Impacts of Low Disturbance Liquid Dairy Manure Incorporation on Alfalfa Yield and Fluxes of Ammonia, Nitrous Oxide, and Methane

Surface applied liquid dairy manure application (i.e., broadcasting) after alfalfa (Medicago sativa L.) harvest is a common practice. Low disturbance manure incorporation (LDMI) may offer multiple benefits including lower ammonia (NH3), greenhouse gas (GHG) and hydrologic nutrient losses compared to broadcast. However, few studies have simultaneously quantified LDMI impacts on alfalfa yield, NH3 and greenhouse gas (GHG) fluxes. We measured NH3, nitrous oxide (N2O), and methane (CH4) fluxes for liquid dairy manure treatments applied to alfalfa plots for broadcast and LDMI over three seasons (2014 to 2016) in central Wisconsin, USA. There were minor differences in alfalfa yield and nitrogen (N) uptake across treatments and years. Shallow disk injection and aerator/band reduced NH3 loss by 95 and 52% of broadcast, respectively, however both substantially increased N2O fluxes (6 and 4.5 kg ha−1 year−1 versus 3.6 kg ha−1 year−1 for broadcast, respectively). The magnitude and timing of N2O fluxes were related to manure application and precipitation events. Average CH4 fluxes were similar among methods and increased with soil moisture after manure application. Results highlight the importance of quantitatively evaluating agri-environmental tradeoffs of LDMI versus broadcast manure application for dairy farms.


Introduction
Dairy manure is an important crop nutrient source, however careful management is needed to optimize nutrient use efficiency and minimize atmospheric and hydrologic losses (overland flow, leaching) associated with land application of manure. Cold climate dairies generate manure year-round but can have limited time windows and fields for application due to the short growing season and other cropping system limitations. Targeting manure applications to hay forages including alfalfa (Medicago sativa) in addition to annual crops like corn and grains provides additional land for manure application, recycles a portion of on-farm manure nutrients and creates multiple application windows after each harvest [1][2][3].
Since there is some risk of stand damage depending on how manure is applied and specific site characteristics (forage regrowth stage, soil moisture/compaction potential), there is considerable uncertainty around the benefits and challenges of applying manure to alfalfa in general. At low application rates and when applied before any regrowth, few negative yield impacts have been noted [1][2][3]. Another concern with applying manure on hay forage crops is forage quality/palatability, however Coblentz et al. [4] found no deleterious effects of applying liquid dairy manure on forage nutritive value. In addition, several studies indicate manure application to stands with optimum soil fertility are (from May 2014 to December 2016). The 10-year average (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) annual temperature and precipitation are 6.6 • C and 793 mm, respectively. The whole field was planted with alfalfa (Medicago sativa) (Nexgrow-6422Q) on 16 May 2013 at 19 kg ha −1 . Fifteen plots (7.3 × 12.8 m) were arranged in a randomized complete block design (3 blocks/replication with 5 treatments) with untreated areas between individual plots to allow equipment maneuvering and routine field operations to the extent possible. Plots were oriented lengthwise with the field and travel direction, perpendicular to the slope.

Treatment Details and Crop Management Practice
The five treatments in the study consisted of four manure application treatments and a no manure control that received triple superphosphate (0-46-0) and potash (0-0-60) at similar total P and K rates as the manure treatments. Both fertilizer and manure treatments were applied 7 August 2014, 30 June 2015 and 14 June 2016 after forage harvests; (second harvest in 2014, first harvest in 2015 and 2016). Manure was sampled in between treatment applications four times during each application period and tested for total N (TN), total P (TP), ammonium-N (NH 4 -N), K, and total solids content [35]. Manure application rate was approximately 74,800 L ha −1 and much higher than desired in 2014 due to a malfunctioning flow meter; in 2015 and 2016 it was 46,750 L ha −1 . Manure application contributed an average of 18 kg P, 100 kg K, 48 kg NH 4 -N, and 98 kg N ha −1 for the study. Detail on individual manure application treatments follows ( Figure S1): (1) Shallow Injection (Inject): 64 cm blades set at a 5 degree angle (Yetter Avenger, Yetter Manufacturing, Colchester, IL, USA), designed to cause minimal soil disturbance created 1.5-2 cm wide slits, manure was applied approximately 8-10 cm deep in these slits which were 30 cm apart. (2) Banded-aerator application (Aerator/Band): Manure was applied in bands about 5 cm wide through steel tubes 90-cm directly behind the tines of a rolling tine aerator (SAF Holland Aerway AWST). Aerator tines (no offset angle used), three per spindle, spaced 19 cm apart along the shaft, penetrated into the soil, creating slots approximately 2-cm × 20-cm at the soil surface narrowing down to a 2-cm wide point at the 18-cm depth. Tine slots were approximately 40 cm apart on center in the direction of travel. Manure slurry entered the slots for increased soil infiltration. (3) Banded application (Band): Manure was applied with the Aerator/Band applicator without the aeration tines, with hoses dragging across the soil surface. Manure bands were about 3-5 cm wide. (4) Broadcast application (Broadcast): Manure was broadcast with the Aerator/Band applicator raised approximately 40 cm above the soil surface so that manure provided complete coverage of the soil.
Harvest measurements were collected 3-4 times a season approximately every 28 days after the initial harvest in early or late June (24 June 2014, 25 June 2015, 9 June 2016) and weighed using a forage plot harvester/mower unit (F935, John Deere, Moline, IL, USA) equipped with digital load cells. Harvest passes were 1 m wide (10 cm cutting height) for each plot and harvest. Separate samples were hand-clipped from alfalfa immediately surrounding the harvest pass (cutting height = 10 cm), dried at 55 • C, and ground to pass a 1 mm sieve. These samples were then analyzed for N by high temperature combustion (Elementar VarioMax CN analyzer, Elementar Americas, Inc., Mt. Laurel, NJ, USA) and total minerals (P, Calcium (Ca), Magnesium (Mg)) after nitric acid digestion by inductively coupled plasma-optical emission spectrometer (ICP-OES) at the University of Wisconsin Soil and Forage Lab following standard procedures [36].

Ammonia and GHG Sampling and Analysis
Ammonia emission was measured in 2015 and 2016 using the dynamic chamber/equilibrium concentration technique a method that is well suited to small replicated plots and successfully used by others [27,37,38]. two types were placed in each chamber and in each ambient sampler holder, one with an acidified filter paper disk directly exposed to the air and the other with the filter paper disk 10 mm below a semipermeable Teflon membrane, requiring NH 3 to diffuse along a 10-mm path to the trap. Ammonia flux was calculated based on the micrometeorological law of resistance (using NH 3 concentrations to estimate required parameters). More detail on the approach, chamber design and flux computations are provided elsewhere [37][38][39][40]. Measurements started immediately after manure application and continued for seven separate periods through the third day. Day 1 measurements started immediately (Time 0) with successive periods starting approximately 1, 3, and 8 h (overnight) after application, followed by two 5-h measurements during Day 2 and a 10-h period on Day 3 (no overnight). Overnight emission between Day 2 and Day 3 was estimated from linear interpolation adjusted for measured temperature and wind conditions [37,41].
Nitrous oxide and CH 4 were measured using the static, vented chamber technique following the GRACEnet protocol [42]. Chambers consisted of stainless steel bases (61 × 38.1 × 10.2 cm) installed centered over a manure band or injection slit where applicable, two per plot on the west side of each plot outside of plot harvester pass locations. Bases were inserted as deep as possible (3.1 cm average height above soil surface to account for surface topography) and were moved and replaced after each cutting and harvest at which time they were alternately placed approximately 0.5 or 1.5 m from plot edge avoiding previously disturbed areas. Insulated and vented (3 mm ID and 40 cm long tubing) stainless steel lids with a height of 15.2 cm were sealed on top of bases during measurement by clipping the tops to the bases, the tops had weather stripping attached along the lip to serve as a gasket. At times when alfalfa was too tall to fit under the lid a 23 cm tall stainless steel, insulated extension was used in addition to the lid. Chamber construction was based on a design from R. Venterea (http://www.ars.usda.gov/pandp/docs.htm?docid=19008 (accessed on 15 April 2013)).
Gas samples were collected by inserting a 10-mL syringe into the chamber top sampling port, removing a sample, and immediately transferring the sample to a 5.9-mL capped, non-evacuated vial containing ambient air. Sample concentrations were later adjusted for the dilution by ambient air. Gas samples were collected four times for each measurement (0, 15, 30, and 45 min) over a 2-to 3-h period, typically between 900 h and 1200 h to approximate the mean daily temperature. Gas fluxes were calculated from the rate of change in concentration over the sampling period using linear regression, adjusted for theoretical flux underestimation [43] resulting from chamber deployment. Measurement began approximately one month prior to manure application in 2014, continued until soil freezing and snowfall and resumed after snowmelt. Sampling was done approximately weekly (more frequently after manure or rain, less frequently later in the season) from manure application 2014 through November 2016. Gas samples were analyzed via gas chromatography using an electron capture detector (micro-ECD) for N 2 O, a flame ionization detector (FID) for CH 4 , and an infrared gas analyzer (IRGA, LiCor 820, Lincoln, NE, USA) for CO 2 (Agilent 7890A GC System, Santa Clara, CA, USA). Annual cumulative gas fluxes were estimated by linear interpolation between sampling times.
Soil bulk density was measured (two 4.8 cm-diam. × 10 cm deep cores per plot) 3-4 times per year at the beginning of each sampling year and after harvest, manure application, or other activities that would be expected to affect bulk density. Bulk density was used in calculating theoretical flux underestimation [43] and adjusting N 2 O fluxes. Volumetric soil moisture (5-cm depth; Delta-T Devices Theta Probe) and soil temperature (5-cm depth; digital soil thermometer) were also measured in each plot during each gas sampling period.
Plots were arranged in a randomized complete block design with manure application method as the main treatment effect. The mixed modeling procedure (proc mixed) of the Statistical Analysis System (SAS) was used to the fixed effect of manure application method with block considered a random effect [44]. Differences in treatment means were performed using linear contrasts at p ≤ 0.10, given the high inherent variability associated with field gas flux measurements. Dependent variables included alfalfa dry matter yield, N uptake, and cumulative NH 3 and GHG fluxes. Dependent variables were tested (proc univariate) for normality and transformed (log 10 or square root) to achieve normality and/or homogeneity of variance as needed. Data are presented as back transformed values to maintain consistency across all variables. Pearson correlation coefficients (proc corr) and linear regression analysis (proc reg) were also performed for select variables.

Weather
The weather in 2014 was on average colder than the 10 year average (20%) ( Table 1), particularly during winter months. Temperatures for 2015 and 2016 were closer to longterm averages and slightly warmer mainly from warmer winter months (wintertime mean temperature was approximately 14% greater than the long-term average for Marshfield, WI). Growing season temperatures were close to long-term averages each year and precipitation was also close to average for 2015, but 31% and 16% greater than the long-term average in 2014 and 2016, respectively. Much of the additional precipitation was in early spring for 2014, both June and September of 2016 had rainfall exceeding the long-term average.

Manure Application Method Effects on Alfalfa Hay Crop Yield
In 2014 and 2016 manure application tended to increase alfalfa yields compared to the no manure control, however response was variable and not always significant ( Table 2). Manure application also generally resulted in greater N removal in 2016, though this too was variable with no clear yield effect associated with manure treatments ( Table 2). A few significant yield differences were noted before manure was applied in 2014 (Table 2) and after manure application in 2015. Inject had significantly lower yield than Broadcast, Band, or control. Compared to other studies, our results suggest relatively minor overall differences in alfalfa forage dry matter yield among all treatments. Some previous LDMI studies in alfalfa and grass hay crops indicate a possible yield reduction with shallow disk injection or banding, often attributed to root/crown or above ground plant damage [13,[21][22][23]33]. Mean N removal by alfalfa far exceeded the amount of annual N applied (from manure) as reported by others [1]. Results suggest a relatively low overall risk of yield reductions from LDMI assuming application is done under appropriate soil conditions and soon after harvest prior to regrowth.

Manure Application Effects on Ammonia Fluxes
Mean NH 3 -N flux rates were similar in 2015 and 2016 with greater flux rates closer to the time of application (Figure 1), as previously demonstrated by other studies [7,11,12,27,33,[45][46][47]. Broadcast application had substantially greater cumulative mean NH 3 fluxes in the first 3 days after application compared to other treatments, followed by Band, Aerator/Band and Inject with flux reductions of 30%, 52%, and 95% of Broadcast, respectively. Band and Aerator/Band impacts on NH 3 flux varied, decreasing NH 3 flux by an average of 18 and 24% of Broadcast in 2015 (not statistically significant) and 43 and 77% of Broadcast in 2016 (significantly lower for Aerator/Band), respectively (data not shown).  Lower NH 3 loss for Aerator/Band than broadcast or band has also been previously reported and is likely related to lower manure surface area contributing to lower NH 3 -N fluxes [22,25,27,48]. Wetter soil conditions in 2016 probably also contributed to deformation of aerator slots (more soil mass was stuck to aerator tines compared to drier conditions), which could have also contributed to more open slot surface area affecting NH 3 loss. Slightly warmer temperatures and higher wind speeds (Table 3) may have also increased Broadcast losses in 2016 since both can be significantly correlated with NH 3 -N losses [48,49]. Despite relatively high temporal and spatial variability, average NH 3 reductions in our study were similar to other trials [7,11,19,26,47,50,51].
Tracking N inputs from manure application permitted estimates of the fraction of applied manure NH 4 -N lost as NH 3 -N. During the 2016 season, approximately 100% of applied NH 4 -N and 50% of TN was lost from Broadcast plots; in 2015, 74 and 36% loss occurred. Similarly, other researchers have reported large N losses from manure application applied after hay crop harvest as NH 3 , ranging from 25 to 78% of applied NH 4 -N loss depending primarily on method of application, weather and soil conditions [7,19,[25][26][27]33]. Misselbrook [26] reported a 99% loss in June versus 58% with March application, further supporting the range of NH 3 loss we observed in our trial. Similarly, our results show the importance of injecting manure to maximize NH 3 retention in the form of NH 4 + but also indicate the importance of accounting for soil moisture and weather conditions (temperature, wind speed, amount/timing of precipitation) to help explain variation in the amount and timing of NH 3 fluxes for individual experiments and among multiple sites or regions.

Manure Application Effects on Nitrous Oxide Fluxes
Mean N 2 O fluxes were larger after manure application, consistent with previous experiments [18,20,26,33,50,51]. In general, larger N 2 O fluxes were associated with Inject and Aerator/Band treatments, whereas the no manure control had the lowest N 2 O fluxes (Figures 2-4). Smaller N 2 O fluxes occurred outside the manure application times and were associated with precipitation and higher soil water contents. Mean cumulative N 2 O-N fluxes were relatively low prior to manure application and during the late summer (Figures 2-4). Larger increases in N 2 O-N flux occurred approximately 16 days after manure application, coincident with the largest observed differences in N 2 O fluxes among treatments ( Figure 5). Given the low NH 3 emission and lack of significance at other times, rainfall events after manure application during the growing season appear to be important times for triggering elevated N 2 O fluxes.
Lower overall fluxes were associated with Broadcast and Band treatments, although Band and Broadcast did not differ significantly from Inject. Aerator/Band had the greatest mean cumulative N 2 O-N flux (mean = 6 kg ha −1 ), primarily from higher fluxes in 2014 and 2015 (though not significantly different from Inject). While the large flux range across years (2.0 to 9.0 kg ha −1 , 3.2 to 6.6 kg ha −1 , and 1.3 to 2.7 kg ha −1 across treatment in 2014, 2015, and 2016, respectively) is undoubtedly related to soil and weather conditions, N input differences from manure probably also contributed, particularly the excessive rate in 2014.
Denitrification and NO 3 − reduction to N 2 O and dinitrogen (N 2 ) in soils is microbiologically mediated with reaction rates related to NO 3 − concentrations, temperature, pH, redox and other physicochemical properties. While soil moisture variation is a known factor influencing N 2 O release, only weak correlations were noted between N 2 O fluxes and soil moisture (r = 0.05, p = 0.05) and temperature (r = 0.09, p = 0.0007) in our study. However, it is clear from other research that N 2 O formation can occur over a range of redox potentials (0 to 400 mV) and pH [52][53][54][55][56]. As previously mentioned, N uptake by alfalfa exceeded N applied annually and probably contributed residual NO 3 -N that was available for periodic microbial reduction to N 2 O.
Agriculture 2021, 11, 750 9 of 15 associated with precipitation and higher soil water contents. Mean cumulative N2O-N fluxes were relatively low prior to manure application and during the late summer (Figures 2-4). Larger increases in N2O-N flux occurred approximately 16 days after manure application, coincident with the largest observed differences in N2O fluxes among treatments ( Figure 5). Given the low NH3 emission and lack of significance at other times, rainfall events after manure application during the growing season appear to be important times for triggering elevated N2O fluxes.  Lower overall fluxes were associated with Broadcast and Band treatments, although Band and Broadcast did not differ significantly from Inject. Aerator/Band had the greatest mean cumulative N2O-N flux (mean = 6 kg ha −1 ), primarily from higher fluxes in 2014 and 2015 (though not significantly different from Inject). While the large flux range across   The cumulative quantity of N2O lost as a fraction of TN applied in our study is similar to Bouwman [29] where a range of 0 to 8% of applied TN was lost as N2O in a review of 180 experiments. Since the rate of manure application in the field to all plots was inadvertently doubled in 2014, this could be considered an outlier; ignoring 2014 data, the range of cumulative N2O losses narrows to 1.1 to 6.1% loss of TN applied.    The cumulative quantity of N2O lost as a fraction of TN applied in our study is similar to Bouwman [29] where a range of 0 to 8% of applied TN was lost as N2O in a review of 180 experiments. Since the rate of manure application in the field to all plots was inadvertently doubled in 2014, this could be considered an outlier; ignoring 2014 data, the range of cumulative N2O losses narrows to 1.1 to 6.1% loss of TN applied.  Cumulative N 2 O losses ranged from 1.1 to 12% of applied N with losses greater in 2014 and lower in 2016. In addition, mean cumulative N 2 O losses from manure treatments ranged from 22 to 100% greater than the no manure control for 2015 and 2016 and 50 to 360% greater than the control in 2014, these results are in line with the 975% greater emissions with injection found in the meta-analysis of Zhou et al. [56].

Manure Application Effects on Methane Fluxes
The cumulative quantity of N 2 O lost as a fraction of TN applied in our study is similar to Bouwman [29] where a range of 0 to 8% of applied TN was lost as N 2 O in a review of 180 experiments. Since the rate of manure application in the field to all plots was inadvertently doubled in 2014, this could be considered an outlier; ignoring 2014 data, the range of cumulative N 2 O losses narrows to 1.1 to 6.1% loss of TN applied.

Manure Application Effects on Methane Fluxes
Compared to N 2 O, average CH 4 fluxes generally peaked sooner after manure application if at all (Figure 6), with larger CH 4 fluxes in 2016 compared to 2014 and 2015. Mean CH 4 fluxes tended to be greater for Aerator/Band and Inject treatments, though not significant. Other studies have also indicated the overriding influence of soil and environmental factors on CH 4 emissions [50,57,58]. Greater losses post-manure application in 2016 could be due to wetter soil conditions in the injection zone that may have decreased soil redox potential. During warm, drier summer periods soils released minimal CH 4 and tended to act as a net sink for CH 4 , particularly in the dry period of the late summer-fall time period. Over the study, temperature and CH 4 fluxes were only weakly correlated (r = 0.14, p < 0.001). Soil moisture content was not correlated with CH 4 fluxes in 2015 (drier season), however they were significantly correlated in 2014 (r = −0.26, p < 0.001) and 2016 (r = −0.13, p = 0.003). Cumulative mean CH 4 fluxes averaged <800 g ha −1 year −1 with no significant application effects. Mean CH4 fluxes tended to be greater for Aerator/Band and Inject treatments, though not significant. Other studies have also indicated the overriding influence of soil and environmental factors on CH4 emissions [50,57,58]. Greater losses post-manure application in 2016 could be due to wetter soil conditions in the injection zone that may have decreased soil redox potential. During warm, drier summer periods soils released minimal CH4 and tended to act as a net sink for CH4, particularly in the dry period of the late summer-fall time period. Over the study, temperature and CH4 fluxes were only weakly correlated (r = 0.14, p < 0.001). Soil moisture content was not correlated with CH4 fluxes in 2015 (drier season), however they were significantly correlated in 2014 (r = −0.26, p < 0.001) and 2016 (r = −0.13, p = 0.003). Cumulative mean CH4 fluxes averaged <800 g ha −1 year −1 with no significant application effects.

Carbon Dioxide Equivalents and Global Warming Potential
Due to the presence of perennial vegetation in the chambers, plant respiration probably contributed to a large portion of CO2 fluxes. The lack of yield differences between treatments is therefore reflected in the few significant differences for cumulative

Carbon Dioxide Equivalents and Global Warming Potential
Due to the presence of perennial vegetation in the chambers, plant respiration probably contributed to a large portion of CO 2 fluxes. The lack of yield differences between treatments is therefore reflected in the few significant differences for cumulative CO 2 flux between treatments within years (data not shown). We did find Inject cumulative flux to be slightly yet significantly lower in 2015 than Broadcast and Band possibly related to significant yield differences after manure application although it remained similar to Aerator/Band and the control. Soil and plant/root disturbance may also have contributed to slightly lower CO 2 flux in other years with Inject although not significant. On average, across years, there were some significant differences of note, Aerator/Band had the highest cumulative flux (33,148 kg ha −1 year −1 ) while Broadcast and Band had intermediate (32,622 and 32,043 kg ha −1 year −1 , respectively), and Inject and control had the lowest (30,244 and 31,269 kg ha −1 year −1 , respectively). Using CO2 equivalents [14] to calculate global warming potential (GWP) (where 265 and 28 are used for N2O and CH4, respectively, and 1% of NH3-N is considered converted to N2O), there were no differences in GWP among treatments for any year. Average GWP followed a similar pattern as CO 2 likely due to the dominance of CO 2 emission compared to other gases and its large influence on the GWP calculation.

Conclusions
Impacts of liquid dairy manure application method after alfalfa harvesting on NH 3 , N 2 O, and CH 4 fluxes using broadcast and LDMI methods were investigated for three field seasons in central Wisconsin on a somewhat poorly drained silt loam soil. Results indicated that application method had a relatively limited effect on dry matter yields. Cumulative NH 3 fluxes were much greater for Broadcast with intermediate losses for Aerator/Band and Band and lowest for the Inject system. Whereas NH 3 fluxes peaked immediately after manure application and approached a steady state after two days, N 2 O fluxes peaked approximately two weeks of application and were triggered by precipitation events. Aerator/Band and Inject had the largest cumulative N 2 O fluxes and were not different. Cumulative N 2 O fluxes for Band and Broadcast were numerically lower on average than Aerator/Band and Inject. Methane fluxes were small in comparison to NH 3 and N 2 O and did not differ by application method. Results show Band mitigated both NH 3 and N 2 O fluxes with intermediate GWP. While Inject maximized NH 3 conservation a portion of this N was lost as N 2 O but also had lower CO 2 fluxes, reducing GWP; Aerator/Band had the greatest N 2 O flux and GWP. Banding and injection of manure to alfalfa stands after harvest for the silt loam soils of central Wisconsin appear to be viable options to increase N use efficiency and mitigate GHG emissions with little impact on overall yield potential. Our results also highlight the trade-offs between NH 3 and N 2 O loss vulnerabilities and a need to account for such management practices in farm nutrient budgeting and developing future nutrient management tools.