The field study was initiated on an annually cultivated field located at SE36-20-04-W3 (50.733107N; 106.424019W). This study site was located approximately 7 km south and 7 km east of the town of Central Butte, within the Rural Municipality of Enfield, No. 194, Saskatchewan, Canada. The site is located within the Brown soil zone but is very close (~2 km) to the boundary between the Brown and Dark Brown soil zone. The soils transition from loam to clay loam in texture and are classified as a Brown Chernozem [10]. The soil map unit in which the site is situated describes the site as consisting predominantly of soils of the Haverhill association. Haverhill association can be described as Brown soils that formed in loamy glacial till. It occurs on undulating and undulating dissected landscapes with gentle slopes. Surface texture is commonly described as loam [11]. Prior to the field trial in 2018, wheat (Triticum aestivum) was grown on the site in 2017, and canola in 2016.

To characterize baseline soil properties at the field trial site, including nutrients before the experiment, pre-seeding composite samples of soil were taken from the 0 - 15, 15 - 30 and 30 - 60 cm depths on May 2018. Ten soil cores were taken in a transect across the site area with a Dutch auger. Soil samples were placed in a cooler for transport back to Saskatoon and then stored in a refrigerator at 5˚C until preparation for chemical analysis. Each soil sample collected was air-dried, sieved and the <2 mm fraction was retained and analyzed for various extractable nutrients (extractable P, potassium (K), nitrate ( NO 3 − )-N, sulfate ( SO 4 2 − )-S) and chemical properties (pH, electrical conductivity, % organic carbon) (Table 1).

Post-harvest soil sampling took place in September, 2018. A composite sample was obtained for each plot by collecting two cores within each plot (in-row and between-row) from all treatments and combining the sub-samples according to depth. A hydraulic punch truck fitted with a 5 cm diameter barrel was used to remove samples at three depth increments (0 - 15, 15 - 30 and 30 - 60 cm). Samples were placed in a refrigerator at 5˚C until preparation for chemical analysis, which consisted of air-drying samples at 30˚C followed by grinding with a Wiley mill to pass a 2 mm sieve. After processing, the samples were stored at room temperature until chemical analysis.

^†P and K = Modified Kelowna extractable PO₄-P and K (Qian et al., 1994). ^‡N = CaCl₂ extractable nitrate, NO₃-N (Houba et al., 2000). ^§S = CaCl₂ extractable sulphate, SO₄-S (Houba et al., 2000). ^¶pH measured in a 1:2 soil: water suspension (Hendershot et al., 2008). ^#EC measured in a 1:2 soil: water suspension (Miller and Curtin, 2008).

The experiment was set up as a randomized complete block design (RCBD) with four replicates of each treatment. Four rates of starter N: 0, 10, 20 and 30 kg N ha⁻¹ were used in the form of a blend of 50% urea (46-0-0) and 50% MAP (11-52-0) with a blend analysis of 28-26-0. In addition to N, this supplied 0, 10, 20 and 30 kg P₂O₅ ha⁻¹. The fertilizer products were obtained from a local agricultural fertilizer retailer (Nutrien Inc). Fertilizer was placed in the seed-row with the lentil and soybean under ~15% seed bed utilization (opener spread/row spacing) using a seeding tool designed and utilized for seeding small scale field plots. Modern varieties were selected of soybean (Glycine max L. cv. NSC Watson RR2Y), and small red lentil (Lens culinaris L. cv. CDC Maxim). Soybean was selected as the crop for the N fixation study as there is limited information on how N fixation in short season soybean responds to starter fertilizer in the prairies. Hard red spring wheat (Triticum aestivum L. cv Brandon) was seeded as the reference crop for soybean N uptake and BNF.

The area of each plot was 3.0 m × 1.0 m with 3 seed-rows per plot. Row length was 3.0 m and row spacing was 25.4 cm (10 inch spacing) as this is a common spacing configuration for these crops in Saskatchewan. At the first-trifoliate leaf stage of the soybean, a 1 m × 1 m subplot was established in each of the soybean and reference crop plots. A 10 atom % excess ¹⁵N-(¹⁵NH₄) (¹⁵NO₃) fertilizer was uniformly applied to each subplot at the rate of 5 kg N ha⁻¹ in a liquid form dissolved in deionized water.

The nitrogen isotope dilution method provides direct evidence for N₂ fixation since the ¹⁵N concentration in the plants exposed to ¹⁵N₂ is greater than the 0.366% natural abundance but is diluted relative to the non-fixing reference crop if BNF occurs [14]. The isotope dilution method introduced by McAuliffe [18] was used to determine N₂ fixed (kg·ha⁻¹) by calculating percentage of N derived from atmosphere (%Ndfa) and multiplying by total plant N calculated for the grain and straw separately:

Amount   of   N 2   fixed ( kg ⋅ ha − 1 ) = %   Ndfa × total N in fixing crop 100 (1)

Further, Hardarson and Danso [4] described that in the absence of any supply of N other than soil and ¹⁵N-labelled fertilizer, both plants will contain the same ratio of ¹⁵N/¹⁴N, since they are taking N of similar ¹⁵N/¹⁴N composition, but not necessarily the same total quantity of N. Therefore, the total amount of N derived from atmosphere (Ndfa, kg N ha⁻¹) in plant biomass was calculated by % ¹⁵N atom excess (Ndff) of fixing and non-fixing crop as:

%Ndfa = ( 1 − % N 15   atom   excess F ( fixing crop ) % N 15   atom   excess NF ( non-fixing crop ) ) × 100 (2)

Above equation can be further classified as,

%Ndff NF + %Ndfs NF = 100

%Ndff F + %Ndfs F + %Ndfa = 100

%Ndff NF %Ndfs NF = %Ndff F %Ndfs F

The total N in the fixing crop can be obtained by multiplying the dry matter by the % N as;

Total   N   in   fixingcrop ( kg / ha ) = Dry   matter   of   each   plant × %   N   of   each   plant

Statistical analyses were conducted using PROC GLIMMIX of SAS, Version 9.4 (SAS Institute Inc, 2017). This study was conducted as an RCBD with a total of four treatments with four replicates. An ANOVA was conducted, with treatments as a fixed effect and block as a random effect. Where applicable, mean comparisons were performed using least significant differences (LSD; equivalent to Fisher’s protected LSD) at a significance level of 0.05. PROC UNIVARIATE was used to determine if the measured values of a treatment were normally distributed.

The mean soybean emergence measured at 19 days after seeding (Table 3) shows that at the 20 kg N, 20 kg P₂O₅ ha⁻¹ seed placed fertilizer rate treatment, the emergence was significantly reduced compared to the 10 kg·ha⁻¹ and control (no fertilizer treatment) (P = 0.0270). Similarly, above 10 kg·ha⁻¹, lentil emergence was significantly lower (P = 0.0272).

^†Reported values are the means of four replicates (n = 4) of each treatment. Within a column, means followed by the same letter are not significantly different at P > 0.05 according to Fisher's protected LSD method. SEM = standard error of mean. CV = Coefficient of variation.

Mean soybean and lentil grain and straw yield for the starter N, P fertilizer applied at different rates is shown in Figure 1 and Figure 2, respectively. For soybean yield (Figure 1), there was a trend for fertilization with starter fertilizer to result in higher soybean yield compared to the unfertilized control, with increasing rate resulting in slight decrease in mean grain yield. However, the effect was not statistically significant (P = 0.3994) among the treatments.

For lentil (Figure 2), mean grain and straw yield followed a similar trend to soybean, with slightly higher mean yield at the 10 and 20 kg N, P₂O₅ ha⁻¹ rates, and showed decrease with the treatments from 10 to 30 kg·ha⁻¹. However, even more so than for the soybeans, the yield differences among treatments were small and not statistically significant (for lentil grain and straw, P = 0.5339 and P = 0.1458, respectively. Overall response to fertilization appeared less in lentil compared to soybean.

Mean soybean and lentil N and P uptake in grain and straw is shown in Table 4. In soybean, the N uptake in grain and straw was not significantly increased by N and P fertilization (P = 0.5227). There was a trend for about 10% - 20% higher mean N uptake in soybean yield components with fertilization, compared to control (no fertilizer). Increasing fertilizer rate had little evident impact on soybean grain or straw yield. For P uptake in soybean grain and straw, mean P uptake in soybean grain for all the rates (10, 20, 30 kg N, P₂O₅ ha⁻¹) was significantly higher (P = 0.0425) than the unfertilized control, but among the fertilizer rates there were no differences. Straw N and P uptake by soybean showed a similar pattern to grain N and P uptake, but straw N uptake of fertilized treatments compared to unfertilized control have increased N uptake by 15% - 20%.

In lentils, both N and P uptake were quite similar among treatments (Table 4) and there was no significant difference in N and P uptake of grain in the different treatments, but a trend to higher N and P uptake in the grain. Straw N uptake was significantly higher in fertilized lentil straw. Compared to N and P uptake increases in soybean, the increases in uptake for lentil with fertilization

^† Reported values are the means of four replicates (n = 4) of each treatment. Within a column, means followed by the same letter are not significantly different at P > 0.05 according to Fisher's protected LSD method. SEM = standard error of mean.

versus without were smaller, only about 4% - 7% increase. Overall, soybeans were more responsive to starter N-P fertilizer in both increased yield and nutrient uptake compared to lentil, especially for P.

The mean % ndfa in soybean grain and straw is provided in Table 5. The % ndfa in soybean grain was significantly lower with higher rates of the N, P₂O₅ urea-MAP fertilizer blend added. A reduction in % ndfa in grain and straw of about 10% can be observed at the 30 kg N, P₂O₅ ha⁻¹ rate compared to the unfertilized control that can be mainly attributed to the effect of the added N. This is expected as N fertilizer addition at higher rates is known to significantly decrease the proportion of N in the legume plant derived from biological fixation of atmospheric N.

In this study, the blend of 50% urea and 50% MAP giving a product blend analysis of 28-26-0, supplied an available N to P₂O₅ ratio of 1:1. The mean soybean and lentil plant emergence counts at 19 days after seeding (Table 3) showed the emergence was reduced by 20% - 25% at 20 and 30 kg N, P₂O₅ ha⁻¹ compared to the 10 kg·ha⁻¹ and control (no fertilizer) treatment. In a previous controlled environment study by Weiseth [8], MAP was placed in a seed-row with soybean, and only rates above 20 kg P₂O₅ ha⁻¹ (40, 60 and 80 kg P₂O₅ ha⁻¹) showed a significant reduction in emergence using a soil similar to that of the current study. Greater apparent injury per unit of P₂O₅ observed in the current study compared

^† Reported values are the means of four replicates (n = 4) of each treatment. Within a column, means followed by the same letter are not significantly different at P ≤ 0.05. The multi-treatment comparisons using Fisher's protected LSD method. SEM = standard error of mean. CV = Coefficient of variation. ^‡ % Ndfa denotes percentage of N derived from the atmosphere.

to Weiseth [8] is explained by use of a urea-MAP blend rather than MAP alone and also drier conditions experienced in the field, especially in May and June (Table 2), compared to optimum moisture conditions in a growth chamber. “Fertilizer burn” is related to salt effect of the fertilizer and ammonia/ammonium toxicity [4]. Free NH₃ is toxic and can move freely through the cell wall, whereas NH₄⁺ cannot, which explains why N fertilizers that produce large amounts of free ammonia in the vicinity of the granule, like urea, contribute more to germination and seedling damage. Therefore, obviously a blend with urea will cause more damage than MAP alone due to hydrolysis of urea into NH₃ and CO₂. It is therefore suggested for both soybean and lentil, that as the blend approaches 50% urea and 50% MAP, the starter N, P₂O₅ rate is reduced to 10 kg N, P₂O₅ ha⁻¹.

Low soil moisture and temperature can also adversely impact emergence along with too much fertilizer placed in the seed-row as discussed above. The current recommended minimum average soil temperature at seeding according to Saskatchewan Pulse Growers [2] is 5˚C and 10˚C for lentils and soybean, respectively. Mean air temperature at site at the time of seeding was ~15˚C and this was above the historical mean for the month but the average precipitation was below the average throughout the growing season (from May to August), about 20% and 4% lower in May and August, respectively (Table 2). This low moisture condition in spring seeding may also have aggravated the salt and ammonia injury potential of the seed-row fertilizer, particularly for soybean. Low precipitation along with high sand content and low organic matter content in the site will lead to greater concentration of the fertilizer salt in the soil solution, reduced ammonia absorption by water and colloids, and thereby lead to more injury at high N rates for both soybean and lentil. This result agrees with the controlled environment study done using soil from same field.

In 2017, the average grain yield per unit area for soybean in the area the study was conducted was reported as 1398 kg·ha⁻¹ [1]. The mean soybean grain yield obtained in this field trial in 2018 is higher than the average, with yield of 2003, 1926, and 1862 kg·ha⁻¹ for the rates of 10, 20, and 30 kg N, P₂O₅ ha⁻¹, respectively. The trend of slight yield reduction with higher rates of seed-row urea-MAP blend may be explained from the emergence reduction and reduced plant stand with the higher rates. The lack of large positive yield responses to fertilization reflects the good ability of the legumes, especially lentil, to access existing P reserves in the soil.

In a meta-analysis of soybean grain yield responses to N fertilization [19], there was a positive response to fertilizer N in about half of the published studies, with the average yield increase from N fertilizer addition in the N-fertilizer responsive studies of 520 kg·ha⁻¹ (n = 154). The magnitude of the response did not significantly differ among N rate categories of 0 - 50, 50 - 100 and >100 kg N ha⁻¹. The results of this 2018 field study indicate that 10 kg N, P₂O₅ ha⁻¹ as starter seed-row placed urea-MAP blend would be optimal for the soybean in terms of maximizing plant stand and yield on this soil under the conditions of the study. This is supported by the results of Gai [3] that used urea as a starter fertilizer at four different N rates (0, 25, 50 and 75 kg N ha⁻¹), with an average yield soybean yield around 3100 kg·ha⁻¹ for all the treatments but no significant difference among the treatments.

As for soybean, the mean lentil grain and straw yield was slightly higher in 10 and 20 kg N, P₂O₅ ha⁻¹ treatments than control, and yield showed a decrease from 10 to 30 kg N, P₂O₅ ha⁻¹, but there was no significant difference among treatments. The mean lentil grain yield in this study: 2610, 2523, and 2334 kg·ha⁻¹ at the rate of 10, 20 and 30 kg N, P₂O₅ ha⁻¹, respectively, was higher than the provincial average yield. Slightly lower mean yield with seed-row rates of N, P₂O₅ above 10 kg·ha⁻¹ for lentils observed in this study may be attributed to fertilizer injury reducing emergence and stand count. Therefore, similar to soybean, 10 kg N, P₂O₅ ha⁻¹ as seed row placed starter 28-26-0 fertilizer for lentil can be considered best agronomic performer.

In soybean, grain N uptake was not significantly increased by fertilization, but there was a trend for increased N uptake by about 10% - 20% with N, P fertilization, compared to the control (no fertilizer). The P uptake in soybean grain at all the rates of urea-MAP blend fertilization (10, 20, 30 kg N, P₂O₅ ha⁻¹) was significantly higher (P = 0.0334) than the unfertilized control but among the fertilizer rates the effect was similar. The increase in P uptake can be explained by the P in the fertilizer contributing to soil available P supplies and additional P uptake by roots that was eventually translocated to the grain. For N, increases in soil available N from fertilization may be countered by reduced biological nitrogen fixation. The trend in increased N uptake by soybean with starter N, P₂O₅ fertilization may reflect additional uptake of soil and fertilizer N without equivalent reduction in N derived from biological nitrogen fixation, as well as enhancement of N uptake associated with P fertilization. There is typically a correlation of nutrient uptake with final crop yield for most plant nutrients [4]. This relationship may exist even for legumes that can fix N biologically if it is not present in the soil. The linear relationship between soybean grain yield and total N uptake with above-ground biomass had a slope of 12.7 kg grain per kg N [19]. Further, it was mentioned that on average, a soybean crop yielding 5000 kg·ha⁻¹ accumulates about 400 kg N ha⁻¹ in its aboveground biomass, and that N must be provided from indigenous soil resources, the biological fixation process, and/or fertilizer.

In lentils, both N and P uptake in yield components were similar among treatments and there was no significant difference in N and P uptake in grain and straw in the different treatments. Compared to lentil, soybeans were more responsive to starter N-P fertilizer in increased nutrient uptake, especially for P. Uptake amounts for lentil similar to that obtained in the current study were observed in a study conducted in the Moist Dark Brown soil zone at Scott, SK in 2014 with grain N and P uptakes of 90.1 kg N ha⁻¹ and 7.8 kg P ha⁻¹, respectively [20]. Corresponding to the lack of large yield response of the soybean and lentil to starter N and P fertilization observed at the field site in this thesis, there were also no visual symptoms of N and P deficiency in any of the soybean and lentil treatment plots throughout the growing season. In this field study, for both soybean and lentil, highest grain N and P uptake was recorded at 10 kg N, P₂O₅ ha⁻¹, which had highest grain yield. This is also the fertilizer rate that produced minimal reduction in emergence.

It has been noted by other researchers that little starter N is typically needed for good pulse yield because they can fix their own N from the atmosphere in the symbiotic relationship with Rhizobium once they establish nodules, and too much available N can reduce the contribution from symbiotic fixation [21]. Concomitant with this concept, the % ndfa in soybean grain (Table 5) was lower with higher rates of N fertilizer, with about 10% reduction in the 30 kg N, P₂O₅ ha⁻¹ rate compared to the unfertilized control. A similar effect was observed for % ndfa in straw but was not significantly different among the treatments. In this work, the highest soybean grain % ndfa with no fertilizer was 81%, with 79%, 72%, and 71% ndfa for 10, 20, and 30 kg N ha⁻¹, respectively. The % ndfa values reported for soybean in the current study tend to be slightly higher than other studies. However, nodulation of soybean on 0 N treatment in the field was visually noted to be good. The % ndfa reported for soybean was 60% in [22], 57% in [19] and 50% in [23] and comparatively lower than the results obtained in this work. Nevertheless, a similar ¹⁵N dilution technique in a Black Chernozmem soil near Rosthern, SK has shown similar results with 74% ndfa for N in soybean grain [20].

A negative exponential relationship was typically observed between N fertilizer rate and N₂ fixation when N was applied in the top 0 - 20 cm of soil or on the soil surface [19]. The % ndfa results gained from the field study in this work supports this, as highest % ndfa was in the unfertilized treatment and lowest in 30 kg N, P₂O₅ ha⁻¹. Therefore, the 10 kg N, P₂O₅ ha⁻¹ rate of 28-26-0, which gave a % ndfa that was not significantly different from the control and produced the highest mean yield and N uptake, may be the best treatment for maximizing yield and contribution of biological fixation from soybean.