Cultivating Salix Viminalis in Agricultural-Riparian Transition Areas to Mitigate Agriculturally Derived N2O Emissions from Potato Cropping Systems on Prince Edward Island

Cultivating shrub willow (Salix viminalis) in agricultural-riparian transition areas has been proposed as a strategy for mitigating elevated riparian nitrous oxide (N2O) emissions in agricultural regions. Nitrogen-based fertilizers are water soluble, enter riparian areas through surface runoff and subsurface lateral flow, and are converted to N2O by incomplete anaerobic denitrification. Salix buffer strips can intercept and recycle fertilizer nitrate (NO3−) into their biomass and/or promote complete denitrification, reducing N2O emissions. We investigated the impact of Salix viminalis buffers on N2O emissions relative to grassed buffers and upslope cultivated fields in potato rotations at 5 research sites across Prince Edward Island (PEI), Canada. Greenhouse gas (N2O, CO2, CH4) flux at the soil-atmosphere interface was measured using non-steady-state static chambers in 2018 and 2019. NO3− exposure, soil temperature, and soil moisture content were quantified. Agricultural-riparian Salix significantly reduced N2O emissions even when high NO3− inputs occurred and following precipitation events. Mean cumulative seasonal reductions of 1.32 kg N2O–N ha−1 (− 0.02 to 6.16 kg N2O–N ha−1) were observed in Salix relative to cultivated fields; however, they were not significantly different than grass. The mean cumulative average global warming potential of Salix was 613 kg CO2e ha−1 lower than cultivated fields, with reductions of up to 2918 kg CO2e ha−1. Differences in N2O flux between vegetation types were the greatest influencing factor. No hot moments of N2O emission were observed in Salix following high rainfall events, which coincided with up to 95% decreases in N2O emissions in Salix relative to cultivated fields.

Abstract Cultivating shrub willow (Salix viminalis) in agricultural-riparian transition areas has been proposed as a strategy for mitigating elevated riparian nitrous oxide (N 2 O) emissions in agricultural regions. Nitrogen-based fertilizers are water soluble, enter riparian areas through surface runoff and subsurface lateral flow, and are converted to N 2 O by incomplete anaerobic denitrification. Salix buffer strips can intercept and recycle fertilizer nitrate (NO 3 − ) into their biomass and/or promote complete denitrification, reducing N 2 O emissions. We investigated the impact of Salix viminalis buffers on N 2 O emissions relative to grassed buffers and upslope cultivated fields in potato rotations at 5 research sites across Prince Edward Island (PEI), Canada. Greenhouse gas (N 2 O, CO 2 , CH 4 ) flux at the soil-atmosphere interface was measured using non-steady-state static chambers in 1 3 Vol:. (1234567890) This includes the retention of agriculturally derived nutrients, agro-chemicals, and sediments as well as removal through processes such as denitrification. As such, management of riparian areas is considered a beneficial management practice (BMP) for water quality protection (Schultz et al., 1995) and they are increasingly mandated in many regions, including Prince Edward Island (PEI), Canada.
Riparian areas are frequently sites of elevated greenhouse gas (GHG) emission (Audet et al., 2013) within the landscape-a direct result of wet conditions and large amounts of slowly decomposing organic matter in riparian soils-and particular concerns exist in agricultural regions. While riparian areas are effective at reducing shallow groundwater nitrate (NO 3 − ) concentrations and preventing it from entering waterways (Young & Briggs, 2005), incomplete denitrification can result in elevated nitrous oxide (N 2 O) emissions (Audet et al., 2013;Drewer et al., 2012;Hellebrand et al., 2008;Jacinthe & Vidon, 2017;Lutes et al., 2016). This is concerning, since N 2 O is a potent greenhouse gas with a global warming potential 298 times that of carbon dioxide (IPCC, 2013). Wet, anaerobic conditions, often prevalent in riparian soils, promote denitrification as a major pathway by which groundwater NO 3 − levels are reduced. Incomplete denitrification results in the release of elevated levels of N 2 O into the atmosphere (Audet et al., 2013;Fisher et al., 2014;Jacinthe & Vidon, 2017). Thus, there is concern that efforts to protect riparian water quality may be at the expense of N 2 O production when NO 3 − is present. Agriculture was responsible for approximately 8.2% of anthropocentric GHG emissions in Canada in 2020 and 75% of national N 2 O emissions (Environment and Climate Change Canada, 2022). Nitrogen fertilizer application, which has increased by 89% since 2005, was responsible for 21% of total agricultural emissions in 2020 and, since 2005, annual GHG emissions from Canadian agricultural soils have increased by 8 Mt CO 2 eq (Environment and Climate Change Canada, 2022). Nitrogen is an essential plant macronutrient that is frequently limiting to plant growth since most soils contain insufficient plantavailable N (Havlin et al., 2013;Zebarth & Rosen, 2007). Fertilization of agricultural soils with synthetic N that is transformed to NO 3 − by the soil biota has contributed to dramatically increased crop yields; however, this has been accompanied by high losses of NO 3 − to the environment via leaching and denitrification (Erisman et al., 2008). In humid and sub-humid climates such as PEI, soil N 2 O emissions in agricultural systems occur mainly due to microbial denitrification (Aulakh et al., 1984;Gauder et al., 2012;Granli & Beckman, 1994;Kavdir et al., 2008). Fertilizer-derived NO 3 − serves as an alternate electron acceptor in microbial respiration under conditions of limited oxygen supply and is a major contributor of agricultural N 2 O emissions . Accumulation of NO 3 − in the soil also inhibits the reduction of N 2 O to N 2 , increasing the ratio of N 2 O/ N 2 resulting from denitrification (Cho et al., 1997a, b;Gillam et al., 2008;Weier et al., 1993;Zebarth et al., 2012). Since water-filled pore space (WFPS) and NO 3 − availability are major controllers of N 2 O emissions Ruser et al., 2001), and NO 3 − is highly mobile within the soil (Havlin et al., 2013), leakage of fertilizer-derived NO 3 − into wetter riparian areas can increase the likelihood of it entering the denitrification pathway and situations favoring incomplete denitrification, resulting in the release of N 2 O. This explains the observance of high N 2 O emissions in riparian areas in some agricultural regions.
PEI, Canada, is the province with the largest potato production area in Canada (33,720 ha in 2016). This represents 20.8% of total PEI cropland and 24.2% of total cropland dedicated to potatoes in Canada (Statistics Canada, 2017). Potatoes require high fertilizer N inputs to produce the tuber yield and size profile desired by the processing industry, in part due to their shallow root systems and corresponding low N use efficiency (Zebarth & Rosen, 2007). Fertilizer rates for potatoes in Atlantic Canada can exceed 200 kg N ha −1 , with up to 55% of this being lost to the environment (Belanger et al., 2001) and leaching losses of NO 3 − from potato fields have been observed to sometimes exceed 171 kg N ha −1 (Zebarth & Rosen, 2007). NO 3 − leaching from potato fields and contamination of both groundwater and surface waterways are known to be problematic on PEI Liang et al., 2019;Zebarth et al., 2015). Management requirements of potatoes, such as the high degree of soil disturbance, as well as soil and climatic conditions specific to PEI (highly permeable sandy surface soils with low natural fertility and organic matter contents), contribute to high fertilization requirements. High annual rainfall, particularly in the non-growing season, also increases Vol.: (0123456789) the potential for NO 3 − leaching losses (Zebarth et al., 2015). As a result, the risk of elevated riparian N 2 O emissions on PEI is great.
Cultivating Salix spp. (Salix) bioenergy buffer strips in riparian areas or downslope field edges in agricultural-riparian transition zones has been proposed as a potential beneficial management practice (BMP) for protecting surface water quality while simultaneously reducing agriculturally derived N 2 O emissions (Borjesson, 1999). It is thought that by intercepting and recycling NO 3 − from upland fertilizer application as it is lost through leaching or surface runoff, downslope riparian Salix buffer strips will remove NO 3 − via plant uptake for biomass accumulation and by complete denitrification, thereby reducing the amount at risk of incomplete denitrification and N 2 O loss in riparian areas. Bioenergy Salix are managed for maximum nutrient uptake and biomass accumulation through regular coppicing on a 3-year cycle (Amichev et al., 2014), which can maximize NO 3 − uptake and consequent N 2 O reductions. Salix have long growing seasons, rapid growth, extensive root systems, continuous soil cover, and a 25-year lifespan, which make them good candidates for limiting soil NO 3 − accumulation by maximizing and sustaining plant uptake (Borjesson, 1999;Bressler et al., 2017;Dimitriou et al., 2012;Kavdir et al., 2008).
The effectiveness of Salix at reducing groundwater NO 3 − levels is well documented, even when large inputs are occurring (Aronsson et al., 2000;Borjesson, 1999;Bressler et al., 2017;Dimitriou et al., 2012;Ferrarini et al., 2017;Heller et al., 2003;Young & Briggs, 2005), and Salix biomass production is directly enhanced by soil NO 3 − supply, indicating that plant uptake is a major removal pathway (Volk et al., 2006). Harvested biomass N removal in Salix averages around 25 kg N ha −1 year −1 (Amichev et al., 2014). Borjesson (1999) estimated that Salix buffer strips planted between annual cropped fields and streams could prevent 70 kg N ha −1 year −1 from entering waterways when nitrate inputs were greater than 15 kg N ha −1 year −1 ; 2/3 of this was due to plant uptake with the remaining 1/3 lost to denitrification (Borjesson, 1999).
The impact of Salix on GHG emissions at the soil-atmosphere interface is less well studied, in part because the spatial and temporal distribution of GHG emissions varies greatly across the landscape (Audet et al., 2013;Bressler et al., 2017;Lutes et al., 2016;Pacaldo et al., 2014a, b;Whitaker et al., 2018). Soil GHG emissions are affected by factors such as soil texture, hydrology, land use and management, climate, soil microbes, and vegetation (Borjesson, 1999;Hellebrand et al., 2008;Lutes et al., 2016). Most studies have found that Salix reduce N 2 O and increase CO 2 emissions at the soil-atmosphere interface compared to other vegetation and/or land use types, while their impact on CH 4 is mixed (Borjesson, 1999;Bressler et al., 2017;Drewer et al., 2012;Gauder et al., 2012;Kavdir et al., 2008;Lutes et al., 2016;Pacaldo et al., 2014b;Smialek et al., 2006;Whitaker et al., 2018). However, most studies have focused on Salix bioenergy crops grown as a field scale replacement for other crops. N 2 O emissions at the soil-atmosphere interface from Salix have been found to vary greatly (Palmer et al., 2014;Whitaker et al., 2018). To date there have been very few studies related to the impact of Salix on shallow soil GHG emissions when grown as buffer strips in riparian areas, and none specifically on their impact in riparian areas downslope of potato cropping systems on PEI. Since wet conditions and large quantities of decomposing organic matter in riparian soils also promote elevated CO 2 and CH 4 emissions, while these gasses are not the primary focus of this study, there is a need to measure the impact of Salix on these other GHG gasses in riparian settings to allow for a more complete GHG accounting.
Previous field studies conducted as part of the Bioeconomy Crop Initiative with the PEI Department of Agriculture on PEI found that Salix viminalis was effective at carbon and nutrient sequestration on field edges bordering riparian zones on PEI, producing higher aboveground biomass yield with greater N concentration than other Salix varieties studied (Gooijer et al., 2011;Government of PEI, 2010). Coppiced Salix viminalis produced annual aboveground biomass yields of up to 23.4 tDM ha −1 year −1 and accumulated N at a rate of 160 kg ha −1 year −1 when both above and belowground biomasses were considered (Pharo, 2012, as cited in Lantz et al., 2014Schroeder, 2019;Wright, 2012, as cited in Lantz et al., 2014. For that reason, Salix viminalis was selected for this GHG study. Since riparian areas within 15 m of waterways are protected on PEI, this study occurred on downslope field edges in agricultural-riparian transition zones, just outside the protected 15-m riparian buffer, as a proxy for the riparian area.
The goal of this study was to determine the effect of Salix viminalis buffers on GHG emissions at the soil-atmosphere interface in agricultural-riparian transition zones downslope of potato cropping systems on PEI. The main objective was to compare soil N 2 O, CH 4 , and CO 2 emissions between agriculturalriparian Salix viminalis buffers, agricultural-riparian grass buffers, and upslope agricultural fields in 3-year potato rotations. This included quantifying the effects of increased distance from the agricultural field and wetting events on buffer N 2 O emissions. A secondary objective was to measure ancillary environmental parameters such as soil moisture, temperature, and NO 3 − availability to help identify controllers of N 2 O emissions. Our hypothesis was that Salix viminalis buffers would reduce N 2 O emissions and elevate CO 2 and CH 4 emissions relative to cultivated fields in potato rotation and grass buffers. Further, we hypothesized the main drivers of N 2 O flux would be soil NO 3 − availability and soil moisture content.

Description of the Study Sites
The experiment was conducted in 2018 and 2019 on Prince Edward Island (PEI), Canada (lat. 46°25′N, long. 63°00′W) at five study sites distributed across the island in agricultural-riparian transition zones and their upslope agricultural fields. PEI is located in the Gulf of St. Lawrence and has a humid-continental climate with a mean annual temperature of 5.7 ℃ and annual rainfall of 887 mm (Government of Canada, 2019a). Total precipitation over the 2018 and 2019 study intervals from early May to mid-November was 815 mm and 835 mm, with average spring (May/ June), summer (July/August), and fall (September/ mid-November) rainfalls of 275, 109, and 431 mm in 2018 and 233, 146, and 456 mm in 2019 (Government of Canada, 2019b). Spring and fall precipitations were higher than average, while summer precipitation was below average in both years. Spring air temperatures were lower than average (10.8 and 11.2 ℃), while summer (19.7 and 19.5 ℃) and fall were above average in both study years (9.9 and 10.3 ℃).
Across all sites, mean soil temperature was 15.7 ℃, with higher recorded temperatures in the cultivated fields (mean 16.3 ℃) than the Salix (mean 15.5 ℃) and grass (mean 15.6 ℃) buffers. Mean soil moisture content was 28.4% with higher overall values recorded in the Salix (mean 30.2%) and grass buffers (mean 32.2%) than the cultivated fields (mean 20.7%).
PEI surface soils tend to be coarse textured (sandyloam), well drained, acidic, low in organic matter with low natural fertility. Soil types at the study sites were mapped to Charlottetown or Aldberry Series with orthic humo-ferric podzols in the Canadian System of Classification being the dominant subgroup, moderately coarse to coarse texture on gently undulating to rolling relief (2-15%). Soils are mapped as well to moderately well drained, although the water table varied from deep to near the soil surface for part of the growing season at the various sites (Mac-Dougall et al., 1988). Underlaying this, at a depth of approximately 1-15 m, is a less permeable basal till layer that can result in a perched water table during periods of high precipitation (Zebarth et al., 2015). The bedrock consists of fractured sandstone. All sites were located adjacent to fields cultivated to potato crop rotations.
The experiment used a randomized complete block design with three treatments: Salix buffer (with 2 location sub-treatments: row 1 and row 3), grass buffer, and upslope cultivated agricultural field. Treatments were replicated three times at each of the five experimental sites. The study sites were established in 2016 (site A and site B), 2017 (site E and site C), and 2018 (site D); therefore, this experiment occurred during years 3 and 4, 2 and 3, and 2 post-establishment (online resource 1, Table I). Each site consisted of 10-m-wide Salix viminalis and grass buffer plots planted on the agricultural field edge adjacent to the riparian zone. Salix viminalis buffers were planted in 4 rows with 2-m row spacing and 0.5-m or 0.75-m inrow plant spacing.
This study occurred prior to any coppicing activity. The grass buffer and grass between the Salix rows were managed by mowing approximately biweekly; however, no fertilizer or irrigation was applied. The agricultural field was managed by the producer as per normal cropping practices for the crop being grown as part of a 3-year rotation including 1 year of potatoes (online resource 1, Table II).

Monitoring of Gas Fluxes
Soil gas fluxes were measured approximately biweekly using the static chamber method as described in Burton et al. (2008) and Burton (2013) from May to November in 2018 and 2019 in order to evaluate treatment effects on GHG emissions. Site A underwent weekly measurements of soil gas fluxes to better evaluate temporal variation in GHG emissions and was selected based on its location which was favorable for more frequent sampling. Measurements captured the entire Salix growth season, from budding in the spring to leaf drop in the fall, over 2 calendar years. Each non-steady-state static chamber consisted of a collar made of polyvinyl chloride (PVC) pipe (diameter 20.3 cm; height 10 cm) with a removable PVC chamber (diameter 20.3 cm; height 15 cm) that fitted over each collar. Chambers were covered with reflective foil-faced insulating bubble wrap, had closed cell foam gaskets consisting of closed-cell foam weather-stripping tape installed around the bottom to ensure an airtight seal with the collar, and contained a butyl rubber suba seal sampling port and plastic air vent tube (diameter 0.4 cm; length 7.5 cm). Total volume of the chamber was 5.0 L, while volume of the protruding collar was approximately 2.0 L but varied throughout the season, covering a soil area of 315 cm 2 .
Twelve collars were installed at each study site with the following distribution: three in the upslope agricultural field perpendicular to the buffer area, three in the grass buffers distributed to reflect the width, three in row 1 of the Salix buffer, and three in row 3 of the Salix buffer. Additionally, in 2019, three extra collars were installed at each study site between the Salix buffers and the agricultural field. Collars were installed in early spring, a minimum of 24 h before initial GHG flux measurements were taken and remained permanently installed at the sites throughout the season. Collars that were damaged or otherwise required replacement throughout the season were also installed a minimum of 24 h before GHG flux measurements were obtained. Field collars were removed during planting and harvest and subsequently replaced. Collars in fields in the potato crop phase of the rotation were placed in the ridge row location. Collars were sunk into the ground with roughly 5 cm protruding and collar height was measured on a monthly basis in order to calculate individual collar headspace volume.
Twenty-milliliter disposable nylon syringes (Becton-Dickinson 14-823-2B) fitted with 25 gauge, 5/8″ length Luer-Lok tip needles (Becton-Dickinson 14-826AA) were used to obtain 20-mL air samples from the headspace of each chamber at 0, 10, 20, and 30 min after deployment. Samples were subsequently injected and stored in 12-mL preevacuated exetainers (Labco Limited 739 W) fitted with butyl rubber septa. Air samples were stored in a cool dark location until they were shipped to the Greenhouse Gas Lab (Dalhousie University, Faculty of Agriculture) in Truro N.S. for analysis.
Gas samples were analyzed for N 2 O, CO 2 , and CH 4 using a Varian Star 3800 Gas Chromatograph (Varian, Walnut Creek, CA) fitted with an electron capture detector (N 2 O), thermal conductivity detector (CO 2 ), and a flame ionization detector (CH 4 ). Samples were transferred from the exetainers to the GC using a Combi-PAL Autosampler (CTC Analytics, Zwingen, Switzerland). As described in Snowdon et al. (2013), the electron capture detector was operated at 350 ℃, 90% Ar, 10% CH 4 carrier gas at 20 mL min −1 , Haysep N 80/100 pre-column (0.32 cm diameter × 50 cm length), and Porapak QS 80/100 mesh analytical columns (0.32 cm diameter × 200 cm length) in a column oven operated at 80 ℃. The pre-column was used in combination with a four-port valve to remove water from samples. The thermal conductivity and flame ionization detectors were operated in series at 130 ℃ and 200 ℃, respectively, pre-purified helium (He) carrier gas at 20 psi, Haysep N 80/100 mesh (0.32 cm diameter × 50 cm length) pre-column, followed by a Porapak QS 80/100 mesh (0.32 cm diameter × 200 cm length) analytical column maintained at 80 ℃.
Individual gas fluxes (F c ) for N 2 O, CO 2 , and CH 4 (g N 2 O-N ha −1 day −1 , kg CO 2 -C ha −1 day −1 , and g CH 4 -C ha −1 day −1 ) were calculated from changes in each gas concentration within the static chamber according to Hutchinson and Livingston (1993): with, as described by Snowdon et al. (2013), dC/dt being the rate of change in N 2 O, CO 2 , or CH 4 concentration (mol mol −1 h −1 ), A being the surface area (m 2 ) of the collar, V c being the total volume (L) of the enclosure (chamber volume plus collar volume), M mol being the molar mass of N 2 O-N, CO 2 -C, or CH 4 -C (g mol −1 ), and V mol being the volume of 1 mol of N 2 O, CO 2 , or CH 4 (L mol −1 ) inside the chamber corrected for temperature using the ideal gas law. The rate of change in concentration in the headspace (dC/dt) was calculated using a simple linear regression of the gas concentrations versus time over the deployment period and all flux values were expressed as g N 2 O-N ha −1 day −1 , kg CO 2 -C ha −1 day −1 , and g CH 4 -C ha −1 day −1 . Positive F c values indicated the soil was emitting the gas, while negative F c values indicated the soil was acting as a sink for the gas. Cumulative growing season gas fluxes were calculated for each individual chamber by linear interpolation between sampling dates . This assumed that gas fluxes measured represented average daily flux and that growing season emissions are a continuous function of daily flux estimates. Cumulative chamber emission values were averaged across chambers within a treatment to determine mean cumulative emission values for each treatment and sub-treatment. N 2 O and CH 4 were converted to CO 2 equivalents (CO 2 e) using conversion factors of 298 and 25 to determine the global warming potential (GWP) of greenhouse gas emissions at the soilatmosphere interface. CO 2 was not included in the GWP calculations since CO 2 contributions to global warming potential in soils where active photosynthesis is occurring are assessed as a change in carbon stocks rather attempting to measure both soil CO 2 emissions and photosynthesis. Changes in C stocks for Salix growth in these plots is dealt with in a subsequent paper.

Soil and Atmospheric Characteristics
Each sampling event included measurements next to each collar of soil surface temperature (depth 6 cm) using a portable soil temperature probe (Cole-Parmer 90,090-06) and surface moisture content (depth 6 cm) using a HH2 moisture meter and attached ML3 ThetaProbe Soil Moisture Sensor (Delta-T devices, Cambridge England). Air temperature (℃) and humidity (%) were also recorded within 10 cm of each collar using an Amprobe Digital Thermohygrometer (model TH-2).
Nitrate exposure (NE) was calculated by obtaining measurements of shallow soil NO 3 − acquired by inserting a single Anion Plant Root Simulator (PRS) probe (Western Ag Innovations, Saskatoon) vertically into the soil within 12 in of each collar. Each anion PRS probe was buried to a depth of 6 cm and had an absorbing surface area of 17.5 cm 2 . Probes were exchanged approximately biweekly over the field season to obtain a continuous assessment of nutrient supply rate, with the three replications being combined for each treatment (field, grass, Salix row 1, Salix row 3) for each of the sites and deployments, and sent to Western Ag Innovations for analysis of NO 3 − , determined colorimetrically by automated flow injection analysis system. Each determination results in a flux or NO 3 − supply rate to the surface of the probe (μg nutrient/10 cm 2 ion-exchange membrane surface area/time of burial). Nitrate exposure (NE) was calculated as the sum of sequential burials at a location over the monitoring period and expressed as μg nutrient/10 cm 2 ion-exchange membrane surface area.
Daily precipitation data for 2018 and 2019 was retrieved from the New Glasgow PEI weather station, the nearest Government of Canada maintained weather station to site A, located 16 km from the site (Government of Canada, 2019b).

Data Analysis
Statistical analysis was performed using JMP software for Windows version 14 (JMP ®, Version 14. SAS Institute Inc., Cary, NC, 1989-2019. Descriptive statistics, one-way analysis of variance (ANOVA), and regression models were used to analyze the data. A one-way ANOVA was done to assess the effect of treatment (Salix buffer, grass buffer, cultivated field) on soil N 2 O, CO 2 − , and CH 4 fluxes. Where ANOVA demonstrated significant treatment effects, the Student t test was used to compare mean GHG emissions between treatments. Statistical significance was determined at p > 0.05. Linear regression was used to determine the relationship between soil GHG emissions and environmental parameters. Linear regression was used to determine the relationship between NE and cumulative N 2 O emissions over the monitoring period. Descriptive statistics were used to qualitatively observe differences between crop type on GHG emissions.

Results
Soils at all research sites were net sources of N 2 O over the monitoring period with observed daily fluxes ranging from − 10.2 g to 400 g N 2 O-N ha −1 day −1 (mean 5.65 g N 2 O-N ha −1 day −1 ). Observed N 2 O flux was often low, with very few emissions occurring below 12 ℃. Greatest N 2 O emissions occurred in the cultivated fields, particularly those cultivated to potatoes, sudangrass, and black peas, with most emission events occurring above 12 ℃. There was a positive relationship between soil moisture content and N 2 O emissions, which was most pronounced in fields cropped to potato (r 2 : 0.122; p = 0.0001), which also had the highest N 2 O emissions (Fig. 1a). Available soil NO 3 − , as measured by NE, explained the greatest amount of the variation in N 2 O emissions (r 2 : 0.582; p < 0.0001) (Fig. 1b).
Cumulative N 2 O flux exhibited considerable between-site and within-site variability. Cumulative seasonal emissions within treatments across all sites and years ranged from − 0.039 to 6.31 kg N 2 O-N ha −1 (mean 0.46 kg N 2 O-N ha −1 ) with higher emissions occurring in the cultivated agricultural fields relative to the grass and Salix buffers ( Fig. 2(a)).
In seven of nine site years, there was a significant effect of treatment on N 2 O flux, with emissions in the cultivated fields being significantly higher than the Salix or grass buffers (p = 0.0001), but no significant difference between the two buffer treatments. Mean cumulative seasonal reductions in N 2 O emissions of 1.3 kg N 2 O-N ha −1 (with maximum reductions of 6.2 kg N 2 O-N ha −1 observed at site A in 2019) were observed in the Salix buffers relative to the cultivated fields, with mean cumulative seasonal reductions in N 2 O emissions of 1.4 kg N 2 O-N ha −1 in the grass relative to the cultivated fields (with maximum reductions of 6.1 kg N 2 O-N ha −1 observed at site A 2019).
There was also a site impact on N 2 O emissions ( Fig. 2(b)). Between sites, crop type impacted N 2 O emissions, with considerable variation between crop types: N 2 O flux was lower in the Salix buffers than the cultivated fields planted to potatoes, sudangrass, and black peas, but not than the fields planted to clover (Fig. 3). N 2 O fluxes in the Salix buffers were similar to fields planted to hay, except at site E in 2018, when N 2 O flux in the Salix was higher than the hay. The cultivated fields planted to potatoes, sudangrass, and black peas were the greatest sources of N 2 O (Fig. 3). Over all sites, mean cumulative seasonal N 2 O emissions were 1.3 kg N ha −1 lower in the Salix buffers than the adjoining cultivated field upslope. When considering only sites with cultivate fields planted to potatoes, sudangrass, or black peas (crops receiving 30 kg N ha −1 or more), mean cumulative seasonal N 2 O emissions were 2.1 kg N 2 O-N ha −1 lower in the Salix buffers than the upslope cultivated field.
CO 2 flux was positively related to soil temperature (r 2 : 0.209, p = 0.0001), with higher emissions occurring in the warmer months of July and August (online resource 1, Fig. 1). No relationship was found between soil moisture content and CO 2 flux. Cumulative CO 2 emissions within the treatments ranged from 1.13 to 12.8 Mg CO 2 -C ha −1 (mean 4.40 Mg CO 2 -C ha −1 across all treatments, sites, and years), with significantly (p = 0.0001) higher emissions occurring in the grass and Salix buffers (4.8 and 5.5 Mg CO 2 -C ha −1 ) than the in cultivated fields (3.3 Mg CO 2 -C ha −1 ). Within the buffer treatments themselves, CO 2 emissions were significantly greater in the Salix treatments than the grass treatments (p = 0.0271), although the effect was less than the difference between either buffer treatment or the cultivated field (p = 0.0001) (Fig. 4(a)). Measured gas fluxes in the cultivated fields revealed that fields planted to sorghum sudangrass, hay, and potatoes had higher CO 2 flux than black peas and clover ( Fig. 3(b)). When broken down by crop type within the potato rotation, CO 2 emissions in the Salix buffers were greater than those in the black peas (4.4 Mg CO 2 -C ha −1 Salix vs. 1.6 Mg CO 2 -C ha −1 black peas), clover (2.1 Mg CO 2 -C ha −1 Salix vs. 1.1 Mg CO 2 -C ha −1 Fig. 2 Median, 25th quartile, 75th quartile, maximum, and minimum cumulative seasonal N 2 O emissions for each a treatment at the 5 research sites over 2 years and b research site in each year of sampling Fig. 3 Median, 25th quartile, 75th quartile, maximum, and minimum cumulative field a N 2 O and b CO 2 emissions for each crop type clover), hay (7.2 Mg CO 2 -C ha −1 Salix vs. 5.0 Mg CO 2 -C ha −1 hay), and potatoes (5.2 Mg CO 2 -C ha −1 Salix vs. 2.3 Mg CO 2 -C ha −1 potatoes), but similar to those from sudangrass (8.1 Mg CO 2 -C ha −1 Salix vs. 7.5 Mg CO 2 -C ha −1 sudangrass). Measured gas fluxes also exhibited considerable between-site variability ( Fig. 4(b)). ANOVA revealed a significant effect of site on measured CO 2 flux, with emissions at site B being significantly higher than all other sites (in both 2018 and 2019; p = 0.0001 to p = 0.0015) and emissions at site D being significantly lower than all other sites (p = 0.0025 to p = 0.0246). 2019 emissions at both site B and site D were higher than 2018 emissions, which could be explained by higher precipitation in 2019. Additionally, site A had significantly higher emissions than site E in 2018 (p = 0.0495), while site E and site C had significantly higher emissions than site D in 2019 (p = 0.0002; p = 0.0025), and site E had significantly higher emissions than site A in 2019 (p = 0.0151). However, relative to the crop types, CO 2 emissions in the Salix buffers were significantly higher than those in the black peas (p = 0.0031), clover (p = 0.0108), hay (p = 0.0001), and the potatoes (p = 0.0001), but not significantly different from the sudangrass.
While CH 4 emissions were detected at all sites, they were extremely low with no obvious trends or drivers. Methane fluxes ranged from − 142 to 174 g CH 4 -C ha −1 day −1 with mean emissions of − 1.2 g CH 4 -C ha −1 day −1 across all sites and years. Soils at all research sites and years were slight net CH 4 sinks except site E and site C in 2019 which were slight net sources.
Nitrate exposure (NE), the cumulative seasonal soil NO 3 − flux, averaged 340 μg/10 cm 2 across all sites and years, with higher NE observed in the cultivated field (mean 948 μg/10 cm 2 ) than the Salix and grass buffers (151 and 109 μg/10 cm 2 , respectively) ( Table 1). The highest observed NE was 2909 μg/10 cm 2 , which occurred in the cultivated field at site A in 2019.
Cumulative N 2 O flux and NE were not observed to change across the Salix buffer zone. The mean NE  across the Salix buffer zones ranged from 162 ug 10 cm 2 in row 1 to 140 ug 10 cm 2 in row 3; however, there was not a consistent decrease between sites. Similarly, while the mean cumulative N 2 O emissions were greater in row 1 of the Salix buffer than row 3 (0.27 kg N 2 O-N ha −1 vs. 0.16 kg N 2 O-N ha −1 ) across all sites, this difference was not significant, and in some instances N 2 O flux was trended to higher in row 3 than row 1 although there was no significant difference. Site A underwent more frequent, weekly, monitoring of GHG emissions to evaluate temporal variations in GHG flux. At that site no hot moments of N 2 O emission were observed in the Salix or grass buffers after wetting events. This is in contrast to the cultivated fields which demonstrated hot moments of N 2 O emission during this same period. Over 2018 and 2019 there were four notable N 2 O emission events occurring on August 21, 2018, June 26 and July 3, 2019, September 12, 2019, and October 10, 2019, all shortly after large precipitation events (Fig. 5). There were no corresponding high N 2 O emission events observed in the Salix and grass buffers on these dates.
When the cumulative N 2 O and CH 4 emissions for each site and treatment were converted to CO 2 e, the overall GHG balances of the cultivated fields were significantly higher than the Salix and grass buffer treatments (p = 0.0076 and p = 0.0052), but there was no significant difference between the Salix and grass treatments (Fig. 6). Despite significantly higher CO 2 emissions in the willows relative to the other treatments, even when CO 2 was included in the CO 2 e calculation, the overall GHG balance of the Salix remained significantly lower than the cultivated fields. However, this fails to consider the increased photosynthesis in the Salix treatment. The overall GWP of the Salix buffers (64.8 Mg CO 2 e/ ha) was 9.5% of the cultivated fields (678 Mg CO 2 e/ ha), but 119% of the grass (54.6 Mg CO 2 e/ha). While the mean average GWP of emissions at the soil-atmosphere interface in the Salix buffers was 409 Mg CO 2 e/ha lower than the cultivated fields across all sites, reductions of up to 1849 Mg CO 2 e/ ha were observed (at site A in 2019).

Nitrous Oxide
This study found that N 2 O emissions in Salix and grass agricultural-riparian buffers were low, in general, and significantly lower than in upslope cultivated fields in potato rotation, which often had larger N 2 O emissions. Both Salix and grass treatments were effective in reducing N 2 O emissions relative to the cultivated field.
Denitrification is known to be influenced by soil temperature and water filled pore space (Smith et al., 2018), which was evident in our study by the correlations, albeit weak, between N 2 O (a product of incomplete denitrification) and soil temperature and moisture content. However, soil NO 3 − can out compete with N 2 O for terminal electron acceptors resulting in incomplete denitrification increasing N 2 O emissions (Gauder et al., 2012;Gillam et al., 2008;Hellebrand et al., 2008). Thus, the accumulation of NO 3 − favors NO 3 − reduction preferentially to N 2 O reduction, the final step of the denitrification reaction, even in the presence of the nosZ enzyme (Audet et al., 2013;Cho et al., 1997a, b;Smith et al., 2018). This results in N 2 O accumulating in the soil and being released to the atmosphere. This relationship is demonstrated in our study by the relatively strong positive relationship observed between soil NE and cumulative N 2 O emissions.
The magnitude of cumulative N 2 O emission can be explained by differences in NE and soil moisture content, which resulted in N 2 O emissions in the agricultural-riparian buffer treatments being significantly lower than the upslope cultivated fields. N 2 O emissions in the cultivated fields were large following fertilizer application and when wetting events occurred, particularly in mid-season when the soils were warm and had high NO 3 − concentration. During these times, buffer N 2 O emissions remained low. This was a result of low soil NO 3 − availability in the buffer zones, as indicated by differences in NE, and higher soil moisture content in the buffer zones which favor the reduction of N 2 O to N 2 (complete denitrification). This points to the efficacy of Salix at absorbing fertilizer-derived NO 3 − transported from agricultural fields, accruing in the Salix biomass and/ or creating an environment conducive to complete denitrification. In either case, this limits shallow soil NO 3 − movement to the adjoining waterway. It is also possible that some of the NO 3 − from the cultivated fields leached to groundwater and/or was leached from the riparian zone before denitrification could occur, but this is unlikely as groundwater is discharging to the adjacent stream through much of the year. Riparian areas, and especially those in agricultural regions, are frequently observed to have elevated N 2 O emissions relative to upslope fields, as a result of N fertilizers accumulating in riparian soils (Billen et al., 2020;Fisher et al., 2014;Gold et al., 2001;Jacinthe et al., 2015). This did not occur in our study, indicating that uptake or denitrification of agriculturally derived NO 3 − in cultivated Salix and grass buffer treatments managed to prevent the buildup of NO 3 − in the soil as indicated by lower NE. The one exception, where elevated N 2 O emissions were observed in the Salix buffers relative to the cultivated field, occurred in 2018 at site E. At this site, N 2 O flux in the Salix was significantly greater than the cultivated field (in hay) and grass buffer for the first few months of the season before declining to statistically similar levels to the grass and field. This difference is related to both the low N 2 O emissions associated with cropped field and increased emissions associated with the relative immature Salix at this site in early 2018, which would impact its ability to take up soil NO 3 − . The site E Salix was planted in late July 2017; therefore, they were less than a year old at the time that data collection commenced for this study in May 2018. At all other sites, Salix were planted as early in the previous season as possible, so the shrubs were in their second (or third) year of growth when data collection for this study began. Previous studies have found that there is an initial delay of 1-2 years before the environmental effects of Salix on water quality and soil GHG emissions take effect (Aronsson et al., 2000;Palmer et al., 2014;Schultz et al., 1995;Smialek et al., 2006). This is thought to be due to immature developing root systems which have limited capacity for NO 3 − uptake (Aronsson et al., 2000). For example, Palmer et al. (2014)  Within the cultivated fields themselves, there was great variability in N 2 O flux depending on what crop within the potato rotation was present and its management. Potatoes had high N 2 O emissions (mean 2.5 kg N 2 O-N ha −1 ), which was expected due to their large N fertilizer requirements. The mean rate of fertilization for potato crops in our study was 184 kg N ha −1 (online resource, Table 2). Estimated emissions for potato fields in our study using the Canadian Tier II regional emission coefficient (0.014) as used in the Canada's National Inventory Report predict cumulative seasonal N 2 O emissions of 2.6 kg N 2 O-N ha −1 , similar those observed in our study. This can be explained by the wet rainy climate of PEI, which could favor denitrification and incomplete denitrification processes from occurring. Given that potatoes require extensive soil disturbance during both planting and harvest, which also coincide with the wettest times of year on PEI, the potential for NO 3 − losses off-field is particularly high during the potato phase of the rotation; however, no correspondingly high N 2 O emissions or soil NO 3 − were observed in the Salix or grass buffers. The potato phase can be seen as the least "environmentally friendly" stage of the rotation in terms of fertilizer losses off-field, and therefore the ultimate test in evaluating the ability of Salix buffers to mitigate fertilizer-derived N 2 O emissions. Fields planted to sudangrass (mean 2.6 kg N 2 O-N ha −1 ) and black peas (mean 0.8 kg N 2 O-N ha −1 ) also had elevated N 2 O emissions relative to Salix/grass buffers, with daily emissions sometimes exceeding those of potatoes These results were unexpected as the purpose of the non-potato crops in the rotation is to mitigate some of the environmental degradation caused by potato cultivation, including keeping nitrogen in the soil and mitigating N 2 O emissions. These high emissions can be explained by crop fertilization and other management factors that resulted in increased soil available NO 3 − , and comparable N 2 O emissions to the potato crop. The sudangrass was planted and fertilized with nitrogen (30 kg N/ha) in mid-August 2018 (warm, moist soil) following field tillage (soil disturbance) and incorporation of the previous hay crop (carbon addition). This resulted in ideal conditions for N 2 O emissions and N 2 O emissions, previously low in 2018, were associated with this event. Most of the N 2 O emissions from the black pea crop occurred early in the 2019 growing season. There are several explanations for this. In 2018 this site was planted to potatoes, and the extremely wet fall in 2018 on PEI resulted in the grower being unable to harvest the potatoes in this field. As a result, the 2018 potato crop was plowed into the ground prior to planting the black peas in spring 2019. The black pea crop was also fertilized at planting (30 kg N/ha). The increased soil organic residues from the potato crop as well as the fertilizer application could have both contributed to elevated early-season N 2 O emissions at this site. These results underscore the importance of management in determining N 2 O emissions. Apparent NO 3 − capturing cover crops such as sudangrass or black pea can result in increased N 2 O emissions if accompanied by N fertilization, especially if fertilization occurs when the soil is warm and wet.
Other field management considerations may explain differences in observed N 2 O emissions between crops within the potato rotation relative to Salix/grass. N 2 O emissions in the salix/grass buffers were similar to fields planted to hay even when crop fertilization occurred. The site D field was planted to hay (timothy/clover) in 2019 and was fertilized (15 kg N/ha). Despite receiving N fertilizer inputs, N 2 O emissions even within the cultivated field were low at this site (mean daily 0.4 g N 2 O-N/ha day). The NE for the cultivated field was low, similar to the Salix and grass treatments indicating that NO 3 − was not accumulating in this cropped field. The N 2 O emissions were lower than other fields that received higher-rate fertilization and had higher NE values. In addition to the nitrogen status of this field, the low emissions can be explained by the perennial crop and the lack of tillage in this field, which has been found to reduce N 2 O emissions relative to tilled fields . The site D field was previously tilled in fall 2017 when the last potato crop was harvested and had been in constant hay production since (barley in 2017, timothy/clover in 2018). A similar situation presented itself at site C, which was planted to timothy/alfalfa in both 2017 and 2018 but received no fertilization. Field N 2 O emissions and NE were low in 2018 (− 0.01 kg N 2 O-N ha −1 )-which was similar to the agricultural-riparian grass and Salix buffers (mean daily − 0.04 and 0.04 kg N 2 O-N ha −1 ).
Our results are comparable to most other studies which have also observed Salix N 2 O emissions to be low, even when receiving fertilizer inputs, and that N 2 O flux in Salix is reduced compared to most other vegetation and/or land use types (Borjesson, 1999;Bressler et al., 2017;Drewer et al., 2012;Gauder et al., 2012;Kavdir et al., 2008;Lutes et al., 2016;Smialek et al., 2006;Whitaker et al., 2018). Some exceptions exist. Similar to our study, Bressler et al. (2017) did not observe significant differences in N 2 O flux between hay and Salix; however, other agricultural crops had elevated N 2 O relative to Salix. Likewise, our results corroborate previous findings that N 2 O fluxes in Salix are thought to be primarily linked to nitrogen fertilizer inputs (Borjesson, 1999;Drewer et al., 2012;Gauder et al., 2012;Heller et al., 2003;Lutes et al., 2016).
However, other studies have found that denitrification is a major process occurring in Salix (Aronsson et al., 2000;Dimitriou et al., 2012) including in 2-to 3-year actively growing Salix riparian buffers in agricultural regions (Ferrarini et al., 2017). As such, the potential for incomplete denitrification and the release of N 2 O exists if NO 3 − accumulates. Since riparian areas tend to be sites of elevated denitrification regardless of vegetation type due to wetter conditions and accumulations of soil organic carbon , the need for further study of the effect of Salix on GHG emissions in these settings exists. Preventing significant amounts of NO 3 − from accumulating, such as when NO 3 − influx is lower than the uptake capacity of Salix, can control incomplete denitrification. While our study indicates that Salix agricultural-riparian buffers can be effective at mitigating agriculturally derived riparian N 2 O emissions on PEI, further study of Salix in a variety of hydrogeomorphically distinct riparian settings is needed to better understand the potential and limitations of this Salix application.

Soil Respiration (CO 2 )
CO 2 emissions at the soil-atmosphere interface are primarily caused by aerobic microbial respiration and root respiration, and are affected by soil moisture content, temperature, and soil organic matter content (Jacinthe & Vidon, 2017;Lutes et al., 2016;Pacaldo et al., 2014a;Smith et al., 2018). Temperature is usually the main determinant of CO 2 flux since microbial activity and root respiration increase at higher temperatures and observed CO 2 flux in our study followed this seasonal trend at all sites and treatments, with highest emissions occurring in July and August. This is consistent with other studies that have found seasonally determined CO 2 flux under Salix in a variety of settings (agricultural, riparian, forest) (Bressler et al., 2017;Gauder et al., 2012;Jacinthe et al., 2015;Lutes et al., 2016;Vidon et al., 2014).
In our study, larger Salix CO 2 emissions relative to other treatments may be explained by leaf decay and the high proportion of fine root hairs in Salix that turnover and decompose approximately once per season (Pacaldo et al., 2013;Rytter, 2001). Increased fine root biomass in soils is positively correlated with soil CO 2 emissions (Hu et al., 2016), and Salix are known to contribute a large amount of decomposing organic carbon to soils (approximately 4.5-10 tonnes of dry matter/ha) through leaf litter and root decomposition (Borjesson, 1999). Annual fine root production of Salix Viminalis in sandy soils has been observed to be 28-50% of the net primary productivity, with most decomposition occurring within a year (Rytter, 2001) although smaller fine roots can turn over up to 4 times in a single year (Borjesson, 1999). Studies have also found that Salix do not appear to cause soil organic matter levels to increase long term, indicating that this carbon is released back to the atmosphere (Pacaldo et al., 2013).
In our study, differences in CO 2 emissions by crop type are consistent with other studies that observed elevated CO 2 fluxes in Salix relative to most, but not all, annual cropping systems (Bressler et al., 2017;Gauder et al., 2012;Pacaldo et al., 2014b). However, the situation is complex: other studies have found that transitioning from annual cropping systems to Salix has decreased soil CO 2 emissions (Borjesson, 1999) and that Salix variety and previous land use can impact observed CO 2 flux (Lutes et al., 2016;Nikiema et al. 2012).
CO 2 flux under Salix was similar to sudangrass, and the larger emissions observed in sudangrass can be explained by a variety of management factors. The sudangrass was cultivated and fertilized in-season, and soil CO 2 emissions can intensify with soil disturbance and N application as a result of enhanced microbial activity and C availability due to increased plant growth (Lutes et al., 2016via Gauder et al., 2012. In our study, cumulative seasonal CO 2 flux was positively correlated with cumulative seasonal available NO 3 − (r 2 = 0.119; p = 0.053). Soil disturbance aerated the soil and mixed in decomposing organic matter from the previous harvested crop, during the warmest months of the year and stimulated microbial activity-enhancing CO 2 emissions.
Vol:. (1234567890) Higher CO 2 emissions at site B can be explained by high rates of fertilization in both study years (online resource, Table II) in conjunction with wetter soil conditions. Site B had the second highest overall mean soil moisture content of 31.9%. Conversely, lower CO 2 emissions at site D can be explained by lower rates of fertilization and dry soil conditions: site D was the driest site overall, with a mean soil moisture content of 20.3% in 2019. CO 2 emissions can intensify with N fertilizer application as a result of enhanced microbial activity and C availability from increased plant growth (Gauder et al., 2012). Additionally, CO 2 flux can be elevated due to more frequent soil tillage that promotes oxidation of soil carbon and decomposition of organic matter (Borjesson, 1999). Site B was in potatoes in 2018, and therefore experienced soil disturbance (tillage) during planting in the spring. This site was not harvested in 2018 due to a particularly wet fall, leaving the potatoes to be tilled into the soil in spring 2019 when the 2019 black pea crop was planted. Soil tillage 2 consecutive years, heavy fertilization, and the increased decomposing organic matter from the discarded potato crop could all have contributed to elevated CO 2 emissions at this site. In contrast, site D did not experience any soil tillage in 2019 or in 2018 (the year prior to monitoring this site). This site was in barley in 2018 and timothy/clover in 2019 when the site was monitored for GHG flux. Lower fertilization rates, lack of soil tillage, and dryer soil conditions could all have contributed to lower CO 2 emissions observed at site D.

Methane
This study showed that across all sites and years, overall CH 4 fluxes at the soil-atmosphere interface were not significantly different in Salix agriculturalriparian buffers than in grass agricultural-riparian buffers or from upslope cultivated agricultural fields in potato rotation, except at site A. Overall, observed CH 4 fluxes were low at all sites and in all treatments, and each site exhibited spatial variability with both sinks and sources of CH 4 occurring within each site on most sampling dates, which is consistent with observations in other Salix studies (Bressler et al., 2017;Jacinthe et al., 2015). However, cumulative seasonal CH 4 fluxes revealed most sites to be slight net sinks except at site E and site C in 2019. In our study, there was no impact of increasing distance from the cultivated field on N 2 O fluxes due to NE being consistently low (< 200 ug N/10 cm 2 ) across the entire width of the buffer zone and much lower than the NE values for the cultivated fields receiving > 30 kg N/ha which ranged 500-3000 ug N/10 cm 2 .
It is well established that Salix is extremely effective at improving soil water quality due to its capacity to significantly reduce groundwater NO 3 − levels even when large amounts of NO 3 − are present or being applied (Aronsson et al., 2000;Borjesson, 1999;Bressler et al., 2017;Dimitriou et al., 2012;Dimitriou et al., 2009;Ferrarini et al., 2017;Heller et al., 2003;Young & Briggs, 2005). Ferrarini et al. (2017) and Young and Briggs (2005) both observed significant reductions in soil NO 3 − concentration across Salix riparian buffers relative to upland agricultural fields. It follows that corresponding decreases in N 2 O emissions could be expected, since NE (NO 3 − accumulation) is an important predictor of N 2 O emissions . In general, it has been estimated that Salix buffer strips planted between annual cropped fields and streams can retain N at a rate of 70 kg N ha −1 year −1 when nitrate inputs are greater than 15 kg N ha −1 year −1 , and that 2/3 of this is due to plant uptake, with the remaining 1/3 lost to denitrification (Borjesson, 1999). Of this 1/3 denitrified N, approximately 3% is thought to be released as N 2 O, representing 1% of nitrate retained by Salix buffers (Styles et al., 2016). In contrast, NO 3 − that is transported to waterways has a 75% risk of being lost as N 2 O (Styles et al., 2016). N 2 O emissions observed at the soil-atmosphere interface reflect N 2 O produced throughout the entire depth of the soil profile. As N 2 O produced at greater depths diffuses upward, it is more likely to be further reduced to N 2 if wet conditions are present. The fact that N 2 O emissions in the Salix buffer were generally low and there was no observed reduction across the buffer zone indicates that Salix were effective creating conditions that prevented incomplete denitrification from occurring. 4.5 The Impact of Wetting Events on N 2 Ofluxes Brief increases of N 2 O production following wetting events are common and often responsible for the majority of soil N 2 O emissions (Audet et al., 2014;Fisher et al., 2014;Hellebrand et al., 2008). Riparian zones in agricultural regions have been found to be particularly at risk for high N 2 O emissions due to the presence and accumulation of high influxes of watersoluble NO 3 − from upland agricultural fertilization following precipitation events Kaushal et al., 2014;Vidon et al., 2010). Denitrification is a major mechanism by which riparian areas act as ecosystem control points for mitigating the transfer of N from upland agricultural land to waterways (Bernhardt et al., 2017;Young & Briggs, 2005). Incomplete denitrification, which releases N 2 O, can occur as a result of the nosZ enzyme (responsible for the final conversion of N 2 O to N 2 ) becoming dormant when soil conditions are dry for an extended period (Robertson & Groffman, 2015;Zaady et al., 2013), but riparian areas seldom experience extended dry periods. In soils where the nosZ enzyme is present, N 2 O reduction has been shown to be controlled by nitrate content and water-filled pore space (Dandie et al., 2008). As a result of high water contents, riparian areas (and by implication, agricultural-riparian buffers) may be more likely to experience elevated instances of N 2 O emissions than upland areas following precipitation events (Audet et al., 2013;Aronsson et al., 2000;Fisher et al., 2014;Jacinthe & Vidon, 2017). Maximum rates of denitrification are often triggered when water-filled pore space exceeds 60-70% and sufficient soil NO 3 − and organic carbon is present (Dandie et al., 2008;Jacinthe & Vidon, 2017).
In our study, there were several instances of elevated N 2 O emissions in the cultivated fields following high rainfall occurrences that coincided with high field NO 3 − availability (observed N 2 O fluxes of 120, 293, 194, and 124 g N 2 O-N ha −1 day −1 ). These were the sites with high NE and cumulative N 2 O emissions across all sites were correlated with NE. However, there were no corresponding high N 2 O emissions observed in the Salix or grass agricultural-riparian buffers (observed grass 6.7, 2.1, 0.5, − 0.7 g N 2 O-N ha −1 day −1 ; observed Salix 4.0, 0.4, 1.4, 0.4 g N 2 O-N ha −1 day −1 ) which had low NE values. This represents a 94-99.9% decrease in N 2 O emissions in the agricultural-riparian buffers during wetting events and suggests this is not a major factor stimulating N 2 O emissions in these systems. The lack of N 2 O spikes in the buffer areas also indicates the effectiveness of this BMP at mitigating elevated riparian N 2 O emissions. Salix would intercept and absorb fertilizerderived NO 3 − for plant uptake as it was being transported from the agricultural field, and the lack of buffer N 2 O emissions following these precipitation events suggests soil moisture conditions were consistently high enough in the buffers to maintain highenough concentrations of activated nosZ enzyme and maintain soil NO 3 − sufficiently low to allow complete denitrification of any remaining NO 3 − . While some other studies have observed elevated N 2 O in Salix following precipitation (Kavdir et al., 2008;Ley et al., 2018;Lutes et al., 2019;Nikièma et al., 2012;Palmer et al., 2014), this seems to be more common in initial and more immature cultivations and is linked to NO 3 − availability. Our findings are in alignment with Bressler et al. (2017) who also failed to observe elevated N 2 O flux in fertilized Salix, and Ley et al. (2018) who observed much lower N 2 O emissions in Salix than an unaltered control in floodplain soils.

Overall GWP of Soil-Atmosphere Interface N 2 O and CH 4 Emissions
Our study found that non-CO 2 GHG emissions at the soil-atmosphere interface in Salix and grass buffers had a significantly lower GWP than upslope agricultural fields and that differences in N 2 O flux between treatments were the main reason for these differences in overall GWP. These findings differ from Bressler et al. (2017), who found that transitioning to Salix from conventional crops on marginal (excessive moisture) cropland had no overall effect on GHG emissions at the soil-atmosphere interface. Their study found that CO 2 equivalents were similar between Salix, hay, and corn, since increased CO 2 emissions in Salix were roughly balanced by increased N 2 O emissions in conventional crops, suggesting that "… in spite of differences for individual gasses, Salix does not provide the ecosystem service of greenhouse gas emission reduction compared to corn or hay" (p137). However, unlike our study, Bressler et al. (2017) incorporated CO 2 flux into GWP calculations. Even when CO 2 emissions are considered, the overall GWP of Salix in our study is still significantly lower than upslope agricultural fields. Other studies looking at the GWP of Salix on both a life cycle and field scale soil GHG emissions have opted to not include CO 2 flux in GWP calculations since Salix are considered a carbon-neutral energy source (Volk et al., 2006) or carbon balance in these situations is generally assessed by measuring carbon stocks (Gauder et al., 2012). Similarly, in their life cycle assessment Volk et al. (2006) found that Salix bioenergy crops have a small GWP (39-52 kg CO 2 equivalents per MWh electricity produced), mainly due to N 2 O and CH 4 emissions. In their field study Gauder et al. (2012) did not include soil CO 2 flux when calculating GWP of different bioenergy crops (including Salix) based on trace gas flux measured at the soil-atmosphere interface. N 2 O was found to have a bigger impact on GWP than CH 4 , and fertilized Salix was found to have a lower (but negative) global warming potential than fertilized miscanthus or maize, while unfertilized Salix had a slightly higher global warming potential (although still negative) than unfertilized miscanthus, but lower than unfertilized maize.

Conclusion
This study found that Salix viminalis and grass grown in agricultural-riparian buffer strips downslope of cultivated fields in potato rotation did not exhibit elevated N 2 O emissions and were successful at preventing near surface NO 3 − transport beyond the buffer zone. Salix buffers had N 2 O fluxes that were an average of 86% lower than upslope cultivated fields (up to 98% lower), representing mean cumulative seasonal reductions of 1.3 kg N 2 O-N ha −1 (up to a maximum observed reduction of 6.16 kg N 2 O-N ha −1 ). Salix buffers also displayed significant cumulative seasonal GHG reductions compared to upslope fields. The mean cumulative average GWP of the Salix buffers was 409 Mg CO 2 e ha −1 lower than the cultivated fields, with reductions of up to 1813 Mg CO 2 e ha −1 observed. There was a large amount of between and within-site variation in soil moisture content observed across all the research sites, and soil water content was positively correlated with N 2 O emissions. Our study did not observe a change in N 2 O emissions across the width of the buffer zone and did not observe elevated N 2 O emissions in buffers following precipitation events; however, these events did result in elevated N 2 O emissions in cultivated fields. While CH 4 emissions were generally low in our study, with no significant differences between treatments, site A was the only one to experience occasional flooding, and these events were marked by elevated CH 4 emissions.
While high N 2 O emissions were anticipated in fields cultivated to potatoes, this study also observed unexpectedly elevated N 2 O flux outside of the potato phase of the rotation, in fields planted to sorghum sudangrass and black peas. These emissions were linked to nitrogen fertilization events occurring later in the growing season when the soil was warm and wet as well as decomposition of the previous years unharvested potato crop at one of the sites. Differences in cumulative N 2 O emissions were primarily related to NO 3 − accumulation as reflected in NE. Our research confirms that cultivating Salix buffer strips on downslope field edges bordering riparian zones is an effective strategy for mitigating agriculturally derived nitrous oxide (N 2 O) emissions in riparian areas on Prince Edward Island. While Salix were no more effective than a grass agricultural-riparian buffer at reducing N 2 O and overall GHG emissions at the soil-atmosphere interface, Salix have the additional benefit of high biomass accrual with associated carbon and nutrient sequestration, topics which will be addressed in a subsequent paper.
Funding This study was funded by the East Prince Agri-Environmental Association through funding provided by Agriculture and Agri-food Canada's Agricultural Greenhouse Gas Program. Funding was also provided by an NSERC Discovery Grant (Burton).

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Conflict of Interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.