Introduction
Organic dairy production is one of the fastest-growing agricultural sectors in the northeastern USA (Kersbergen, Reference Kersbergen2007; Jemison et al., Reference Jemison, Darby and Reberg-Horton2012). Despite growth, farmer profit margins are variable (Parsons, Reference Parsons2016) and purchased organic grain remains a top expense (Jemison, Reference Jemison2008; Hoshide et al., Reference Hoshide, Halloran, Kersbergen, Griffin, DeFauw, LaGasse and Jain2011; Kersbergen et al., Reference Kersbergen, Anderson, Criner and Davis2013). In response, dairy producers seek methods that optimize the production of farm-grown, high-quality perennial forages to support year-round dairy herd feed needs (Sanderson et al., Reference Sanderson, Soder, Muller, Klement, Skinner and Goslee2005; Hoshide et al., Reference Hoshide, Halloran, Kersbergen, Griffin, DeFauw, LaGasse and Jain2011).
Perennial legume/grass forage must be high yielding to be economical and ‘high-quality’, as defined by high protein content, digestibility and intake, for the efficient conversion into milk (Waldo and Jorgensen, Reference Waldo and Jorgensen1980). Key factors affecting forage yield and quality include maturity at harvest (Buxton, Reference Buxton1996; Juskiw et al., Reference Juskiw, Helm and Salmon1997), harvest interval (Sheaffer et al., Reference Sheaffer, Martin, Lamb, Cuomo, Jewett and Quering2000; Kallenbach et al., Reference Kallenbach, Nelson and Coutts2002; Brink et al., Reference Brink, Hall, Shewmaker, Undersander, Martin and Walgenbach2010) and forage composition (Waldo and Jorgensen, Reference Waldo and Jorgensen1980). Legumes have higher intake than grasses (Shaver et al., Reference Shaver, Satter and Jorgensen1988) and greater potential to increase milk production than grass alone (Bertilsson et al., Reference Bertilsson, Dewhurst, Tuori, Wilkins and Paul2002; Dewhurst et al., Reference Dewhurst, Fisher, Tweed and Wilkins2003). Despite these benefits, the yield potential of legumes tends to decline within 3–5 years of production depending on the species, soil pH and drainage (Griffin, Reference Griffin2004), and may result in a shift toward less palatable weed species (Jemison, Reference Jemison2008) or grasses. The relatively short duration of high-quality perennial forages underscores the need for reliable methods of stand re-establishment.
To replace a degraded perennial forage crop, organic dairy producers in temperate regions may plant small grains because they modify the cropping environment in favor of stand re-establishment. The deep tillage (i.e. moldboard plowing) associated with establishing grain crops may suppress perennial weeds (Liebman and Dyck, Reference Liebman and Dyck1993) that can be problematic. By re-establishing a forage crop by ‘underseeding’ (seeding a secondary crop into or with a cash crop) the grain serves as a ‘nurse-crop’ and the forage has weed suppression capabilities as a ‘living-mulch’ (Sheaffer and Seguin, Reference Sheaffer and Seguin2003; Hiltbrunner et al., Reference Hiltbrunner, Liedgens, Bloch, Stamp and Streit2007). Small grains can supplement the feed ration (Rotz et al., Reference Rotz, Roth and Stout2002), or, if higher quality standards are met, dairy producers may earn greater profits by selling grain off-farm on the expanding local, organic bread flour market (Nass et al., Reference Nass, Papadopolous, MacLeod, Caldwell and Walker2002).
Organic food-grade wheat must have a protein concentration of 120 g kg−1 to satisfy flour milling standards or it will receive a reduced price (Mallory and Darby, Reference Mallory and Darby2013). However, growing commercially viable organic bread wheat can be challenging because of insufficient crop available nitrogen (N) (David et al., Reference David, Jeuffroy, Henning and Meynard2005; Olesen et al., Reference Olesen, Askegaard and Rasmussen2009) and lack of effective weed control methods (Kolb and Gallandt, Reference Kolb and Gallandt2012). Furthermore, these factors tend to magnify a well-known negative relationship between grain yield and grain crude protein (CP) (Fowler, Reference Fowler2003) making it critical to identify the opportunities and limitations for bread wheat production within organic dairy cropping sequences.
The availability of N during key periods of wheat growth is a major factor supporting bread wheat yield and CP (Dawson et al., Reference Dawson, Huggins and Jones2008; Casagrande et al., Reference Casagrande, David, Valantin-Morison, Makowski and Jeuffroy2009; Mallory and Darby, Reference Mallory and Darby2013). Early-season N uptake tends to influence grain yield and CP, whereas later N uptake influences CP more than yield (Eilrich and Hageman, Reference Eilrich and Hageman1973; Gooding et al., Reference Gooding, Gregory, Ford and Ruske2007). Dairy (Bos taurus) manure may be a practical N amendment, yet studies show limited success in achieving adequate grain protein content for winter wheat (Mallory and Darby, Reference Mallory and Darby2013) and varied success attaining desired grain yields in spring wheat (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014) due to factors such as the amounts of inorganic and organic N contained in the manure (Cabrera and Gordillo, Reference Cabrera, Gordillo and Hatcher1995). Matching N availability with crop N demand is an important factor and yet the mineralization rate of organic N can be low and variable depending on application timing (Mallory and Darby, Reference Mallory and Darby2013), environmental conditions (Ellert and Bettany, Reference Ellert and Bettany1992) and agronomic practices (Douglas and Magdoff, Reference Douglas and Magdoff1991).
Alternatively, studies suggest that N mineralization from legume/grass residues may better match wheat N uptake than manure applications. Olesen et al. (Reference Olesen, Askegaard and Rasmussen2009) found that grain protein of organic winter wheat increased more following N2-fixing crops than it did after receiving manure. In other studies, wheat following leguminous crops showed increased N uptake (Debaeke et al., Reference Debaeke, Aussenac, Fabre, Hilaire, Pujol and Thuries1996), and grain yield, versus following non-legume crops. Legume species, age, management (grazing versus cutting) and time of incorporation, are important factors dictating the amount and timing of N release (Torstensson, Reference Torstensson1998; Fageria, Reference Fageria2007). Further research is needed to understand how bread wheat production standards are affected by perennial forage stands managed for high-quality feed production, which at the time of incorporation may be in decline and with less legume content.
Prior crop also affects the weed species and abundance in the cropping environment. For instance, Casagrande et al. (Reference Casagrande, David, Valantin-Morison, Makowski and Jeuffroy2009) found lower weed density when wheat followed cultivated annual crops compared with perennial fodder crops. In the northeastern USA, early-spring germinating annual weeds such as wild mustard (Sinapis arvensis) can be especially challenging for spring wheat, whereas perennials such as quack grass (Elytrigia repens) and Canada thistle (Cirsium arvense) can be problematic for both wheat types (Mallory et al., Reference Mallory, Bramble, Williams and Amaral2012). Prior crops also can encourage diseases. Corn and wheat crop residues are the primary inoculum host for the fungal pathogen Fusarium head blight (FHB) (Bai and Shaner, Reference Bai and Shaner2004), which can produce a mycotoxin in the grain, making it unsuitable for human and/or animal consumption at certain concentrations.
Currently, the economic and environmental implications of incorporating small grains on dairy farms have been analyzed only through whole-farm modeling studies (Rotz et al., Reference Rotz, Roth and Stout2002; Abreu et al., Reference Abreu, Hoshide, Mallory, Roche, Oliveira, Kersbergen, Lana and Fonseca2016) as opposed to field studies. Including high-quality bread wheat in crop sequences has the potential to improve whole-farm profitability at relatively low risk. If grain does not meet bread-flour quality standards, it could potentially be used on-farm as feed thereby reducing purchased concentrate costs, or easily sold based on the high demand for organic feeds. Economic review is needed to evaluate how bread wheat performs in conjunction with the other crop goals of a typical organic dairy located in the northeastern USA.
In this study, we aimed to determine if hard red winter and spring wheat can be grown profitably for the bread flour market in dairy crop rotations and serve to re-establish a high-quality legume/grass perennial forage stand for milk production. We evaluated five representative 3-year, feed-based, annual crop sequences that included bread wheat and ended with a new perennial forage seeding, and compared them with the regional ‘standard practices’ of reseeding perennial forage after two years of corn silage or maintaining continuous perennial forage production. The specific objectives were to: (1) determine how prior crop affects wheat N uptake, grain yield and grain quality; (2) assess whether rotating out of perennial forage into annual crops and then back into perennial forage improves forage quality; and (3) evaluate the overall profitability of the crop sequences.
Materials and methods
Study site and experimental design
A field trial with 3-year crop sequences was conducted at two sites at the University of Maine Smith Farm Forage and Crop Research Facility (44°90′N lat., 68°69′W long.) in Old Town, ME, USA. The first site was initiated in 2010 and ended in 2012. A second site was initiated in 2011 and ended in 2013. Before trial initiation, all plots were in a perennial forage mixture of 40% alfalfa (variety WL 319HQ), 40% perennial ryegrass (Lolium perenne ‘Tivoli’) and 20% timothy (Phleum pratense ‘Glacier’) that had been established in the spring of 2009 at site 1 and in the spring of 2010 at site 2. At both sites, the perennial forage was established using a Massy Harris 2.4 m width drill (ACGO Co., Duluth, GA) and seeded at a rate of 29 kg ha−1. During the seeding year, the perennial forage was harvested once during late August (2009) at site 1 and not at all at site 2.
The trial was conducted on soils with clayey substratum with slow permeability, typical to many northeast organic dairy farms (Hoshide et al., Reference Hoshide, Halloran, Kersbergen, Griffin, DeFauw, LaGasse and Jain2011). The soil at site 1 was a Suffield very fine sandy loam (coarse-silty over clayey, mixed, active, mesic Dystric Eutrudepts) with, a pH of 6.3, 4.5% organic matter, 6.8 kg ha−1 soil test phosphorus (P) (Modified Morgan Extraction), 172.2 kg ha−1 soil test potassium (K) and 25.2 kg ha−1 soil test sulfur (S), based on 2,241,702 kg ha−1 of soil in a plow layer (16.9 cm deep) as determined per the University of Maine Soil Testing Service (Hoskins, Reference Hoskins1997). The soil at site 2 was a mix of Suffield very fine sandy loam and Melrose fine sandy loam (coarse-loamy over clayey, mixed over illitic, superactive, frigid Oxyaquic Dystrudepts) with a pH of 6.1, 3.9% organic matter, 5.3 kg ha−1 soil test P, 155.5 kg ha−1 soil test K and 23.4 kg ha−1 soil test S.
Each trial was arranged in a randomized complete block design with four replications and individual plots were 10.7 × 4.9 m.
Treatments consisted of six 3-year crop sequences that included annual crops and ended in a new perennial legume/grass forage, plus one control that remained in continuous perennial legume/grass forage production (Table 1). Two of the sequences included winter wheat in the second year (varieties Harvard in 2011 and AC Morley in 2012) preceded in the first year by either an early-season, 83-day variety of corn silage (Masters Choice [MC-468]) (C–WW–PFc) or one-year-old perennial legume/grass forage (PFa–WW–PFc). The change in variety for winter wheat was due to seed availability. Three of the sequences included spring wheat in the second year (variety Glenn) preceded in the first year by either a full-season, 94-day variety of corn silage (MC-490) (C–SW–PFc), soybean (variety Tundra) (S–SW–PFc) or year-old perennial legume/grass forage (PFa–SW–PFc). In crop year 2, wheat was ‘underseeded’ with a perennial forage mix of 60% red clover (Trifolium pratense ‘Mammoth’), 20% perennial ryegrass and 20% timothy, hereafter referred to as ‘PFc’. Clover was used instead of alfalfa in the new perennial forage for the shorter rotations. The sixth sequence (C–C–PFc), modeled after the standard practice used by New England farmers to re-establish a new perennial forage stand, included two years of full-season corn silage followed by winter triticale (× Triticosecale ‘Trical 336′) undersown with PFc. Thus, in crop year 3, the first six sequences were in new perennial forage stands in their first production year, while the continuous perennial forage control, PFa–PFa–PFa, was a fourth-year perennial forage stand.
Descriptions of the crop sequences for an organic dairy crop sequence trial conducted in Maine from 2010 to 2013.
1 C, corn; WW, winter wheat; PFc, perennial forage established during crop year 2 was (60% red clover, 20% perennial ryegrass and 20% timothy); PFa, perennial forage established prior to crop year 2 was (40% alfalfa, 40% perennial ryegrass and 20% timothy); SW, spring wheat; S, soybean.
Management practices and crop management
Fields were managed organically for the three years preceding trial initiation, though the land was not certified under the NOP (National Organic Program). Table 2 outlines the dates for field operations and sampling by year. Composite soil samples were collected prior to planting to determine P, K and S levels (Hoskins, Reference Hoskins1997). The perennial forage control was typically harvested three times each growing season, on average every 55 days. During forage harvests, a forage subsample was collected with a handheld sickle bar mower (BCS 725, BCS America, Portland, OR), after which the remaining area was cut with a flail mower and the remaining biomass was raked off the plots by hand. Prior to planting annual crops the existing perennial forage stand was cut with a flail mower and cuttings were left on the ground. Primary tillage was performed using a 450 International three-bottom moldboard plow (Navistar International Co, Lisle, IL). Seedbed preparation was performed with a Kuhn S.A. rototiller Type EL62-210 (Kuhn North America Inc., Brodhead, WI) or a 2.4 m width Bush Hog disk harrow (Bush Hog Inc., Selma, AL), followed by a Perfecta® II Field Cultivator (Unverth Mg. Co, Inc., Kalida, OH). In plots that received solid dairy manure, the manure was applied before field cultivation and was incorporated within 4 h. Manure application rates varied by crop treatment and the N credit from the incorporated forage (Table 3). Nitrogen rhizodeposition of the legumes was not considered resulting in the 0 values for PF.
Summary of field operations and sampling in an organic dairy crop sequence trial conducted in Maine from 2010 to 2013.
1 Primary tillage consisted of one pass with a moldboard plow.
2 Seedbed preparation consisted of one pass of a rototiller or disc harrow, and field cultivator exclusive of manure applications.
3 Manure application and incorporation with a field cultivator.
4 60% red clover, 20% perennial ryegrass and 20% timothy.
5 Spring wheat had to be reseeded due to a crop establishment failure.
6 Only performed for spring wheat following perennial forage in treatment PFa–SW–PFc.
7 Perennial forage re-seeded due to triticale/PFc crop failure.
8 Except C–C–PFc, which was not harvested until the second and third forage harvests.
Estimated available N from preplant manure and perennial forage incorporated prior to crop establishment in crop years 1 and 2 of an organic dairy crop sequence trial conducted in Maine from 2010 to 2013.
1 C, corn; WW, winter wheat; PFc, 60% red clover, 20% perennial ryegrass and 20% timothy; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
2 No fertility sources applied during crop year 3.
3 Estimated available N was calculated as 35% of the total organic-N and 50% of the total inorganic N for dairy manure (Gale et al., Reference Gale, Sullivan, Cogger, Bary, Hemphill and Myhre2006).
4 Estimated N credit of 78 kg N ha−1 for perennial forage with >25% legume (Hoskins, Reference Hoskins1997).
Corn and soybean crops were planted with a John Deere 700 Max Emerge four-row crop planter (John Deere, Moline, IL) at 81-cm row spacing and target densities of 7.4 viable seeds m−2 for corn and 37.5 viable seeds m−2 for soybeans. Corn and soybeans were cultivated between the fourth and seventh leaf stages with a 183 International Vibra Tine/Vibra Shank® four-row cultivator (Navistar International Co, Lisle, IL). Corn, both early- and full-season varieties, were harvested by hand with rice knives when the corn was well dented. Soybeans were harvested by hand at physiological maturity when 95% of pods expressed their genetically mature color and were processed using a Wintersteiger small-plot combine (Ried, Austria).
Hard red winter wheat was drill seeded at 16.5-cm row spacing and a target density of 355 viable seed m−2. Wheat was underseeded with PFc by frost seeding the following spring using a using a Scotts® Turf Builder Classic Drop Spreader (The Scotts Company LLC, Marysville, OH) at a rate of 29 kg ha−1. In crop year 2, hard red spring wheat was seeded using the same drill and row spacing as for winter wheat with a target density of 333 viable seed m−2 and underseeded the same day with PFc using the same equipment mentioned above at a rate of 25 kg ha−1 except that the seed was lightly raked in by hand. Both spring and winter wheat were harvested with the small-plot combine. Forage was seeded after full-season corn harvest by seeding PFc at a rate of 42 kg ha−1 with drill-seeded triticale at a target density of 440 viable seeds m−2. In crop year 3, newly established perennial forage was harvested using the same method above at three harvest intervals unless otherwise noted.
Measurements and analytical procedures
Forage sampling
Perennial forage yield and quality were measured at all harvests, with the exception of forage quality at harvest 2 in site 1, crop year 3 and in site 2, crop year 2. Two, 6.1 × 1.1 m strips per plot were mowed 5.1 cm above the soil surface. Samples from the two strips were bulked, representing a total sample area of 13.4 m−2. The bulk sample was weighed within 2 h and a 10%-by-weight subsample was weighed, dried for 7 days at 50°C and reweighed to determine moisture. Yield was corrected to a dry matter (DM) basis. Subsamples were ground (2 mm mesh) and a 113-g sample was submitted to Dairy One Laboratory (Ithaca, NY) for Near Infrared Reflectance (NIR) spectroscopy to determine perennial forage nutrient composition. In crop year 3, an additional 10%-by-weight subsample was taken of the harvested perennial forage and separated into legume, grass or broadleaf weed fractions. Fractions were weighed fresh, dried for 7 days at 50°C and reweighed. The yield of each fraction was corrected to a DM basis.
Wheat sampling
Wheat above-ground biomass and N uptake were measured at the soft dough stage, Zadoks growth stage (GS) 85 (Zadoks et al., Reference Zadoks, Chang and Konzak1974), when plants are considered to have accumulated their greatest biomass. Four 0.30 × 0.33 m quadrats per plot were cut 2.5 cm above the soil surface and bulked, representing a total sample area of 0.40 m−2 of plot. Samples were separated to wheat, perennial forage and weed fractions, which were dried for 7 days at 50°C and weighed. Wheat tissue was ground (2 mm mesh) and a 250-mg subsample was analyzed for total N concentration by combustion using a Leco CN2000 analyzer (Leco Corp., St. Joseph, MI). Plant N uptake was calculated as plant aboveground biomass × %N. Wheat grain was harvested at physiological maturity from a 1.4 × 9.1 m area and the number of spikes per bulked sample was recorded. Grain was cleaned (Clipper M2B, Commodity Traders International, Trilla, IL) and grain moisture was determined (GAC 2100®, DICKEY-john Corp., Auburn, IL). Fresh weights of cleaned grain were recorded and adjusted to 135 g kg−1 grain moisture. A 350-g subsample of grain from each plot was ground (2 mm mesh) and submitted to Dairy One Cooperative, Inc. (Ithaca, NY) for determination of N concentration by NIR. Grain protein concentrations were calculated by dividing the reported CP (NIR analysis) by 6.25 (Kjeldahl determination) and multiplying by 5.7 (AACC International, 2010). Concentrations were corrected to 120 g kg−1 grain moisture. Concentrations of DON were determined on a 10-g grain subsample using Veratox DON 2/3 Quantitative test (NEOGEN Corp., Lansing, MI), which has a detection range of 0.5–5 ppm.
Corn silage and soybean sampling
Corn silage and soybean crop yields were measured by cutting the crops by hand at 15 cm above the soil surface from two 4.6 × 0.8 m lengths of row from the center of the plots, representing a total sample area of 7.4 m2. Corn stalk number and fresh weights were immediately recorded in the field. A ten-stalk subsample was taken from both rows combined. The subsample was processed through a Sears Craftsman chipper shredder (Sears Mgt. Co, Hoffman Estates, IL), weighed, dried for 7 days at 50°C and reweighed to determine moisture and the original bulk yield was corrected to 30% DM. Soybeans were cut 9.0 cm above the soil surface and whole plants were processed using the small-plot combine and weighed fresh. Subsamples were dried for 7 days at 50°C and reweighed to determine moisture and soybean yield was corrected to 13% DM.
Statistical analysis and evaluation models
Data were analyzed with the statistical program RStudio (R Core Team, 2015) using a mixed model analysis of variance (ANOVA) with replication as a random effect and year and treatment as fixed effects. The data were analyzed by year due to the presence of year X treatment interactions in the majority of response variables. The ANOVA assumption of equal variance was verified using Brown Forsythe's test using the ‘car’ package (Fox and Weisberg, Reference Fox and Weisberg2011). Residual values were also used to assess normal distribution using the Shapiro–Wilk Normality test. When residuals did not conform to equal variances and normality, a Box-Cox power transformation was used using the ‘MASS’ package (Venables and Ripley, Reference Venables and Ripley2002). If the ANOVA produced a significant F test (P < 0.05), preplanned orthogonal contrasts were used for means separation using the ‘multcomp’ package (Hothorn et al., Reference Hothorn, Bretz and Westfall2008). A repeated measures analysis was used to analyze perennial forage yield over time and pairwise comparisons were performed using the Holm Adjustment using the ‘stats’ package (R Core Team, 2015). We used MILK worksheets versions 2000 (Shaver et al., Reference Shaver, Lauer, Coors, Schwab and Hoffman2000) and 2013; (Undersander et al., Reference Undersander, Combs, Shaver and Hoffman2013) to estimate potential milk yield Mg ha−1 based on the yield and quality parameters of perennial forage and silage. The MILK worksheets use prediction equations that combine forage yield and quality into a single term to provide an estimate of milk produced per area (Shaver et al., Reference Shaver, Undersander, Schwab, Hoffman, Lauer, Combs and Coors2001). As described by Shaver et al. (Reference Shaver, Undersander, Schwab, Hoffman, Lauer, Combs and Coors2001), MILK uses forage analyses to estimate energy content using a modification of the National Research Council (2001) summative approach, and DM intake from neutral detergent fiber (NDF) and in vitro NDF digestibility to predict milk production per Mg of forage DM. Forage DM yield multiplied times the milk produced per Mg of forage DM provides an estimate of the milk produced per hectare.
Economic analysis
A partial revenue budget was used to compare cropping sequences under a 3-year revenue scenario assuming all crops were sold off-farm because feed value projections for all crops was not possible given the evaluation models available. Crops valued at off-farm feed prices included haylage, corn silage, feed-grade wheat and feed-grade soybeans. Wheat grain with CP of 120 g kg−1 or greater received food-grade prices. Seed and machinery costs and crop pay prices were from 2014 and their sources are included in Table 4. Machinery costs were based on the operations performed in this study with machinery scaled to meet the needs of an assumed ‘medium scale’ 120-head milking herd, on a 200 ha organic dairy farm. For this study the assumption was made that farmers have the machinery, equipment and storage capacity needed to successfully grow and market grains. This, however, may not be the case and would add substantial capital costs. For profitability assessments, a 15% yield loss due to respiration and fermentation was assumed for corn silage and haylage and both were adjusted to 35% DM (R. J. Kersbergen, personal communication, March 4, 2016).
Seed costs, machinery costs and pay prices from 2014 used to partial budget for the 3-year crop sequences compared in a study in Maine.
1 Albert Lea Seed House, Albert Lea, MN, 2014.
2 Northeast Agriculture Sales Inc., ME, 2014.
3 Lazarus (Reference Lazarus2014).
4 Mower conditioner, 2.7 m (27.7 US$ ha−1), hay rake, 2.7 m (13.7 US$ ha−1), forage harvester, pull-type (Pickup head) (54.3 US$ ha−1).
5 USDA Agricultural Marketing Service (2016).
6 R. J. Kersbergen, personal communication, March 2014.
Weather conditions
Monthly mean temperature and rainfall events for the four years of the trial are presented in Table 5. Monthly mean temperatures were typical for this site relative to 30-year averages with the exception of 2010 when early spring temperatures were warmer than the average. In 2011, spring rainfall was heavy, especially in April when rainfall exceeded the average by 50 mm, and caused a spring wheat crop establishment failure with reseeding delayed until early June (Table 2). Late-season precipitation also was higher than usual, exceeding the average by 138 mm in August. In 2012, June rainfall was 50 mm above average, affecting early-season spring wheat growth, followed by low rainfall in July that impacted grain fill for spring wheat. In 2013, the early spring was relatively dry but precipitation was greater than average from late May to June and then again in August.
Monthly mean temperature and rainfall from May through September in 2010–2013 at the University of Maine Research Farm, Old Town, Maine compared with average climate data for 1981–2010.
1 NA, not applicable as experiment data collection was completed on 21 August.
2 Precipitation data were not available for 26 April 2012.
Results and discussion
Corn and soybean yields
Corn silage and soybean yields in crop year 1 (data not shown) were variable compared with regional targets for organic production, which are approximately 31 and 48 Mg ha−1 for 83- and 93-day corn varieties, respectively (Darby et al., Reference Darby, Cline, Gervais, Cummings, Madden and Harwood2012a, Reference Darby, Cummings, Burke, Harwood and Monahan2014) and 3.1 Mg ha−1 for organic soybeans (Darby et al., Reference Darby, Harwood, Domina, Madden, Cummings and Gervais2012b, Reference Darby, Blair, Cummings, Harwood, Madden and Monahanc). In this trial, corn silage that followed perennial forage in the first year of the crop sequences yielded 32 and 31 Mg ha−1 for early-season corn and 43 and 48 Mg ha−1 for full-season corn in 2010 and 2011, respectively. Second year corn silage in the C–C–PFc treatment yielded 32 and 26 Mg ha−1 in 2011 and 2012, respectively. Soybean following perennial forage yielded 1.7 and 1.5 Mg ha−1 in 2010 and 2011, respectively. Low soybean yields were possibly associated with selecting a variety with early maturity.
Wheat biomass and N uptake
Spike density, aboveground wheat biomass and plant N uptake at peak biomass were significantly affected by wheat type in both years (Table 6) such that sequences with winter wheat were always greater than spring wheat. Spike density was on average 28% greater in sequences with winter wheat. Average wheat biomass was similar in both years and sequences with winter wheat produced more than three times, or 8786 kg ha−1 more biomass than those with spring wheat. Due to an earlier planting date, winter wheat experiences greater accumulated temperatures (Thorup-Kristensen et al., Reference Thorup-Kristensen, Salmerón Cortasa and Loges2009), and has a longer duration of DM production than spring wheat (Entz and Fowler, Reference Entz and Fowler1991). Similarly, averaged over the two site-years, winter wheat N uptake was 42%, or 34 kg N ha−1, greater than spring wheat N uptake.
Wheat spike density, aboveground biomass of wheat, perennial forage and weeds, wheat N uptake at soft dough stage (GS85 1 ) and ANOVA results for spring and winter wheat grown in the organic dairy cropping systems trials conducted in 2011 and 2012 in Maine.
1 Zadoks scale for growth staging cereals (Zadoks et al., Reference Zadoks, Chang and Konzak1974).
2 C, corn; WW, winter wheat; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
3 Wheat aboveground biomass in 2011, weed aboveground biomass and wheat N uptake data in 2012 were transformed for the ANOVA; back-transformed values are presented.
4 Weed biomass not collected in WW treatments because biomass was insignificant.
*, **, ***, Significant at P < 0.05, P < 0.01 and P < 0.001, respectively.
In 2011, a prior crop effect was evident in spring wheat sequences. Spring wheat following perennial forage exhibited 31% more spikes m−2, 1767 kg ha−1 more biomass and 32 kg ha−1 more N uptake than following full-season corn or soybeans (Table 6). This effect may be based on differences in the amount of available N and the timing of N availability due differences in the manure versus the legume–grass forage (Olesen et al., Reference Olesen, Askegaard and Rasmussen2009). Slower release of N from the perennial forage may have mirrored wheat N demand more closely and preempted loss from early season excessive rainfall compared with the inorganic N from manure. Furthermore, similarities in weed and forage biomass among these treatments suggests competition had no influence on N uptake.
Biomass of perennial forage and weeds were highly variable during both years. Coefficients of variation (CV) of perennial forage biomass were 36 and 32% during 2011 and 2012, respectively, and weed biomass was even more variable (CV = 50 and 55%), which has been observed in other studies of organic wheat systems (David et al., Reference David, Jeuffroy, Henning and Meynard2005; Casagrande et al., Reference Casagrande, David, Valantin-Morison, Makowski and Jeuffroy2009). Weed biomass in the winter wheat treatments was not appreciable in 2011 and was on average 123 kg ha−1 less than spring wheat treatments in 2012. Greater aboveground biomass and vigorous fall and early spring growth of winter wheat can suppress weed establishment and make it more competitive against weeds than spring wheat (Beres et al., Reference Beres, Harker, Clayton, Bremer, Blackshaw and Graf2010). To give spring wheat a competitive advantage against weeds early springtime seeding is recommended (Mallory et al., Reference Mallory, Bramble, Williams and Amaral2012). In this study, challenging springtime site conditions in 2011 delayed spring wheat seeding and put the crop at a disadvantage against weeds. Regardless of the amount of weeds present, winter and spring wheat had equal forage biomass, suggesting that the timing of underseeding had no effect on perennial forage re-establishment. Lastly, there was no evidence that prior crop influenced weed biomass, indicating weed pressure following annual crops was similar to perennial forage despite repeated cultivation of the annual crops in crop year 1. This finding suggests higher than anticipated weed suppression in the perennial forage.
Wheat grain yield and quality
The majority of treatment differences found in grain yield, quality and DON levels at harvest were based on wheat type (Table 7). Winter wheat yields were on average 4.2 Mg ha−1 more than spring wheat with consistent effects across years. Studies have shown differences in the agronomic performance between the two wheat types. In a study comparing winter to spring wheat in Saskatchewan, Entz and Fowler (Reference Entz and Fowler1991), found winter wheat produced on average 36% higher yields than spring wheat, which they attributed to greater kernel weight, harvest index and kernels per unit area. While these yield components were not measured in this study, spike density at peak biomass was always greater in winter wheat than in spring wheat (Table 6), which may have contributed to higher grain yields observed.
Wheat grain yield, crude protein and DON 1 at grain harvest and ANOVA results for spring and winter wheat grown in organic dairy cropping systems trials conducted in 2011 and 2012 in Maine.
1 Deoxynivalenol.
2 C, corn; WW, winter wheat; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
3 Grain yield data in 2011 were transformed for the ANOVA; back-transformed values are presented.
*, **, ***, Significant at P < 0.05, P < 0.01 and P < 0.001, respectively.
Beyond inherent performance differences, the yield effects observed here reflect the influence of seeding time and soil conditions more than other factors such as available N. Waterlogged spring soil conditions during both trial years delayed spring wheat seeding by over two weeks on average compared with spring wheat seeded in a nearby organic variety trial on sites with well-drained fine-sandy loam soil the same years (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014). Late planting caused the grain filling period to coincide with warmer, drier weather which shortened the duration of grain fill to the extent that yield was reduced by an average of 2.4 Mg ha−1 compared to yields of the same variety grown in the nearby trial (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014). Similar results were found in a spring wheat planting date study in Ottawa, Canada, where wheat planted ‘late’ on 19 May had a 15–45% lower grain yield in three out of four site years compared with wheat planted on approximately 28 April and 9 May (Subedi et al., Reference Subedi, Ma and Xue2007). In contrast to spring wheat, average winter wheat yields in the current study either exceeded or were equivalent to yields in the local comparison (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2015).
Crude protein differed by wheat type during both years (Table 7). Spring wheat had consistently higher CP than winter wheat, making bread wheat quality more attainable: averaging 145 versus 96 g kg−1, respectively, across crop sequences and years. In bread wheat production, high CP is often negatively correlated with grain yield (Terman et al., Reference Terman, Ramig, Dreier and Olson1969; Terman, Reference Terman1979; Loftier et al., Reference Loftier, Rauch and Busch1985; Fowler, Reference Fowler2003). Spring wheat CP averaged across the years was 11 g kg−1 higher than published CP values for ‘Glenn’ with higher yields, likely because grain yields were low (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014). In contrast, winter wheat CP was, on average, 24 g kg−1 below bread flour market standards. Mallory and Darby (Reference Mallory and Darby2013) found difficulties in achieving desired winter wheat CP with preplant manure alone, but suggested higher CP could be accomplished with topdressing and variety selection. Variety selection is a top factor that determines CP (Fowler et al., Reference Fowler, Brydon, Darroch, Entz and Johnston1990), and both ‘AC Morley’ and ‘Harvard’ tend to be high yielding varieties with lower protein potential (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2015). Selecting varieties for high protein potential is, however, usually coupled with lower grain yield (Fowler, Reference Fowler2003), but may be justified in organic production systems to meet required standards.
In 2012, winter wheat following perennial forage had 15 g kg−1 higher CP than winter wheat following early-season corn silage. This result suggests either that N was lost from manure or that the mineralization of organic N from forage occurred later than that from manure. Despite an increase in CP, the perennial forage–winter wheat sequence was still 13 g kg−1 below bread flour market standards. There were no measurable differences in N uptake between these two treatments at soft dough.
DON levels were within US-FDA guidelines with the exception of spring wheat in 2011 (Table 7). In this year, spring wheat was acceptable as livestock feed though it exceeded the 1 ppm guideline for human consumption (US-F.D.A., 2010) and had on average 1.34 ppm more DON than winter wheat. Warm moist weather that coincides with wheat flowering stimulates the development of the disease and wheat infection (Goswami and Kistler, Reference Goswami and Kistler2004). Over 90 mm of rainfall occurred at the time of flowering in late July into early August (data not shown), making conditions conducive for infection. Varietal effects likely had little influence on the FHB infection because in other New England studies, ‘Glenn’ has shown low DON levels compared with other spring wheat varieties (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014).
Perennial forage yield and botanical composition
Yields of the continuous perennial forage often exceeded the target yields for our region, which are 5.6–6.7 Mg ha−1 on a DM basis based on three cuts per year (R. J. Kersbergen, personal communication, June 21, 2014). In crop year 1, average yields were 6.3 Mg ha−1 (sum of two cuts) and 8.8 Mg ha−1 (sum of three cuts) for sites 1 and 2, respectively, over all treatments in perennial forage (P = 0.44, P = 0.59) (data not shown).
At site 1, total yield of the continuous perennial forage control remained statistically constant over all three crop years (P = 0.339; 7.1, 7.5 and 8.5 Mg ha−1). In contrast, at site 2 total yield changed over time (P = 0.001; 9.2, 10.0, 4.6 Mg ha−1), decreasing substantially from years 2 to 3. Our ability to assess perennial forage over time may have been confounded by factors including weather, harvest intervals and manure applications during crop year 1 (Table 3) which probably influenced yields.
Perennial forage re-established with wheat effectively increased total forage yield by 0.8 Mg ha−1 at site 1 (2012) and 1.7 Mg ha−1 at site 2 (2013) relative to the continuous perennial forage control (Table 8). Re-establishing perennial forage with triticale after corn compared with wheat had variable performance, yielding considerably lower than all other treatments at site 1 (2012) but equal to the perennial forage re-established with wheat at site 2 (2013). The use of a full-season corn in this rotation prohibited the timely seeding of the perennial forage at both sites. A better comparison to the wheat rotations would have been a corn rotation that included an early-season corn variety in the second year before perennial forage establishment.
Perennial forage composition and total average DM yield, and ANOVA results for forage harvested during crop year 3 in organic dairy cropping systems trials in 2012 and 2013 in Maine.
1 C, corn; WW, winter wheat; PFc, 60% red clover, 20% perennial ryegrass and 20% timothy; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
2 Grass data from 2012, and weed data from 2012 and 2013 were transformed for the ANOVA; back-transformed values are presented.
3 PFa–PFa–PFa.
4 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc, PFa–SW–PFc and C–C–PFc.
5 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc and PFa–SW–PFc.
*, **, ***, Significant at P < 0.05, P < 0.01 and P < 0.001, respectively.
Newly established perennial forage had greater legume biomass on average than the perennial forage control. In 2012 and 2013, respectively, legumes comprised 82 and 54% of the total forage biomass for the first-year forage stands compared with 25 and 36% legume biomass in the perennial forage control, which was in its fourth year of production. The perennial forage control was seeded with a mix containing 20% less legume than the PFc mixture, and the legume was alfalfa, which is known to have reduced longevity in poorly-drained soils and at a pH below 6.5–7 (Griffin, Reference Griffin2004). However, it is likely we would have found similar legume biomass by crop year 3 had the control been seeded with the PFc mix because, although red clover may better tolerate a lower pH and wet soil conditions (Griffin, Reference Griffin2004), it has low persistence and production is generally diminished by years 2–3 (Kunelius et al., Reference Kunelius, Durr, McRae and Fillmore2006; Frankow-Lindberg et al., Reference Frankow-Lindberg, Halling, Hoglind and Forkman2009). Interspecific competition with grass, which is a primary threat to alfalfa during the first 3 years of production, could also explain reduced legume biomass of the perennial forage control (Beuselinck et al., Reference Beuselinck, Bouton, Lamp, Matches, McCaslin, Nelson, Rhodes, Sheaffer and Volenec1994). However, an increase of grass biomass at the decline of legumes was evident only in 2012, when the perennial forage control exceeded the grass yield of annual crop sequences by an average of 4.6 Mg ha−1. In the same year, weed biomass was almost two times higher in the perennial forage control than in the new perennial forage. Though, this trend was not consistent over years. Weed abundance in perennial forages tends to be highly variable and influenced by many factors including forage species evenness and forage stand productivity (Tracy et al., Reference Tracy, Renne, Gerrish and Sanderson2004; Sanderson et al., Reference Sanderson, Soder, Muller, Klement, Skinner and Goslee2005). Dominant species in this study included perennial broadleaf weeds such as dandelion (Taraxacum offcinale) and broadleaf plantain (Plantago major L.).
There also was evidence that establishing a new perennial forage crop 1 year after plowing in a perennial forage stand may favor grasses over legumes. In 2012, new perennial forage following spring wheat after perennial forage (PFa–SW–PFc) contained on average almost four times more grass and 22% less legume biomass than forage established after two years of annual crops (C–SW–PFc and S–SW–PFc). This may be due to higher availability of soil N following a perennial forage crop versus after two years of annual crops. Nitrogen fixation from legumes is limited by soil inorganic N levels, and as a result, legumes do not compete well with grasses in high soil N environments (Eriksen et al., Reference Eriksen, Vinther and Søegaard2004). However, this effect was observed in only one year.
Perennial forage quality and projected milk yield
Perennial forage chemical composition on a DM basis and projected milk yield per ha from MILK2013 are presented by year in Table 9. Milk yield in 2012 was based on only two perennial forage harvests because samples from the second of three harvest dates were damaged in storage and unsuitable for quality and MILK2013 analysis. At site 1, milk yields remained statistically constant (P = 0.127) over all three crop years (9.8, 12.3 and 9.0 Mg ha−1) (MILK2000). At site 2, the change in yield over crop years approached significance (P = 0.067), (13.0, 11.3 and 9.2 Mg ha−1).
Average perennial forage chemical composition on a DM basis, projected milk yields and ANOVA results for perennial forage harvested during crop year 3 in organic dairy cropping systems trials conducted in 2012 and 2013 in Maine.
1 CP, Adjusted crude protein; NDF, neutral detergent fiber; dNDF, digestible neutral detergent fiber; ADF, acid detergent fiber.
2 C, corn; WW, winter wheat; PFc, 60% red clover, 20% perennial ryegrass and 20% timothy; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
3 Adjusted crude protein data from 2012, and acid detergent fiber from 2013 were transformed for the ANOVA; back-transformed values are presented.
4 PFa–PFa–PFa.
5 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc, PFa–SW–PFc and C–C–PFc.
6 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc and PFa–SW–PFc.
*, **, ***, Significant at P < 0.05, P < 0.01 and P < 0.001, respectively.
The perennial forage quality and projected milk yields from treatments following wheat varied by year compared with the continuous perennial forage control. In 2012, milk yields from new perennial forage following wheat did not differ from the continuous perennial forage control, which is surprising considering the new forage following wheat had greater legume biomass compared with the control (Table 8). Previous studies show cows fed legume/grass mixtures have greater DM intake and milk production compared with grass silage alone (Bertilsson et al., Reference Bertilsson, Dewhurst, Tuori, Wilkins and Paul2002; Dewhurst et al., Reference Dewhurst, Fisher, Tweed and Wilkins2003). As well, the new perennial forage following wheat had some chemical traits that would indicate increased milk yield relative to the control, such as 12% lower NDF content, which is negatively correlated with intake (Shaver et al., Reference Shaver, Satter and Jorgensen1988). These milk yield results were likely due to a lack of total forage yield differences (Table 8) and digestible NDF, which was 16% higher in the perennial forage control. Though grasses have higher NDF values (Buxton, Reference Buxton1996), the digestibility of the NDF of grasses is higher than that of legumes (Buxton and Redfearn, Reference Buxton and Redfearn1997).
In 2013, perennial forage established after wheat yielded 2.6 Mg ha−1 more milk than the perennial forage control. In this instance, perennial forage quality was relatively similar among treatments, but perennial forage following wheat had on average 1.7 Mg ha−1 greater biomass yields than the control.
Re-establishing perennial forage earlier with wheat always produced greater milk yields than with perennial forage established after corn because the sequence with wheat had greater forage yield in 2012 and greater quality in 2013.
Profitability
Gross income, production costs and net 3-year revenue of off-farm sales for crop sequences are shown by site in Table 10. Spring wheat always achieved the 120 g kg−1 CP industry standard for bread flour and was evaluated as food-grade wheat, whereas winter wheat was evaluated as feed due to low CP. Spring wheat showed the potential for unacceptable DON levels and associated risks; however, DON results were disregarded for this analysis because the variety ‘Glenn’ has shown consistently low DON levels in regional trials (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2014).
Three-year gross income from off-farm sales, production costs, net revenue based on yield data and ANOVA results from organic dairy cropping systems trials conducted from 2010 to 2013 in Maine.
1 C, corn; WW, winter wheat; PFc, 60% red clover, 20% perennial ryegrass and 20% timothy; PFa, 40% alfalfa, 40% perennial ryegrass and 20% timothy; SW, spring wheat; S, soybean.
2 PFa–PFa–PFa.
3 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc, PFa–SW–PFc and C–C–PFc.
4 C–WW–PFc, PFa–WW–PFc, C–SW–PFc, S–SW–PFc and PFa–SW–PFc.
*, **, ***, Significant at P < 0.05, P < 0.01 and P < 0.001, respectively.
Net 3-year revenue of sequences with annual crops was never greater than the continuous perennial forage control. This was due to low spring wheat yields and competitive returns of the perennial forage control. The perennial forage control had both relatively high yields and low production costs, which when averaged across sites were US$ 291 ha−1 less than treatments with annual crops.
Despite being excluded from the higher bread flour pay price, winter wheat treatments had an average of US$ 741 ha−1 greater net returns than spring wheat treatments due to substantially higher yields. Differences in revenue by wheat type highlight the potential economic challenges and opportunities associated with planting bread wheat in the organic dairy farm setting. Our results indicate planting a spring grain on poorly-suited soils may be an economic risk, while winter wheat may be economically viable as feed, or as bread flour if additional management strategies such as variety selection and/or topdressing are used to increase protein levels. For instance, varieties with high protein potential managed with organic practices in the northeastern USA have had CP values that ranged from 100 to 130 g kg−1 (Mallory et al., Reference Mallory, Darby, Molloy and Cummings2015). Regarding topdressing, Mallory and Darby (Reference Mallory and Darby2013) found that topdress applications of 22 kg N ha−1 Chilean nitrate (CN) or dehydrated poultry litter (DPL) at the boot stage increased CP from 92 to 105 g kg−1 with CN and to 100 g kg−1 with DPL, which are levels accepted on the local artisan bread flour market. If winter wheat had been topdressed in this trial with CN costing US$ 256 ha−1 (includes bulk fertilizer sidedressing US$ 27 ha−1, 2014 Pennsylvania Custom Rates), projected net revenues would increase by US$ 185 ha−1 on average. Similarly, with DPL costing US$ 486 ha−1, net revenues would increase by US$ 108 ha−1 on average. Despite projected revenue increases with the topdress addition, there were no new measurable treatment differences (data not shown). Dairy producers must identify within the context of their farm if the cost–benefit of a CP-boosting technique is economical.
Rotz et al. (Reference Rotz, Roth and Stout2002) suggested that the straw generated from wheat must be used either on-farm as bedding or sold to make wheat a profitable enterprise on dairy farms. Straw yields were not formally measured in this trial, though based on an estimated 111 bale ha−1 regional average (R. J. Kersbergen, personal communication, 2011) priced at US$ 5 per bale (bulk quantity costs; J. Dyer, personal communication, September 19, 2014) and a US$ 34 ha−1 cost for baling hay (Lazarus, Reference Lazarus2014), off-farm revenue associated with wheat production could increase by US$ 521 ha−1. With straw included, annual crop sequences with wheat would return US$ 429 and 328 ha−1 more than the perennial forage control at sites 1 and 2, respectively (P < 0.01 and P < 0.001) and US$ 546 and 512 ha−1 more than the continuous corn sequence at each site (P < 0.01 and P < 0.001).
Lastly, the prices used in these profitability estimates were from one year (Table 4) and results could vary for other years. For instance the 7-year (2008–2014) average organic food grade wheat price was US$ 487 Mg ha−1 and prices fluctuated by 72% from the low of US$ 290 Mg ha−1 in 2010 to the high of US$ 694 Mg ha−1 in 2014 (USDA-AMS, 2016). As well, the 7-year average organic milk price was US$ 706 Mg ha−1 and prices fluctuated by 16% from the low of US$ 666 Mg ha−1 in 2009 to the high of US$ 774 Mg ha−1 in 2014 (Parsons and McCrory, Reference Parsons and McCrory2011; Parsons, Reference Parsons2012, Reference Parsons2016).
Conclusion
The opportunity for organic dairy producers to use bread wheat as a cash crop and cost-effective method to re-establish high-quality perennial forage prompted our efforts to evaluate bread wheat under field conditions and in conjunction with other crops. Rotating out of perennial forage and into wheat was a consistently viable strategy for improving perennial forage yields and increasing legume biomass compared with the standard practices. In terms of profitability, the crop sequences that included wheat showed no financial advantage over the perennial forage control as evaluated. However, rotations that included winter wheat would likely be more competitive if topdressing practices or varietal choice were used to increase grain protein levels, or if straw were sold or used on the farm. Rotations that included spring wheat were limited in the years tested by delayed wheat planting due to wet conditions and soil types. Given the prevalence of heavier soils on many northeast dairy farms (Hoshide et al., Reference Hoshide, Halloran, Kersbergen, Griffin, DeFauw, LaGasse and Jain2011), proper site selection would be critical for profitable spring grain production.
While our profitability results are important in terms of understanding potential off-farm revenues, our analysis was limited in that it did not account for the quality of crops other than bread wheat. To capture the on-farm value of crops and to better understand the value of bread wheat, future analyses should both translate feed and perennial forage crops into milk yield and include more than a few years of crop yields. A profitability analysis that more accurately depicts the end use of crops on dairy farms will further guide dairy producers on the potential values and risks of entering the bread flour market.
Acknowledgements
The authors would like to thank Hannah Griffin and the MAFES Analytical Laboratory for their technical assistance. This work was supported by the USDA National Institute of Food and Agriculture Organic Agriculture Research and Extension Initiative under Agreement no. 2009-51300-05594, “Enhancing Farmers' Capacity to Produce High Quality Organic Bread Wheat”, and by Hatch Grant no. ME08001-10 from the USDA National Institute of Food and Agriculture.
