Ammonia Emissions, Exposed Surface Area, and Crop and Weed Responses Resulting from Three Post-Emergence Slurry Application Strategies in Cereals

Abstract

Ammonia (NH₃) emissions resulting from the field application of livestock slurry has both negative human health and environmental impacts. However, decreasing the exposed surface area (ESA) of slurry upon application can reduce NH₃ volatilization by limiting its atmospheric exposure. In the present study, three strategies for depositing slurry within a growing crop were studied, including: 1. standard trailing hoses (SA_hose), 2. trailing shoes (SA_shoes), and 3. the combination of rigid tines and trailing shoes (SA_tines+shoes). Application methods interact with the soil to varying degrees and were evaluated within the context of contemporary weed management practices, namely in cereals receiving inter-row hoeing. SA_hose, SA_shoes, and SA_tines+shoes were compared in three coinciding experiments that assessed slurry ESA, NH₃ emissions, and crop and weed effects. SA_tines+shoes resulted in smallest ESA, 70–72% and 61–66% less than SA_hose and SA_shoes, respectively. However, in only one of three site–years did SA_shoes and SA_tines+shoes reduce NH₃ emissions compared to SA_hose, by 46% and 29%, respectively. Crop yields, nitrogen (N) accumulation in crop biomass, and intra-row weed biomass were unaffected by the placement method. In heavily crusted soils, the SA_tines+shoes prototype worked well; however, the functional differences among placement strategies were not great enough to detect crop and NH₃ effects.

1. Introduction

Ammonia (NH₃) emissions resulting from the land application of animal manure constitutes a significant agricultural pollutant. The direct and indirect effects of NH₃ pollution are detrimental to both human and environmental health. Once airborne, NH₃ contributes to the formation of fine particulate matter in the atmosphere [1] and the development of disease in humans upon inhalation [2]. Wet and dry deposition of atmospheric nitrogen (N) also adds to the acidification and eutrophication of terrestrial and aquatic ecosystems [3,4]. Within the EU-28, approximately 93% of NH₃ emissions can be traced to agricultural activities [5], with 26% of total emissions stemming from animal manure application to soils in 2018 [6]. The EU has set substantial goals for reducing NH₃ emissions by 19% within the 25-year period from 2005 to 2030 [7]. Achieving these objectives requires a multifaceted approach focused on improved animal rearing techniques, on-farm manure handling, storage, and processing, as well as the development of field application strategies that reduce NH₃ volatilization.

Methods known to effectively mitigate ammonia emissions during land application of liquid manure (slurry) include strategies that reduce atmospheric exposure by increasing infiltration rate and reducing exposed surface area (ESA) [8,9]. As slurry infiltrates into the soil, the quantity of aboveground N susceptible to loss decreases, thus reducing emission potential. NH₃ volatilization occurs at the interface between exposed slurry and the atmosphere; therefore, when aboveground ESA is reduced or infiltration rate increases, emissions decrease.

Existing methods for low-emission slurry application include surface application with trailing hoses, shallow application with trailing shoes, and sub-surface application with open and closed slot injection [10]. Open slot injection places slurry in a channel that is not covered by soil, while closed slot injection places slurry below the soil’s surface, so it is not exposed to the atmosphere. Closed slot injection is considered the most effective strategy for reducing NH₃ emissions [11,12]. When implemented before planting a cereal crop, injection improves yields by reducing atmospheric N loss and by placing fertilizer near the crop’s rooting zone [13]. An associated increase in crop vigor may also aid in the suppression of weeds, further contributing to increased yields [14,15]. However, injection also has its drawbacks. Weed response may differ among species with contrasting root morphologies; Melander et al. [16] report that upon fertilizer injection, the biomass of shallow-rooted species are reduced, while deeply rooted species benefit. Closed slot injection is also highly disruptive, rendering its implementation impractical within a growing crop, where significant damages are not offset by improved N conservation and crop acquisition [17]. Therefore, closed slot injection is best suited for bare field settings preceding sowing, and open slot injection is most applicable in grasslands. Ample horsepower is also required to accommodate high draught forces associated with pulling injectors; this limits implement width and further contributes to crop damages and soil compaction resulting from additional tractor passes [18]. Finally, a tradeoff in volatile organic compound emissions must be considered. Compared to surface application, closed slot injection reduces NH₃ emissions by approximately 61% but increases emissions of the potent greenhouse gas, nitrous oxide (N₂O), by 196%; this “pollution swapping effect” is described in greater detail by Emmerling et al. [19].

Methods less disruptive than injection also exist, suitable for post-emergence application within a growing cereal crop. Camera and GPS guidance technologies have made the accurate placement of liquid manure between narrowly spaced rows possible, using tools that either deposit at the surface or interact shallowly with the soil. These strategies also demand less power during operation and can therefore be used at increased working rates, covering a wider surface area with each tractor pass [18].

In Northern Europe, trailing hoses represent the standard method for placing liquid manure in a growing cereal crop. Hoses drag across the soil surface, depositing slurry in bands between crop rows, reducing ESA and NH₃ volatilization compared to broadcast, full-width spreading [20]. Placing slurry beneath the canopy of a growing crop has also been shown to reduce NH₃ emissions [9,21,22] and is, therefore, a method well suited to springtime application in established winter cereals. However, trailing hoses are prone to side-to-side shifting during operation and may bounce on top of the soil, causing some splashing; thus, hoses are not ideal when targeting fertilizer placement in close proximity to crop rows. In addition, when compared to trailing shoes, hoses may produce higher ESA and NH₃ emissions [23].

A variety of trailing shoe designs are now available on the market. Typically, these shoes function by causing some shallow soil disturbance, creating a lightly cultivated zone or depression ahead of where the slurry is placed. Cultivation immediately preceding slurry deposition reduces NH₃ loss by increasing infiltration rate [24]. In addition, some shoes are designed to funnel slurry into a narrow channel, reducing ESA and NH₃ emissions [25,26]. Trailing shoes are more rigid in their construction compared to hoses, preferable for targeting placement near crop rows. However, if substantial soil crust is present, trailing shoes may not function as intended, riding on top of the soil without causing any disturbance; the present study investigates whether combining the operation of rigid tines and trailing shoes will help resolve this issue.

When operating any slurry placement tool within a growing crop, the risk of inflicting damages that may negatively affect yields exist simultaneously alongside the potential positive effects, including N conservation, improved infiltration, and improved crop uptake. Therefore, when comparing slurry application strategies, it is necessary not only to evaluate resulting ESA and NH₃ emissions; one must also assess crop and weed responses.

In the present study, three slurry application methods are compared. First, standard surface application with trailing hoses. Second, application with trailing shoes designed to create a shallow furrow ahead of where the slurry is placed. Third, a novel prototype applicator composed of rigid tines, followed by trailing shoes. Tines were added to break through any hardened crust that has formed on the soil’s surface; field conditions typical for loam- and clay-type soils in spring, when slurry is spread in over-wintered cereals. Application methods are evaluated within the inter-row hoed cereal system, an effective weed management strategy gaining popularity among organic growers in Northern Europe [16,27]. Hoeing is a robust and highly effective method for controlling inter-row weeds [28]; however, hoeing has little effect on weeds growing in the intra-row zone, capable of negatively affecting crop yields [29].

This study aims to evaluate how the three slurry application methods described affect crop and intra-row weed response, NH₃ emissions, and slurry ESA. It is hypothesized that similar crop yields, biomasses, and grain quality responses will be observed across slurry application treatments. It is anticipated that damages resulting from more disruptive placement strategies (trailing shoes and tines plus trailing shoes) will be offset by increased N uptake due to improved infiltration and targeted placement. While total intra-row weed biomass may not differ among application strategies, individual weed species are expected to respond differently depending on key morphological traits. Taprooted weeds are hypothesized to experience improved growth and inferior control with targeted placement tactics (trailing shoes and tines plus trailing shoes), while weeds with shallow fibrous roots will experience reduced growth. Finally, greater slurry NH₃ emissions and ESA are expected to result from the standard trailing hoses, and these measures will be reduced with the trailing shoes and further reduced with the tines plus trailing shoes. The present study contributes a comprehensive comparison of three low-draft slurry placement methods well suited for post-emergence use in cereals; one prototype and two pre-existing designs are evaluated. Coinciding experiments allow results to be contextualized within the interdependent dynamics of slurry ESA, NH₃ emissions, plant N uptake, and crop and weed effects.

2. Materials and Methods

Separate experiments were required to evaluate the effect of liquid manure application strategy on crop and intra-row weed response (EXP_CROP+WEED), NH₃ emissions (EXP_NH3), and slurry ESA (EXP_ESA). In 2019, one replication of EXP_CROP+WEED (EXP_CROP+WEED.A) and EXP_NH3 (EXP_NH3.A) was performed simultaneously at the Foulum Research Center, Denmark (56.49° N, 9.58° E) in a field possessing a loamy-sand-soil texture and using spring barley (Hordeum vulgare L., cv. RGT Planet) as the test crop (SITE_A). From 2019 to 2020, experiments were performed at the Flakkebjerg Research Center, Denmark, using winter wheat (Triticum aestivum L., cv. Cheriff) as the test crop. In fields possessing a sandy-loam-soil texture, one replication of EXP_CROP+WEED (EXP_CROP+WEED.B), two replications of EXP_NH3 (EXP_NH3.B1 and EXP_NH3.B2), and one EXP_ESA trial (EXP_ESA.B) were performed (SITE_B; 55.32° N, 11.39° E). A second repetition of EXP_CROP+WEED (EXP_CROP+WEED.C) was implemented at a nearby site (SITE_C; 55.33° N, 11.41° E). Spring barley and winter wheat were sown uniformly across experiments and sites. At SITE_A, spring barley was sown at 24 cm row spacing and a target density of 300 plants m⁻². At SITE_B and SITE_C, winter wheat was sown at a 25 cm row spacing and a target density of 300 plants m⁻².

Table 1. Overview of experiments performed in 2019 and 2020 across sites.

Year	Site	Soil Type	Soil Texture				Test Crop	Experiments Performed
			Sand (%)	Silt (%)	Clay (%)	Humus (%)
2019	Tjele, DK (SITEA)	Loamy sand	89	5	4	2	Spring barley	Crop and weed response (EXPCROP+WEED.A)
								Ammonia emissions (EXPNH3.A)
2019−2020	Flakkebjerg, DK (SITEB)	Sandy loam	61	18	19	2	Winter wheat	Crop and weed response (EXPCROP+WEED.B)
								Ammonia emissions (EXPNH3.B1 and EXPNH3.B2)
								Exposed Surface Area (EXPESA.B)
	Flakkebjerg, DK (SITEC)	Sandy loam	73	11	14	2	Winter wheat	Crop and weed response (EXPCROP+WEED.C)

provides an overview of experiments performed across site–years.

2.1. Crop and Weed Response (EXP_CROP+WEED)

2.1.1. Experimental Design and Treatment Specifications

For all EXP_CROP+WEED trials, three factors were evaluated within a full factorial, randomized complete block design with four replications. Experimental factors included pig slurry application method, N rate, and weed management treatment. In total, three post-emergence slurry application methods were studied, including (i) standard trailing hoses (SA_hose), (ii) trailing shoes (SA_shoes), and (iii) a prototype model comprising a rigid tine, designed to break through the soil’s crust, followed by a trailing shoe (SA_tines+shoes). Two N rates, calculated as kg NH₄-N ha⁻¹ of pig slurry, were evaluated, (i) 100 kg ha⁻¹ (N₁₀₀) and (ii) 50 kg ha⁻¹ (N₅₀). Finally, two weed management treatments were evaluated, (i) weedy plots, receiving no herbicide treatment (WMT_weedy), and (ii) weed-free plots, receiving herbicide treatment (WMT_weed-free). All plots received inter-row hoeing with 16-cm shares. Plot dimensions were 3 by 13 m at SITE_A and 2.5 by 12.5 m at SITE_B and SITE_C. A total of 96 plots were included in each experiment; 3 slurry application methods * 2 N rates * 2 weed management treatments * 4 blocks. Table S1 provides a record of the implementation date for each EXP_CROP+WEED field operation, as well as crop, surrogate weed, and ambient weed growth stages.

Two surrogate weeds with differing root morphologies were sown together in a strip, horizontally through each plot’s center. Surrogate weeds included white mustard (Sinapis alba L., cv. Lotus), possessing a deep taproot, and Italian ryegrass (Lolium multiflorum Lam., cv. Fabio), exhibiting shallow fibrous root growth. The target population of both surrogate weed species S. alba and L. multiflorum was 30 plants m⁻²; therefore, the combined target density of both surrogate weed species was 60 plants m⁻². At SITE_A, surrogate weeds were sown in strips 3 m wide within four days of planting the crop and established successfully (Table S1). At SITE_B and SITE_C, surrogate weed sowing was attempted in late March 2020. The planter was raised during operation to avoid damaging the growing winter wheat crop, dropping seeds onto the soil’s surface; however, both S. alba and L. multiflorum failed to establish due to a lack of precipitation following planting. In place of surrogate weeds, two prevalent and naturally occurring ambient weed species were studied instead. Ambient weeds studied at SITE_B and SITE_C included mayweed (Matricaria L.), possessing a deep taproot, and annual bluegrass (Poa annua L.), exhibiting shallow fibrous root growth.

Inter-row hoeing was performed twice, twenty-one days apart at SITE_A (Table S1). At SITE_B and SITE_C, hoeing was performed twice on the same date, and in opposite directions, due to difficult cultivation conditions caused by hardened crusted soils. At SITE_A in WMT_weed-free plots, herbicide was applied twice, on 14 May and 15 June 2019, due to poor control following the first application (tetribenuron-methyl, Nuance WG, 3 g active ingredient (a.i.) ha⁻¹, Cheminova; tribenuron, Nuance WG, 2.89 g a.i. ha⁻¹, Cheminova; florasulam, Starane XL, 0.63 g a.i. ha⁻¹, Dow AgroSciences; fluroxypyr, Starane XL, 25 g a.i. ha⁻¹, Dow AgroSciences; diflufenican, DFF, 15 g a.i. ha⁻¹, Bayer; spreading adhesive, Contact, 200 g a.i. ha⁻¹, UPL Euro). At SITE_B and SITE_C, one herbicide application was made on 5 May 2020 (metsulfuron-methyl, Nicanor SG, 4 g a.i. ha⁻¹, Cheminova; metsulfuron, Nicanor SG, 3.86 g a.i. ha⁻¹, Cheminova; spreading adhesive, Agropol, 150 g a.i. ha⁻¹, UPL Europe).

Slurry was applied within the growing cereal crop in the inter-row approximately 8 cm from crop row centers. Different trailing shoe models were used when implementing the SA_shoes and SA_tines+shoes treatments at SITE_A versus SITE_B and SITE_C. In addition, differing prototypes combining the operation of a rigid tine and trailing shoe were used in SA_tines+shoes plots at SITE_A versus SITE_B and SITE_C. Bomech trailing shoes (Bomech B.V., Albergen, Netherlands) were employed at SITE_A. The arm connecting the Bomech shoe to the toolbar is fabricated from steel, as is the V-shaped hull that runs along the bottom of the shoe responsible for creating a shallow soil opening where the slurry is deposited. The Bomech shoe can respond to greater downward pressure owing to its rigidity; therefore, when operating in lighter soils at SITE_A (Table 1), the tool functioned as expected, creating a furrow in the soil while placing slurry in SA_shoes plots. The prototype used in SA_tines+shoes plots at SITE_A was designed, so the tine and the Bomech shoe operated together, mounted as a single unit on the same toolbar (Figure S1). It was observed that the tine flexed and bounced abruptly during operation while breaking through deeper soil layers; thus, the Bomech shoe moved with the tine, resulting in some splashing of slurry. Vogelsang trailing shoes (Vogelsang GmbH & Co. KG, Essen/Oldb., Germany) were used at SITE_B and SITE_C (Video S1). The arm connecting the Vogelsang shoe to the toolbar is composed of rigid plastic and, therefore, more flexible than the Bomech model. When operating at SITE_B and SITE_C in heavy soils with significant surficial crusting, the Vogelsang shoes could not penetrate the soil crust on their own. In SA_shoes plots, the Vogelsang shoes rode evenly on top of the soil, depositing slurry without creating a depression. The prototype used in SA_tines+shoes plots at SITE_A was altered for use at SITE_B and SITE_C. The rigid tine and the Vogelsang trailing shoe were mounted on separate toolbars, thus operating independently (Video S2). The tine was able to break through the hardened soil crust, and the Vogelsang shoe rode smoothly behind in the narrow opening created by the tine while depositing slurry. For reference, Video S3 shows slurry application with trailing hoses (SA_hose); this method was used at SITE_A, SITE_B, and SITE_C.

Slurry analyses were performed using standard methods for dry matter content [30], total nitrogen [31], and total ammoniacal N (TAN) [32]. Viscosity measurements were performed using a rotational viscometer (DV-II + P Viscometer, Brookfield, AMTAK, Middleboro, MA, USA) with an LV-2 spindle at 10 RPM and slurry temperature of 19 ± 0.8 °C. All analysis results and application rates for EXP_NH3 and EXP_ESA can be found in

Table 2. Slurry properties and application rate used in EXP_NH3 and EXP_ESA. Standard errors are displayed in parentheses.

Experiment	N Rate Treatment	Application Rate	Application Rate	Dry Matter	Total N	Ammoniacal N	Viscosity	pH
		kg m−2	g NH4-N m−2	%	g L−1	g L−1	cP
EXPNH3.A	N100	2.3	3.54 (±0.11)	4.93 (±0.26)	2.50 (±0.08)	1.55 (±0.05)	−	7.77 (±0.03)
EXPNH3.B1	N100	3.0	10.57 (±0.54)	3.27 (±<0.01)	4.16 (±0.05)	3.52 (±0.18)	106 (±8)	7.82 (±0.04)
EXPNH3.B2 and EXPESA.B	N100	3.0	11.28 (±0.32)	6.46 (±0.03)	4.85 (±0.04)	3.76 (±0.11)	1004 (±45)	7.51 (±0.02)

. Achieved slurry application rates in EXP_CROP+WEED.A, N_50, and N₁₀₀ treatments were 2.3 kg m⁻² (3.54 g NH₄-N m⁻²) and 4.6 kg m⁻² (7.08 g NH₄-N m⁻²), respectively; and in EXP_CROP+WEED.B and EXP_CROP+WEED.C, N₅₀ and N₁₀₀ treatments were 1.5 kg m⁻² (5.29 g NH₄-N m⁻²) and 3.0 kg m⁻² (10.57 g NH₄-N m⁻²), respectively. Following application, the slurry surface pH was measured with a flat-tip pH electrode (Orion^TM 8135BN ROSS^TM, Combination Flat Surface pH Electrode, Fischer Scientific, Loughborough, UK). Measurements were taken as close to the slurry–air interface as possible. Results can be seen in Figure S2 in Supplementary Materials.

2.1.2. Data Collection

To avoid edge effects, samples were not collected from the outermost rows or from the first 50 cm at the top and bottom of each plot. Achieved in-field crop, surrogate weed, and ambient weed densities (plants m⁻²) were measured before herbicide application, inter-row hoeing, or slurry application. Crop density (plants m⁻²) was calculated from four counts of 1-m length row per plot. Intra-row surrogate and ambient weed counts were made within 8 by 100 cm quadrats (0.08 m²), centered on random sections of crop row. Intra-row ambient weed density (plants m⁻²) was recorded individually for each species observed within four quadrats per plot. At SITE_A, counts of intra-row surrogate weed density (plants m⁻²) were made within two quadrats per plot, positioned in the surrogate weed strip.

After barley surpassed principal growth stage six (flowering and anthesis), intra-row crop, surrogate weed, and ambient weed biomass samples were cut from six 8 by 100 cm quadrats per plot. Quadrats were centered on crop rows, encompassing only the intra-row zone. Having been treated with herbicide, surrogate weed strips were absent in WMT_weed-free plots; thus, the distribution of the six biomass samples taken within each plot was slightly different for WMT_weed-free and WMT_weedy. In WMT_weed-free plots, all six biomass cuts were made randomly throughout the plot and combined into a single sample. In WMT_weedy plots, four biomass cuts were made outside the surrogate weed strip, and two cuts were made inside the surrogate weed strip; biomass samples taken from these zone were evaluated separately. All samples were sorted into four categories. At SITE_A, plant biomass was divided into (i) spring barley, (ii) S. alba, (iii) L. multiflorum, and (iv) ambient weeds. At SITE_B and SITE_C, biomass was divided into (i) winter wheat, (ii) Matricaria, (iii) P. annua, and (iv) other ambient weeds. All biomass samples were dried for at least 24 h (h) at a temperature of 80 °C and then weighed.

Following the collection of plant biomass dry weights, biomass samples from WMT_weedy plots were pulverized. Sub-samples of ground plant tissue were analyzed to determine percent N using the Dumas combustion procedure [33,34]. Information on N concentration was employed to calculate N accumulation (g N m⁻²) among the four previously described plant categories.

Before harvesting the crop, the surrogate weed strip in the plot’s center was cut away. Final plot length and the number of harvested rows were recorded, and crop yields (kg ha⁻¹) were calculated based on the final plot area. Harvested grain was cleaned, and final yield weights were standardized to 15% moisture content.

Grain quality measures were obtained after harvest. Moisture content (%), protein content (%), and bulk density (kg hL⁻¹) were measured using a near-infrared spectroscopy analyzer (Infratec^TM 1241 Grain Analyzer, Foss A/S; Buchmann et al., 2001). Thousand kernel weight (g) was acquired by weighing four samples of 200 grain kernels per plot.

2.1.3. Data Analysis

JMP^® software, version 14.3.0 (SAS Institute Inc., Cary, NC, USA) was used to analyze EXP_CROP+WEED data. Each site (SITE_A, SITE_B, and SITE_C) was evaluated separately. All crop effects were analyzed using a three-way analysis of variance (ANOVA); subsequent means comparisons were performed using Tukey’s HSD tests. In the three-way ANOVA, fixed variables consisted of main effects, weed management treatment (WMT), nitrogen rate (N), slurry application method (SA) and interaction effects, WMT*N, WMT*SA, N*SA, and WMT*N*SA; block served as a random variable.

Data from WMT_weedy plots were analyzed to evaluate intra-row weed effects using a two-way ANCOVA. Subsequent means comparisons were performed using Tukey’s HSD tests. In the two-way ANCOVA, N, SA, and N*SA served as fixed variables, and block served as a random variable. The corresponding measure of intra-row weed density (plants m⁻²) served as the continuous explanatory variable when analyzing each class of weed biomass data. For example, when analyzing P. annua biomass (g m⁻²) data, P. annua density (plants m⁻²) serves as the continuous explanatory variable. It is important to note that intra-row weed density measures were recorded before implementing inter-row hoeing, herbicide application, or slurry application; therefore, variables are not confounded and meet the assumption of independence.

Anderson–Darling tests were applied to residuals to ensure a normal distribution, and Levene’s tests were performed to ensure equality of variance (α = 0.05). Where required, log₁₀, log₁₀ (x + 1), square root, and square root (x + 1) transformations were performed to meet model assumptions.

2.2. Ammonia Emissions (EXP_NH3)

2.2.1. Experimental Design, Treatment Specifications, and Data Collection

In plots adjacent to the EXP_CROP+WEED trials, relative NH₃ emissions were compared among the three slurry application strategies in a randomized complete block design with three replications. The experiment was repeated a total of three times, once at SITE_A (EXP_NH3.A) and twice, in separate locations, at SITE_B (EXP_NH3.B1 and EXP_NH3.B2). EXP_NH3.A was initiated on 15 May 2019, the same day that slurry was applied in EXP_CROP+WEED.A. Experiments at SITE_B were started on 17 April 2020, the day that slurry was applied in EXP_CROP+WEED.B (EXP_NH3.B1), and one week later on 24 April 2020 (EXP_NH3.B2). A target application rate of 100 kg NH₄-N ha⁻¹ was used; however, a computational error resulted in a low application rate of 35.4 kg NH₄-N ha⁻¹ at EXP_NH3.A (Table S2).

Ammonia emissions were measured using a system of nine dynamic chambers (wind tunnels) and continuous online measurements with a cavity ring down spectroscopy (CRDS) instrument (G2103 NH₃ Concentration Analyzer, Picarro, CA, USA). See Pedersen et al. [23] for a detailed description of the wind tunnel system and instrumentation used to measure NH₃ emissions in the present study. To assess SA_hose, SA_shoes, and SA_tines+shoes treatments, equipment described in Section 2.1.1 was operated within plots without applying slurry, creating the soil disturbance associated with each application strategy; liquid slurry was then applied manually using a watering can, and measurements commenced immediately thereafter. Within each tunnel, NH₃ concentration in the air was logged every 1 to 2 s. In addition, background concentrations of NH₃ were measured at the air intake of each tunnel. The total retention time for each tunnel was 104 min and the total measuring time for each tunnel was 113 h. Before beginning experiments, recovery throughout the system was tested with a standard NH₃ gas and was found to be 96% at minimum.

2.2.2. Data Analysis

An average of the last 30 s of measurements for each measuring cycle was used for emission calculations. An average of the three backgrounds was subtracted for each measurement cycle. Concentrations lower than the detection limit (three times the standard deviation of the background measurements) were set to zero. The NH₃ flux (F_NH3 [g min⁻¹ m⁻²]) was calculated from the NH₃ concentration (C), the volumetric airflow rate (q) and the emitting area (A) (Equation (1)).

$$F_{N{H}_3}=\left(C\times q\right)/A$$

(1)

The cumulative emissions were calculated using the trapezoid rule [35].

Differences among treatments were analyzed for each experiment individually using a one-way ANOVA with treatment as the independent variable. In each experiment, the nine wind tunnels were divided into three blocks, each containing one tunnel with each treatment, with a randomized block design. A single wind tunnel was used as an observational unit, and the cumulative NH₃ was used as the response variable. Tukey’s HSD test (confidence interval of 95%) was used to investigate differences among treatments. Homogeneity of soil and crops within the experimental area was assumed.

2.3. Exposed Surface Area (EXP_ESA)

2.3.1. Experimental Design, Treatment Specifications, and Data Collection

Positioned next to EXP_CROP+WEED.B and performed in conjunction with the EXP_NH3.B2 experiment, the achieved slurry ESA among SA_hose, SA_shoes, and SA_tines+shoes treatments were studied. The experiment, EXP_ESA.B, possessed randomized complete block designs with three replications. Exposed surface area of the slurry following application was measured by employing methods developed and presented by Pedersen et al. [36]; fluorescent dye was used to detect and quantify slurry on top of the soil, and images collected within dark chambers were analyzed to determine ESA (m² m⁻²). Exposed surface area measurements were taken every 10 min for a period of 60 min, followed by every 20 min for a period of 80 min, resulting in a total of 10 images per dark chamber. To mimic SA_hose, SA_shoes, and SA_tines+shoes treatments, the same methods described in Section 2.2.1 were used; soil disturbance was replicated by operating equipment within treatment plots, and then slurry was applied by hand using a watering can.

2.3.2. Data Analysis

The time decay of slurry ESA was analyzed using a generalized linear mixed model defined with the Gamma distribution and the logarithmic link function. Responses were observed as change in slurry ESA per square meter (m² m⁻²) over time for each dark chamber; the model contained two fixed effects: one factor representing the dark chamber and a continuous explanatory variable (regression) representing the logarithm of the time shifted.

According to the model, ESA at time t (with t = 0, 0.015, 0.334, 0.484, 0.667, 0.834, 0.984, 1.35, 1.65, 1.984, 2.317, and 2.6 h) in the b^th block, is represented by the random variable A_bt, and is conditional on the residual random component U_bt, which is Gamma distributed with conditional expectation given by:

$$log\left\{E\left(A_{bt}|U_{bt}=u\right)\right\}=\alpha +\beta \left(t+\gamma \right)+u \mathrm{for} \mathrm{all} u=ℝ,$$

(2)

which is equivalent to:

$$E\left(A_{bt}|U_{bt}=u\right)=\exp \left(\alpha \right){\left(t+\gamma \right)}^{\beta }\exp \left(u\right)=A\left(t\right)\exp \left(u\right).$$

(3)

Model specification is completed by stating that the random components U₁₀,…,U_2.66 are identically distributed with $U_{10}~N\left(0, {\sigma }_r^2\right)$, and for b = 1, 2, and 3,

$$Corr\left(U_{bt}, U_{bs}\right)=\exp \left(-\tau \left|t-s\right|\right), \mathrm{with} \tau >0$$

(4)

Moreover, the residual random component corresponding to each dark chamber is independent per construction. In this way, we account for dependencies between the observations made at the same dark chamber at different observation times.

The time shift parameter γ was estimated by maximizing the likelihood profile, obtained with marginal models, and defined with fixed values of γ. Additionally, initial ESA (obtained at time t = 0) and final ESA (obtained at the last observation time, t = 2.66) were estimated as $\exp \left(\alpha \right){\gamma }^{\beta }$ and $\exp \left(\alpha \right){\left(\gamma +2.66\right)}^{\beta }$, respectively. We used parametric bootstrap techniques, with 1000 bootstrap replicates, to construct confidence intervals and treatment comparison tests, for the parameter and the quantities calculated with the parameters referred above.

3. Results

3.1. Crop and Weed Responses

Achieved crop densities at SITE_A, SITE_B, and SITE_C were 240, 242, and 363 plants m⁻², respectively. At SITE_A, surrogate weeds S. alba and L. multiflorum were established at densities of 20 and 28 plants m⁻², respectively. At SITE_B and SITE_C, surrogate weeds failed to establish due to lack of moisture; therefore, ambient weeds, P. annua and Matricaria were studied, each occurring naturally at an average density of 8 and 11 plants m⁻² at SITE_A, respectively, and 64 and 17 plants m⁻² at SITE_B. Ambient weed densities at SITE_A SITE_B and SITE_C were 207, 103, and 33 plants m⁻², respectively. Ambient weed populations at each site, characterized by the three most predominant species were, at SITE_A, 46% Common lambsquarters (Chenopodium album), 23% European field pansy (Viola arvensis), and 12% P. annua, at SITE_B, 33% Matricaria, 23% P. annua, and 19% Stickywilly (Galium aparine), and at SITE_C, 63% P. annua, 17% Matricaria, and 8% Quackgrass (Elymus remens).

Winter wheat yields (kg ha⁻¹) were on average 11% and 7% greater in WMT_weed-free plots compared to WMT_weedy, and 35% and 54% greater in N₁₀₀ plots compared to N₅₀ at SITE_B and SITE_C, respectively (

Table 3. The effect of weed management treatment (WMT_weedy and WMT_weed-free), nitrogen rate (N₁₀₀ and N₅₀), and slurry application method (SA_hose, SA_shoes, and SA_tines+shoes) on spring barley grain yield (SITE_A) and winter wheat grain yield (SITE_B and SITE_C), presented at 15% moisture content. Data from each site were analyzed separately. A connecting letters report displays differences among treatments, whereby means not sharing a similar letter are significantly different (p < 0.05). Significant main effects and interactions are highlighted in bold text (p < 0.05). Standard errors are displayed in parentheses.

			Crop Yield
Treatment			SITEA			SITEB			SITEC
			kg ha−1
WMTweedy	N50	SAhose	5171	(±96)	a	4272	(±385)	d	4255	(±308)	d
WMTweedy	N50	SAshoes	5389	(±304)	a	4090	(±510)	d	4484	(±230)	d
WMTweedy	N50	SAtines+shoes	4799	(±265)	a	4505	(±176)	d	3812	(±296)	d
WMTweedy	N100	SAhose	5661	(±152)	a	6107	(±413)	abc	6913	(±432)	ab
WMTweedy	N100	SAshoes	5248	(±492)	a	6425	(±462)	ab	6542	(±503)	ab
WMTweedy	N100	SAtines+shoes	5584	(±174)	a	5971	(±426)	bc	5945	(±347)	bc
WMTweed-free	N50	SAhose	5439	(±333)	a	4962	(±188)	cd	4532	(±621)	d
WMTweed-free	N50	SAshoes	5448	(±305)	a	5183	(±335)	bcd	4322	(±565)	d
WMTweed-free	N50	SAtines+shoes	5307	(±226)	a	5166	(±210)	bcd	4606	(±610)	cd
WMTweed-free	N100	SAhose	5520	(±115)	a	7351	(±109)	a	7466	(±489)	a
WMTweed-free	N100	SAshoes	4608	(±500)	a	6117	(±311)	abc	6703	(±240)	ab
WMTweed-free	N100	SAtines+shoes	5593	(±620)	a	6102	(±367)	abc	6560	(±445)	ab
ANOVA			p
WMT			0.9444			0.0005			0.0289
N			0.4692			<0.0001			<0.0001
SA			0.3426			0.3807			0.0229
WMT * N			0.0851			0.142			0.6252
WMT * SA			0.3335			0.2243			0.232
N * SA			0.0239			0.0653			0.1603
WMT * N * SA			0.9225			0.0444			0.8225

). Across treatments at SITE_C, SA_hose resulted in winter wheat yields 11% greater than SA_tines+shoes. However, differences were not significantly different (p > 0.05) at the other sites.

Crop biomass was consistently affected by nitrogen rate across site–years (Table S2). On average, spring barley biomass (g m⁻²) was 12% greater at N₁₀₀ compared to N₅₀ at SITE_A, and winter wheat biomass (g m⁻²) was 27% and 38% greater at N₁₀₀ compared to N₅₀ at SITE_B and SITE_C, respectively. At SITE_B, winter wheat biomass was also 7% greater in WMT_weed-free plots than WMT_weedy plots, and SA_tines+shoes produced 13% less biomass than SA_hose and SA_shoes.

Weed biomass (g m⁻²) was largely unaffected by the treatment variables studied. Nitrogen rate and slurry application strategy did not affect the biomass of taprooted surrogate weeds, S. alba at SITE_A, or Matricaria at SITE_B and SITE_C (

Table 4. The effect of nitrogen rate (N₁₀₀ and N₅₀) and slurry application method (SA_hose, SA_shoes, and SA_tines+shoes) on intra-row taprooted Sinapis alba (SITE_A) and Matricaria biomass (SITE_B and SITE_C). Data from each site were analyzed separately. A connecting letters report displays differences among treatments, whereby means not sharing a similar letter are significantly different (p < 0.05). Significant main effects and interactions are highlighted in bold text (p < 0.05). Standard errors are displayed in parentheses.

		Sinapis alba				Matricaria
Treatment		SITEA a				SITEB a				SITEC a
		Trans.		Back-Trans.		Trans.		Back-Trans.		Trans.
		g m−2
N50	SAhose	1.429	(±0.049)	1.048	a	10.6	(±3.3)	144	a	4.16	(±0.25)
N50	SAshoes	1.313	(±0.184)	0.825	a	5.6	(±4.0)	79	a	2.63	(±0.53)
N50	SAtines+shoes	1.811	(±0.120)	2.325	a	4.4	(±1.9)	29	a	6.77	(±0.67)
N100	SAhose	2.311	(±0.072)	4.358	a	5.3	(±3.2)	58	a	5.76	(±0.53)
N100	SAshoes	2.171	(±0.138)	3.769	a	7.6	(±2.3)	72	a	6.47	(±0.81)
N100	SAtines+shoes	1	(±0.145)	0.063	a	9	(±1.7)	89	a	5.61	(±0.19)
ANCOVA		p
N		0.9665				0.2499				0.4086
SA		0.7098				0.9347				0.8002
N * SA		0.5513				0.1062				0.1736
Weed density (no. m−2) b		0.3356				<0.0001				0.0461

^a Data underwent a square root (x + 1) transformation before analysis; transformed means (Trans.) and standard errors are presented along with back-transformed means (Back-trans.). Treatment comparisons made throughout the text of the present article are calculated from back-transformed means. ^b Sinapis alba density (no. m⁻²) was used as a covariate in the analysis for SITE_A, and Matricaria density (no. m⁻²) was used in the analysis for SITE_B and SITE_C.

). The biomass of P. annua, possessing shallow fibrous roots, was 66% greater at N₁₀₀ compared to N₅₀ across slurry application treatments at SITE_B. However, no significant effects were observed for P. annua at SITEc or L. multiflorum at SITE_A (

Table 5. The effect of nitrogen rate (N₁₀₀ and N₅₀) and slurry application method (SA_hose, SA_shoes, and SA_tines+shoes) on intra-row fibrous-rooted Lolium multiflorum (SITE_A) and Poa annua biomass (SITE_B and SITE_C). Data from each site were analyzed separately. A connecting letters report displays differences among treatments, whereby means not sharing a similar letter are significantly different (p < 0.05). Significant main effects and interactions are highlighted in bold text (p < 0.05). Standard errors are displayed in parentheses.

		Lolium multiflorum				Poa annua
Treatment		SITEA a				SITEB			SITEC
		Trans.		Back-Trans.
		g m−2
N50	SAhose	7.68	(±0.369)	58	a	41.9	(±11.2)	b	3.6	(±2.2)	a
N50	SAshoes	7.316	(±0.170)	53	a	41.2	(±7.2)	b	6.4	(±3.6)	a
N50	SAtines+shoes	8.305	(±0.113)	68	a	43.2	(±12.0)	b	1.5	(±1.0)	a
N100	SAhose	8.825	(±0.200)	77	a	54.7	(±17.0)	ab	5.2	(±4.7)	a
N100	SAshoes	5.567	(±0.320)	30	a	51.4	(±9.0)	ab	3.8	(±2.6)	a
N100	SAtines+shoes	11.535	(±0.058)	132	a	103	(±15.7)	a	4.3	(±3.3)	a
ANCOVA		p
N		0.4783				0.0003			0.2946
SA		0.2622				0.2172			0.1661
N * SA		0.4053				0.0741			0.3421
Weed density (no. m−2) b		0.5357				0.0009			<0.0001

^a Data underwent a square root (x + 1) transformation before analysis; transformed means (Trans.) and standard errors are presented along with back-transformed means (Back-trans.). Treatment comparisons made throughout the text of the present article are calculated from back-transformed means. ^b Lolium multiflorum density (no. m⁻²) was used as a covariate in the analysis for SITE_A, and Poa annua density (no. m⁻²) was used in the analysis for SITE_B and SITE_C.

). The intra-row biomass of naturally occurring ambient weeds was also unaffected by treatments across sites (

Table 6. The effect of nitrogen rate (N₁₀₀ and N₅₀) and slurry application method (SA_hose, SA_shoes, and SA_tines+shoes) on intra-row ambient weed biomass (SITE_A, SITE_B and SITE_C). Data from each site were analyzed separately. A connecting letters report displays differences among treatments, whereby means not sharing a similar letter are significantly different (p < 0.05). Significant main effects and interactions are highlighted in bold text (p < 0.05). Standard errors are displayed in parentheses.

		Ambient Weed Biomass
Treatment		SITEA			SITEB			SITEC
		g m−2
N50	SAhose	55.6	(±12.6)	a	5.6	(±3.3)	a	23.7	(±14.7)	a
N50	SAshoes	19.6	(±4.5)	a	12.8	(±6.9)	a	9.4	(±5.0)	a
N50	SAtines+shoes	55.6	(±19.9)	a	5.2	(±5.1)	a	32.4	(±18.8)	a
N100	SAhose	39.8	(±6.3)	a	45.6	(±19.5)	a	30.1	(±19.1)	a
N100	SAshoes	75.8	(±23.0)	a	28	(±11.9)	a	38.6	(±18.4)	a
N100	SAtines+shoes	47.1	(±25.8)	a	18.3	(±8.5)	a	28.4	(±11.1)	a
ANCOVA		p
N		0.422			0.1378			0.8259
SA		0.9613			0.6026			0.9894
N * SA		0.1178			0.9928			0.6211
Weed density (no. m−2) a		0.6991			0.0334			0.0059

^a Ambient weed density (no. m⁻²) was used as a covariate in the analysis. Surrogate weeds, Sinapis alba, and Lolium multiflorum, were not included in the ambient weed density measure for SITE_A, and ambient weeds, Matricaria and Poa annua, were not included for SITE_B and SITE_C.

).

Nitrogen accumulation (g N m⁻²) among crop and weed plant biomass components is presented in Table S3. At SITE_A, N capture in spring barley was on average 20% greater in N₁₀₀ plots compared to N₅₀. However, statistically significant differences were not present among other plant biomass components samples from SITE_A (i.e., L. multiflorum, S.alba, and ambient weeds). At SITE_B, winter wheat within the SA_shoes treatment had 38% more g N m⁻² than SA_tines+shoes. Across slurry application methods, winter wheat N accumulation was 65% and 68% greater in N₁₀₀ plots compared to N₅₀ at SITE_B and SITE_C, respectively. A greater degree of difference in N accumulations for N₁₀₀ versus N₅₀ among winter wheat compared to spring barley may be linked to the 49% increase in achieved NH₄-N application rates at SITE_B and SITE_C versus SITE_A. At SITE_B, N in P. annua and ambient weed components were 90% and 335% greater at N₁₀₀ than N₅₀, respectively, while no differences among these groups were present at SITE_C.

Grain quality measures were somewhat affected by treatment variables across site–years. At SITE_A, grain protein (%) was on average 13% greater in N₁₀₀ than N₅₀ plots; also, grain protein was on average 7% and 8% greater in SA_shoes compared to SA_tines+shoes and SA_hose, respectively (Table S4). At SITE_B and SITE_C, on average in N₁₀₀ plots, grain bulk density was 1% and 2% greater, grain protein was 7% and 11% greater, and thousand kernel weight was 3% and 5% greater than N₅₀ at SITE_B and SITE_C, respectively.

3.2. Ammonia Emissions

In experiment EXP_NH3.A, there were significantly lower cumulative emissions following slurry application by SA_tines+shoes and SA_shoes compared to SA_hose (p = 0.0007). However, in experiments EXP_NH3.B1 and EXP_NH3.B2, there were no significant differences between the treatments. Cumulative emissions can be seen in

Table 7. Cumulative NH₃ emission as a percentage of applied total ammoniacal N (TAN) and g N m⁻² 113 h after field application of 22.8 tonne ha⁻¹ pig slurry to a spring barley crop on loamy sand (EXP_NH3.A at SITE_A) and 30 tonne ha⁻¹ pig slurry to a winter wheat crop on sandy loam (EXP_NH3.B1 and EXP_NH3.B2 at SITE_B). A connecting letters report displays differences among treatments, whereby means not sharing a similar letter are significantly different (p < 0.05). Standard errors are displayed in parentheses.

Site	Experiment	Treatment	TAN		g N m−2
			%
SITEA	EXPNH3.A	SAhose	16.60	(±0.74)	0.59	(±0.02)	a
		SAshoes	9.14	(±0.54)	0.32	(±0.02)	b
		SAtines+shoes	11.81	(±0.89)	0.42	(±0.03)	b
SITEB	EXPNH3.B1	SAhose	31.25	(±2.67)	3.30	(±0.25)	a
		SAshoes	24.93	(±1.08)	2.64	(±0.03)	a
		SAtine+shoes	29.42	(±2.50)	3.11	(±0.23)	a
	EXPNH3.B2	SAhose	42.42	(±1.69)	4.79	(±0.15)	a
		SAshoes	41.97	(±1.05)	4.62	(±0.05)	a
		SAtines+shoes	39.39	(±3.90)	4.44	(±0.43)	a

, and flux over time can be seen in Figure S3.

Following application, slurry surface pH showed the same trends for all three application methods (Figure S2). Within the first two hours, pH increased approximately one unit; thereafter, it dropped below the initial value. At the time of final pH measurements (3.5–4.5 h after application), no liquid was observed at the soil surface.

3.3. Exposed Surface Area

Slurry ESA immediately after application, and 2.66 h after application, differed among all treatments. Both initial ESA (t = 0 h) and final ESA (t = 2.66 h) were greatest in the SA_hose treatment, second largest in SA_shoes, and were lowest in SA_tines+shoes (

Table 8. Parameter estimates used in Equations (2) and (3) to analyze EXP_ESA.B data. Average slurry exposed surface area (ESA, m² m⁻²) immediately after (t = 0 h) and 2.66 h after (t = 2.66 h) the application of 30 tonne ha⁻¹ of pig slurry to winter wheat in sandy loam soil are also estimated and presented for three application methods (SA_hose, SA_shoes, and SA_tines+shoes). Lower and upper 95% confidence interval limits are shown in parentheses. A connecting letters report displays differences among parameters and estimates, whereby means not sharing a similar letter are significantly different (p < 0.05).

Treatment	α			β			γ			ESA (t = 0 h)			ESA (t = 2.66 h)
										m2 m−2			m2 m−2
SAhose	−0.659	(−0.670, −0.648)	a	−0.010	(−0.016, −0.004)	a	0.001	(−0.007, 0.010)	b	0.552	(0.536, 0.568)	a	0.512	(0.504, 0.520)	a
SAshoes	−0.895	(−0.905, −0.886)	b	−0.013	(−0.020, −0.006)	a	0.019	(0.005, 0.033)	a	0.430	(0.422, 0.439)	b	0.430	(0.398, 0.408)	b
SAtines+shoes	−1.907	(−1.965, −1.850)	c	−0.028	(−0.086, −0.031)	b	0.013	(−0.083, 0.109)	a	0.167	(0.132, 0.203)	c	0.145	(0.131, 0.158)	c

; Figure 1). Initial ESA at t = 0 h was 3.3 and 2.6 times greater in SA_hose and SA_shoes compared to SA_tines+shoes, respectively. Final ESA at t = 2.66 h was 3.5 and 3.0 times greater in SA_hose and SA_shoes compared to SA_tines+shoes, respectively.

4. Discussion

Achieved winter wheat yields were low compared to the regional average. Within N₁₀₀ plots, yields were 29% and 25% less than the 2020 mean reported for Zealand, Denmark, at SITE_B and SITE_C, respectively [37]. Spring barley yields, however, were comparable in 2019; across N₁₀₀ plots, achieved yields were only 6% lower than the reported average for West Jutland, Denmark [37]. While placing slurry within a growing spring barley crop is not a recommended practice, the application of 100 kg N ha⁻¹ is representative of a typical rate for organic farms, with conventional rates ranging from 140 to 150 kg N ha⁻¹. Reduced winter wheat yields are partly explained by having employed low N rates throughout the experiment; recommended rates for winter wheat range between 170 and 200 kg N ha⁻¹ [38]. In addition, splitting fertilizer placement into two springtime events, typically applying a mineral fertilizer first and slurry second, has shown to improve yields. Compared to a single application event, Olesen et al. [39] report the three-year yield average of winter wheat increasing by 2 to 4% when an equivalent N rate is split. Reduced N rates of 50 and 100 kg N ha⁻¹ and one application event were chosen to enhance resource competition among crop and weeds, and to simplify the interpretation of treatment outcomes.

Nitrogen rate had the most consistent effect on crop growth across site–years. Crop biomass increased with elevated N input at all sites, with winter wheat demonstrating a stronger response compared to spring barley. This result is logical, given that post-emergence fertilizer placement is ideal for meeting N demand of winter wheat [40], while pre-emergence fertilizer placement is best suited for spring barley [41]. At SITE_B and SITE_C, the accumulation of N in winter wheat biomass was consistently greater in N₁₀₀ plots compared to N₅₀, as were crop yield, grain protein, bulk density and TKW. However, a yield response was not present in spring barley, and among quality parameters, only grain protein was greater at N₁₀₀. The absence of yield response could be due to the late application of slurry. Improved N capture in spring barley helps to explain the increase in crop biomass and grain protein at SITE_A. By comparing the difference in N uptake between N₅₀ and N₁₀₀, apparent N recovery (ANR) of NH₄-N in applied slurry could be estimated. When Pedersen et al. [8] applied a mineral fertilizer on the same soil type ANR was 80% for spring barley. In spring barley at SITE_A, average ANR was 41%, and in winter wheat at SITE_B, average ANR was only 47%, indicating a significant loss of NH₃. Corresponding NH₃ emission rates observed at SITE_A and SITE_B ranged from 9 to 17% and 25 to 42%, respectively. At SITE_C, ANR was 73% indicating only small losses of NH₃; high utilization of NH₄-N, despite some NH₃ loss, could be attributed to the mineralization of organic N in the pig slurry [8].

It is important to note that the field conditions at SITE_A differed considerably from SITE_B and SITE_C. Soil texture at SITE_A was lighter, possessing a higher percentage of sand (Table 1), and because spring barley was employed as the test crop, field cultivation preceding sowing further loosened the soil shortly before slurry was applied. As was hypothesized, standard trailing hoses were found to produce the greatest total NH₃ emissions at SITE_A; however, the effect of combining tine and trailing shoe did not reduce emissions compared to trailing shoes alone. This lack of difference is partially explained by the absence of a surface crust on light loose soils, limiting the functional difference between SA_shoes and SA_tines+shoes treatments (see Section 2.1.1). Soil texture is known to affect NH₃ emissions; sandy soils possess a greater slurry infiltration capacity than heavier clay-type soils, leading to decreased emissions. Both sandy soils [25,42] and cultivation preceding slurry application are known to increase slurry infiltration capacity and decrease NH₃ emissions [9,43]; in the case of SITE_A, reducing the functional difference between SA_hose versus SA_shoes and SA_tines+shoes.

At SITE_A, a decrease in atmospheric N loss associated with reduced NH₃ emissions in combination with targeted placement in SA_shoes and SA_tines+shoes treatments did not translate into increased N uptake in spring barley as expected. More invasive slurry placement strategies may inflict yield penalties by causing root damage; however, among the tactics causing some shallow soil disturbance (SA_shoes and SA_tines+shoes), no yield losses were observed. It can be interpreted that substantial spring barley damages were not imparted by SA_shoes or SA_tines+shoes strategies. If significant damages had been acquired, yield loss would have to be offset by improved N capture. Instead, crop yield, N accumulation, biomass, and grain quality measures did not differ among slurry application treatments, with the only exception being that SA_shoes produced higher grain protein compared to other application strategies.

Soils at SITE_B and SITE_C had a much higher clay content than SITE_A (Table 1). Because winter wheat was employed as the test crop, a lack of cultivation in the spring and low precipitation preceding slurry placement resulted in heavily crusted soils (Figure S4). Trailing shoes did not operate as expected, riding on top of the hardened surface crust without creating a depression, therefore, decreasing the functional difference between SA_hose and SA_shoes treatments. Results from Pedersen et al. [23] are supportive of the notion that soil type can impact a slurry placement method’s ability to reduce NH₃ emissions; improved abatement by trailing shoes compared to trailing hoses were observed in coarse sand only, whereas no effects of placement method were observed in heavier soils, including a loamy sand and sandy loam. Meanwhile, in the present study, the prototype combing the operation of tines and trailing shoes operated as intended, with the straight tine breaking through the soil crust and the shoe riding behind depositing slurry into the soil break created (see Section 2.1.1 and Video S2). Despite SA_tines+shoes functioning as designed and in opposition to the expected outcome, no significant differences in cumulative NH₃ emissions were observed among slurry application strategies. Although, when ordered from highest to lowest emissions, slurry application treatments performed as expected in EXP_NH3.B2, with SA_hose > SA_shoes > SA_tines+shoes, and deviated slightly from our hypothesis, in EXP_NH3.A and EXP_NH3.B1, SA_hose > SA_tines+shoes > SA_shoes. Accordingly, there were no significant differences in crop N uptake between application methods.

SA_tines+shoes reduced initial (t = 0 h) and final (t = 2.66 h) slurry ESA in EXP_NH3.B. Slurry was spread at SITE_B and SITE_C following a long period without precipitation; therefore, cracks that developed under dry conditions in heavy clay soils provided a pathway for liquid slurry to funnel into when deposited in a band at the soil’s surface (Figure S4). While SA_tines+shoe functioned as intended, producing the lowest initial and final ESA, it was not enough to reduce total NH₃ emissions (Table 7). Under field conditions where hardened crusted soils are present, SA_tines+shoes shows promise for reducing ESA and limiting NH₃ volatilization compared to SA_shoes and SA_hose, if large cracks have not developed providing an alternate route for rapid infiltration using simpler tools. However, the formation of large fissures in clay soils is common especially in overwintered crops following the freeze–thaw cycles of winter [44]. When using surface application under these conditions, a larger slurry ESA may, in fact, increase the infiltration rate by broadening the surface area whereby liquid slurry is able to access and infiltrate into surficial cracks. Thus, an important subject of future research would be to investigate the use of slurry placement strategies under a variety of soil conditions.

It was hypothesized that crop damages caused by the shallow disturbance of SA_shoes and SA_tines+shoes treatments would be offset by improved N uptake, resulting in similar yields across slurry application methods. At SITE_B, greater N uptake was observed for SA_shoes compared to SA_tines+shoes, and while this translated into increased biomass in SA_shoes, grain yield effects were not observed among treatments. At SITE_C, SA_hose produced greater yields than SA_tines+shoes, but differences in N uptake and winter wheat biomass were not observed among slurry application strategies. Again, there is little evidence to suggest that the SA_shoes or SA_tines+shoes treatments inflicted significant crop damages, and contrary to the expected outcome, at SITE_B, SA_tine+shoe reduced N uptake in winter wheat. As expected, slurry placement strategies did not have an effect on grain quality parameters at SITE_B or SITE_C, and ambient weed biomass did not differ among slurry application treatments.

Weed effects were observed at SITE_B only; increasing N rate resulted in greater N accumulation in P. annua and other ambient weeds; P. annua biomass was also greater in N₁₀₀ plots. The observation of both positive and null effects across site–years may be accounted for by differences in weed species’ response to N rate [45]. The dominant ambient weed at SITE_B was G. aparine, a species whose growth is consistently enhanced by increasing N [46]; while ambient weed biomass at N₁₀₀ was 2.9 times greater than N₅₀ at SITE_B, significant differences were not observed due to high variability. As N rate increases, a positive growth response for P. annua has also been documented [47,48]. However, it is surprising that no P. annua response was observed at SITE_C, where P. annua accounted for 63% of the ambient weed population.

Weed species were expected to respond differently to slurry placement treatments tested; however, this effect was never observed. In studies where species-specific effects are present, differences in slurry placement depth among treatments are greater; for example, when depositing slurry at the soil’s surface is compared to deep injection. A possible explanation for our having observed no response difference among shallow-rooted and taprooted weeds is that those treatments evaluated do not differ enough in slurry placement depth, either interacting shallowly or on top of the soil’s surface. Upon injecting liquid fertilizer to a depth of 5 cm, Melander et al. [16] observed improved growth in deeply rooted ambient weed, Tripleurospermum indorum, and surrogate weed, Sinapis alba, as well as suppressed growth in shallowly rooted weed, Stellaria media. Reduced growth in S. media is likely due partly to having placed fertilizer below the weed’s rooting zone, leading to decreased N capture, while improved growth in T. indorum and S. alba may be linked to increased N capture from having deposited fertilizer within these weeds’ rooting zone. Results from Petersen [13] support this notion; when injecting slurry to a depth of 5 to 15 cm was compared to broadcast spreading followed by incorporation, injection resulted in a 58% increase in ambient weed N capture. In the present study, SA_shoes and SA_tine+shoes were expected to improve slurry infiltration into the soil column, resulting in increased N accumulation by taprooted weeds. The SA_hose treatment places slurry at the soil’s surface and was expected to have a slower infiltration rate, thus increasing accumulation among shallow rooted weeds. In opposition to our hypotheses, increased N uptake and plant biomass were not observed among shallowly rooted surrogate weeds, L. multiflorum or P. annua, for the SA_hose treatment, nor among taprooted surrogate weeds, S. alba or Matricaria, in plots receiving SA_shoes or SA_tines+shoes treatments. Results from the present study indicate that compared to SA_hose, slurry application with SA_shoes or SA_tines+shoes have neither positive nor negative implications for intra-row weed management within the hoed cereal system.

Inter-row hoeing was performed throughout the present study so that crop and weed response to slurry placement strategies could be assessed under contemporary management practices. Hoeing is an effective management tactic [28], and the advance of precision guidance technologies has eased implementation and increased adoption within the last decade [49]. It may be hypothesized, however, that weed response to slurry placement treatments may become less pronounced when inter-row weeds are controlled, and assessments focus solely on the response of intra-row weeds. If inter-row weeds are not controlled, spreading slurry in a wide band between crop rows becomes much less efficient than when slurry is purposefully and precisely placed close to the crop row; in this scenario, weeds growing in the inter-row zone will have greater access to available N by proximity when compared to crop plants growing in the intra-row. Therefore, if weed effects are measured across both the inter- and intra-row zones, we would expect to see significant differences between banded and targeted placement strategies [14,15]. In contrast, when hoeing is performed, and only intra-row weeds remain, the response of crop and weeds to slurry placement strategy becomes more reliant on slurry infiltration and targeted placement within the soil column rather than targeted slurry placement on the soil’s surface [16]. While it was expected that SA_shoes and SA_tines+shoes would have greater effects on weeds throughout the present study, our implementation of inter-row hoeing and focus on intra-row weeds may also help to explain the general absence of weed effects.

5. Conclusions

Observations made in the present study illustrate the considerable impact that soil conditions have on slurry deposition outcomes. In the presence of soils possessing a hard surficial crust, the successful operation of a prototype device was observed, whereby the combined operation of rigid tines and trailing shoes (SA_tines+shoes) reduced the initial and final slurry-exposed surface area when compared to standard trailing hoses (SA_hose) and trailing shoes alone (SA_shoes). Thus, the present study provides proof of concept for the prototype design in terms of developing a low draft, post-emergence slurry application device that is capable of successfully operating as intended in heavily crusted soils. However, the slurry placement methods tested were evaluated under the pretext that the adoption of new technologies must be justified by reduced ammonia (NH₃) emissions and improved crop growth. Therefore, the results ultimately do not warrant advocating the use of SA_shoes or SA_tines+shoes over SA_hose across a range of soil textures and conditions. No significant effects on crop or weeds were detected among treatments, and compared to SA_hose, NH₃ emissions were reduced by SA_shoes and SA_tines+shoes in only one of three site–years when operating in loose sandy soil. Therefore, deeper slurry placement within the soil column is likely necessary to consistently mitigate NH₃ emissions and illicit a positive crop response. This information may be of use to equipment manufacturers and researchers who wish to further develop and evaluate post-emergence slurry placement strategies. Future research on this topic should be conducted across a wide range of soil conditions or be considered site-specific.