins and Wilcut2007) reported similar findings, showing that cotton lint yields were comparable when PRE and POST-directed herbicide applications were made regardless of a mid-POST application.
Trends for economic returns among treatments were similar to that of cotton lint yield. Although inclusion of a PRE treatment increased costs, greater returns were observed when the PRE was implemented with one or two POST application timings compared to programs without a PRE (Table 6). This was more evident when POST timings were delayed until 4 WAP, compared to timings at 2 WAP. Excellent weed control was obtained with the PRE herbicide program, and the POST application 2 WAP was not warranted in most cases, as no weeds had emerged. Highest economic returns were observed when the PRE herbicide program was included with POST application timings of 4 followed by 5 WAP and 5 WAP alone (Table 6). This can further be attributed to the differences in weed interference at critical growth stages (Buchanan and Burns Reference Buchanan and Burns1970). There was no difference in economic returns when herbicides were applied at least three times, irrespective of PRE herbicide treatment. These data show the importance of timely herbicide applications and herbicide program cost in relation to cotton lint prices. When cotton lint prices are low, it would be more cost-efficient to ensure timely herbicide applications compared to a delay, as this would help prevent follow-up herbicide applications. As cotton lint prices increase, there may be more flexibility in costs associated with increased herbicide use or applications.
Table 6. Influence of POST application timing with and without PRE herbicides on economic net return.a
a Means within pricing columns (comparison of No PRE and PRE) followed by the same letter are not different according to Fisher’s Protected LSD test at P ≤ 0.05. Data are pooled over four environments.
b Abbreviation: WAP, weeks after planting.
Total POST herbicide programs can be successful in some situations (Askew and Wilcut Reference Askew and Wilcut1999; Burke et al. Reference Burke, Troxler, Askew, Wilcut and Smith2005; Culpepper and York Reference Culpepper and York1998; Jordan et al. Reference Jordan, Frans and McClelland1993). However, timely applications are critical when soil-applied herbicides are not included (Culpepper and York Reference Culpepper and York1999; Vann et al. Reference Vann, York, Cahoon, Buck, Askew and Seagroves2017a). These data show that excellent Palmer amaranth control was achieved when three or more timely POST applications were utilized. Similar weed control can be obtained with sequential applications of glyphosate plus dicamba at 4 and 5 WAP. However, cotton lint yields could be reduced as a result of early-season weed interference. Although effective at the time, POST-only weed management programs most likely contributed to a more rapid evolution of herbicide resistance due to less diversity of herbicide mechanisms of action and spraying of larger weeds resulting in partial control (Beckie Reference Beckie2006). Furthermore, soil weed seedbank dynamics should be included in weed management programs (Buhler et al. Reference Buhler, Hartzler and Forcella1997; Norsworthy et al. Reference Norsworthy, Korres and Bagavathiannan2018). Control of larger weeds is obtainable, but weed seed contribution is unknown when weeds are not completely controlled. This research demonstrates that even when adequate weed control is obtained with larger weeds, early-season weed interference can still adversely affect cotton yield. The use of PRE herbicides can offset missed early-season weed control efforts by providing similar yields and economic returns compared to timely, early-POST applications. Timely POST herbicide applications contribute to greater weed efficacy that leads to higher yields, and though not quantified in this study, reduce weed seed contribution to the soil seedbank. Yield was consistently greater across all POST treatments that included PRE herbicides. Economic returns were greater among treatments that included PRE applications, except for the four POST application program, where returns were similar. Management tactics aside from herbicides should be better understood. Future research should explore how herbicide frequency and timings in combination with other control factors such as cover crops, tillage, and rotation of herbicide-resistant crop varieties can be used most efficiently.
Acknowledgments
The authors gratefully acknowledge the North Carolina Agricultural Foundation for providing partial funding for this research. The authors would like to thank the staff at the Central Crops Research Station and Upper Coastal Plain Research Station for technical assistance. No conflicts of interest have been declared.
