Do cover crops on the Canadian prairies affect soil nitrogen cycling?

Abstract For one of Canada’s most important regions of crop production—the prairies—it’s uncertain if cover crops can be successfully integrated into rotations; if so, will soil nitrogen (N) cycling be influenced to benefit main crops? To address these gaps, we compared a crop rotation with cover crops (CC) vs. without cover crops (LR) from 2018 to 2021 in Saskatoon, SK. The main crops were grown in sequence of wheat–canola–potato–pea; the cover crops included red clover, berseem clover/oat mix, fall rye, and a brassica cover. Yield and aboveground biomass were collected each year and analyzed to determine crop yield and N use efficiency (NUE). Soil N availability was monitored in various ways, that is, by assessing pre-plant soil nitrate, soil inorganic N (SIN) supply rate, and potentially mineralizable N (PMN). We found that the influence on soil N dynamics was restricted to the non-growing season where cover crops reduced SIN supply rate and nitrate content compared to the conventional practice without cover crops. Yet, rotations with vs. without cover crop did not differ in crop NUEs, yields, or in-season N dynamics. We found some evidence that diversifying rotations with cover crops may help the system to function more like perennial systems in terms of regulating N in the long run; but had limited impact during the three years studied. To ensure that cover crops are effective and functional on the prairies, innovative design approaches are needed to adapt cover crops to reach soil health goals under prairie conditions.


Introduction
Sustainable agriculture is envisioned as a food system in which food is not only nutritious and accessible to all, but also that natural resources are managed in a way that maintains ecosystem functions, supporting current and future human needs. Achieving sustainable agriculture is therefore an enormously complex challenge, especially considering the pressure to intensify food production to meet the demands of a growing population. Due to this demand, there is a heavy reliance on nitrogen (N) inputs to agricultural lands for sustained food production (Janzen et al. 2003;Baligar et al. 2015;Heffer and Prud'homme 2016). Nitrogen fertilizer is long known to increase food production, and without it, crop yields will decrease as soil N reserves decline (Janzen et al. 2003). However, the uptake of N by plants is generally less than half of the N applied, depending on the crop species and conditions of the soil (Perchlik and Tegeder 2017) leaving a considerable portion in the soil and at risk for N loss and negative environmental consequences. Better understanding and managing N dynamics is necessary for improving agricultural sustainability (Janzen et al. 2003;Martinez-Feria et al. 2018;Fu et al. 2019).
Cover cropping is a promising management practice that can influence soil N cycling (Dabney et al. 2001;Blanco-Canqui et al. 2015). Cover crops can improve agroecosystem functions by enhancing crop, nutrient and water use efficiencies (Dabney et al. 2001;Finney et al. 2016), soil carbon (C) sequestration and greenhouse gas mitigation (Blanco-Canqui et al. 2015). Cover crops can help to supply N for cash crops as well as reduce residual N levels during periods prone to N losses by efficiently utilizing and cycling excess N (Basche et al. 2014). This is achieved through the capacity of cover crops to increase the absorption and utilization of N during periods that would otherwise be fallow, and subsequently releasing N (ideally in time for the next crop) by mineralisation. In theory, cover crops can help to reduce the need for superfluous N fertilization to cash crops by contributing to the potentially mineralizable N pool during the subsequent growing season (Taveira et al. 2020). Better understanding N dynamics with and without cover crops will improve our ability to manage N additions to croplands and to balance crop yields and quality whilst conserving the environment.
The practice of cover cropping in Canada is increasing, but adoption has been greater in eastern Canada than in western Canada (Statistics Canada 2016). In the prairie regions of Canada, the short growing season, coupled with dry periods presents a challenge to successfully implementing the practice of cover cropping. In Saskatchewan, one of the dominant regions of crop production in Canada, climate is typically classified under semi-arid conditions, with extreme fluctuations between seasons, that is, long, cold winters and short, hot summers (Floate and McGinn 2010). Growing cover crops, therefore, may not be viable due to the relatively short growing window and limited water availability. However, a recent voluntary survey conducted across the Canadian prairies found that 281 farmers grew over 41 000 ha of cover crops (either as full season or shoulder season cover crop), in 2020 even in the driest regions of Saskatchewan and Alberta (Morrison and Lawley 2021). To better understand the functions, benefits, and drawbacks of cover cropping on the prairies, research is needed--and yet there is currently a paucity of cover crop research for this region. Based on the cover crop literature from other regions (Dean and Weil 2009;Blackshaw et al. 2010;Halde et al. 2014;Jahanzad et al. 2016), one would hypothesize that cover crops improve crop N use efficiency (NUE) by immobilizing N during the non-growing season and increasing soil N availability during the growing season; however, it is not certain that this hypothesis would hold true in our semi-arid and short growing season region. Although difficult, it might be possible to overcome these challenges by selecting the right cover crop species coupled with advanced agronomic management. However, more research is necessary to address this gap in knowledge and understanding. This study therefore investigated the viability of establishing a wide range of cover crops in rotations on the prairies and its effect on soil N dynamics, productivity, and crop NUE.

Study site and climate
A multi-year crop rotation trial was initiated in 2018 at the University of Saskatchewan's North Management Area (52 • 09 22.7 N, 106 • 36 28.8 W) in Saskatoon, on a Dark Brown Chernozem. At this site, the soil texture is a sandy loam (53.7% sand, and 18.6% clay), has organic matter of 3.7%, soil organic carbon of 2.20%, and soil total N of 0.23%, a pH of 6.9, cation exchange capacity of 14.5 meq 100 g −1 , and soil bulk density of 1.35 g cm −3 . The climate in Saskatoon is semiarid with seasonal climate fluctuating between long cold winters and short warm summers. The mean annual temperatures range from 1 to 5 • C and total annual precipitation between 300 and 500 mm (Saskatchewan Research Council Climate Reference Station, Saskatoon). Weather data was collected from an on-site climate station for the duration of the study.

Experimental design and field management
The experiment comprised a fully phased crop sequence of wheat-canola-potato-pea with and without cover crops (denoted CC and LR), respectively. Also included was a perennial alfalfa treatment (PR) and a short rotation of wheat-canola (SR), serving as treatment checks. The selection of cover crop species for a specific region relies on adequate assessment and knowledge of a diverse group of species i.e., integrating information about the cover crop species themselves, the growing conditions during the cover crop window, and crop rotation (the crops preceding and following the cover crop). Thus, cover crops (red clover, Trifolium pratense; oat + berseem clover, Avena sativa + Trifolium alexandrinum; fall rye, Secale cereale; tillage radish + mustard, Raphanus sativus + Brassica juncea) were selected to encompass the wide range of species (legumes, non-legumes and grasses) as well as overwintering vs non-overwintering categories commonly used as cover crops across Canada. All treatments (Table 1) were established on 6 m x 6 m plots using a randomized complete block design with four replicates. A total of fortyfour plots were used for this experiment. The period of study for this research was from spring 2019 to spring 2021.
In spring, 12, 7, and 13 May of 2019, 2020, and 2021, respectively, the plots were seeded with the main crops (wheat, Triticum aestivum; canola, Brassica napus; potato, Solanum tuberosum; pea, Pisum sativum; alfalfa, Medicago sativa). Seeding rates for all main crops are shown in Table 2. Grain/oilseed crops were seeded using a small plot drill, with 30 cm between rows. When seeding pea, a nodulator was applied at rate of 3 kg ha −1 . For potato, a single row planter was used, seed pieces were planted at a depth of 20 cm, with a row spacing of 1 m x 0.3 m. For all plots, N fertilizer (urea) was broadcast on the soil surface, and rates were determined by conducting a pre-plant soil test (0-60 cm depth) and averaging the N fertilizer recommendation for each species from treatments without cover crops (determined by AgVise Laboratories). In 2019 and 2020, respectively: 142 and 186 kg N ha −1 was applied to wheat, 118 and 92 kg N ha −1 was applied to canola, 146 and 131 kg N ha −1 was applied to potato, and 25 and 0 kg N ha −1 was applied to pea. Alfalfa plots did not receive any fertilizer. The crops received other nutrients such as potassium, phosphorus and sulphur as needed according to the soil-test recommendations.
The entire field received irrigation after planting to enhance germination, and subsequent irrigation was applied as needed based on weather forecast. Approximately 10-18 cm of irrigation water was applied during the growing season. Potato plots were hilled after germination. Potato and pea plots were sprayed with Decis insecticide at 150 mL ha −1 to control Colorado beetles, and Vertisan at 600 mL ha −1 to control black spot, respectively in 2020. In 2019, pea plots were sprayed with other fungicides, matador at rate of 740 mL ha −1 and priaxor at 490 mL ha −1 .
Cover crop (red clover, Trifolium pratense; oat + berseem clover, Avena sativa + Trifolium alexandrinum; fall rye, Secale cereale; tillage radish + mustard, Raphanus sativus + Brassica juncea) seed was drilled following main crop harvest on 6 September and 19 August--except for red clover, which is broadcast under-sown into wheat during the growing season on 25 June and 2 July in 2019 and 2020, respectively, to allow enough time for establishment (Table 2). During 2020, herbicide resistant mustard was seeded with tillage radish as a strategy to overcome potential residual injury from the herbicide applied earlier that year. Cover crops received some supplemental irrigation (<1 cm) in September  to support germination. In spring, before seeding of main crops, overwintering cover crops biomass was collected, after which, cover crops were terminated with CleanStart at 2471 mL ha −1 in 2019 and Roundup at 3311 mL ha −1 in 2020.

Main crop sampling
At grain/oilseed crop harvest, crop biomass samples were collected from near the centre of each plot, in two 0.25 m 2 areas. The aboveground shoots were clipped at ground level and grain separated from the crop residue. Ahead of potato harvest (∼1 week), plant tops were mechanically mowed to allow for tuber skin-set. For potato yield, samples were collected by digging up tubers from a representative central area (0.25 m 2 ) twice per plot. Alfalfa plots were combined, and residues removed leaving the stubble in the plots each year. For all plant samples, fresh weights were recorded, and dry weights measured after oven drying the plant tissue at 60 • C until constant dry weights were obtained. Grain samples were threshed, and yield and residue samples were sep-arately ground to pass through a 1 mm sieve using Wiley grinder and stored in 8.5-dram polypropylene clear snap cap vials for %C and %N analysis.

Cover crop sampling
After cover crop emergence (1-3 weeks after planting), plant counts were recorded in two 0.25 m 2 areas per plot; plant heights are recorded twice during cover crop growth, once after emergence and again before biomass sampling in early October. Biomass samples were collected (early October) by clipping all vegetation at the soil surface within three 0.25 m 2 areas per plot; fresh weights were recorded before tissues were oven-dried at a temperature of 60 • C until constant dry weights were obtained. The dry samples were ground to pass through 1 mm sieve using blender and stored in 8.5 dram polypropylene clear snap cap vials for %C and %N analysis.

Plant tissue %C and %N analysis
Plant tissue samples (from the main crops and cover crops) were analysed for C and N concentrations using a LECO CN628. Briefly, about 0.10-0.15 g of plant tissues were weighed into tin foil capsules, sealed, and placed in the sample carousel for % C and %N determination using the combustion method.

Soil inorganic N release rates
For key periods throughout the year, soil N availability was monitored using ion exchange resin strips to quantify soil inorganic N (SIN) (NH 4 + and NO 3 − ) (Schoenau et al. 1993;Qian and Schoenau 2002). The ion exchange resin strips imitate the supply of nutrients to plant roots and function by exchanging ions between the soil medium and the exchange site of the membrane. The resin strips come in two forms; i) a cation resin strip, which has a negatively charged membrane to attract and adsorb positively charged nutrients, and ii) an anion resin strip, with a positively charged membrane to attract and adsorb the negatively charged nutrients. The exchange site of the resin strips acts as a sink for ions when buried in a soil (Schoenau et al. 1993). The resin sheets (AMB-SS and CMB-SS membranes from ResinTech Inc., West Berlin, NJ) were cut into 2.5 cm × 12 cm strips. For each strip, the top 1.5 cm was taped off with marine duct tape, leaving an exposed area of 2.5 cm × 10.5 cm for burial into the soil. Prior to burial, the cation and anion strips were pre-conditioned by soaking in 0.5 M HCl for 1 h and in 0.5 M NaHCO 3 − solution for 3 h, with solution changed every hour. The strips were buried by making a slit in the soil and placing two pairs of strips (positive and negative) about 10 cm apart in each plot. To ensure good contact, the soil was pressed firmly around the strips and area flagged for easy identification. Strips were left in the soil for 14 days during key periods (spring, summer, fall). To capture overwinter N dynamics, resin strips were placed in the soil just prior to freeze-up, left overwinter, and removed in spring at snowmelt (capturing the ∼October-April period). Strips due for removal were retrieved from the field by placing all four strips from each plot together in a Ziploc bag and brought to the lab. The strips were cleaned from all soil debris by rinsing in de-ionized water.

Soil inorganic N extraction
The cleaned strips (2 pairs) from each plot were placed in containers containing 140 mL of 2 M KCl solution (35 mL per strip) and placed on a shaker at 55 rpm for 1 h to desorb nutrient ions (similar to the method described by Qian and Schoenau 2007). The extract was filtered through Whatman No. 42 paper into 16-dram polypropylene clear snap cap vials and stored in a freezer (approx. −15 • C) prior to analysis. The N supply rate was determined using air-segmented (continuous) flow analysis with a SEAL AA3 HR chemistry analyser (SEAL Analytical, Kitchener, ON). The concentration obtained was transformed into rate of absorption per area of 4 strips (0.0105 m 2 ) per number of days buried (mg N m −2 duration of burial in days −1 ).

Potentially mineralizable N
In spring 2020, soils were sampled from each plot at depths of 0-15 and 15-60 cm to determine potentially mineralizable N (PMN). The PMN pool was determined by incubating soil samples anaerobically for 7 days followed by KCl extraction to determine NH 4 + -N concentration. Subsamples of field moist soil (5 g) were used to determine pre-incubated NH 4 + -N using 2 M KCl solution. This was followed by submerging 5 g soil samples in 10 mL de-ionised water and incubating for 7 days at a temperature of 37 • C (Curtin and Campbell 2007). Incubated samples were diluted with 3.33 M KCl solution, shaken at 120 rpm for 30 minutes and supernatant filtered using Whatman No. 42 filter paper into vials and stored for analysis. The NH 4 + -N concentration for the pre-incubated and incubated samples were determined using air-segmented (continuous) flow analysis with a SEAL AA3 HR chemistry analyser (SEAL Analytical, Kitchener, ON). Mineralized N was estimated by deducting pre-incubated NH 4 + -N from incubated NH 4 + -N. Potential mineral N values obtained were used to estimate the total N potentially mineralized over a growing season.

Estimation of crop nitrogen use efficiency (NUE)
The crop NUE indices were estimated using plant dry weights in kg ha −1 and pre-plant available soil N (kg ha −1 ). Soil available N (Sn) was calculated by summing pre-plant available soil N, fertilizer N, and the potentially mineralizable N estimates derived from incubated soil samples. The various NUE indices explored were calculated using Eq. 1-4: N harvest index (NHI) estimates portioning of plant N between yield and vegetative parts: N uptake efficiency (NUpE) measures how much N is taken up by the plant relative to soil N: Apparent N recovery (ANR) measures crop N removed from the field as yield N based on available N: Soil yield N use efficiency (NUEyield) estimates yield potential based on available N : where Yn is yield N, Pn is aboveground plant N, Yw is yield weight, and Sn is soil N. The N contents were expressed in kg N ha −1 ; weight was expressed as kg ha −1 (dry weight basis).
Statistical analysis SAS (SAS Institute, Inc., University edition, Cary, NC) was used to perform the analyses of variance (ANOVA). The MEANS procedure was used for descriptive statistics, UNI-VARIATE procedure for normality check, and LEVENE for testing homogeneity of variances in SAS; all response variables which did not meet normality test were log or square root transformed. Where normality test was not met by transformations, the data set was double checked for outliers and when points determined were erroneous (i.e., by confirming with field logbooks for issues) they were removed. All data transformed were back-transformed for presentation.
The ANOVAs were conducted using the MIXED procedure, and means comparisons were conducted using Tukey-Kramer's tests. A significance level of α = 0.05 was used. Each year was analyzed separately. To determine if cover crops in the rotations significantly influenced all SIN metrics, analysis was conducted to compare the LR to the CC rotation for each crop phase and year of the study. For the mixed model, crop rotation was considered a fixed effect and replicate was considered a random effect for all response variables (crop yield, crop NUE, SIN dynamic, and PMN). There were no significant interactions, thus means comparisons were conducted for the main effect of crop rotation within each crop phase.
To test if cover crops--as part of a soil health management strategy--influenced soil N dynamics, regression analyses were conducted. The crop rotation treatments (Table 1) were assigned an index value from 1 to 4 based on near continuous soil cover, where in theory: (1) indicates a relatively poor management practice for soil health, a short rotation with only two crops in a 2 year rotation; (2) indicates an incrementally better management strategy for soil health, a longer rotation with four species in a 4 year rotation; (3) indicates an incrementally better management strategy for soil health, with four species in a 4 year rotation plus four shoulder-season cover crops; and (4) indicates the best theoretical management strategy for building soil health, a perennial cropping system. The GLM procedure was used to test for two biologically meaningful relationships between the soil health management index and the soil N response--that being either a linear or quadratic response (cubic models were not considered as these were not deemed biologically meaningful). A significance level of α = 0.05 was used.

Weather conditions
The monthly precipitation and temperature fluctuated over the study period with differences in the amount and distribution of rainfall (Table 3). Certain periods experienced relatively average precipitation, close to the 30 year normal (i.e., March 2020 and May 2021); but other periods were much drier than normal. For example, some of the driest periods were May and August in 2019; July and August 2020; and June and July 2021. The average temperature was 4.2, 3.4, and 4.4 • C in 2019, 2020, and 2021, respectively.
Cover crop biomass production, N content, and C/N ratio Cover crop biomass, N content and C:N ratios differed by species (Table 4). Cover crop biomass (dry matter) production ranged from 99-437 kg ha −1 in fall of 2019 to 90-428 kg ha −1 in fall 2020, contained between 4 and 16 kg N ha −1 each year. In fall 2019, red clover produced the most biomass and highest N content and differed significantly from the mixture of oat and berseem clover which had the lowest biomass and N content but their C:N ratios were similar. At cover crop termination in spring 2020, red clover and fall rye accumulated 296 and 287 kg dry matter ha −1 , respectively. A similar trend was observed in fall of 2020, where red clover produced the greatest amount of biomass and accumulated the most N and was significantly different from other species while rye on the other hand, produced the lowest biomass and N content. By spring 2021, rye performed better in terms of biomass and N content than red clover.

Pre-plant soil nitrate content in rotations with vs. without cover crops
Only after three years did cover crops reduce pre-plant soil NO 3 − content, that is, prior to the pea in the rotation wheat-canola-potato-pea (P = 0.040), and prior to the wheat in the rotation canola-potato-pea-wheat (P = 0.029) with a tendency towards reduced N content prior to the canola in the rotation potato-pea-wheat-canola (P = 0.073) (Fig. 1). These SIN results show that NO 3 − levels were most influenced by cover cropping especially when canola and potato occurred early in the rotation. Before 2021, there were some indications--albeit weak--that cover cropping was lowering NO 3 − contents compared to without cover cropping at pre-plant; however, significant differences were not detected.

Pre-plant soil nitrate content as influenced by soil health management
The soil health management index was related to pre-plant SIN only in 2021 (P = 0.014) (Fig. 2). In 2021, as the soil health management index increased from 1 to 2 (i.e., the short rotation to the long rotation), pre-plant SIN was gradually increased; but thereafter decreased for the higher indices of 3 and 4 (the cover cropped rotation, and perennial alfalfa).

Supply rate of soil inorganic N in rotations with vs. without cover crops
Generally, there was a cyclical pattern towards greater SIN supply rates in the spring followed by lower rates in the summer and fall. Including cover crops in the rotation significantly influenced SIN supply rates only during the fall season in 2019 for the canola phase (P = 0.0137) and in 2020 for the pea (P = 0.0007) and the wheat phase (P = 0.0010), and never for spring or summer (Fig. 3). Notably, the cover crop effect was apparent after harvest for all grain/oilseed crops (i.e., canola in 2019, pea in 2020, and wheat in 2020) but never for potato.
The rotation starting with wheat had lower SIN supply rates with vs. without cover crops in the fall after canola  Table 4. Mean cover crop biomass, nitrogen content, and C:N ratio ± standard error at fall before freeze-up and at spring before termination.
Year Cover crop Biomass (kg ha −1 ) N content (kg ha −1 ) C : N r a t i o

Fall 2018
Red clover n/a n/a n/a Oat/berseem 8 n/a n/a Rye 59 3 9 Tillage radish n/a n/a n/a

Spring 2019
Red clover 37 1.4 11 Rye 215 10 9 Oat/berseem n/a n/a n/a Tillage radish n/a n/a n/a Oat/berseem n/a n/a n/a Tillage radish n/a n/a n/a Oat/berseem n./a n/a n/a Tillage radish n/a n/a n/a P = 0.0408 P = 0.0158 P = 0.7943 Note: Within each column, means followed by different letters are significantly different at α < 0.05 with P-values bolded and italicized. Fall 2018 and spring 2019 samples were not analysed due to incomplete data and are represented by n/a when data were not available or cover crop is not a winter hardy species. Red clover was seeded earlier than the other cover crops. Fig. 1. Mean pre-plant soil nitrate content (NO 3 − ) levels in the 0-60 cm depth as influenced by cover crops over study period (2019-2021). The soil inorganic N content beginning in each rotations represent the N level at the start of the experiment in 2018. Afterwards results are based on the period prior to seeding the main crop in any year and only the 4-year long rotations with and without cover crops are compared. CC indicates the cover cropped long rotation; LR indicates the long rotation without cover crops. Markers with red astericks at each sampling point are significantly different (P < 0.05), while red plus (+) sign shows a tendency for significance (P > 0.05 but < 0.1). Pre-pea in 2019 was log transformed whilst pre-canola in spring 2020 was square root transformed for analysis. The crop sequence is W (wheat)-C (canola)-Po (potato)-P (pea).

Fig. 2. Relationship between pre-plant soil NO 3
− content in the 0-60 cm depth as influenced by soil health management index from 2019 to 2021. Numbers 1-4 on the x-axis represent the soil health management index. The lowest number is assigned to rotation with the poor soil health management and the highest number is assigned to the rotation with a better soil health management strategy. 1 = short rotation, 2 = long rotation without cover crop, 3 = long rotation with cover crop, and 4 = perennial rotation with alfalfa. The regression lines represent significant quadratic relationships. harvest in the second year of the rotation (wheat-canolapotato). By the third year when potato was grown, no significant differences were observed. The rotation beginning with canola had lower SIN supply rates with vs. without cover crops in the third year of the rotation, after pea harvest (canola-potato-pea). The rotation beginning with potato also had lower SIN supply rate with vs. without cover crops by the third year (fall 2020) after wheat harvest (potato-pea-wheat). The rotation starting with pea did not show any cover crop influence on SIN during the study period.

Soil inorganic N supply rate as influenced by soil health management
The SIN supply rate was not significantly related to soil health management until the later periods, that is, summer 2020--after 2 years (P = 0.061) and spring 2021-approaching three years (P = 0.0002) and (surprisingly) never in the fall of 2019 and 2020 (Fig 4). When the relationship was detected, the trajectory was for SIN supply rate to increase along with the soil management index from 1 to 2 (short rotation to long rotation); but, thereafter decrease as management had greater soil cover (3 and 4, the long rotation with cover crops, and the alfalfa system, respectively).

Fig. 3.
Mean soil inorganic N supply rate (NH 4 + and NO 3 − ) in the top 10 cm depth, as influenced by rotation from fall 2019 to spring 2021. Only the 4 year long rotations with and without cover crops are compared. CC indicates the cover cropped long rotation; LR indicates the long rotation without cover crops; markers with red astericks at each sampling point are significantly different, otherwise are not significantly different. Spring and summer are period during the growing season for any specific crop phase while fall periods are during the non-growing season after crop harvest. Data for wheat in Spring 21 was log-transformed for analysis. The crop sequence is W (wheat)-C (canola)-Po (potato)-P (pea).

Fig. 4. Relationship between soil inorganic N (NH 4
+ and NO 3 − ) supply rate in the top 10 cm as influenced by soil health management index. Numbers 1-4 on the x-axis are the soil health management. The lowest number is assigned to rotation with the poor soil health management and the highest number is assigned to the rotation with a better soil health management strategy. 1 = short rotation, 2 = long rotation without cover crop, 3 = long rotation with cover crop, and 4 = perennial rotation with alfalfa. The regression lines represent significant quadratic relationships.

Overwinter soil inorganic N dynamics
The supply rate of SIN overwinter (October-April) was not influenced by rotation in the two winter periods studied (Fig. 5). Numerically, the cover cropped rotations tended to have lower SIN supply rates, but this difference was not significant. The influence of soil health management on overwinter SIN supply did not show any significance in both 2019/2020 and 2020/2021 (Fig. 6).

Potentially mineralizable N dynamics
Potentially mineralizable N was not significantly influenced by rotation in the period studied, spring 2020 (Table  5) --despite the numerical differences that pointed towards greater PMN with vs. without cover crops. Soil health management, on the other hand, was strongly related to PMN (P = 0.007) in a quadratic form (Fig. 7). The PMN levels were higher for the improved management indices of 3-4 (repre- Fig. 5. Mean over-winter soil inorganic N (SIN) (NH 4 + and NO 3 − ) supply rate as influenced by cover crops overwinter (Fall 2019 to Spring 2020 and Fall 2020 to Spring 2021), mg N m −2 181 days −1 . Only the 4-year long rotations with and without cover crops are compared. CC indicates the cover cropped long rotation; LR indicates the long rotation without cover crops; no significance difference were detected between treatments at P < 0.05. Overwinter SIN data after canola in 2020/2021 was transformed for analysis. Overwinter is the period from soil freeze-up (end of October) to spring thaw (April). The crop sequence is W (wheat)-C (canola)-Po (potato)-P (pea).

Fig. 6.
Relationship between overwinter soil inorganic N supply rate as influenced by soil health management index. Two overwinter periods are shown (Fall 2019-Spring 2020 and Fall 2020-Spring 2021). Numbers 1-4 on the x-axis denotes the soil health management index. The lowest number is assigned to rotation with the poor soil health management and the highest number is assigned to the rotation with a better soil health management strategy. 1 = short rotation, 2 = long rotation without cover crop, 3 = long rotation with cover crop, and 4 = perennial rotation with alfalfa. The absence of regression line represents no significant quadratic relationship. Overwinter is the period from soil freeze-up (end of October) to spring thaw (April). senting long rotation with cover crops and the perennial alfalfa system) compared to the poorer indices of 1-2 (representing the short rotation and long rotations).

Crop yields and NUE in rotation with vs. without cover crops
In both 2019 and 2020, crop yields were not significantly different with vs. without cover crops (Table 6) --except for pea where there was a tendency for lower yields with cover crops (P < 0.1 but > 0.05). Crop NUE did not differ by rotation in any year (Table 7). The NHI ranged from 29% to 86%; NUpE  Fig. 7. Relationship between soil potentially mineralizable N in the top 15 cm and soil health management index. The lowest number is assigned to rotation with the poor soil health management and the highest number is assigned to the rotation with a better soil health management strategy. 1 = short rotation, 2 = long rotation without cover crop, 3 = long rotation with cover crop, and 4 = perennial rotation with alfalfa. The regression line represents a significant quadratic relationship.

Discussion
Cover crop production in a short season growing region: low biomass and N content, and opportunities for improvement Generally, the aboveground biomass produced in our study was lower than that observed in other studies conducted under similar climates (Ruis et al. 2019;Farzadfar et al. 2021). Cover crop performance is known to vary with climatic conditions, even within temperate regions (Ruis et al. 2019). For example, more humid areas with precipitation >750 mm produce greater cover crop biomass compared to semi-arid regions with less precipitation (Coombs et al. 2017;Ruis et al. 2019).
Despite the relatively low biomass levels, we were able to successfully establish cover crops in a prairie crop rotation-in that some level of growth was observed in the shoulderseasons. However, in our study, the levels of biomass were far below the "general goal" of 1 Mg ha −1 that is typically recommended . Biomass, N contents, and C:N ratios differed by the type of cover crop and how it was incorporated. Red clover was the only cover crop that was under-sown during the growing season (between wheat rows), and it generally accumulated more biomass and N content than the rest of the shoulder-season cover crops. The biomass produced in a study involving similar cover crops i.e., red clover, rye, oilseed radish and oat, observed greater biomass accumulation in red clover than the other cover crops (Vyn et al. 2000). This pattern is similar to another study where 31 kg N ha −1 was accumulated in cover crop tissues when under-sown compared to 25 kg N ha −1 when sown after harvest (Doltra and Olesen 2013). In our study, the winterhardy species (red clover and fall rye) accumulated greater biomass than non-hardy species (oat/berseem clover mix and tillage radish/mustard mix).
The legume cover crop (red clover) produced the greatest N content (i.e., an average of 16 kg N ha −1 in 2020) whereas a lower amount of 4.5 kg N ha −1 was accumulated in the legume-grass mix (oat/berseem clover). This result is similar to others who reported 88% more N accumulation in red clover compared to when it was mixed with grasses (Nyiraneza et al. 2021). However, the average N content observed for red clover in our study was much lower compared to others who found between 34 to 102 kg N ha −1 as cover crop N within temperate, but more humid, climates (Gentry et al. 2013;Coombs et al. 2017). Among the cover crop species we studied, we found that red clover had the greatest N contents. However, this may not always be the case, as fall rye was a close contender. The similarity in N content produced in red clover and fall rye is an indication that grass cover crops can scavenge N and accumulate as much N in the tissues as legumes when considerable biomass is produced.
Overall, cover crops performed rather differently depending on the year and growing conditions, supporting our hypothesis that cover crop biomass accumulation and N content will differ by species and how it is incorporated in the rotation. To support better establishment of cover crops on the prairies in the future--and to sufficiently accumulate N-we recommend that researchers explore early-seeding techniques, such as under-seeding in-season, and selecting hardy species/varieties. Cover crops reduce soil inorganic N levels and supply rates during the non-growing season, but not during the growing season During the fall, the SIN supply rate in the top 10 cm was generally lowered by including cover crops--the effect being most pronounced after canola, pea, and wheat; when Table 7. Mean crop N use efficiency (NUE) indices of NHI (nitrogen harvest index), NUpE (nitrogen uptake efficiency), ANR (apparent nitrogen recovery), and NUE yield (yield efficiency) for wheat, canola, potato, and pea in 2019 and 2020 ± standard error (SEM). oat/berseem clover, tillage radish, and red clover were growing, respectively. This result was not surprising because as the cover crops were growing, they were taking up SIN thereby reducing the SIN supply rates, as likewise observed by others (Dean and Weil 2009;Lapierre et al. 2022). When interpreting the SIN results from the ion exchange strips, it is important to note that the measurements indicate an apparent supply rate rather than an actual supply rate; that the supply rate is also a function of soil moisture. The rye cover crop after potato harvest did not have a pronounced effect on reducing the SIN supply rate in the 0-10 cm; however, the SIN supply rate in the fall-2020 after potato harvest was already relatively low (likely due to the high NHI of potato tubers in 2020) which may have precluded a cover crop effect postharvest. High amounts of SIN are often present at pea harvest (Arcand et al. 2013); thus, it is possible that a cover crop effect on SIN during the fall depends on the amount of SIN remaining at harvest. The ability for cover crops to reduce N available for loss during the post-harvest period has been observed in other studies, especially if the levels of SIN would have otherwise been high, had a cover crop not been grown (Coombs et al. 2017;Kaye et al. 2019).
Other than the post-harvest fall period, there was limited evidence that SIN supply rates were influenced by cover crops--be that spring, summer, or overwinter. However, if cover crops influence the 0-10 cm soil SIN supply rates in the fall, then it is possible that this translates into different N dy-namics in deeper depths, i.e., altering N supply and N movement of deeper depth layers over time. Lapierre et al. (2022), in their study, observed lower nitrate content in the deeper soil layers (20-140 cm depth) when cover crops were present compared to the absence of cover crops. The lower SIN supply rate for the cover cropped rotations observed for the postharvest fall periods was translated to the subsequent spring period when actual soil nitrate contents (pre-plant SIN) were measured. Results from a cover crop study showed that N supply from resin was significantly correlated with SIN in spring prior to cover crop termination (Kaye et al. 2019). Although the spring SIN supply rates in the 0-10 cm depth did not significantly differ with or without cover crops, the 0-60 cm SIN contents at pre-plant were generally reduced with vs. without cover crops in rotation--the effect being most pronounced in 2021 after three years of cover cropping, and when heavy N feeders like canola and potato occurred early in the rotation. A similar presumption was made by Vyn et al. (2000) when they reported lower nitrate concentrations after harvest of wheat. The reduction in 0-60 cm SIN content at this timepoint indicates cover crop N uptake and immobilization before the cover crop tissues have started to decompose/remobilize N. Through this process, it is possible that cover cropping will increase the PMN pool over a longer period of time. Accordingly, the PMN results pointed towards higher N supply with vs. without cover cropping, but after two years of including cover crops in rotation the difference remained in-significant. Perhaps a longer period of study will reveal larger differences.
It is critical to acknowledge that fertilizer N applications directly increase the SIN supply (Lapierre et al. 2022) to the main crops, and in our study the N fertilizer was applied at seeding--the rotations with and without cover crops receiving the same amount of fertilizer (depending only on crop type). As such, the application of fertilizer N might have masked any cover crop effect on SIN supply during the summer period. The amount of biomass produced by cover crops generally dictates its influence on the soil N (Ruis et al. 2019) so other plausible explanations for why cover crops did not markedly influence SIN supply rate in the top 10 cm depth in spring and summer may be due to the relatively low cover crop biomass and (or) the principal factors responsible for the process of N movement, decomposition, and plant N uptake.
Cover crops as part of a wider soil health management practice regulated N cycling with time Indirectly, certain indicators can be used as a measure of soil functions to determine soil health and quality (Allen et al. 2011) and agricultural practices such as crop rotation, cover cropping and organic matter additions to the soil are known to overall improve soil health (Idowu et al. 2008;Chahal et al. 2021). A practice such as one that promotes soil surface cover is recognized as an essential physical soil health indicator (Allen et al. 2011). In that context, we assumed that crop rotations with near continuous soil cover will represent better management practice in terms of influencing N dynamics. In evaluating the relationship between soil health management index and soil N metrics--be that SIN supply rates, NO 3 − contents, and PMN--they all pointed towards a similar pattern: over time, the cover cropped rotation and perennial system helped to redirect N dynamics by reducing SIN availability during the sampling time points but improved the pool of N that would be potentially mineralizable. This indicates that these systems have potential to reduce N losses. The greater chance of agricultural soils becoming sources of nitrate pollution can therefore be minimized under such systems.
Furthermore, the similarity in N dynamics for the cover crop and perennial system can be linked to their characteristic of having near continuous soil cover. This raises interesting questions on if including cover crops in annual crop rotations can help cropping systems to function more like perennial systems. Although a perennial alfalfa cropping system is not diverse in terms of the number of species, it provides continuous soil cover and associated benefits. Tighter N cycling might be achieved by "perennializing" annual cropping systems via the inclusion of cover crops. In a meta-analysis study that looked at how crop rotations influence nutrient cycling, it was concluded that crop rotations diversity, particularly ones that include cover crops improves soil C and N pools (McDaniel et al. 2014). A 6.3% and 12.5% gain in SOC was observed for cover cropped and perennial rotations respectively, when SOC was evaluated in these systems compared to a grain-only system (King and Blesh 2018). In another study that evaluated microbial biomass C and N (MBC and MBN) in different cropping systems, higher MBC and MBN was associated with a 4 year crop rotation (corn-soybean-grain cropalfalfa) than a 2 year rotation with just corn-soybean (King and Hofmockel 2017). Chahal et al. (2021) also reported that diversifying cropping systems with cover crops such as red clover and perennials like alfalfa overall increased soil health indicators. Improved soil health is therefore often associated with diversified and perennialized cropping systems--and regulating N is no exception.
Crop yield and NUE was not influenced by the presence of cover crops Crop yields and NUE were not significantly influenced by having cover crops included in the rotation. Several conflicting reports for how cover crops influence crop productivity have been reported, as studies have shown increased (Farzadfar et al. 2021), decreased (Abdalla et al. 2019, and no change (Dozier et al. 2017) in yields when cover crops were included in the rotation. In a recent survey conducted in the prairies, 24% of farms reported an increase or no change in farm profits when cover crops were included, compared to 4% which saw a reduction in farm profit (Morrison and Lawley 2021). Research suggests that the impact of cover crops on soil properties and crop productivity to a large extent are variable in the short term (Blanco-Canqui et al. 2015;Ruis et al. 2019). Gentry et al. (2013) explained that the timing of sufficient rainfall is a crucial factor in regulating the flush of soil N mineralization and cover crop biomass decomposition. In their case, sufficient rainfall occurred in August after a period of dry conditions in July; this led to a flush of soil N mineralization, corresponding to rapid corn N uptake and remobilization of N into yield portions. Furthermore, the effect of cover crop on yields can be variable, depending on whether it is a legume, non-legume, grass, or a legume-non-legume mixture. For example, legume and non-legume cover crops reduced grain yields whilst a mixture of legume-non-legume increased yields (Doltra and Olesen 2013;Abdalla et al. 2019). Further, a study conducted in Ontario determined N dynamics and corn yields in a cover crop-corn rotation, and reported a positive effect of red clover and alfalfa cover crops on corn yields in one of the 2 years, likely owing to N credits associated with the N-fixing cover crops (Coombs et al. 2017). The N returned to the soil from the crop/cover crop residues serves as a source of nutrients for next crop if mineralization and plant N demand are synchronized. However, Taveira et al. (2020) traced crop residue N into subsequent crops and found that the yield benefits associated with diversified crop rotations may not always be tied to improved N supply from crop residues; instead, possibly due to how diversified rotations impact the N supply from indigenous soil N pools. Likewise in our study, a considerable amount of N may have been supplied by indigenous soil N pool; together with the N fertilizer supply, the cover crop impact on soil N dynamics was not sufficient to translate into crop NUE differences. It is important to emphasize that to trace the fate of N sources and pools, isotope labelling techniques would be useful, and we recommend that future research use labelling techniques to provide more comprehensive understanding of the fate, partitioning, and use of N sources.

Conclusion
Cover crops are multifunctional and can provide several agronomic, environmental, and economic benefits when properly managed. It is said that N cycling can be improved using cover crops, leading to a reduction in potential N losses and further minimizing excessive N fertilizations. Yet, studies have shown several conflicting results of cover cropping, globally. This study was not an exception, as some findings partially supported the hypothesis that cover crops influence soil N cycling but not sufficiently enough to induce changes in crop productivity. Cover crop biomass and N content differed by species; for a short-season growing region like the prairies, selecting the most suitable cover crop is an important choice. During the non-growing season, only after three years did cover crops demonstrate potential for reducing soil inorganic N availability for loss. However, there was limited evidence that cover crops influenced N dynamics during the growing season. As a next step, adjusting N fertilizer applications in conjunction with cover cropping may help improve crop NUE. When cover crops are viewed as part of a wide range of soil health management strategies, then differences began to emerge. There was a greater chance for cover cropped rotations and perennialized systems to modulate soil N overtime. More innovative agronomic research is needed to find ways of practically integrating cover crops into prairie crop rotations--because, unlike other regions, there are unique barriers associated with short-growing seasons with arid and cold climates. Overall, our results point towards key avenues offering promise for improving the practicality of cover cropping on the prairies, such as early-seeding techniques (i.e., under-sowing in-season) and selecting fastgrowing hardy species and varieties. Crop breeding efforts for developing better cover crops for the prairies would be worthwhile.