A non-irrigated research study was conducted during two growing seasons (2013–2014) in the US Midwestern Corn Belt region. In 2013, field experiments were established at the Purdue University Agronomy Center for Research and Education (ACRE) near West Lafayette—IN (Lat 40.486675° Lon 87.004635°, elevation 216 m) and at Pinney-Purdue Agricultural Center (PPAC) near Wanatah—IN (Lat 41.445113° Lon 86.943464°, elevation 222 m). At ACRE, soil was a Chalmers silty—clay loam (Fine—silty, mixed, superactive, mesic Typic Endoaquolls) and at PPAC soil was Sebewa loam (Fine—loamy over sandy or sandy—skeletal, mixed, superactive, mesic Typic Argiaquolls). In 2014, one experiment was established at ACRE (Lat 40.493593°, Lon 86.493593°, elevation 216 m). Soil type was a Raub—Brenton complex (Fine-silty, mixed, superactive, mesic Aquic Argiudolls).

Maize field trials were established following soybean [Glycine max (L.) Merr.] at all three site-years. In 2013, tillage operations prior to planting maize involved fall chisel plow followed by a spring field cultivator. For the 2014 site, fall plus spring strip-tillage on the no-till soybean stubble preceded maize planting (with precision guidance in all operations).

Experiments were planted with a four—row precision planter (Almaco SeedPro 360) with 76 cm row spacing, to achieve a final plant density of ~79,000 plants ha⁻¹. In 2013 plot lengths were shorter (6.7 m) than in 2014 (13.7 m) due to seed supply limitations. Plant populations were evaluated in 5.3 m row length sections of all four rows at the V5 growth stage.

Weather data was acquired on a daily basis from planting to harvest (1 April—May to 31 October 2013 and 2014) from automated weather stations operated by the Indiana State Climate Office proximately located to the research sites. The field experiment at ACRE in 2013 was planted on 14 May and harvested 10 October. The total growing season precipitation was 385 mm; maximum and minimum daily average air temperatures were 26.9 and 14.6°C, respectively, for the entire growing season (Table 1). The field experiment at PPAC in 2013 was planted 5 June and harvested 30 October. Total growing season precipitation was 626 mm; maximum and minimum daily average temperatures were 24.7 and 12.1°C, respectively, for the entire growing season (Table 1). The field experiment at ACRE in 2014 was planted on 25 April and harvested 25 September. Total precipitation was 592 mm and maximum and minimum daily temperatures averaged 26.2 and 13.6°C throughout the growing season (Table 1). Weed pressure was minimal at all locations.

The two N rate treatments were 0 (0N), representing soil indigenous N pool and high N stress, and 112 kg N ha⁻¹ (112N) as a moderate level of stress; the latter was sidedressed at the V4 stage between corn rows with a DMI Nutri-Placr at 10–12 cm soil depth as Urea Ammonium Nitrate (UAN) (28-0-0). The 0N plots received the same machine pass to avoid variations in plant growth performance due to soil compaction.

Comparisons were made between four modern hybrids varying in NUE with a similar 114 relative maturity range (Dow AgroSciences, Inc., Indianapolis, IN) and one historic hybrid from the 1970's. In this paper, results will be presented as combined means of genotypes to place emphasis on the “multi-stage pulse labeling” technique; a future paper will focus on the methodology's ability to separate hybrid differences.

At all sites, the plot arrangement was a split-plot experimental design, consisting of six replications and two treatment factors. Nitrogen rates were the main plots and hybrids were the sub-plots. The experiment consisted of six replications (because of cost constraints and small plot size, only three replications were used at a time for ¹⁵N application).

Statistical analyses were executed with SAS GLM (SAS Institute, 2014) at each growth stage and each year separately. The whole unit error (N rate) was pooled with the subunit error (Hybrid) for all ANOVAs because it was not significant for the majority of analyses (p > 0.25). A combined location analysis of variance was performed in 2013 for each variable for all growth stages where the error variances were homogeneous for the majority of the stages (p > 0.01). Data presentation in this publication is confined to the combined means of both locations in 2013 (as there were no significant treatment differences between locations in a majority of the observations), and the mean of ACRE in 2014. Location by treatment interactions were pooled with the experimental error when not significant (p > 0.01). Fisher's protected least significant difference (LSD) test was used to compare treatment means for each growth stage where the corresponding ANOVA F-test was significant (p ≤ 0.05).

Differences between growth stages were analyzed with Tukey's HSD test (p ≤ 0.05) using the R program (R Development Core Team, 2005).

Near the beginning of the growing season, single-row ¹⁵N microplots were identified in three replications and were comprised of five (2013) or six (2014) consecutive plants at the appropriate density with sufficient undisturbed border plants (i.e., beyond 1 m from the microplot) so as to maintain a uniform canopy surrounding each plant group.

Microplots were labeled with 3.17 kg of ¹⁵N ha⁻¹ in 2013 and 2.12 kg of ¹⁵N ha⁻¹ in 2014 as Ca (¹⁵NO₃)₂ containing 98 atom % ¹⁵N (SIGMA-ALDRICH Co., St. Louis, MO). The labeled fertilizer provided 0.041 or 0.027 g of ¹⁵ N plant⁻¹, respectively, for 2013 and 2014 growing seasons which represents a very small portion relative to the main N rate of 112 kg ha⁻¹ (equivalent to 1.42 g of N plant⁻¹). The ¹⁵N rate was lowered in 2014 to reduce costs after we learned that we could detect ¹⁵N concentration differences in all plant components during the entire 2013 season. Our ¹⁵N rates were similar to the 3.0 kg of ¹⁵N ha⁻¹ applied by Ta and Weiland (1992) with a much lower plant density of ~60,000 plants ha⁻¹.

Injection of ¹⁵N occurred in multiple plant development stages (from late vegetative stage to near the end of the grain filling period (Table 1). Prior to each time of ¹⁵N application, the labeled fertilizer was diluted in water (~300 ml of water per gram of Ca (¹⁵NO₃), placed in 30 ml plastic syringes, properly sealed, and taken to the field. In the field, screwdrivers were used to make holes ~15 cm deep into the soil at a distance of 15 cm perpendicular to the row on both sides of each plant in the microplot. The ¹⁵N was injected into the hole.

Immediately after the ¹⁵N application, PVC pipes (0.6 diameter and 30 cm length) were installed on top of each hole and 0.8 l of water was applied (for a total of 1.6 l per plant). The added water was intended to ensure immediate availability of labeled nitrate-N to the plant roots.

Application dates for ¹⁵N during the late vegetative stages were determined based on when 50% of plants in the entire experiment had fully expanded leaf collar for each targeted stage. Application timing during reproductive stages occurred when over 50% of plants were in early R1 (ER1) (50% silking), R1 (90% silking), R2 (kernel blister), R4 (kernel dough), R5 (kernel dent) (Abendroth et al., 2011). Development stages of biomass harvesting varied slightly among locations and years (Table 1). In general, development stages refer to the time of plant biomass harvest 4–6 days after the ¹⁵N application (except for the V16 stage at ACRE in 2013 that indicates the time of ¹⁵N application).

At ACRE in 2013, plants treated with ¹⁵N at the V16 stage were in the field for 6 days (from ¹⁵N application 17 July to biomass harvesting 23 July) and reproductive development was so rapid that 90% of plants in the entire experiment had completed silking by the time of biomass harvest. Plants treated with ¹⁵N at the ER1 stage were harvested 4 days later (23 July). Subsequently, the microplots labeled as R1 received ¹⁵N when 90% of plants in the experiment had completed silking. For PPAC in 2013, we achieved better separation of the growth stages close to flowering (in part because of cooler temperatures). Thus, because the V16-stage samples at ACRE also covered the tassel formation period we decided it was still appropriate to combine the V16 results at ACRE with the V16 at PPAC. However, ER1 stage was evaluated only at ACRE and the R1 stage represents the combined means for ACRE and PPAC in 2013.

For both locations in 2013, the six actual biomass harvesting times and further ¹⁵N evaluations were generally completed at the same growth stages except that the V14 harvesting only occurred at PPAC and the early R1 harvesting occurred only at ACRE. Therefore, the 2013 results from V16, R1, R2, R4, and R5 stages consistently represent the means of the two locations (Table 2). In 2014, when there was only one location (ACRE), one of the pre-R1 stage treatments was substituted for a duplicate R1 application time that was then not harvested until the R6 stage (Table 1).

Destructive aboveground biomass of maize was collected from the three (2013) or four (2014) middle plants in the previously selected micro-plots from 4 to 6 days after the ¹⁵N application. One additional micro-plot application was implemented in 2014 whereby plants receiving ¹⁵N at R1 stage were not harvested until the R6 stage (R1/R6). Plant biomass was removed by cutting at the base of the stems with the use of a lopper. Plants were partitioned into stems + leaf sheaths + tassels (stems), leaf blades (leaves), husks + shank + silks (husks), and either whole ear (prior to R4) or cobs and kernels at R5 in 2013, and at R4, R5, and R6 stages in 2014. Stems were chopped into sections immediately after harvest and leaf removal. All samples were dried for 7 days at 60°C, weighed and coarse ground. Stems, leaves and husks were ground to pass a 2 mm screen and cobs and kernels were ground to a pass 4 mm screen size. Sub-samples were taken to the isotope laboratory to proceed with the mass spectrometry analysis.

In order to estimate ¹⁵N uptake (¹⁵Nu) in plant components, N concentration (Nc) and delta ¹⁵N (δ¹⁵N ‰) were determined in the mass spectrometer for each sample. Nitrogen concentration represented the amount of ¹⁴N + ¹⁵N in the sample. Delta ¹⁵N (δ¹⁵N ‰) was the proportional amount of ¹⁵N in parts per mil in the sample relative to an international isotope standard of Air (0.366 atom % ¹⁵N) (Shearer and Kohl, 1986; Fry, 2006; Van Cleemput et al., 2008).

Additionally, δ¹⁵N ‰ is also described as the ¹⁵N atom% excess and can be calculated in the mass spectrometer through the isotopic ratio (IR =¹⁵N/¹⁴N) in the compounds. Absolute amounts of isotopic N cannot be simply estimated due to its atomic unit (Mariotti, 1983). Therefore, the linear relationship between ¹⁵N contents and δ¹⁵N values allows accurate estimation based on the isotope ratio (IR) approach in the mass spectrometer (Fry, 2006).

The isotopic ratio in a standard (IRstrd) represents the amount of ¹⁵N in the standard relative to the ¹⁵N abundance in the atmospheric dinitrogen (N₂) defined as 0.366 atom % ¹⁵N (Faust, 1981; Shearer and Kohl, 1986; Dawson et al., 2002; Fry, 2006; Van Cleemput et al., 2008). The isotopic ratio in the sample (IRsmp) represents the quantity of ¹⁵N in the sample relative to the amount of ¹⁵N in a given standard.

The equations used to calculate ¹⁵Nu in plant components are as follows:

The isotopic ratio in the samples (IRsmp) was calculated through the proportional ¹⁵N abundance (δ¹⁵N) in the plant samples relative to the ¹⁵N abundance in the atmospheric dinitrogen (N₂) defined as 0.0036764 ‰ (IRstrd) (Fry, 2006; Van Cleemput et al., 2008) (Equation 1): (1)IRsmp=(δN151000+1)*0.0036764

Results on the fate of the recently applied ¹⁵N fertilizer in the above-ground plant components are presented using three parameters: total ¹⁵N uptake per unit area (¹⁵Nu, kg ha⁻¹), the proportional allocation of the total ¹⁵N uptake (¹⁵Np, %) in each separate plant component, and plant ¹⁵N recovery efficiency (¹⁵Nrec, kg kg⁻¹¹⁵N applied) of the total labeled fertilizer applied (3.2 or 2.1 kg of ¹⁵N ha⁻¹) at each application time for 2013 and 2014, respectively. Hybrid differences in ¹⁵Nu and ¹⁵Np of individual plant components, as well as ¹⁵Nrec, for each stage will not be discussed in detail in this paper but the specific results for each hybrid, and the full main, sub and interaction statistics for each stage are available in the Supplementary Tables S3 and S4. Results were averaged across hybrid treatment for each N rate (0 and 112N). Supplementary Figure S1 demonstrates hybrid impacts on ¹⁵N allocation to plant components at R1 and R5 stages in 2013.

In 2013, at the V14 stage, mean ¹⁵Nrec (averaged across five hybrids) represented 0.25 and 0.30 kg kg⁻¹ of the total ¹⁵N applied, respectively for 0 and 112N (Table 2 and Figure 1) and ~50% of that ¹⁵Nu was accumulated in the leaves and 50% in the stems (Figure 2). The maximum ¹⁵Nrec within the 6-day period following ¹⁵N application was observed near the critical period bracketing silking; at the V16 stage plants recovered almost half of the ¹⁵N applied (~0.45 kg kg⁻¹) for both N rates, and ~50% of that ¹⁵Nu accumulated in the leaves followed by 35% in the stems, 15% in the husks (Table 2 and Figure 2). Only a negligible amount was found in the ear shoots that were just beginning to enlarge (data not shown). Nitrogen rate had no impact on ¹⁵Np in leaves; however, the 112N treatment increased ¹⁵Np to stems, while decreasing ¹⁵Np to husks. At the ER1 stage (when ~50% of plants in the entire experiment had visible extruded silks), maize plants recovered almost 0.40 kg kg⁻¹ of the ¹⁵N fertilizer applied, and of this ¹⁵N uptake about 48% was allocated to the leaves, 36% to the stems, and 15% to the husks (Table 2 and Figure 2). At this ER1 stage, plants that were stressed due to the lack of N applied at 0N partitioned significantly more ¹⁵Np to the leaf and husk components than those at 112N (Table 2 and Figure 2). Similar patterns of ¹⁵Np between V16 and ER1 stages was probably due to overlapping days of ¹⁵N application and biomass harvesting at the ACRE site. Reducing the time from ¹⁵N application to biomass harvesting lowered total ¹⁵Nu at ER1 as expected (Figure 2). By dividing total ¹⁵Nu by number of days that plants remained in the field (4–6) after ¹⁵N application, it was observed that there was a similar daily ¹⁵Nu of ~0.30 kg of ¹⁵N ha⁻¹ day⁻¹ between those two stages at ACRE location (data not shown).

In 2013, at the end of R1 stage (i.e., when >90% of plants completed silking), the ¹⁵Nrec declined, relative to previous stages, to about 0.32 and 0.34 kg kg⁻¹ of the ¹⁵N applied, respectively, for 0 and 112N (Table 2 and Figure 1). At the R2 stage, plants recovered similar amounts of ¹⁵N as the R1 stage, but more ¹⁵Nu was partitioned to the ears (35 and 27% at 0 and 112N respectively) at the R2 stage than at R1. Additionally, at the R2 stage leaves and ears received equal ¹⁵Np in the 0N treatment (~35%). In the R4 stage it was observed that, out of 0.25–0.33 kg kg⁻¹ of the ¹⁵N recovered by plants (at 0 and 112N, respectively), about 58 and 51% of the accumulated ¹⁵Nu was allocated into the ears, followed by 26–33% to leaves (Table 2 and Figure 2). Although the R5 stage presented the lowest total ¹⁵Nu of the growing season, plant ¹⁵Nrec was about 0.18 or 0.28 kg kg⁻¹ of the ¹⁵N applied, respectively, for 0 and 112N rates. Approximately 68% of this late-stage ¹⁵N uptake was allocated to the kernels, followed by ~17% to the leaves, 11% to stems, and 3% to the husks.

In 2014, ¹⁵Nrec was nearly constant across all growth stages from V15 to R5 stage (Figure 3). There were larger differences across N rates this year than in 2013. At the R1 stage mean ¹⁵Nrec was about 0.21 and 0.42 kg kg⁻¹, at 0 and 112N respectively (Table 3 and Figure 3), and, when averaged over both N rates, ~40% of this ¹⁵Nrec accumulated in the leaves, 25% in the stems, 11% in the husks, and a significant amount of 23% was allocated to the ears (Figure 4). At the R2 stage, plants recovered about 0.24 and 0.44 kg kg⁻¹ of the ¹⁵N applied, at 0 and 112N respectively, allocating 31% to the leaves, 17% to the stems, and 44% to the ears on average of both N rates (Table 3 and Figure 4). Even with a smaller ¹⁵N fertilizer rate applied in 2014, plants recovered ~5% more of ¹⁵Nu at R2 in 2014 than in 2013, and a smaller proportion of ¹⁵Nu was partitioned to the vegetative components while higher amounts of ¹⁵Np were found in ears in 2014 than in 2013. In contrast to 2013, at the R4 stage in 2014 ¹⁵Nu allocation to the ears (60%) was not statistically different for both N treatments. At R5, lower ¹⁵Np was found in the ears (62%) at both N rates in 2014 than in 2013.

For the N rate factor, plant ¹⁵Nrec was almost always N rate dependent (Tables 2, 3). Furthermore, results suggest a strong influence on both ¹⁵N accumulation and the ¹⁵N allocations to the sink and source organs. A higher N rate generally helped increase ¹⁵Nu by maize plants at all development stages. Even though leaves accumulated higher amounts of ¹⁵N at the moderate N rate of 112N, before the flowering period in 2013 it was observed that ¹⁵Np to leaf components was higher with 0N treatment than at 112N. Conversely, both ¹⁵Np and ¹⁵Nu in stem components were higher at 112N than with the 0N. In the same year, plants with more N deficiency stress allocated more ¹⁵N to their reproductive ear organs at 0N than with the 112N rate. Except for the R5 stage, ¹⁵Np to the ears (cobs + kernels) following ¹⁵N application during reproductive stages was higher at 112N. In 2014, ¹⁵Nu accumulation as well as the proportional allocation of ¹⁵Nu to leaves was higher at 112N than at 0N for the entire season. In contrast to the previous year, the allocation of the ¹⁵Nu to the stems, husks and cobs was significantly larger at the 0N than with the 112N during the reproductive period.

Through the “multi-stage pulse labeling” ¹⁵N method it was also possible to estimate the extent to which plants uptake N during the growing season (Figures 5, 6). The amount of ¹⁵N taken up per daily cumulative GDD was calculated by dividing total plant ¹⁵Nu (kg ha⁻¹) (Tables S1, S2) by the GDD °C (base 10°C) units accumulated from ¹⁵N application to plant biomass harvest (Table 1).

In 2013, the 112N treatment increased amounts of ¹⁵Nu per GDD after the flowering period, from R2 to R5 stages, relative to the 0N treatment. However, in 2014, ¹⁵Nu per GDD was higher at the 112N rate in comparison to the 0N rate for all stages.

For both years, the period of greatest ¹⁵Nu per GDD occurred from V14 to R4 stage, and the lowest ¹⁵Nu per GDD was observed at the R5 stage for both N rates (Figures 5, 6). In 2013, although plants showed higher ¹⁵Nrec at the V16 relative to the reproductive stages (R2, R4, and R5) (Figure 1), the actual rate of ¹⁵Nu accumulation per GDD was similar across stages, except that ¹⁵Nu per GDD was much lower at the R5 stage (Figure 5).

For both years, the 112N rate significantly increased dry matter accumulation (DM) per GDD in comparison to the 0N treatment (Figures 7, 8). There was no interaction between growth stages and N rate (data not shown). In 2013, the highest rate of DM accumulation was observed from V6 to the R4 stage with about 20 kg ha⁻¹ per GDD on average of both N rates (0 and 112N) (Figure 7). In 2014, rates of DM accumulation per GDD were relatively consistent from V6 to R5 at ~20 kg ha⁻¹ per GDD °C (Figure 8), or for a longer period relative to 2013. Additionally, the apparent lower DM accumulation from R2 to R4 stage could be explained by the lower GDD accumulation during that period.

Our results suggest that plant N uptake is strongly driven by both the sink strength and source availability. In 2013, plants took up similar rates of the ¹⁵N applied up to the R4 stage and DM accumulation began to decrease from R4 to R5 stages. In 2014, with possibly more acute soil N deficiency, ¹⁵Nu per GDD decreased even with a continual DM accumulation up to the R5 stage.

In the “long term” labeling method implemented in 2014, the amount of ¹⁵Nu accumulated during the grain filling period of the ¹⁵N applied at silking (R1) was measured at physiological maturity in the duplicate R1R6 microplots (Table 3 and Figures 9, 10). At maturity ¹⁵Nrec was dramatically affected by the sidedressed N treatments; plants recovered ~0.45 and 0.72 kg kg⁻¹ of the ¹⁵N fertilizer applied at R1, respectively, for 0 and 112N (Figure 9). Although the proportional allocation of the ¹⁵Nu (¹⁵Np) in the whole ear (cob + kernel) was not affected by the N rate, more ¹⁵Np to the kernels was observed with 112N (Table 3 and Figure 10). Overall, averaged over both N rates, ~77% of that accumulated ¹⁵N uptake at flowering (from the ¹⁵N application at R1) was partitioned to the kernels at maturity, followed by ~7% to the leaves, stems, and cobs and 4% to the husks. In the pulse labeled method, total plant ¹⁵Nrec averaged ~0.21 and 0.42 kg kg⁻¹ of the ¹⁵N applied in the ~5 day period at R1; thus the remaining ¹⁵Nrec (0.24 and 0.30 kg kg⁻¹) that was found at R1R6 was the ¹⁵Nu that was recovered during the rest of the grain filling period.

We also used the net change approach to calculate the post-flowering ¹⁵N uptake per unit area (post¹⁵N, kg ha⁻¹) as (Total plant ¹⁵Nu at R6 minus Total plant ¹⁵Nu at R1), to estimate remobilized ¹⁵Nu per unit area (¹⁵N rem, kg ha⁻¹) as (Total plant ¹⁵Nu at R1 minus Stover ¹⁵Nu at R6), and to estimate the ¹⁵N remobilization efficiency (¹⁵NremEF, kg kg⁻¹) as (Total plant ¹⁵Nu at R1 minus Stover ¹⁵Nu at R6 divided by Total plant ¹⁵Nu at R1) (Table 4). Stover was the sum of all plant components except kernels. Because we estimated the ¹⁵Nu at the silking period (multi-stage pulse labeling method) (Table 3), it was also possible to differentiate the proportion of the ¹⁵Nu in the grain that originated from the ¹⁵Nu accumulated during the reproductive period [i.e., post-flowering ¹⁵N uptake (post¹⁵N)], to the amount of ¹⁵Nu remobilized (¹⁵Nrem) from silking to maturity (Table 4).

Pan et al. (1986) calculated Post¹⁵N uptake (Post¹⁵N) in maize using the net change approach; however, they sampled plants at the R1 stage on the same day that ¹⁵N was applied for the “long term” evaluation. Therefore, it might not be possible to estimate post¹⁵N and ¹⁵Nrem by the net change approach in our study since we applied ¹⁵N at the same time for both pulse labeled and long term approaches. Thus, from this method of calculation, the post¹⁵N represented about 54–42% of the total ¹⁵Nu accumulated from R1 to R6, and the ¹⁵Nrem accounted for 22–36% of the total ¹⁵Nu at R6, respectively, for 0 and 112N treatments. For both sources of ¹⁵Nu to the grain it was found that post¹⁵N contributed more at 0N than with 112N. Post ¹⁵N averaged ~70 and 53% of the grain ¹⁵Nu at 0 and 112N, respectively, while ¹⁵N rem contributed ~30–47% to grain ¹⁵Nu at R6, respectively, for 0 and 112N. The higher post¹⁵NuGr at 0N that was calculated by the net change from R1 to R6 may be explained by the larger effect of the N rate on the actual quantities of ¹⁵Nu.

In contrast to most studies using labeled N, this research describes the potential implementation of the “multi-stage pulse labeling” ¹⁵N application in field-grown maize; this approach involves tracking ¹⁵N uptake and identifying changes in the proportional allocation of the recent ¹⁵N taken up into plant components at multiple maize development stages when field maize is grown under limited N supply. It is well-known that plants with superior ability to remobilize N to sink components yield more (Cliquet et al., 1990a), and that the higher ability of modern hybrids to continue accumulating N after the flowering period with efficient N allocation to its sink organs has increased both NUE and yield (Ciampitti et al., 2013). For these reasons, we regard the use of the labeled N technique to accurately estimate the fate of the recent N uptake in specific plant components over time, and especially during reproductive growth stages, to be a potentially powerful phenotyping tool.

For this 2-year study using the “multi-stage pulse labeling” technique, and after combining means of five hybrids, at two N rates (0 and 112N), it was observed that prior to flowering the distribution of short-term ¹⁵N uptake to leaves and stems was not affected by the N rate treatment as much as the actual amount of ¹⁵N uptake (except for V16 stage in 2013). Paponov and Engels (2005) also observed no effect of N rates (873 and 2,626 mg of N per plant) on ¹⁵N allocation to plant components during the vegetative and reproductive period. In our study, at late vegetative stages, maize plants primarily allocated ¹⁵Nu to their main sinks such as leaves and stems followed later by the husks and immature earshoots. Approximately 50% of the total ¹⁵N uptake was allocated to leaves for both N rates (0 and 112N). In contrast to our results, Cliquet et al. (1990b) analyzed N allocation in the below and aboveground maize plant components (8 days after ¹⁵N application) and found that stems were a stronger sink component than all plant organs at late vegetative stage; out of the total plant new N uptake, stems accumulated ~45% of the recent N taken up, followed by the leaf blades (22%) and roots (18%). For both N rates in our study, stems and husks appeared to act as transitional destinations of the ¹⁵Nu, and ears became an increasingly powerful sink component. Stems appeared to largely function as a “pipeline” where the ¹⁵N passed through accumulating primarily into the leaves from V14 to ~R2 stage at first and then to the ears at ~R2 or R4 and R5 stages. Ma et al. (1998) also confirmed the movement of the ¹⁵N to major sink component around the silking period. In their study, 3 days after ¹⁵N application (by stem infusion) at silking, plants accumulated ~40% of the new N in reproductive organs vs. 30% to the leaves and ~18% to the stems. However, in contrast to our plant partitioning procedure, they included husks as a reproductive organ (which may have resulted in their higher allocation values to reproductive plant parts at R1).

During the grain filling period, the moderate N rate (112N) in our study helped to maintain N contents in maize plants longer in the season, especially in the leaf components, while the 0N treatment showed the dynamics of ¹⁵Np in plant components deficient in N. Thus, our moderate N supply (112N) increased absolute amounts of ¹⁵Nu in the plants but it did not necessarily increase the allocation of ¹⁵Nu to the sink organs. Maize plants increased the allocation of the most recent ¹⁵Nu to the developing organs under the 0N rate compared to the 112N rate. Likewise, Paponov and Engels (2005) acknowledged the sink power of the ears over the vegetative plant parts at low N rates.

In agreement with our study, Rajcan and Tollenaar (1999) examined the effects of source: sink ratio on the N uptake during the reproductive period and demonstrated that after flowering the accumulation of N gradually declined in the vegetative plant parts with more severe declines occurring at lower N rates, while N accumulation increased into the ears. Our results show that the primary destination of the most recent ¹⁵Nu at both N rates was similar during the vegetative period but higher to the leaf component at R1 (~42%) and R2 stages (~38%), then increased to the ears at R4 (~55%), and kernels at R5 stages (~70%). The latter overall pattern of N partitioning in the plant was also confirmed by other authors using different labeling approaches. Crawford et al. (1982) provided ¹⁵N to the plants by nutrient solution just once at the time of pollen shed followed by multiple harvests. Yoneyama et al. (2003) used a leaf feeding labeling method to assess N assimilation in several crops including maize. According to the latter research, in general, N transportation within plants is driven by the two elements of growing plant components and reserve organs, and once N is absorbed in the reserve organs and mature leaves it will probably be reallocated to a developing organ.

The main N treatment influenced the amounts of plant ¹⁵N uptake and allocation of ¹⁵N at all growth stages; however, it did not alter the overall vegetative distribution of the ¹⁵Nu within the plants, and leaves always accumulated more ¹⁵N than stems. The multi-stage pulse labeling N approach measured the distribution of the recent N uptake by the plants, demonstrating that even at late grain fill plants were actively taking up N and mainly allocating the most recent N to the ears and/or kernels. Therefore, comparison between N treatments in this study was crucial to observe plant response to the N supply until late in the season and also to examine how plants change allocation patterns when they are source limited or stressed due to lack of N. For both N treatments, the largest amount of ¹⁵Nu and ¹⁵Nrec within 4–6 days of application was observed at the V16 stage and the lowest at the R5 stage. However, rates of ¹⁵N uptake per GDD°C were fairly consistent across stages, and only declined at R5. Correspondingly, ¹⁵Nrec was as high as 0.46 (kg kg⁻¹) at 112N to as low as 18 (kg kg⁻¹) at 0N. The latter suggests that further investigations in late vegetative stage or late reproductive stages may help to identify plant processes related to its N uptake efficiency.

Our results indicate that plant organs with lower Nc showed higher ¹⁵Np to the same component, and therefore ¹⁵Np in the plants operated as a signal of plant organs with the most deficiency in Nc. Paponov and Engels (2005) also found the same tradeoff with higher allocation of the ¹⁵Np to grain with reduced grain Nc under low N supply. Detailed evaluations of the proportional allocation of the ¹⁵Nu in plant components in our study suggested the late vegetative stage (V15) is a good stage to indicate relative plant total N status and variation in genotypic responses (data not shown). Indeed, ¹⁵Np to the leaves at V15 was strongly correlated with the responsiveness of ¹⁵Np to N rate treatments for the same plant component later in the season (data not shown).

In summary, for both years, leaves were the main sink component up to the kernel set period (R2) and, from R4 stage onward, more than 50% of the ¹⁵N absorbed accumulated in the ears. However, lower soil N availability and, consequently, higher N stress in 2014 prompted higher partitioning of ¹⁵Np to ears already at the R2 stage. Plant components that demonstrated higher ¹⁵Np had more apparent N deficiency.

There are many important factors that should be taken into account for a successful field study using a labeled N approach. Adequate plant sample size in field experiments using labeled N is crucial to identify genotypic responses for ¹⁵Nu partitioning to plant components. In a greenhouse study, Pan et al. (1986) addressed the large plant biomass variation in total N concentration or N content determination in plant components of five plants harvested at silking. Our studies were based in the field, and microplots were stringently selected for equal-density representations from the middle 3 or 4 plants. In 2014, after increasing sample size to four plants per growth stage and decreasing the ¹⁵N fertilizer rate by a third, we were still able demonstrate efficient assessment of ¹⁵Nu dynamics in the plants. Although harvesting more plants is even more preferable, resource costs of labeled N application and analyses are significant barriers.

The experiment layout and distance of plant zones that are used to analyze labeled or unlabeled treatments is critical (Van Cleemput et al., 2008). Several authors have suggested the use of plastic films, chambers, or tarpaulin materials to avoid problems with ¹⁵N percolation and with cross-contamination among treatments (Anhar, 2005). In the present study, we tried to represent the natural soil conditions in open environments using microplots within the main plots with no physical barrier. Microplots were installed at least 1 m from each other; the distance and the very low rates of ¹⁵N we utilized helped to circumvent contamination of plant zones by labeled N from elsewhere, and also helped prevent growth rate distortions arising from the extra sunlight in remaining plants harvested for biomass in other zones at a later time. The labeled N was carefully applied into the soil close to the root zones (~15 cm) to avoid direct contact of the ¹⁵N applied with the plant canopy and that, plus the water application, ensured immediate access of the labeled nitrate to growing roots.

The source of the labeled fertilizer used in the present study [Ca (¹⁵NO₃)₂ 98 atom%¹⁵N] was intended to make nitrate-N (NO3--N) immediately available to the plants and to minimize possible interactions of ammonium N sources (NH4+-N) to the negatively charged soil colloids. Roots can assimilate N by different mechanisms depending on the N source (Yoneyama et al., 2003), and several studies have observed that maize does not have a preferential uptake between mineral N sources (NO3-/NH4+-N) (Reddy and Reddy, 1993; Crozier et al., 1998). However, Pan et al. (1986) found higher accumulation of the labeled (NH₄)₂SO₄ surface applied at planting in the grain when compared to the labeled Ca(¹⁵NO₃)₂ source which was applied 26 and 52 days after planting. Our study assumed that all labeled N taken up by plants was distributed to a uniform depth on both sides of the maize plants with the expectation that N flux into plant components over time should be consistent for the 3 or 4 consecutive plants that were removed from the 5- to 6-plant micro-plot zones.

The amount of labeled fertilizer applied per plant [0.16 g (2014) or 0.24 g (2013)] was enough to enable instrument detection of the ¹⁵N into all plant components, and yet not so much that it changed the dominant soil plus fertilizer N supply pool that met the majority of the maize plant N requirements. The “spike” of the highly enriched labeled fertilizer injected into the soil allowed a precise ¹⁵N determination in the mass spectrometer. The additional water after each time of ¹⁵N application was crucial to increase the N availability to plants via mass flow while reducing the risk of soil-mineral interactions from the fertilizer ¹⁵N applied with soil microorganisms. Labeled N (¹⁵NO⁻³) losses via leaching were expected to be minimal with the 1.6 l of water applied unless there was an intensive rain event between the pulse application and biomass harvest.

In the isotope laboratory, highly enriched plant materials required the use of additional “blank” and standard samples to monitor and avoid ¹⁵N carryover between labeled materials during the mass spectrometer analysis. Thus, lower quantities of labeled fertilizer were applied in 2014 because we realized after the first year that extremely enriched plant samples increased the risk of subsequent sample contamination during testing. High costs (i.e., >$35,000 with the labeled fertilizer alone in the 2-year period) made it even more imperative to utilize a lower practical rate of ¹⁵N in the second year.

The natural abundance of isotopic N in the atmospheric dinitrogen (N₂) is considered equal to 0.366 atom %¹⁵N in a ratio (¹⁴N:¹⁵N) equal to 273:1 (Boaretto et al., 2004; Schepers and Raun, 2008). Similar values were found in the present study by sampling control plants in all hybrid and N rate treatments at the R4–R5 growth stages. These “control” values were used to calculate atom %¹⁵N excess (At%¹⁵NExcs) in plant components for each labeled sample as the amount of ¹⁵N which exceeded that represented in the non-labeled plant tissues originating from the natural abundance in the soil. The At%¹⁵NExcs was estimated by subtracting atom%¹⁵N (At¹⁵N) of each labeled plant component by the N natural abundance found in the present study (0.3683 or 0.3677 Atom %¹⁵N) (Equation a). In this work, amounts of pre-existing ¹⁵N in the soil based on natural abundance were neglected; however, the latter approach confirmed that soils were not labeled before the current field experiment. Additionally, discrimination from the uptake of natural abundant ¹⁵N to the ¹⁵N fertilizer was not important to estimate in our evaluations of short-term ¹⁵N accumulation in the plants because the large 98% enrichment of the fertilizer ¹⁵N applied provided confident identification of the isotope N taken up by plants.

The use of isotopic N has been acknowledged for its accuracy as N tracer for several decades (Van Cleemput et al., 2008). The multifaceted use of ¹⁵N in open systems is, when properly conducted, highly effective for N balance investigations in real, as distinct from simulated, soil-plant systems under different G × E × M interaction treatments. In this work, through mass spectrometry analysis, we estimated the δ¹⁵N in aboveground plant components to calculate uptake, partitioning, allocation, and recovery of N throughout the season. Observations of high correlations between ¹⁵N and total N allowed for detailed evaluations of maize plant response to labeled N additions in the root zone at multiple stages of development and at different N rates. After decades evaluating variations of plants δ¹⁵N in both laboratory and field experiments using ¹⁵N feeding methods, Yoneyama et al. (2003) also showed a strong association of δ¹⁵N with N dynamics in several crops.

Several techniques have been implemented using ¹⁵N in either open or controlled environments (field, greenhouses, and chambers) during short- or long-term methods of ¹⁵N application (Duete et al., 2009; Ning et al., 2017). However, few studies have invested in field studies using highly enriched ¹⁵N to address the allocation of N in plant components spanning the reproductive period or other intervals of the growing season in maize.

The “long term” labeled N application has occasionally been implemented as a one-time alternative to differentiate and more precisely estimate post-flowering N uptake and N remobilization from late vegetative stage or silking periods to physiological maturity. This method has helped to quantify N flux within the plants over the period of interest. Mae and Ohira (1981) and Ta and Weiland (1992) calculated the ¹⁵N remobilization by subtracting the ¹⁵N uptake in the stover at R6 by the ¹⁵N in the stover at R1. However, ¹⁵N remobilization can also be estimated as the difference between the ¹⁵N accumulated in the stover at maturity minus the whole plant ¹⁵N uptake at silking (Pan et al., 1986). The latter calculation was used in the present study.

When we compared ¹⁵Nu accumulated in the R1R6 zones by the more traditional “long term” evaluation method of ¹⁵N application vs. the results of the Pan et al. (1986) study, we observed a much higher level (~77%) of ¹⁵Nu accumulated in the kernels at R6 vs. the lower allocation (~51%) of ¹⁵N uptake to the grain in the Pan et al. study (Table 3). This may be partially explained by their lower total plant ¹⁵N recovery (~75%) when compared to ours (~83%), by their use of older and different genotypes that preferentially accumulated ¹⁵N to the stalk and root components, and by their shorter post-flowering period of 33 days in comparison to our 66 days. Furthermore, Pan et al. (1986) found that post ¹⁵N accounted for 42% of the total ¹⁵Nu accumulated in the plant at physiological maturity, while we found that post ¹⁵N contributed 54 and 42% to the total ¹⁵Nu accumulated at R6, for 0 and 112N treatments (Table 4). Similarly to our results, the review study of Ciampitti and Vyn (2013) demonstrated that post-flowering N uptake can contribute up to about 56% of the total plant N uptake at R6 stage.

Results from Ma et al. (1998) generally agreed with our current findings for the “long term” ¹⁵N accumulation. While we found that 77% of ¹⁵Nu in the kernels was taken up from R1 to R6 stage, their results from a one-time ¹⁵N application show 65–75% of infused ¹⁵N was accumulated in the kernels. However, they estimated 59 and 82% as the amount of remobilized N going to the grain from vegetative components, and a much lower contribution of the post-silking ¹⁵N uptake to the total ¹⁵Nu at R6 (ranging from 18 to 41%, which was similar to our 42% in the 112N treatment). Bertin and Gallais (2000) indicated a range of 35–65% of the N accumulated in the grain originated from N remobilization. We observed that remobilized ¹⁵N accounted for 22 and 36% of the total ¹⁵Nu accumulated at R6, for 0 and 112N rates (Table 4).

Furthermore, in attempting to integrate the isotope technique and breeding approaches, Gallais et al. (2007) estimated the contribution of the N remobilization and post N uptake in several maize inbred lines using a one-time ¹⁵N application. In their 3-year experiment, comparing different methods of labeled N application, they found that ~83% of that ¹⁵Nu was accumulated into the kernels at R6, slightly higher than our ¹⁵N proportional allocation ~77%. They also estimated ~25% of the total ¹⁵Nu at maturity originated from the post-flowering N uptake and about 62% was coming from the ¹⁵N remobilization. Our study found that post ¹⁵N accounted for ~50% of the total ¹⁵Nu accumulated at R6, and about 30% of that ¹⁵N accumulated at R6 originated from the remobilized ¹⁵N (Table 4). The difference in results of these two detailed studies may be explained by our ¹⁵N application at silking while their ¹⁵N application was at the beginning of stem elongation period (which enabled a better estimate of the accumulation in vegetative components), and by their use of ¹⁵N fertilizer with much lower enrichment to test an even larger number of genotypes.

This “long term” labeled approach has been broadly used to assess ¹⁵N uptake dynamics at multiple plant development stages (Friedrich and Schrader, 1979; Cliquet et al., 1990a,b; Deléens et al., 1994; Ma et al., 1998; Ning et al., 2017). When the influence of the ¹⁵N application at planting and at anthesis on improvements of N harvest index (NHI) in wheat (Triticum aestivum L.) was studied, Wuest and Cassman (1992) showed that ¹⁵N applied at flowering improved NHI (0.89) by almost 20%. The latter suggests a possibility that the ¹⁵N injection at the R1 stage overestimates the real amounts partitioned to the grain during the reproductive period.

Apart from injection method into the soil, Hertenberger and Wanek (2004) suggested that ¹⁵N stem infiltration and leaf feeding were the most effective labeling techniques to analyze N in thick-stem plant and grasses species, respectively. More recently, Putz et al. (2011) has advanced the termed “in-situ” labeling technique in plant species cultivated in greenhouse. Through the use of paint brush, they applied 2 mg of ¹⁵N Urea (98 atom% ¹⁵N) onto the leaves once a day during 5 days and plants were collected at short interval of hours after labeling.

Hence, in contrast to the techniques previously described by authors as “long term” and “short term” labeling approaches, our modified “multi-stage pulse labeling” technique ¹⁵N was used as a tracer to evaluate the fate of recently applied ¹⁵N into maize plants. The “multi-stage pulse labeling” technique used very small absolute amounts of highly enriched ¹⁵N fertilizer (98 atom% ¹⁵N) and different plants were routinely exposed to the labeled N for short intervals (just 4–6 days) mainly during the reproductive period. Our approach was able to accurately estimate plant N use in maize and to detect the dynamics of the in-season ¹⁵N fertilizer applications at development-stage-specific levels with a special focus on the reproductive period. This technique has high potential as an additional tool for phenotyping evaluation of genotypes with superior N use efficiency.

Although our emphasis was on applying ¹⁵N to maize plants during reproductive stages, this research observed a positive association of the allocation of ¹⁵Nu at late vegetative stage with the allocation of ¹⁵Nu during grain filling. Therefore, we propose that three N plant development stages (V15, R2, and early R5) are promising growth stages to evaluate potential hybrids with superior NUE. Future research studies using the multi-stage pulse labeling approach for phenotyping purposes with limited budgets should prioritize sampling a higher amount of plants (at least four) rather than increasing the rates of labeled N fertilizer beyond those used in our studies.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.