Digestate Application on Grassland: Effects of Application Method and Rate on GHG Emissions and Forage Performance

Abstract

The application of digestate as a fertilizer offers a promising alternative to synthetic inputs on permanent grasslands, with benefits for productivity and environmental performance. This four-year study evaluated the impact of two digestate application methods—disc injection (I) and band spreading (S)—combined with four dose variants (0, 20, 40, and 80 m³·ha⁻¹), including split-dose strategies. Emissions of ammonia (NH₃), carbon dioxide (CO₂), and methane (CH₄) were measured using wind tunnel systems immediately after application. Vegetation status was assessed via Sentinel-2-derived Normalized Difference Vegetation Index, Normalized Difference Water Index, and Modified Soil Adjusted Vegetation Index, and agronomic performance through dry matter yield (DMY), net energy for lactation (NEL), and relative feed value (RFV). NH₃ and CO₂ emissions increased proportionally with digestate dose, while CH₄ responses suggested a threshold effect, but considering solely the disc injection, CH₄ flux did not increase markedly with higher application rates. Disc injection resulted in significantly lower emissions of the monitored fluxes than band spreading. The split-dose I_40+40 variant achieved the highest DMY (3.57 t·ha⁻¹) and improved forage quality, as indicated by higher NEL values. The control variant (C, no fertilization) had the lowest yield and NEL. These results confirm that subsurface digestate incorporation in split doses can reduce emissions while supporting yield and forage quality. Based on the findings, disc injection at 40+40 m³·ha⁻¹ is recommended as an effective option for reducing emissions and maintaining productivity in managed grasslands.

1. Introduction

Despite all the technological advances made by humanity, contemporary agriculture continues to face one of its most pressing challenges. The fundamental question remains how to feed an ever-growing population (projected to reach 9.8 billion by 2050, according to the United Nations [1]) while addressing climate change, land degradation, soil fertility decline, and resource depletion—within the overarching framework of sustainability [2,3]. Sustainability has become a defining theme of our era. There is a growing trend to quantify carbon footprints at increasingly detailed scales—even down to the level of individual food items or production units. While globally, agriculture accounts for around 13.5% of greenhouse gas emissions [4], farming in the European Union contributes less than 11% [5]. Although most EU countries have reduced agricultural GHG emissions over the past three decades, further reductions are needed to meet ambitious EU climate goals, according to 2030 climate targets [4,6].

In the search for sustainable agricultural solutions, alternatives to conventional mineral fertilizers are being explored, and digestate could be one of them. Digestate is an organo-mineral material that remains after the anaerobic digestion of organic feedstocks, such as manure, municipal waste, or crop residues, in the production of biogas and can be recycled back to soil as a conditioner and fertilizer, thereby improving the environmental sustainability of the entire energy supply chain [7,8]. In contrast to mineral fertilizers, digestate can be more environmentally sustainable in terms of GHG emissions associated with the production of synthetic fertilizers [9]. Additionally, the use of mineral fertilizers may carry environmental risks—for example, phosphate fertilizers derived from phosphate rock often contain cadmium (Cd), a heavy metal that can accumulate in soil and living organisms, cause harm, and enter the food chain through fertilizer use [10]. This further highlights the importance of safe nutrient recycling using organic materials such as digestate. In addition, a previous study found that the effects of digestate application were more favorable than those of mineral fertilizers, suggesting that digestate could be not only a valuable organic fertilizer but also a substitute for mineral fertilizers [11]. As part of the shift toward greater sustainability, the number of biogas plants in Europe nearly doubled between 2010 and 2020, which raises the increasingly relevant question of how to manage the large volumes of digestate produced. [12,13]. Moreover, it represents an effective application of circular economy principles [14]. Biogas energy offers farmers three key benefits: efficient organic waste management, climate change mitigation through GHG reduction, and the production of high-quality organic fertilizer—while also reducing unpleasant odors [15]. Digestate is a key by-product, with around 180 million tons produced annually in the EU, with an average content of nitrogen (2–5 kg.m⁻³) and phosphorous (0.5–1.5 kg.m⁻³) [16]. Despite these agronomic advantages, digestate composition can vary considerably based on feedstock origin and processing methods. It may contain undesired substances such as heavy metals, pathogens, antibiotics, or microplastics, raising environmental and regulatory concerns [17,18]. While some pollutants are currently regulated, emerging contaminants remain largely unaddressed in many national frameworks [17]. Given its chemical composition, digestate from agricultural biogas plants can be regarded as a multi-nutrient fertilizer [19]. To ensure good yields, it is necessary to provide an adequate amount of nutrients. However, the amount of digestate applied must be carefully controlled to prevent nutrient leaching and contamination of surface and groundwater. Meeting three criteria (environmental efficiency, agronomic efficiency, and economic efficiency) is essential for optimizing the nutrient intake by plants and the amount of fertilizers applied, including those from digestate [20]. Digestate application increases soil levels of carbon, nitrogen, phosphorus, potassium, and magnesium. Over the long term, it also improves physical soil properties and water infiltration [21,22].

However, not only the quantity but also the method of digestate application is critical for its environmental impact. Surface broadcast using a splash plate is still widely practiced by smallholders [23], but is associated with high ammonia (NH₃) losses—up to 31% of total nitrogen applied [24]. In contrast, low-emission application methods such as band application (e.g., trailing hose or shoe) and shallow injection substantially reduce nitrogen losses by 40–50%, mainly by limiting the surface area exposed to the atmosphere. Among these, shallow injection is generally more effective than trailing shoe. Injection methods have also been presented to reduce ammonia emissions to as little as 1.8 mg·m⁻²·h⁻¹, compared with 5.8 mg·m⁻²·h⁻¹ for splash plate techniques [25]. Band application, due to its balance of technical feasibility and environmental benefits, remains widely adopted digestate application methods in Europe [26,27].

This study focuses on permanent grasslands, which are essential for agriculture and ecosystems alike—covering more than one-third of the EU’s agricultural area—but also for human life, as they contribute to essential ecosystem services [28,29]. In contrast to the extensively studied effects of digestate application on arable land, its use on permanent grasslands remains notably under-researched—despite the importance of these ecosystems for agriculture, biodiversity, and climate resilience. The limited data available, such as that from Holátko et al. [30], suggest promising effects of digestate in these systems, yet comprehensive long-term studies comparing application methods and rates are still lacking. As Pawlett et al. [31] emphasize, there is a significant research gap regarding nutrient uptake and environmental performance of digestate in permanent grassland systems. In this study, two digestate application techniques—disc injection and band spreading—were evaluated on permanent grassland over a four-year period. The main objective was to compare their environmental and agronomic performance across increasing digestate rates. Specifically, the study aimed to test the following hypotheses: (a) emissions of greenhouse gases (GHG) and ammonia (NH₃) increase proportionally with rising digestate rates regardless of application method; (b) the disc injector results in lower emissions than the band spreader due to deeper incorporation and reduced surface exposure; and (c) splitting the total digestate dose into two even applications has a more favorable impact on vegetation health and biomass yield compared to applying the full amount at once. This approach is particularly relevant in grassland systems, where continuous plant cover and nutrient dynamics differ from those in arable fields; split applications may align more closely with plant nutrient demand, while reducing the risk of leaching and volatilization losses.

2. Materials and Methods

2.1. Experimental Site and Grassland Management

The four-year field experiment was conducted near the town of Čechtice in the Central Bohemian Region of the Czech Republic (49°37′57.4″ N, 15°01′54.1″ E, 478 masl.). The soil type in the experimental area is Mesobasic Cambisol with sandy loam soil texture class (based on the USDA soil texture triangle). A detailed characterization of the soil’s chemical properties is provided in

Table 1. Chemical characteristics (mean ± st. dev.) of experimental fields [30].

Soil Property	Value
Total carbon (%)	4.3 ± 0.7
Total nitrogen (%)	0.4 ± 0.1
C/N ratio (-)	10.6 ± 0.6
K (mg·kg−1)	345.0 ± 95.9
Ca (mg·kg−1)	2674.8 ± 525.6
Mg (mg·kg−1)	321.3 ± 47.2
P (mg·kg−1)	34.3 ± 9.6
pH (-)	6.7 ± 0.2

. Based on data from the Czech Hydrometeorological Institute, the experimental site is characterized by an average temperature of 9 °C (with seasonal averages of 17.1 °C in summer and 1.3 °C in winter) and an average annual precipitation of 583 mm (normal between 1991 and 2020).

Digestate was produced at a farm’s biogas plant that primarily processes maize silage and cow slurry. Prior to filling the application tanker, a mixed sample was taken annually from the storage facility, where the digestate had been homogenized, for laboratory analysis. The chemical composition of the applied digestate is presented in

Table 2. Chemical composition of applied digestate [30].

Year	DM	Ntot	P	K
	(%)		(g kg−1 DM)
2019	4.50 ± 0.12	31.96 ± 0.69	7.72 ± 1.07	80.37 ± 0.17
2020	5.27 ± 0.21	26.20 ± 1.05	10.24 ± 0.37	31.71 ± 1.27
2021	5.30 ± 0.35	31.52 ± 1.72	8.39 ± 1.40	68.62 ± 0.56
2022	5.37 ± 0.38	37.47 ± 1.75	8.36 ± 0.32	89.41 ± 1.22

.

In 2019, eight semi-operational experimental variants of 0.6 ha apiece were established, with three replications for each. The variants differed in the rate and method of digestate application, while one variant was a control. The detailed schedule of mowing events and digestate applications throughout the experimental period is presented in

Table 3. Mowing and digestate application schedule [30].

Number of Mowing	Term of Mowing	Term of Digestate Application
I.	2 June 2019	7 June 2019
II.	31 July 2019	8 August 2019
III.	1 November 2019
I.	21 May 2020	9 June 2020
II.	7 July 2020	7 August 2020
III.	31 October 2020
I.	2 June 2021	8 June 2021
II.	11 August 2021	2 September 2021
III.	9 November 2021
I.	1 June 2022	9 June 2022
II.	18 July 2022	25 July 2022
III.	10 October 2022

. Digestate was consistently applied after the first and second cuts each year to evaluate both single and split application strategies. The timing of applications was aligned with typical management practices for temperate grasslands and adjusted annually to match local weather and vegetation conditions. Digestate was applied into the soil using a machine VT4556 (Vredo Dodewaard B.V, Dodewaard, The Netherlands) equipped with injector ZB3 8448 (Vredo Dodewaard B.V, Dodewaard, The Netherlands) till a depth of 4 cm or to the soil surface using a standard hose applicator. The control experimental treatment was left devoid of digestate fertilization. Digestate was applied using two different technologies: a disc injector (variants labelled as I) and a band spreader (variants labelled as S). The total dose of digestate (in m³·ha⁻¹) was expressed directly in the treatment label as first dose + second dose. A control treatment without digestate application was included and was labelled as C. The maximum digestate application rate was set in accordance with the legislative requirements of the Czech Republic.

After the first cutting in 2021 and after the second cutting in 2022, emissions were measured immediately following digestate application, as concentrations typically decrease over time post-application. To enable a comprehensive comparison of gas fluxes between the two application methods across all applied rates, three additional replicates using the band spreader were conducted at 0 m³·ha⁻¹, 40 m³·ha⁻¹, and 80 m³·ha⁻¹ in addition to the standard 20 m³·ha⁻¹ rate. The gases monitored included NH₃, CO₂, and CH₄, measured using the INOVA 1412 gas analyzer (INNOVA Air Tech Instruments, Ballerup, Denmark), paired with an INOVA 1309 multiplexer to facilitate simultaneous sampling across variants. Wind tunnels were placed over each replication on the experimental plot to capture gas fluxes, with specialized tubing channelling the sampled air to the analyzer. Concentrations were measured in the air entering and exiting the wind tunnels, and fluxes were calculated based on the differences between these values. Each wind tunnel measured 50 × 35 cm. The inlet was fitted with an anemometer, while the outlet contained an adjustable fan to maintain a steady airflow of approximately 1 m s⁻¹. A thermometer inside each tunnel recorded temperature and humidity throughout the measurements. The values presented in

Table 4. Temperature and humidity values (mean ± st. dev.) during flux measurement on the two measurement dates; lowercase letters denote significant differences at a probability level of 0.05 between the dates.

Flux Measurement Date	Temperature °C	Humidity %
8 June 2021 after cutting 1	25.795 a ± 2.281	56.56 b ± 13.84
25 July 2022 after cutting 2	33.286 b ± 2.904	41.10 a ± 10.39

display significant differences between the two measurement dates, with the flux measurement conditions in 2022 being warmer and drier than those in 2021.

Wind tunnels were repositioned within each variant every hour, with three positions tested per variant to achieve three replicates. At each position, gases were measured at least five times per hour, yielding a minimum of 15 measurements per variant. Data were continuously stored on a PC. The flux, “J”, was calculated in units of mass per area per time (µg m⁻² min⁻¹) using Equation (1) [32]:

$$J={\displaystyle \frac{v·A_t·\left(C_{out}-C_{in}\right)}{A_s}},$$

(1)

where “v” is the mean air velocity in the wind tunnel (m min⁻¹), “A_t” is the cross-sectional area of the ventilation openings (m²), “C_out” and “C_in” are the concentrations of the monitored gases in the outgoing and incoming air (µg m⁻³), and “A_S” represents the footprint area of the wind tunnel (m²). The calculated concentrations were subsequently converted into minute-based flux rates. Dixon’s Q test was applied at a 0.05 significance level to identify and exclude outliers.

2.2. Crop Status

Satellite data from Sentinel-2 (European Space Agency; ESA) were utilized to assess crop status. The Sentinel-2 mission’s data are highly suitable for agricultural applications, as it offers free access to high spatial and temporal resolution data covering most regions of the world [33]. The vegetation indices were selected based on their suitability for use in the assessment of permanent grasslands. The Normalized Difference Vegetation Index (NDVI) is one of the most widely used indices for assessing plant health and correlates very well with the biomass of permanent grassland [34,35]. In terms of assessing potential water stress, the Normalized Difference Water Index (NDWI) was selected, as it is highly correlated with the NDVI but provides a higher sensitivity to drought stress than the NDVI [36]. Finally, the Modified Soil Adjusted Vegetation Index (MSAVI) was selected; this index may provide an improved vegetation signal compared to traditional indices, as it is more robust to soil noise [37]. These indices were used to assess vegetation status across all experimental treatments and years, providing insights into crop development and stress conditions. The formulas for vegetation indices calculations are given in

Table 5. Vegetation indices used for crop stand assessment.

Vegetation Index	Abbreviation	Formula	Reference
Normalized Difference Vegetation Index	NDVI	${\displaystyle \frac{NIR-RED}{NIR+RED}}$	[38]
Normalized Difference Water Index	NDWI	${\displaystyle \frac{NIR-SWIR}{NIR+SWIR}}$	[39]
Modified Soil Adjusted Vegetation Index	MSAVI	${\displaystyle \frac{2NIR+1-\sqrt{{\left(2NIR+1\right)}^2-8\left(NIR-RED\right)}}{2}}$	[40]

RED—red reflectance band; NIR—near infrared reflectance band; SWIR—shortwave infrared reflectance band.

.

The data were accessed via Google Earth Engine (Google LLC, San Francisco, CA, USA) and further processed using qGIS (Open Source Geospatial Foundation, Beaverton, OR, USA) and Statistica 14 software (TIBCO, Palo Alto, CA, USA).

While greenhouse gas emissions (CO₂, CH₄) and ammonia (NH₃) volatilization were measured across all digestate rates and both application methods, vegetation-related parameters (spectral indices, net energy for lactation, real feed value, and biomass yield) were assessed for the disc injector only at all rates and for the band spreader solely at the 20+20 m³ ha⁻¹ dose. This specific dose was selected for the band spreader, as it reflects common practice in digestate application on permanent grasslands in the region. This approach enabled method comparison under conditions representative of actual farm practice, while allowing a broader exploration of dose effects with the disc injector.

From the viewpoint of fodder quantity and quality, the experimental variants were assessed through dry matter yield (DMY), net energy for lactation (NEL), and relative feed value (RFV). The NEL is defined as the energy contained in the milk and contributed by milk nutrients. Prediction of NEL is crucial because it is directly related to the productivity of dairy cows [41]. The NEL is calculated based on the milk energy estimated from the composition and energy content of the energy-yielding nutrients (i.e., lactose, fat, and protein). RFV is the most widely used system to predict the quality of forages fed to ruminants. RFV grades forage according to their predicted digestible dry matter intake (DDMI), the product of dry matter intake (DMI), and the percentage of digestible dry matter (DDM) [42]. Both the NEL and RFV were calculated according to the methods of Třináctý [43].

Statistical analysis of data was performed using Statistica 14 software (TIBCO, Palo Alto, CA, USA). Significant differences in fluxes (factors: date; digestate rate; application method), vegetation indices (factors: vegetation index type; date; variant), and quantitative and qualitative harvest parameters (factors: year; cutting; variant) were determined through factorial analyses of variance (ANOVA) and Tukey’s HSD post-hoc tests with a 95% confidence interval. Simple linear regression for gas fluxes and for quantitative and qualitative harvest parameters was performed in order to analyze their dependence on the digestate application rate.

3. Results

Figure 1 presents the measured fluxes of ammonia (NH₃), carbon dioxide (CO₂), and methane (CH₄) in relation to digestate application rate (0, 20, 40, 80 m³ ha⁻¹), application method (disc injector vs. band spreader), and year of application (2021 vs. 2022). When evaluated separately, year-to-year differences in average fluxes were evident across all gases and application techniques. In 2021, average emissions under disc injector application reached 1176.2 µg m⁻² min⁻¹ for NH₃, 37,106.1 µg m⁻² min⁻¹ for CO₂, and 1488.5 µg m⁻² min⁻¹ for CH₄. For the same year, band (surface) application resulted in substantially higher values: 3494.1 µg m⁻² min⁻¹ for NH₃, 131,723.9 µg m⁻² min⁻¹ for CO₂, and 2402.6 µg m⁻² min⁻¹ for CH₄. In 2022, disc injector application resulted in 803.6 µg m⁻² min⁻¹ for NH₃, 50,594.0 µg m⁻² min⁻¹ for CO₂, and 2057.5 µg m⁻² min⁻¹ for CH₄, while band application again led to higher emissions: 2530.8 µg m⁻² min⁻¹ for NH₃, 173,878.2 µg m⁻² min⁻¹ for CO₂, and 2643.2 µg m⁻² min⁻¹ for CH₄. These differences confirmed that, in both years and across all gases, surface application (band spreader) consistently resulted in substantially higher gaseous losses compared to subsurface incorporation (disc injector). Specifically, surface application increased emissions by 197.1–214.9% for NH₃, 243.7–255.0% for CO₂, and 28.5–61.4% for CH₄, depending on the year. A factorial ANOVA confirmed that both digestate rate and application method had statistically significant effects on all three gas fluxes (p < 0.001).

A factorial ANOVA revealed that NH₃ emissions were significantly affected by application method (F(1, 287) = 171.54, p < 0.001, η² = 0.374), digestate rate (F(3, 287) = 142.35, p < 0.001, η² = 0.598), and application date (F(1, 287) = 18.57, p < 0.001, η² = 0.061). Strong interaction effects were found for method × rate (F(3, 287) = 55.17, p < 0.001), while method × date and rate × date were marginal or moderately significant (p = 0.053 and p = 0.001, respectively). NH₃ emissions increased sharply at higher application rates, with band spreading causing consistently higher fluxes than injection. A summary of factorial ANOVA results is presented in

Table 6. Summary of factorial ANOVA results for NH₃, CO₂, and CH₄ fluxes; partial η² represents effect size; bold font denotes significant effects/interactions.

Flux	Effect	F (df1; df2)	p-Value	Partial η2
NH3	Method	171.54 (1; 287)	<0.001	0.374
	Rate	142.35 (3; 287)	<0.001	0.598
	Date	18.57 (1; 287)	<0.001	0.061
	Method × Rate	55.17 (3; 287)	<0.001	0.366
	Method × Date	3.76 (1; 287)	0.053	0.013
	Rate × Date	5.49 (3; 287)	0.001	0.054
	Method × Rate × Date	1.06 (3; 287)	0.368	0.011
CO2	Method	367.28 (1; 287)	<0.001	0.561
	Rate	111.28 (3; 287)	<0.001	0.538
	Date	24.33 (1; 287)	<0.001	0.078
	Method × Rate	46.66 (3; 287)	<0.001	0.328
	Method × Date	6.22 (1; 287)	0.013	0.021
	Rate × Date	1.52 (3; 287)	0.209	0.016
	Method × Rate × Date	1.22 (3; 287)	0.302	0.013
CH4	Method	31.91 (1; 287)	<0.001	0.100
	Rate	14.28 (3; 287)	<0.001	0.130
	Date	9.34 (1; 287)	0.002	0.032
	Method × Rate	1.39 (3; 287)	0.247	0.014
	Method × Date	1.54 (1; 287)	0.215	0.005
	Rate × Date	0.65 (3; 287)	0.584	0.007
	Method × Rate × Date	1.6 (3; 287)	0.189	0.016

.

Figure 1a presents ammonia (NH₃) fluxes in relation to digestate rate, application method, and year. NH₃ emissions increased significantly with the rising digestate rate in both years, with the most pronounced increases observed between 20 and 40 or 80 m³ ha⁻¹. The band application method consistently resulted in higher NH₃ emissions compared to the disc injector, with statistically significant differences observed for all rates except for the control (0 m³ ha⁻¹) and 20 m³ ha⁻¹. In 2021, no significant difference was found between the control and 20 m³ ha⁻¹, while all higher rates (40 and 80 m³ ha⁻¹) resulted in significantly increased emissions. In 2022, although no statistically significant differences were found between 0 and 20 m³ ha⁻¹ or between 40 and 80 m³ ha⁻¹, a significant difference was observed between the lower group (0–20 m³ ha⁻¹) and the higher group (40–80 m³ ha⁻¹). This suggests a possible threshold response in NH₃ emissions between 20 and 40 m³ ha⁻¹. Notably, the NH₃ flux following 80 m³ ha⁻¹ via band application exceeded three times the value recorded under disc injector at the same rate. Across both years, average NH₃ fluxes increased from approximately 420 µg m⁻² min⁻¹ (control) to over 3300 µg m⁻² min⁻¹ at the highest dose.

For CO₂ emissions (Table 6), all main effects were highly significant: method (F(1, 287) = 367.28, p < 0.001, η² = 0.561), rate (F(3, 287) = 111.28, p < 0.001, η² = 0.538), and date (F(1, 287) = 24.33, p < 0.001, η² = 0.078). The method × rate interaction (F(3, 287) = 46.66, p < 0.001) and method × date (F(1, 287) = 6.22, p = 0.013) were also significant, indicating that surface application intensified emissions under specific rates and timings.

Figure 1b displays carbon dioxide (CO₂) emissions in response to digestate rate and application method. CO₂ fluxes increased significantly with the rising digestate rate, and the effect of application method was statistically significant (p < 0.001). In both years, the band application method consistently resulted in higher CO₂ emissions than the disc injector unit. In 2021, Tukey’s HSD test revealed significant differences primarily between the lower-rate group (0–20 m³ ha⁻¹) and the higher-rate group (40–80 m³ ha⁻¹). In contrast, the 2022 dataset showed significant differences among all individual digestate rates, indicating a more continuous dose–response effect. When both years were evaluated together, statistically significant differences were found among all application rates, confirming a robust and consistent increase in CO₂ emissions with increasing digestate input.

CH₄ emissions were influenced by all three main effects (Table 6): method (F(1, 287) = 31.91, p < 0.001, η² = 0.100), rate (F(3, 287) = 14.28, p < 0.001, η² = 0.130), and date (F(1, 287) = 9.34, p = 0.002, η² = 0.032). No interaction terms reached statistical significance (p > 0.18), suggesting a consistent CH₄ response pattern across treatments. Slight increases were observed at higher doses, particularly under surface application.

Figure 1c shows methane (CH₄) fluxes in relation to digestate rate and application method. When data from both years were combined, the digestate rate had a significant effect on CH₄ emissions (p < 0.05), with a significant difference observed between the control (0 m³ ha⁻¹) and all other application rates. However, no significant differences were found among the individual rates of 20, 40, and 80 m³ ha⁻¹, indicating a possible threshold-like response. In 2021, CH₄ emissions increased with digestate rate under band application, and significantly higher fluxes were recorded at 40 and 80 m³ ha⁻¹ compared to the same rates under disc injection. This trend was not observed in 2022, where no statistically significant differences were found between application methods at any individual digestate rate. While CH₄ emissions under band application tended to be higher than under disc injection, this effect was less consistent than for NH₃ and CO₂. In 2022, CH₄ fluxes after 80 m³ ha⁻¹ applied via band exceeded 2600 µg m⁻² min⁻¹, while values under disc injection remained lower in absolute terms but not significantly different.

Figure 2 displays linear regressions between the digestate rate and the fluxes of NH₃ (a), CO₂ (b), and CH₄ (c), evaluated separately for application method and year. In all cases, emissions increased with higher digestate rates, though the strength of this relationship varied. In 2021, the increase was more gradual, especially for CH₄ and CO₂. In 2022, steeper trends were observed, particularly for ammonia under the band applicator, where R² exceeded 0.65. For CO₂, the band spreader in 2022 showed the strongest fit (R² = 0.7186). CH₄ fluxes showed consistent linearity only under the band method. These results confirm that both the rate and method of application strongly influenced gaseous emissions, with year-specific conditions further amplifying the response. All regressions, except for CH₄ under band spreader in 2022, showed R² values greater than 0.25, which in this context indicates a moderately strong relationship due to the high variability inherent in the field-based experimental setting.

Figure 3 shows the seasonal development of vegetation indices (NDVI, MSAVI, NDWI) for individual treatments (digestate rate and application method) during the years 2022 (a), 2021 (b), 2020 (c), and 2019 (d). Statistical comparisons of index values at specific time points are provided in Tables S1–S3 (see Supplementary Materials). In all four years, NDVI and MSAVI responded both to digestate application and biomass regrowth following mowing events. In particular, in 2022 (see Figure 3a), vegetation index values after the first cut remained considerably lower across all variants, likely due to dry conditions and limited regrowth. This trend was most pronounced for the NDWI index, which reached its lowest values in the I_80+0 variant and indicated increased water stress. The highest values of all three vegetation indices were recorded in 2022 during the period preceding the second cut, particularly in the I_40+40 and I_0+80 variants, where MSAVI reached values of up to approximately 0.95. From an overall perspective on vegetation development, the most unfavorable variant in terms of plant status appeared to be I_80+0, which exhibited the lowest mean values of all monitored vegetation indices (NDVI, MSAVI, NDWI) in all years except 2019. In 2019, the lowest values were recorded in the I_40+0 variant. Conversely, the highest mean values of NDVI and MSAVI were observed across years in the variants C, I_40+40, and I_0+80, suggesting that not only the digestate rate but also the timing and distribution of its application played a crucial role in the dynamics of vegetation development. Additionally, the I_20+20 m³·ha⁻¹ dose applied via disc injector consistently resulted in higher annual mean values of all vegetation indices—except for NDWI in the first year—when compared to the same dose applied using the band applicator. The NDWI index, which reflects plant water status, reached the highest values in 2019 and 2020 in the control treatment (C), whereas in subsequent years, its development followed a similar trend to NDVI and MSAVI—peaking in variants with split or late digestate application.

A three-way ANOVA (

Table 7. Summary of factorial ANOVA results for dry matter yield (DMY), net energy for lactation (NEL), and relative feed value (RFV); partial η² represents effect size; bold font denotes significant effects/interactions.

	Effect	F (df1, df2)	p-Value	Partial η2
DMY	Variant	16.58 (7, 288)	<0.001	0.287
	Cutting	4077.33 (2, 288)	<0.001	0.966
	Year	1236 (3, 288)	<0.001	0.928
	Variant × Cutting	8.31 (14, 288)	<0.001	0.288
	Variant × Year	1.01 (21, 288)	0.455	0.068
	Cutting × Year	567.35 (6, 288)	<0.001	0.922
	Variant × Cutting × Year	1.2 (42, 288)	0.194	0.149
NEL	Variant	8.68 (7, 288)	<0.001	0.174
	Cutting	2492.75 (2, 288)	<0.001	0.945
	Year	631.28 (3, 288)	<0.001	0.868
	Variant × Cutting	8.63 (14, 288)	<0.001	0.296
	Variant × Year	1.02 (21, 288)	0.443	0.069
	Cutting × Year	273.85 (6, 288)	<0.001	0.851
	Variant × Cutting × Year	1.42 (42, 288)	0.053	0.172
RFV	Variant	8.92 (7, 288)	<0.001	0.178
	Cutting	791.33 (2, 288)	<0.001	0.846
	Year	297.71 (3, 288)	<0.001	0.756
	Variant × Cutting	7.09 (14, 288)	<0.001	0.256
	Variant × Year	8.61 (21, 288)	<0.001	0.386
	Cutting × Year	717.19 (6, 288)	<0.001	0.937
	Variant × Cutting × Year	7.21 (42, 288)	<0.001	0.512

) revealed significant main effects of variant (F(7, 288) = 16.58, p < 0.001, η² = 0.287), cutting number (F(2, 288) = 4077.33, p < 0.001, η² = 0.966), and year (F(3, 288) = 1236, p < 0.001, η² = 0.928) on dry matter yield (DMY). Significant two-way interactions were observed for variant × cutting (F(14, 288) = 8.31, p < 0.001, η² = 0.288) and cutting × year (F(6, 288) = 567.35, p < 0.001, η² = 0.922), while the variant × year interaction was not significant. The three-way interaction was also non-significant (F(42, 288) = 1.2, p = 0.194). These results highlight strong temporal and phenological variation in DMY responses, especially across cuts. Tukey HSD post hoc tests showed that split applications, particularly the I_40+40 treatment, outperformed single-dose options across multiple years.

Figure 4 presents the dry matter yields for the individual digestate rate and application method variants across years and mowing events. The highest values were recorded for the 40+40 and 0+80 variants applied using the disc applicator, reaching 3.57 t ha⁻¹ and 3.55 t ha⁻¹, respectively. While the 0+80 variant involved a high single dose applied after the second cut, the 40+40 variant represented a balanced distribution of nutrients throughout the season and showed the most stable yield performance across individual cuts. When focusing on individual cuts, the first cut (except for the year 2020) was generally the most productive, contributing the most to total seasonal yield across all treatments. The 80+0 variant, despite consistently showing the lowest vegetation index values, still achieved a relatively high yield (3.49 t ha⁻¹). This discrepancy indicates a potential trade-off between short-term vegetation stress and cumulative seasonal productivity. In contrast, the control (C) variant reached the lowest average yield (3.10 t ha⁻¹), reflecting the absence of nutrient input in the form of digestate.

For net energy for lactation (NEL), all three main effects were again highly significant (Table 7): variant (F(7, 288) = 8.68, p < 0.001, η² = 0.174), cutting (F(2, 288) = 2492.75, p < 0.001, η² = 0.945), and year (F(3, 288) = 631.28, p < 0.001, η² = 0.868). Interactions between variant × cutting (F(14, 288) = 8.63, p < 0.001, η² = 0.296) and cutting × year (F(6, 288) = 273.85, p < 0.001, η² = 0.851) were significant. The three-way interaction showed borderline significance (F(42, 288) = 1.42, p = 0.053). This indicates that NEL was affected by both agronomic treatment and phenological stage, with seasonal patterns further shaped by interannual climate variation.

Figure 5 presents NEL across all treatments. The highest average NEL value was recorded for the 40+40 variant applied via disc injector (17.39 GJ ha⁻¹), followed by 0+40 and 0+80, both also using the disc injector. The control variant (C) exhibited the lowest average NEL (15.32 GJ ha⁻¹), with a statistically significant difference compared to digestate-treated variants (p < 0.05). NEL was significantly influenced by both the year and the cutting number (ANOVA, p < 0.001 for both). The first cut consistently reached the highest NEL (24.15 GJ ha⁻¹), exceeding the second cut by 13.78 GJ ha⁻¹ and the third by 8.23 GJ ha⁻¹ (Tukey HSD, p < 0.001). Similarly, the highest interannual values were observed in 2020 and 2021, while 2019 showed the lowest mean NEL.

In the case of relative feed value (RFV), all main effects were significant (Table 7): variant (F(7, 288) = 8.92, p < 0.001, η² = 0.178), cutting (F(2, 288) = 791.33, p < 0.001, η² = 0.846), and year (F(3, 288) = 297.71, p < 0.001, η² = 0.756). All interactions were also significant: variant × cutting (F(14, 288) = 7.09, p < 0.001), variant × year (F(21, 288) = 8.61, p < 0.001), cutting × year (F(6, 288) = 717.19, p < 0.001), and the three-way interaction (F(42, 288) = 7.21, p < 0.001). These results underscore a complex interplay among treatment, harvest timing, and climatic year, with RFV showing the highest sensitivity to multi-factor interactions among the tested parameters.

Figure 6 illustrates the relative feed value (RFV), a composite indicator reflecting forage digestibility and potential dry matter intake. The highest RFV was observed in the 20+20/band spreader treatment (95.9), followed by the 0+40/disc injector (94.1) and 80+0/disc injector (93.4) variants. Although the 40+40 treatment achieved the highest overall dry matter yield, its RFV was slightly lower (92.4), indicating that biomass quantity and forage quality are not directly correlated. Interestingly, the 80+0/disc injector variant, despite consistently exhibiting the lowest vegetation index values, still maintained a relatively high feed value. Statistical analysis confirmed that RFV significantly differed between years and individual cuts (ANOVA, p < 0.001). The third cut consistently recorded the highest RFV values, with a mean difference exceeding 15 units compared to the first cut (Tukey HSD, p < 0.001). In the first experimental year, a statistically significant difference in RFV was observed between the 20+20 variants applied using the band spreader and the disc injector (t-test, p = 0.02). However, this difference was not observed in subsequent years

Figure 7 illustrates the results of linear regression analysis between digestate dose (applied prior to each cut) and selected agronomic parameters. A consistent and statistically significant positive relationship (p < 0.001) was observed for dry matter yield (DMY) across all years and both cuts, with R² values ranging from 0.2914 to 0.7882, indicating moderate to very strong relationships. For net energy for lactation (NEL), the relationship was statistically significant in all cases (p < 0.05), with R² values between 0.1315 and 0.8060. While some models showed strong or very strong effects (e.g., R² = 0.8060 in 2019, cutting 2), others showed lower explanatory power (e.g., R² = 0.1315 in 2021, cutting 2). In contrast, the relationship between digestate dose and relative feed value (RFV) was less consistent and generally weaker. Statistically significant relationships of moderate strength were observed in four cases: R² = 0.4450 (2019, cutting 3), R² = 0.3312 (2021, cutting 2), R² = 0.3172 (2021, cutting 3), and R² = 0.2904 (2022, cutting 2). While most other cases remained weak or non-significant (R² < 0.25), this indicates a more variable response in terms of forage digestibility and the influence of additional sources of variability not directly related to nutrient input.

4. Discussion

According to the IPCC Sixth Assessment Report [35], methane (CH₄) of biogenic origin has a global warming potential (GWP₁₀₀) of 27. Nitrous oxide (N₂O), although not measured directly in this study, has a GWP₁₀₀ of 273 and is widely recognized as the most potent greenhouse gas associated with fertilized soils. Although ammonia (NH₃) is not classified as a greenhouse gas and has no assigned global warming potential under the IPCC framework, it plays a significant indirect role in climate dynamics. NH₃ volatilization contributes to secondary N₂O formation following atmospheric deposition, making it a relevant factor in the broader context of indirect greenhouse gas emissions [44]. Moreover, ammonia is closely linked to air quality degradation—over 81% of global NH₃ emissions originate from agricultural sources, particularly livestock production and fertilizer use [45]. These emissions contribute to acidification and eutrophication and represent a major pathway of nitrogen loss from agroecosystems [46]. Immediately after digestate application, peak greenhouse gas fluxes typically occur, as confirmed by previous studies examining the temporal dynamics of CO₂ and CH₄ in soils [47,48]. Therefore, this study was designed to investigate the immediate gas fluxes of NH₃, CO₂, and CH₄ in response to different digestate application methods and rates.

The volatilization of NH₃ was markedly influenced by both the applied dose and the digestate application technique. Lower emissions were consistently observed with disc injection compared to band spreading, with the disparity increasing at higher application rates. These findings align with those reported by Nicholson et al. [24]. In the study by Pedersen and Nyord [49], it was demonstrated that the use of disc injection led to a 38% reduction in ammonia emissions compared to the trailing shoe application. In our study, an even more substantial reduction was observed: disc injection reduced NH₃ emissions by more than 66% on average, despite the comparison being made against band application of digestate. The results of our study support the EU’s objectives under the NEC Directive to significantly reduce ammonia emissions, by up to 30% in some member states by 2030, as disc injection reduced NH₃ losses by more than two-thirds compared to band spreading. These findings underscore the critical importance of incorporating digestate into the soil as an effective strategy to mitigate ammonia volatilization. Czubaszek and Wysocka-Czubaszek [50] evaluated a dose of 30 m³·ha⁻¹, with a peak CO₂ emission of 211,248 µg·min⁻¹·m⁻². It should be noted that the application was carried out using the splash plate method. Although this specific dose was not used in our experiment, linear interpolation between the 20 and 40 m³·ha⁻¹ variants using band spreading resulted in an estimated CO₂ emission of 156,033 µg·min⁻¹·m⁻², which is 26.14% lower. The same study reports a maximum measured CH₄ flux of 2934 µg·m⁻²·min⁻¹ [50], which, when recalculated for a dose of 30 m³·ha⁻¹, closely matches our results of 2620.7 µg·m⁻²·min⁻¹. However, it should be noted that in our case, the digestate was applied using the band spreading method, not the splash plate technique. Korba et al. [51] reported that there were no statistically significant differences in CH₄ flux among application rates of 0, 20, and 40 m³·ha⁻¹ of digestate when applied using a disc injector. These findings are consistent with the results of our study—although a significant difference between 0 and 20 m³·ha⁻¹ was observed in 2021, this pattern did not persist when data from both years were evaluated together. Previous studies have found that CH₄ emissions following digestate application tend to be lower than those of other greenhouse gases, particularly CO₂ and N₂O [47,52]. This aligns with our results, which showed CH₄ fluxes to be consistently lower than CO₂ emissions across all treatments and application rates. The threshold-like response observed in our study may be related to microbial activity or soil conditions that influence the nonlinear character of CH₄ emissions. Soil moisture thresholds may play a crucial role in regulating microbial activity relevant to CH₄ dynamics, as suggested by a recent study [53]. Another important mechanism involves the interaction between functional genes and soil properties, particularly particle size and cation exchange capacity, which have been identified as major drivers of CH₄ fluxes [54]. The application of digestate has also been shown to induce short-term changes in microbial activity and alter the composition of soil microbial communities [55]. Furthermore, CH₄ emissions tend to be significantly higher in fine-textured soils and under elevated temperature conditions [56]. The threshold behavior in CH₄ fluxes following digestate application is most likely driven by the combined effects of microbial processes and soil physical–chemical properties, as evidenced by previous studies. Further investigation into these interactions would be valuable for understanding and managing methane dynamics in grassland systems. While CH₄ emissions were generally lower than those of CO₂ across all treatments, their potential contribution to climate forcing should not be underestimated, particularly in light of methane’s high GWP₁₀₀.

Although remote sensing is a highly useful and efficient tool for assessing vegetation status, there are still a relatively limited number of studies focusing on grassland production monitoring in Europe. This may be due to the high management intensity and the spatial and temporal heterogeneity typical of permanent grasslands [57]. Nevertheless, studies focusing specifically on digestate application on permanent grasslands are even scarcer. Dodin et al. [58], however, investigated digestate application on winter wheat using a trailing hose applicator and found that the most significant decrease in spectral reflectance occurred on the first day after application. Moreover, digestate was more easily detectable on sparsely developed canopies, while detection became increasingly limited as vegetation cover advanced. Möller and Müller [59] stated in their study that incorporating digestate into the soil is more beneficial in terms of nutrient availability. Our results support this, with the 20+20 m³·ha⁻¹ dose applied via disc injector consistently showing higher annual mean values of vegetation indices compared to the same dose applied using the band applicator, apart from NDWI in the first year, which may reflect the index’s higher sensitivity to short-term water stress and canopy moisture status rather than to biomass accumulation or vegetation greenness. Previous studies by various authors have demonstrated that digestate can serve as a full-value alternative to mineral nitrogen fertilizers on permanent grasslands without negatively affecting either biomass yield or forage quality [60,61,62]. When considering the cumulative yield across all years and harvests, the control variant (C), which received no digestate, was the only treatment with a significantly lower biomass yield compared to all fertilized variants. This finding is in agreement with Kovačić et al. [63], who reported a positive yield response of permanent grassland to digestate fertilization. From an overall perspective, the same dose of 20+20 m³·ha⁻¹ of digestate resulted in a higher biomass yield when applied via disc injector, although this difference was not statistically significant, which supports the findings of Holátko et al. [30]. From the perspective of nutrient supply, it is important to recognize that the nutrient content of the applied digestate varied between years. For example, in 2020, the lowest total nitrogen dose was applied (1.38 kg·m⁻³), yet this year recorded the highest average DMY. This outcome was likely driven by the most favorable climatic conditions observed during the four-year study period, highlighting that weather factors can sometimes outweigh nutrient inputs in determining biomass production. As highlighted by De Boever and De Campeneere [64], evaluating forage solely by its yield is insufficient; key qualitative parameters such as energy content and digestibility must also be considered, as they directly influence the nutritional value of the feed, particularly for high-producing dairy cows. Overall, the highest NEL values were observed in the first cuts, which supports the findings of Hrabě and Knot [65], who demonstrated that later harvest dates reduce the concentration of nutrients contributing to net energy availability in forage dry matter. Koláčková et al. [66] found that digestate application does not negatively affect forage nutritive value, which is in agreement with our findings, where the highest overall NEL values were recorded for the I_40+40 variant, i.e., 80 m³·ha⁻¹ applied in split doses. However, this variant exhibited only average values in terms of RFV, with the highest overall RFV recorded for the S_20_20 treatment. This discrepancy may reflect underlying shifts in forage composition, such as increased fiber and reduced protein content, reinforcing the importance of assessing both indicators. The above-mentioned outcome is consistent with the observations of Sezmiş and Gürsoy [67], who observed that a potential discrepancy may occur between these indicators and recommended that both RFV and NEL should be considered for a comprehensive evaluation of forage quality. These findings highlight that energy-based (NEL) and fiber-based (RFV) indicators reflect different dimensions of forage quality and should be interpreted together. Importantly, the lack of detrimental effects on forage quality, even at the highest digestate rates, underscores its potential for safe, high-rate application in grassland systems. This reinforces the role of digestate in sustainable nutrient strategies and carbon farming initiatives, where agronomic stability is as crucial as emission reduction.

5. Conclusions

This four-year study confirmed that both the rate and method of digestate application significantly influence greenhouse gas emissions, vegetation development, and forage quality on permanent grasslands. Hypothesis (a) was partially supported: emissions of ammonia (NH₃) and carbon dioxide (CO₂) increased proportionally with higher digestate rates, while the response of methane (CH₄) emissions suggests a possible threshold effect—elevated fluxes were observed in all fertilized treatments compared to the control, but further increases in dose did not significantly affect emission levels. Hypothesis (b) was confirmed, as disc injection resulted in lower emissions than surface band spreading, particularly at higher doses. Hypothesis (c) was also supported: split applications improved vegetation status, stabilized biomass yield, and enhanced forage energy value compared to high single-dose applications.

The combination of disc injection and split-dose fertilization proved to be the most effective strategy for reducing emissions while maintaining high agronomic performance. Specifically, split-dose disc injection at 40+40 m³·ha⁻¹ emerged as the most sustainable approach, offering a balanced solution for productivity and environmental protection. This aligns directly with the objectives of EU sustainability frameworks such as the European Green Deal and the Farm to Fork Strategy. For future research, further investigation into the threshold response of CH₄ emissions is recommended.