Online Article and more

Spring wheat (Triticum aestivum L.) is a major crop in the northern Great Plains that is generally grown following a 21-mo fallow period. A 12-yr study was conducted to determine the effects of tillage system [conventional-till (CT), minimum-till (MT), and no-till (NT)], N fertilizer rate (34, 67, and 101 kg N ha^-1), and cultivar (Butte86 and Stoa) on spring wheat yields within a dryland spring wheat (SW)–winter wheat (WW)–sunflower (Helianthus annuus L.) (SF) rotation. Grain yield responses varied with tillage system, N fertilizer rate, cultivar, and year as indicated by significant tillage x N rate x year and N rate x cultivar x year interactions. In years with >260 mm total plant available water (TPAW) but <400> yields were greater than those with CT at the highest N rate, with similar trends at the medium and low N rates. When TPAW exceeded 400 mm, grain yields for CT were generally greater than for NT at the medium N rates. The greatest 12-yr average grain yield (1727 kg ha^-1) was obtained with NT and application of 101 kg N ha^-1. Grain yields were lowest during years when TPAW was <300> N treatments. Cultivars responded similarly to N fertilization in years with >300 mm TPAW, with Butte86 yielding more than Stoa in 6 out of the 12 yr. Soil NO₃–N levels increased in the root zone following three consecutive drought years, but had declined to initial year levels by the end of the study. These results indicate that farmers in the northern Great Plains can produce SW following SF in annual cropping systems that do not include a fallow period, particularly if NT or MT systems are used with adequate N fertilization.

Abbreviations: CT, conventional-till • MT, minimum-till • NT, no-till • PAW, plant-available water • SW, spring wheat • TPAW, total plant-available water • WW, winter wheat • SF, sunflower

INTRODUCTION

IN the semi-arid northern Great Plains, plant-available water (PAW) and soil erosion are major factors limiting agricultural production. Therefore, farmers need to manage crop residues and tillage to control soil erosion and effectively store and use the limited precipitation received for crop production. No-till and minimum-till systems are effective steps in efficiently saving more precipitation for crop production (Aase and Schaefer, 1996; Halvorson, 1990b; Peterson et al., 1996; Tanaka and Anderson, 1997).

The traditional crop–fallow system of farming uses water/precipitation inefficiently as evidenced by the development of dryland saline-seeps in the northern Great Plains (Halvorson and Black, 1974). The solution to the saline-seep problem is to crop more intensively with efficient use of precipitation for crop production (Halvorson, 1990a). Saline-seep areas have been controlled and returned to crop production by growing alfalfa (Medicago sativa L.) and/or by annual cropping of the seep recharge area (Halvorson, 1984; Halvorson and Reule, 1980).

Deibert et al. (1986) suggested that farmers in the northern Great Plains need to use more continuous cropping and less crop–fallow to attain more efficient use of limited water supplies. Peterson et al. (1996) and McGee et al. (1997) point out that MT and NT fallow systems have a high percentage of the soil water in the profile recharged by the first spring following harvest. Continuing the fallow period for an additional 5 to 12 mo is very inefficient and costly. Therefore, cropping more intensively than crop–fallow is needed to efficiently use the water stored by NT and MT systems. Improved precipitation-storage efficiency with MT and NT allows producers the option of cropping more intensively than with crop–fallow (Halvorson and Reule, 1994; Peterson et al., 1996). Black et al. (1981) reported more efficient water use with more intensive cropping systems. Halvorson and Black (1985) reported crop yields that were generally >80% of 2-yr SW–fallow yields when grown in an annual cropping system with adequate N and P fertilization. Aase and Reitz (1989) and Aase and Schaefer (1996) reported that annually cropped SW with NT was more profitable and productive than SW–fallow in a 356 mm precipitation zone in northeast Montana.

Hall and Cholick (1989) reported varying responses of SW cultivars to tillage systems and a need to select cultivars for use under NT conditions. Most SW cultivars developed for use in the northern Great Plains have been developed using crop–fallow systems and low residue, CT environments.

Information is limited on the successes of more intensive dryland cropping systems in the northern Great Plains that include MT and NT management systems and a deep rooted crop, such as SF, in the rotation. In addition to more efficient water use, more intensive MT and NT cropping systems have the potential to be more profitable (Dhuyvetter et al., 1996) and reduce soil erosion potential (Merrill et al., 1999). This study was undertaken to determine the effects of tillage system (CT, MT, NT), N fertilizer rate (34, 67, and 101 kg N ha^-1), and cultivar (Butte86 and Stoa) on SW grain yields within a dryland SW–WW–SF rotation.

Methods and materials
The study was initiated in 1984 on a Temvik–Wilton silt loam soil (fine-silty, mixed, superactive, frigid Typic and Pachic Haplustolls) located near Mandan, ND. Surface soil pH was 6.4 , soil organic carbon was 21.4 g kg^-1, and soil test P was 20 to 26 mg kg^-1 in the spring of 1984 (Black and Tanaka, 1997). Data collection was from 1985 through 1996. An annual cropping rotation, SW–WW–SF, was managed under three tillage systems, CT, MT, and NT (Halvorson et al., 1999a, b). Nitrogen fertilizer was applied in early spring each year as a broadcast application of NH₄NO₃ at rates of 34, 67, and 101 kg N ha^-1, except for 1991 and 1992 when no N was applied because of a build-up of residual soil NO₃–N due to drought conditions and low yields from 1988 to 1990. Phosphorus fertilizer was applied broadcast at a rate of 40 kg P ha^-1 at the beginning of the study in October 1983. Soil test P levels in the 0- to 15-cm depth averaged 16 mg kg^-1 in 1991 and 11 mg kg^-1 in 1996. Two SW cultivars with good yield potential, Butte86 and Stoa, were used throughout the study. Each main block of the study was 137.2 by 73.1 m in size. Tillage plots (45.7 x 73.1 m) were oriented in a north–south direction, N plots (137.2 x 24.4 m) in an east–west direction across all tillage plots, and cultivars (22.9 x 73.1 m) in a north–south direction within tillage plots and across all N plots. The smallest plot with the combination of all variables was 22.9 by 24.4 m. Triplicate sets of plots were established to allow all phases of the rotation to be present each year. Experimental design was a strip-strip-split plot, with tillage and N rate treatments stripped and cultivar as subplots with 3 replications.

The CT treatments were generally not tilled in the fall after SF harvest but were disked once to a depth of 8 to 12 cm in the spring prior to SW planting. Surface residue cover was usually <30%> not tilled in the fall after SF harvest but were undercut once in the spring with a sweep plow at a shallow depth (<7.5 cm) prior to SW planting. Surface residue cover was usually 30 to 60% after planting. No-till treatments were not tilled following SF harvest and received one application of glyphosate [N-(phosphonomethyl)glycine] herbicide just prior to SW planting in 1985, 1986, 1989, 1990, 1993, and 1994. Surface residue cover was generally >60% after planting. No preplant herbicides were applied in the other years. Residue cover estimates were based on visual observations using experience with photographic measurements made of residue cover in adjacent SW–fallow plots (Merrill et al., 1995). Spring-applied herbicides were used to control broadleaf and grassy weed species within the growing SW crop. Tillage treatments, agronomic operations, and herbicides applied to the SF and WW crops are described by Halvorson et al. (1999a, 1999b).

The SW was generally planted in early May at a seeding rate of about 3.2-million seeds ha^-1 with a NT disk drill with 17.8-cm row spacing. The plots were harvested in mid- to late- August each year by hand cutting SW samples for grain yield determination from two 1.5-m² areas within each plot (1985–1993). In 1994 through 1996, grain yields were determined from a 50-m² area with a plot combine. Grain yields are expressed on a 120 g kg^-1 water content basis.

Soil samples, one 3-cm diameter core per plot, were collected for gravimetric soil water and NO₃–N analyses from one cultivar plot for each tillage and N fertilizer treatment each spring (April) before N fertilization. Samples were collected in 30-cm increments to a depth of 120 cm. Soil NO₃–N was determined for each depth increment by autoanalyzer (Lachat Instruments, 1989; Technicon Industrial Systems, 1973) on a 5:1 extract/soil ratio using 2 M KCl extracting solution (1985–1992) and a 0.01 M CaSO₄ extracting solution (1993–1996). Volumetric soil water content was estimated from gravimetric soil water measurements using a soil bulk density of 1.42 gm cm^-3 for the profile (Black and Tanaka, 1997). Total plant available water was estimated as the sum of spring soil PAW in the 0- to 120-cm profile plus growing season precipitation (April through August). Spring soil PAW was estimated by subtracting the lowest measured soil water content (152 mm) in the 0- to 120-cm profile following SW harvest during the 12-yr study from soil water contents in the 0- to 120-cm soil profile each spring. Precipitation was measured from April through October each year with a recording rain-gauge at the site. November through March precipitation was estimated from the U.S. Weather Bureau measurements made at the Northern Great Plains Research Laboratory at Mandan, ND, which was located approximately 5 km northeast of the site.

Analysis of variance procedures were conducted using SAS statistical procedures (SAS Institute, 1991) with years treated as a fixed variable. All differences discussed are significant at the P = 0.05 probability level unless otherwise stated. A least significant difference (LSD) was calculated only when the analysis of variance F-test was significant at the P = 0.05 probability level.

Results
Soil Nitrate Nitrogen
Spring soil NO₃–N levels varied significantly with tillage system, N rate, and year with significant tillage x year, N rate x year, and tillage x N rate x year interactions. Spring soil NO₃–N levels (0- to 120-cm depth) associated with the tillage x N rate x year interaction are reported in Table 1 . No differences were observed in spring soil NO₃–N levels among tillage treatments when compared over N rates and years from 1985 through 1988. In 1989, spring soil NO₃–N levels were significantly greater with MT than with NT and CT for the highest N rate. In 1990, spring soil NO₃–N levels were greater with CT than NT at the highest N rate. In 1991, CT and MT had a higher level of soil NO₃–N than NT at the low N rate. At the medium N rate, CT had a higher soil NO₃–N level than MT and NT. At the high N rate, MT had a higher level of soil NO₃–N than CT and NT. In 1992, CT had a higher level of soil NO₃–N than MT and NT at the low and high N rates. At the medium N rate, soil NO₃–N was greater with CT than MT. In 1993, there were no differences in soil NO₃–N among tillage treatments at the low N rate, with CT having higher levels than NT at the medium N rate and CT and MT having higher levels than NT at the highest N rate. In 1994, MT had a higher level of soil NO₃–N than NT at the lowest N rate, with CT and MT having higher levels than NT at the medium N rate and CT having higher levels than MT and NT at the highest N rate. In 1995 and 1996, no differences were observed among tillage treatments for each of the N rates. The data in Table 1 show that spring soil NO₃–N in the soil profile had increased considerably following the drought years of 1988 through 1990, which experienced poor WW and SF yields and reduced N requirements (Halvorson et al., 1999a, b). The trend was for higher levels of soil NO₃–N with increasing N rates from 1990 through 1994, with spring soil NO₃–N levels in 1995 and 1996 approaching levels similar to those in 1985 at study initiation.

The cultivar x year interaction effects on grain yields are shown in Fig. 7 . Grain yields were greater for Butte86 than for Stoa in 1985, 1986, 1991, 1992, 1993, and 1995 (6 out of the 12 years of this study). Stoa grain yields equaled those of Butte86 during the other six years. In this study, Butte86 generally reached the heading and grain filling growth stages about 6 d earlier than Stoa, thereby escaping some of the effects of plant diseases and late summer drought stress. This may partially explain the greater yields with Butte86 over Stoa in some years. The overall impact of SW cultivar on grain yields in this study was small.

Summary
The results of this study show that SW grain yields following SF in rotation are generally enhanced using MT and NT systems compared with CT during most years with adequate N fertility. Grain yields tended to be greatest with NT compared with CT during those years with <400> treatments generally produced greater SW yields than NT, particularly at low N rates. Leaf spot disease pressure was greater during the wetter years and at the low N rate (Krupinsky et al., 1997, 1998). The results show that during extremely dry years (e.g., 1988 through 1990), reduced tillage treatments did not store enough additional water to significantly enhance yield potential over that of the CT system. Responses to N fertilization were insignificant during these dry years, which resulted in increased residual spring soil NO₃–N levels. Responses to N fertilization were similar for both cultivars during the average and wetter years. Butte86 grain yields were greater than those of Stoa in 6 out of 12 years, three of which were years with >400 mm TPAW and two years with 300 to 400 mm TPAW. Grain yields were similar for both cultivars for all other years. The highest 12-yr average grain yield was obtained with the highest N rate and NT.

Spring wheat yield responses in this study are in agreement with annual cropping SW yields reported by Aase and Schaefer (1996) using NT, Halvorson and Black (1985), and Black et al. (1981) in northeastern Montana. In 6 out of the 12 years in this study, spring wheat yields exceeded the average 2-yr, SW–fallow yields (1737 kg ha^-1) reported for five southcentral North Dakota counties near the study site for 1989 through 1996 (Beard and Hamlin, 1995; Beard and Waldhaus, 1997). In two years, yields were 77 and 87% of SW–fallow yields; in two years, yields were 54% of SW–fallow yields; and in two years, yields were about 34% of SW–fallow yields.

These results indicate that farmers in the northern Great Plains can successfully produce SW following SF in annual cropping rotations that do not include a fallow period. During production periods with low soil water recharge following sunflower in rotation, farmers may want to consider producing a crop with lower water-use requirements than spring wheat to avoid uneconomical spring wheat yields. Fallow following sunflower should be considered as the last alternative because of a high soil erosion potential.SAS Institute Inc 1991

Legume cover crops can supply all or most of the N required by a subsequent crop if legume biomass is of sufficient quantity and N mineralization is approximately synchronous with crop demand. Three 2-yr crop rotation cycles were conducted on a Lamoine silt loam (fine, illitic, nonacid, frigid Aeric Epiaquept) soil in Maine to (i) evaluate biomass and N accumulation of alfalfa (Medicago sativa L.), winter rye (Secale cereale L.), and hairy vetch (Vicia villosa Roth subsp. villosa) plus winter rye cover crops; (ii) determine sweet corn (Zea mays L.) response to legume and fertilizer N sources in a barley (Hordeum vulgare L.)–sweet corn rotation; and (iii) assess the accuracy of the presidedress soil nitrate test (PSNT) and leaf chlorophyll N test (LCNT) for distinguishing N-responsive and nonresponsive sweet corn. Both legumes accumulated more N than rye grown alone, although total biomass was similar. Sweet corn following rye always exhibited a linear response to N fertilizer (up to 156 kg N ha^-1), but generally exhibited no response to added N following either alfalfa or hairy vetch plus winter rye (VR). Both PSNT and LCNT were 75% accurate in identifying plots responsive to additional fertilizer N. The legume cover crops grown were able to replace all or nearly all of the N fertilizer required by a subsequent sweet corn crop, with fertilizer replacement values (FRVs) of 58 to 156 kg N ha^-1 in a short-season environment. These cover crops are a viable alternative source of N, greatly reducing or eliminating the need for N fertilizer.

Abbreviations: FRV, fertilizer replacement value • LCNT, leaf chlorophyll N test • PSNT, presidedress soil nitrate test • VR, hairy vetch plus winter rye • DM, dry matter

INTRODUCTION

PRODUCTION systems that utilize N-fixing legumes as a primary N source for subsequent nonlegume crops include full-season green manure crops, interseeded legumes, and cover crops. Each of these options present significant management challenges to producers interested in reducing fertilizer N inputs or producing crops without fertilizer N. For example, full-season green manure crops, while potentially having the greatest impact on soil quality and pest/weed cycles (Altieri, 1995; Biederbeck et al., 1998), may not be economically viable because of the loss of income from that field for an entire growing season. Interseeding legumes with or into a standing crop is common in small grains like wheat (Triticum aestivum L.) and oat (Avena sativa L.). Bruulsema and Christie (1987), Hesterman et al. (1992), and Stute and Posner (1995) demonstrated that this system can result in significant contributions of N to a subsequent corn crop. However, interseeding into a widely spaced row crop like corn or soybean [Glycine max (L.) Merr] may require specialized equipment and additional field operations, and competition between the interseeded and main crops also can be problematic (Kumwenda et al., 1993).

Cover crops, generally grown over the winter between harvest of one crop and planting of a subsequent crop, can overcome at least some of the obstacles associated with green manure and interseeded crops. It is well-documented that cover crops can supply sufficient N for production of a subsequent grain crop with little or no supplemental N fertilizer. McVay et al. (1989) found minimal corn yield response to N fertilizer following hairy vetch or crimson clover (Trifoilium incarnatum L.) cover crops, compared with corn following wheat. Burket et al. (1997) compared sweet corn yield following clover (Trifolium pratense L.), rye, and rye plus pea (Pisum sativum L.) cover crops, finding that both legume cover crops replaced approximately 150 kg fertilizer N ha^-1. Cover crop N contributions to vegetable crops also can be substantial. Stivers and Sheehan (1991) evaluated the use of cover crops for tomato (Lycopersicon esculentum var. esculentum), demonstrating that yields were similar using cover crops and fertilizer N sources. There were minimal responses to fertilizer N following legume cover crops. Skarphol et al. (1987) used legume cover crops to supply N for snap bean (Phaseolus vulgaris L.). Bean yields following legume cover crops were similar to yields obtained with 90 kg fertilizer N ha^-1 without a legume cover crop.

The use of cover crops in cool, northern climates can present additional production challenges, including: (i) limited opportunity for cover crop seeding and establishment; (ii) potentially small accumulation of cover crop biomass and N, especially if the cover crop is seeded after the main crop harvest; and (iii) the rate of cover crop decomposition and N mineralization, a function of both cover crop composition and temperature, may not keep pace with subsequent crop demand. In fact, these criteria of successful establishment, sufficient biomass and N accumulation, and synchrony between N mineralization and crop demand must be met for a cover crop to be viable in any crop system. The availability of reliable in-season N tests (either tissue- or soil-based) would better allow timely decisions on supplemental N applications following cover crops. Our research addresses these potential constraints to cover crop use in short-season conditions, evaluating cover crop options within a 2-yr barley–sweet corn rotation. The use of a cool-season small grain as the first-year crop allows additional flexibility for cover crop seeding and establishment; cover crops can either be interseeded with the barley or sown after barley harvest. In both cases, there is usually sufficient time for cover crop establishment. Sweet corn, as the test crop, responds to legume and fertilizer N application. It can be grown in a short (90–100 d) season environment, although economic return is usually less for later planted crops. The PSNT (Magdoff et al., 1984; Fox et al., 1989) developed for field corn has been successfully used in sweet corn production (Heckman et al., 1995), while the accuracy of the LCNT (Blackmer and Schepers, 1995) has not been assessed for sweet corn. The specific objectives of our research were to evaluate: (i) alfalfa, winter rye, and VR cover crop biomass and N accumulation; (ii) response of a subsequent sweet corn crop to cover crop and fertilizer N application; and (iii) accuracy of PSNT and LCNT in predicting sweet corn response to cover crop and fertilizer N sources.

Materials and methods
Research was conducted from 1992 to 1995 at the University of Maine Sustainable Agriculture Research Farm in Stillwater, ME (44°56' N, 68°42' W). Soil type was a Lamoine silt loam (USDA-SCS Soil Survey Staff, 1992), on a 0 to 2% slope. Soil nutrient levels in 1992, from analysis by the University of Maine Analytical Laboratory, were: pH 6.3 (1:1, soil:water), cation exchange capacity (CEC) 8.8 cmol kg^-1, 9.1 kg P ha^-1, 215 kg K ha^-1, 323 kg Mg ha^-1, and 2276 kg Ca ha^-1, determined using a pH 3.0 1 M NH₄Ac extractant and inductively coupled plasma emission spectroscopy. Soil organic matter was 45 g kg^-1, measured by loss on ignition. Phosphorus was applied as triple superphosphate (0–46–0 N–P–K) and K was applied as KCl (0–0–60 N–P–K), before barley and sweet corn planting in each rotation cycle, according to soil test recommendations from the University of Maine Analytical Laboratory. In 1994 and 1995, Zn (2.25 kg ha^-1) was applied prior to planting sweet corn, as soil Zn levels were less than 1.0 mg kg^-1 soil. Three rotation cycles (2-yr each) were conducted, beginning in 1992 (Cycle I), 1993 (Cycle II), and 1994 (Cycle III). Barley and cover crop establishment occurred during Year 1 and cover crop incorporation and sweet corn production during Year 2 of each rotation cycle. Calendar dates for field operations and crop and soil sampling are provided in Table 1 and climate information is shown in Table 2 . Barley was seeded at 125 kg ha^-1, using a grain drill without packer wheels, at 20 cm row spacing. Alfalfa (cv. Saranac) was broadcast (13.5 kg ha^-1) immediately after barley planting on appropriate plots (3.25 by 10.7 m) and the entire experimental area was cultipacked to enhance seed-to-soil contact. Barley was harvested but yields were not measured. Following barley harvest, straw was incorporated using a tractor-mounted rototiller, except in plots interseeded with alfalfa. Winter rye (cv. Aroostook; 112 kg ha^-1) and hairy vetch (cv. Madison; 56 kg ha^-1) plus winter rye (56 kg ha^-1) were then planted using a grain drill, with 20 cm row spacing. Newly planted plots were then cultipacked.

The experimental design was a randomized complete block with four replications. Treatments were arranged factorially, with the two factors being cover crop and N rate applied to sweet corn. Preliminary analysis of data was accomplished via analysis of variance (ANOVA) with main effects (rotation cycle, cover crop, N rate applied to sweet corn) and interactions. Significant interactions between rotation cycle and treatment (cover crop, N rate) led us to present data for each rotation cycle separately. Cover crop biomass and N accumulation were compared using least significant difference (LSD) only if ANOVA indicated significant treatment differences at = 0.05. To further partition cover crop and N rate effects on sweet corn, single degree of freedom contrasts were utilized to evaluate response to N fertilizer rate (N linear, N quadratic), compare cover crops (legume vs. rye, alfalfa vs. VR), and evaluate cover crop by N rate interactions. Regression equations for N rate response were developed only if linear or quadratic contrasts were significant within that cover crop system. Regression equations were calculated using treatment means, as they are intended to describe the general relationship between N rate and crop yield. Following Hesterman et al. (1992), FRVs were calculated for legume cover crops only if sweet corn yield following the legume cover crop was significantly higher than yield following winter rye without N fertilizer.

Main effects (cover crop, N rate) and interactions between main effects were evaluated on three parameters: marketable ear yield ha^-1, total ear yield ha^-1, and LCNT. Total ear yield was used as an estimate of overall crop productivity. The LCNT is expected to integrate the plant response to both cover crop and fertilizer N sources.

Results and discussion
Preliminary ANOVA using a combined dataset (across rotation cycles) indicated significant cycle by cover crop interactions for most cover crop parameters. For this reason, rotation cycles were analyzed separately. Cover crop biomass in the fall was generally quite small [<0.5>-1; data not shown]. Total cover crop biomass just prior to incorporation in late May ranged from approximately 3.7 to 6.9 Mg DM ha^-1 (Table 3) , demonstrating that these cover crops are well adapted to a moderately cool growing environment. Total biomass for rye and VR was similar to that found in more southern locations, including Maryland (Clark et al., 1994; Shipley et al., 1992), North Carolina (Rannells and Wagger, 1996), and Georgia (McVay et al., 1989). In Cycles I and III, VR produced significantly more above-ground biomass than either alfalfa or rye grown alone. In Cycle II, the proportion of hairy vetch in the mixture was very low (<10%> date and harsh winter conditions without prolonged snow cover. As expected, root biomass accumulation by alfalfa was higher than either annual cover crop option (rye, VR).

Cover crop N content (the product of biomass accumulation and tissue N concentration) varied widely, from 52 to 209 kg N ha^-1 (Table 4) . In Cycles I and III, both legume cover crops had higher N content than rye grown alone, generally accumulating two to three times as much N. In these two rotation cycles, the N content of alfalfa and VR were similar, as were above- and below-ground tissue N concentration. The N content we observed for alfalfa (105–174 kg N ha^-1) is higher than the 20 to 75 kg N ha^-1 found by Hesterman et al. (1992), although they incorporated the alfalfa 3 to 4 wk earlier. Stute and Posner (1995) found that dormant alfalfa accumulated 40 to 50 kg N ha^-1 during the seeding year when underseeded with oat, but did not provide estimates for spring regrowth immediately prior to incorporation. Hairy vetch herbage, when grown alone, generally contains 35 to 45 g N kg^-1 DM (McVay et al., 1989; Skarphol et al., 1987) and mineralizes very rapidly when incorporated into the soil. Growing this legume in mixture with a small grain generally dilutes tissue N concentration for the herbage as a whole. Even with this dilution effect, VR accumulated as much N as alfalfa and had similar tissue N concentration (Table 4). As mentioned above, hairy vetch growth during Cycle II was very limited, thus tissue N concentration and N content were similar to rye grown alone. Rye is a standard cover crop in many areas because of its winterhardiness and tolerance of late planting. As we found in our research, it can produce a substantial amount of biomass as a winter cover crop, generally from 2.5 to 5.5 Mg DM ha^-1 in northern climates (Tollenaar et al., 1993, in Ontario; Kuo et al., 1996, in Washington). However, tissue N concentration is commonly 10 to 15 g kg^-1 or less, especially after seedhead emergence, so total N accumulation may be low.

Conclusions
Sweet corn following alfalfa or VR generally did not respond to additional N fertilizer, as it did following rye grown alone. Alfalfa and VR cover crops supplied all or nearly all of the N required by sweet corn, with FRV ranging from 58 to 156 kg N ha^-1. In one rotation cycle (Cycle II), however, severe winter conditions limited survival of both legume cover crops, resulting in no significant N contribution from the legumes. Both the PSNT and LCNT, using previously established critical values of 25 mg kg^-1 (Heckman et al., 1995) and 43.5 SPAD units (Jemison and Lytle, 1996), respectively, could identify responsive sites with an accuracy of about 75%.

The end-of-season stalk NO₃ test has been used to determine N sufficiency in corn (Zea mays L.). Nitrate concentration is commonly determined with flow-injection analysis (FIA), which is accurate but uses hazardous chemicals and is time-consuming. Use of a simpler method of NO₃ determination, such as the NO₃ specific ion electrode (SIE), may save time and costs, and reduce hazards. The objective of this study was to compare estimates of stalk NO₃ concentration by FIA and NO₃ SIE. For FIA, NO₃ was extracted with 2 M KCl, and the extract was filtered before analysis. For SIE, NO₃ was extracted with 0.04 M (NH₄)₂SO₄, and the extract was analyzed without filtration. The slope of the linear regression between concentrations estimated by SIE and FIA did not differ from 1.0. Use of the NO₃ SIE, compared with FIA, reduces costs, sample processing, and use of hazardous chemicals.

Abbreviations: FIA, flow-injection analysis • SIE, specific ion electrode

INTRODUCTION

THE END-OF-SEASON corn stalk NO₃ test was proposed and advocated by Binford et al. (1990) as a method of determining if excessive or insufficient N was available to the corn crop during the latter part of the season. In the test, 20-cm segments of corn stalks (between 10 and 30 cm above the soil) are collected from several plants (10), dried, ground, and analyzed for NO₃–N. Nitrate N concentrations less than about 700 mg kg^-1 plant tissue indicate that available N limited grain yield; NO₃–N concentrations above 2000 mg kg^-1 indicate that excessive amounts of N were available to the crop (Binford et al., 1992). Other researchers have evaluated the proposed test and concur that when end-of-season stalk NO₃ concentrations are great (>2000 mg kg^-1), excessive levels of N were available to the crop (Varvel et al., 1997). These studies suggest that the end-of-season corn stalk NO₃ test can be used as a postmortem to determine if yield-limiting or excessive N was present. Historical knowledge of crop N need may be used by producers to guide future fertilizer-N management, thereby improving profitability and reducing environmental degradation.

In the initial publications on use of the end-of-season stalk NO₃ test, Binford et al. (1990, 1992) reported using the MgO–Devarda alloy steam-distillation procedure (Keeney and Nelson, 1982) and the Lachat¹ flow-injection procedure (Lachat Instruments, Milwaukee; Method 12-107-04-1-B) to determine NO₃ concentration in aliquots of filtered extracts prepared by shaking known weights of ground stalk material for 30 min in 100 mL of 2 M KCl. Though accurate, these analytical procedures are expensive, time-consuming, and employ hazardous chemicals (strong acids and bases and Cd).

Given that the goal of the stalk NO₃ test is to determine if stalk NO₃–N concentrations are less than 700 mg kg^-1 or greater than 2000 mg kg^-1, it seems logical that a somewhat less accurate procedure could provide essentially the same information, with the possibility of saving time and laboratory resources and avoiding safety and environmental hazard issues. A candidate procedure that is less expensive and less time-consuming, but may be less accurate, is the use of a NO₃ SIE. The object of this study was to compare stalk NO₃ concentration determined by the flow-injection method and NO₃ SIE techniques.
Materials and methods
Shortly after physiological maturity, stalk samples were collected from 10 corn plants in a crop sequence x inbred line x N rate experiment initiated to determine the optimum rate of N fertilizer application for hybrid seed production fields (Wilhelm and Johnson, 1997). Twenty-two (Table 1) of these samples were selected for use in this study to compare methods of determining stalk NO₃ concentration. Samples were selected a priori to represent the range of treatment combinations in the study, and therefore were assumed to provide samples covering the range of stalk NO₃ concentrations found in producers' fields.
Stalk segments were 10 to 20 cm in length and came from the base of the stalk, from 0 to 25 cm above the soil surface. At sampling time, all plants in a 3.1-m segment of row were cut at the soil surface and moved to the field edge. Ten of these plants were selected at random and a stalk segment was taken from each. Each stalk segment was composed of one node and one internode (Fig. 1) . Individuals collecting the samples estimated the fraction of total length of internode between the lowest node and the cut end of the stalk on each sampled plant. The length of internode above the lowest node needed to represent the complement of the fraction below the node was estimated and the stalk cut at that point. In the example shown in Fig. 1, about 0.3 of the internode below the lowest node remained on the stalk as it was removed from the field. To collect the equivalent of one internode, 0.7 of the internode above the lowest node was estimated and the stalk cut at that point. In so doing, each stalk segment was composed of one node and one internode, but part of the internode portion of the sample came from the internode below the node and part from the internode above the node. This sampling procedure was used so that differences in NO₃ concentration between node and internode tissue and differences in length of internodes would not influence estimates of the stalk NO₃ concentration. Stalk segments were dried at about 60°C and ground with a Wiley mill to pass a 2-mm screen before extraction and NO₃ analysis.

In this paper we will use the term FIA to mean the automated procedure for NO₃ analysis defined by Lachat Instruments (Milwaukee, WI; Method 12-107-04-1-B). This procedure is a modification of the Griess–Ilosvay method (Keeney and Nelson, 1982). Nitrate was extracted by shaking a 0.25-g sample of ground stalk tissue for 30 min with 100 mL of 2 M KCl. Extraction media were filtered through Whatman No. 1 paper before analysis with the flow-injection procedure.

For the NO₃ SIE method, 0.25 g of stalk tissue was shaken with 50 mL of 0.04 M (NH₄)₂SO₄ for 30 min. This extraction medium was chosen because it is one of many possible weak salt solutions that could be used to extract NO₃ from plant tissue and is the solution used in the outer chamber of the reference electrode. If water were used as the extraction medium, equal parts of extractant and 0.08 M (NH₄)₂SO₄ would be combined to determine NO₃ concentration with the NO₃ SIE. By using 0.04 M (NH₄)₂SO₄, the need to filter the media was also eliminated, because the electrode could be placed directly into the extraction medium to determine NO₃ concentration. Reference and NO₃ SIE (Orion Research, Boston) were placed directly into the agitating extraction media and electrometer readings observed. Readings were recorded after sequential additions of 1-mL aliquots of NO₃ interference suppressor [0.0378 M (Al₂SO₄)₃, 0.0109 M Ag₂SO₄, 0.0257 M H₃SNO₃, and 0.0210 M H₃BO₃] produced no change in meter output. Several ions can influence the accuracy of NO₃ concentration estimates made with NO₃ SIE. The NO₃ interference suppressor was used to eliminate interference from organic anions (aluminum sulfate), halogens, cyanide and sulfide ions (silver sulfate), nitrite (sulfamic acid), and carbonate and bicarbonate ions (boric acid; Orion Research, 1980).

For both analytical methods, NO₃–N concentration in stalk tissue was calculated from a standard curve (NO₃–N on log scale) developed from known standards ranging in NO₃–N concentration from 0 to 20 mg kg^-1. For the FIA, standards were prepared in 2 M KCl; for the NO₃ SIE, in 0.04 M (NH₄)₂SO₄. Analysis of variance, regression analysis, and t-tests were used to determine if the two methods differed in their estimates of NO₃ concentration and how the differences affected interpretation of the end-of-season stalk NO₃ test.

Results and discussion
To be useful as an alternative method for assessing stalk NO₃ concentration, the NO₃ SIE method must have two characteristics. First, mean values must be similar to those found by methods assumed to be the standard (FIA). Secondly, estimates of NO₃ concentration must be repeatable. We will address the second question first. Though we expected FIA to provide more precision than the NO₃ SIE, mean standard deviations (3 extractions and analyses on each of 22 samples) for the two methods were similar; 37.5 mg NO₃–N kg^-1 for FIA and 44.3 mg NO₃–N kg^-1 for the NO₃ SIE. Sample NO₃–N concentrations ranged from about 100 to 5300 mg kg^-1. These standard deviations values may seem large; however, when they were converted to coefficients of variation and expressed as percent of the mean, the precision of both methods was very acceptable (1.5% for FIA and 1.8% for NO₃ SIE). Visual examination of the relationship between standard deviations and means (Fig. 2) appears to show a stronger association between these parameters for the NO₃ SIE than for FIA. However, when linear correlation coefficients were computed the reverse was found: For the NO₃ SIE method, ; for the FIA method, . This apparent contradiction was caused by the strong influence of five samples that showed very little variation with the NO₃ SIE (i.e., the five points falling on the x-axis in Fig. 2). When these points were removed, results of the correlation analysis agreed with our visual assessment. The recalculated correlation coefficient for the NO₃ SIE method was . The reason for several points having no variation is largely an artifact of the use of a digital electrometer to measure output from the NO₃ SIE. The meter cannot display very small differences between samples. Therefore, the meter readout was the same for all samples and the variation was calculated to be zero. The purpose of the stalk NO₃ test is to determine if NO₃–N concentrations are less than 700 mg kg^-1 or greater than 2000 mg kg^-1. Therefore, the inability to detect small differences between samples and a strong correlation between the mean and standard deviation of measurements (undesirable characteristics for analytical procedures) have little bearing on the usefulness of the technique.

In conclusion, these data indicate that stalk NO₃–N concentration estimated by the two methods may differ slightly. The strong relationship between results produced by the methods indicates that any discrepancy between methods would be small and within the requirements for the end-of-season stalk NO₃ test. In addition, savings in terms of equipment costs and time for sample preparation could be substantial. Use of hazardous chemicals is also eliminated: There is no need for strong acids and bases, nor for the carcinogen Cd.