This study took place at Bison (45˚30'N, 102˚33'W) and Onida, SD, USA (44˚42'N, 100˚15'W) in 2014 and was reduced to just the Onida site (44˚35'N, 100˚05'W) in 2015. This research was conducted on-farm with the Onida site separated by approximately 24 km between years. Selected soil characteristics by site at the initiation of this research are listed in Table 1. The experiment was arranged in a complete randomized block design with four replications for each treatment. The main effect consisted of five N treatments using a target application rate of 90 kg∙ha⁻¹. The N treatments were as follows: Control (0N), 90 kg N ha⁻¹ applied at planting (90AP), 90 kg N ha⁻¹ with N-(n-butyl)-thiophosphoric triamide (NBPT) urease inhibitor (90AP + NBPT), 90 kg N ha⁻¹ split-applied with 50% at planting and 50% applied at the V4 (four true leaves at least 4 cm in length) growth stage (90 SplitV4) and 90 kg N ha⁻¹ split-applied with 50% at planting and 50% applied at the R1 (terminal bud formation) growth stage (90 SplitR1). All fertilizer N was banded at planting as a urea-ammonium-nitrate solution (UAN, 28-0-0). The split-application was dribbled approximately 7 cm off each row and applied by hand. The determination of growth stages were evaluated using the stages developed by Schneiter and Miller [18] .

Sunflower (Mycogen Seeds MY8H456CL, Size 3, Indianapolis, IN) was planted with a no-till grain drill (Model 750, John Deere Co., Moline, IL) at a population of 4.1 plants m⁻² on 9 and 12 June, 2014 for Bison and Onida, respectively. In 2015, the Onida site was planted on 10 June. Thiamethoxam (3-[(2- Chloro-1,3-thiazol-5-yl)methyl]-5-methyl-N-nitro-1,3,5-oxadiazinan-4-imine) (Syngenta, Wilmington, DE) was applied as an insecticide seed treatment at a rate of 0.25 mg a.i. per seed. Each plot consisted of four rows planted at 76.2 cm

between rows and 9.1 m long. For weed control, Sulfentrazone (N-{2,4-Dichlo- ro-5-[4-(difluoromethyl)-3-methyl-5-oxo-4,5-dihydro-1H-1,2,4-triazol-1-yl]phenyl}methanesulfonamide) (FMC, Philadelphia, PA) and glyphosate (N-[phos- phonomethyl] glycine) were applied at planting for each site.

Fertilizer recovery was evaluated using ¹⁵N-labeled UAN (1.366‰ ¹⁵N atom excess) applied in a micro-plot (3.9 m²) established within the center of each plot. This plot size was assumed to be sufficient to eliminate border effects based on research from other crops [19] . Labeled N was applied by hand to mimic application in the larger plot to ensure compatibility. At physiological maturity, four plants were hand-harvested at the soil surface from the center of each microplot. Due to an unusually early frost (September 8, 2014) accurate biomass (lamina) samples were only obtained from the Onida location in 2015. The plants were separated into component parts (grain, head, stalk and leaves), dried at 70˚C for 72 hours, weighed, and ground to pass a 0.5 mm sieve. Nitrogen content and ¹⁵N atom excess were determined using an NC1500 (Carlo Erba, Milan, Italy) automated dry combustion analyzer coupled to an Isoprime (Micromass, Beverly, MA) mass spectrometer. All prepared samples were run in duplicate. The fraction of N derived from fertilizer (Ndff) was calculated as:

and

where F is the fraction of total N uptake derived from ¹⁵N enriched fertilizer, A_s is the atom % ¹⁵N measured in the harvested plant sample, A_f is the atom % ¹⁵N in the enriched fertilizer,
A_r is atom % ¹⁵N of the reference harvested plant material from non-¹⁵N enriched control plots and E_f is the total uptake of ¹⁵N enriched fertilizer and N_u is the total N uptake by the plant or plant component [20] .

The study was arranged in a 5 × 2 × 2 split-split-block factorial arrangement with the N treatment randomly applied within each of four replications. Grain yield, oil content, N uptake and residual N were analyzed statistically as a linear mixed-effects ANOVA model with Satterthwaite’s approximation for denominator degrees of freedom using the lme4( ) and lmerTest( ) modules [4] [21] in R statistical package [1] . All of the data were obtained from the harvested microplots and results are presented at this scale. Nitrogen treatment, a categorical factor with 5 levels and Location, a categorical factor with 2 levels, were analyzed as fixed effects. To account for the high variability between the years of the study and difference in the number locations, Year was analyzed as a random effect, which assumes a different baseline for measured indicators based on each year. Interactions were assessed statistically for each combination of the fixed effects.

Levene’s test was used to check for homogeneity of variance. Residual and Q-Q plots were applied to examine data normality. Further assumptions of the linear package were verified using the gvlma( ) based on Pena and Slate [22] . Significance was determined at P ≤ 0.05 (unless otherwise stated) with means separation determined by Tukey’s Honest Significant Difference method. All linear relationships were analyzed with the lm( ) function using the R statistical package.

In general, the timing of N supply did not have a significant effect on grain yield, N uptake or oil yield. A single application at planting was as effective as a split- application. Hocking and Steer [23] noted that maximum N uptake in sunflower is the period between floret initiation and anthesis, which implies a necessity for adequate N supply just prior to floret initiation to ensure maximum yield. In 2014 in particular, planting was influenced by above average rainfall and cooler average temperatures (Table 2). This likely impacted seed germination, emer gence, and stand uniformity. Overall, grain yield was significantly increased by N treatments. Across all treatments, average grain yield was increased by approximately 65%. Within N treatments, however, grain yield was not statistically different (Table 3). Similar inconsistent results have been recorded in maize

*Statistical significance at 0.05; **Statistical significance at 0.01; ***Statistical significance at 0.001; NS, not significant. †Effects are compared within each column and main effect. Mean values followed by the same letter are not significantly different at P ≤ 0.05.

from numerous possible contributing factors including soil type, weather, rotation and application method [24] . Likewise, N uptake was increased in a similar manner for all N additions; again with no statistical differences among N timing treatments. Moreover, percent N in grain followed a similar trend, but the addition of N did not significantly increase N concentrations in any of the treatments (Table 3).

Oil content did not differ by N treatment but did vary significantly by location with Bison averaging 36.3% and Onida averaging 44.6% (Table 4). Moreover, the yearly effect (data not shown) was significant (P < 0.001) at Onida with oil content averaging 41.1% in 2014 and 47.7% in 2015. However, for oil yield (oil content x grain yield) both N treatment (P = 0.02) and Location (P < 0.001) were statistically significant effects, following a similar pattern as overall grain yield trends. Figure 1 shows the large disparity in oil yield between the two study sites. Due to the heavy rainfall early in the growing season in 2014, oil production was extremely low, particularly at the Bison site.

In general, applying fertilizer N increased oil yield. Because oil content was not materially different between treatments, this effect was largely due to the increased grain yield. Contrary to our initial hypothesis, a later N application did not increase either oil content or yield. The addition of a urease inhibitor was as effective as a split-application and in fact produced the highest oil yield on aver-

†Effects are compared within each column and main effect. Mean values followed by the same letter are not significantly different at P ≤ 0.05.

age. However, this difference was not statistically different than other N addition methods.

Based on ¹⁵N analysis, approximately 27% of the N in the grain was derived from the UAN fertilizer, which indicates a relatively large reliance upon soil N for final grain N content. The addition of a urease inhibitor significantly increased average fertilizer uptake by nearly 6% to 32.7%. When compared to the standard practice of applying UAN at planting, the urease inhibitor showed a trend toward increased yield as well, which improved the overall efficiency of the fertilizer.

Moreover, the sources for the final N concentration in the grain are a mix of N derived directly from the soil and N mobilized by the photosynthetic apparatus, with little contribution from the stalk [23] . Early studies determined that uptake of N versus mobilization of N to the sunflower grain varied from 57% and 43%, respectively, for sunflower plants with high N supply to 25% and 75%, respectively, for sunflower plants with low N supply [23] [25] .

Excepting grain, location was not a significant treatment effect for Ndff in any of the analyzed plant components (Table 5). This result suggests that fertilizer N uptake to the other plant components is relatively stable over broad environ

Statistical significance at 0.10; *Statistical significance at 0.05; **Statistical significance at 0.01; ***Statistical significance at 0.001; NS, not significant. †Effects are compared within each column and main effect. Mean values followed by the same letter are not significantly different at P ≤ 0.05.

mental conditions. However, N derived from fertilizer, as a percent of total N in the component part, varied significantly by application timing (Table 5). In general, as N was applied later in the season, the concentration of fertilizer N in each plant component decreased. For example, a split-application significantly reduced the Ndff in the leaves as would be expected based on the timing of the N supply and the growth stage of the plant. Conversely, the addition of a urease inhibitor at planting increased the proportion of fertilizer N in all plant components by 3.7% - 7.7% on average over the other N treatments.

In theory, split-applying N provides a benefit over an at-planting application because the plant can better compete with early season environmental losses (i.e. leaching, denitrification, etc.) [8] . Based on the lack of yield response in the current results, it appears that the target N application rate (90 kg∙ha⁻¹) was sufficient to maintain yield despite any early season losses.

This is likely due to a substantial reliance upon soil N for the plant’s needs. When averaged across all treatments, fertilizer met only 30% of any sunflower component’s needs. By applying half of the total N fertilizer at planting in the split-application treatments, the plant appears to simply shift its N source to soil reserves. However, this appears to come at the expense of relying more on soil N for its needs. Therefore, the resulting yields are similar between treatments, but they are achieved through different N sources. This dynamic likely played a large role in the location differences. Bison had much lower starting soil N and low organic matter content leaving this site with much fewer N reserves for the plant to draw upon (Table 1). However, these results must be regarded carefully. A full accounting of N use efficiency must factor in total N uptake, which requires measurement of total biomass. As mentioned previously, due to an uncharacteristically early frost in 2014, a significant portion of the leaf matter was sloughed off and thus unaccounted for.

However, as shown in Figure 2, there is a strong linear correlation between grain yield and Ndff. Moreover, the correlation between individual plant component Ndff is similarly robust (Figure 3). This suggests that as grain yield or biomass increases, so too does fertilizer N uptake and hence, FNUE. Based on 2015 data at Onida, the average whole-plant FNUE was 57% (unpublished). In comparison, Scheiner et al. [7] found an overall FNUE of 51% with a target N application rate of 75 kg∙ha⁻¹.

In the drier climate at Onida, the Ndff from the 90AP + NBPT treatment did not differ from the 90AP treatment but was significantly greater than either split-application. Conversely, at Bison where rainfall was high early in the growing season, the Ndff from the 90AP + NBPT treatment was significantly greater than the 90AP, but not significantly different than the split-applications (Figure 4). These results suggest that there was likely some increased environmental N losses through increased rainfall and that the urease inhibitor likely protected the UAN against these conditions by reducing the conversion to NO₃-N

throughout the early growing season, which allowed for greater fertilizer N uptake. However, due to a lack of data with respect to split-application and slow- release fertilizers on sunflower, these results should be verified through additional trial replications across a broader environmental gradient.