Grain and starch granule morphology in superior and inferior kernels of maize in response to nitrogen

Maize (Zea mays L.) contributes approximately 55% of China’s grain production. The effects of nitrogen (N) on maize grain morphology and starch granules remain elusive. In this study, a field experiment in clay loam soil was conducted using three maize hybrids (Suyu 30, Suyu 20, and Suyu 29) and four N levels (0, 360, 450, and 540 kg ha−1) in 2010 and 2012. The results indicated that increased grain length and width, starch granule number, surface area, and volume, was associated with the application of 450 kg ha−1 of N. Differences between superior (ear base) and inferior (apical) grains decreased under highest yield treatments. The effects of N levels on inferior grains was more than that on superior grains. The starch granules of superior grains showed more polygonal, and bigger shape than inferior grains. The results revealed that N levels affected size and morphology of starch granules and grains. The application of 450 kg N ha−1 resulted in larger-sized starch granules and less difference between superior and inferior grains.

most of these studies mainly focused on the quality and physicochemical characteristics of starch, and information on the relationships between grain morphology and maize yield in superior and inferior grains is limited. Hence, this study was conducted to determine the relationship between yield and grain morphology in superior and inferior grains receiving various N levels and planted at a density of 82,500 plants ha −1 .

Results
Effect of N rates on Grain Yield and Morphology. The application of various N levels resulted in a significant increase in yield of the three maize hybrids, Suyu 20, Suyu 29 and Suyu 30 ( Fig. 1). As the N level increased, the yield of the three varieties presented single-peak curves, reaching peak values with 450 kg N ha −1 in both 2010 and 2012. Tables 1 and 2 illustrate that that 450 kg N ha −1 markedly improved the morphological characteristics of superior and inferior grains (except grain thickness) compared to 0 kg N ha −1 . At 450 kg N ha −1 , the variation in the grain density and 1,000-grain weight in both superior and inferior grains was minimal. The variance values of grain density and 1,000-grain weight of the three hybrids (Suyu 20, Suyu 29 and Suyu 30, respectively; the same as follows) were 0.5, 2.1, and 15.3 g L −1 and 82.8, 33.5, and 50.3 g, respectively. ANOVA results showed that the effect of N levels on inferior grains was greater than that on superior grains for grain density and 1,000-grain weight. For example, compared to 0 kg N ha −1 , the application of 450 kg N ha −1 resulted in an improved grain density and 1,000-grain weight in inferior grains from the three hybrids, which measured 2.3, 1.5, and 1.0 and 1.9, 1.9, and 2.1 times higher than those of superior grains, respectively. On average of four nitrogen levels, the higher geometric volume of superior and inferior grains of the three hybrids were 1.6, 1.7, and 1.8 and 1.4, 1.5, and 1.6 times higher than those of volume of grain, respectively, suggesting that sinking strength was still an important factor in generating a high-yield maize population. In terms of the F-value, no significant differences in the effect of N levels on grain thickness were detected for superior grains over inferior grains. And, significant Cultivar × N treatment interactions were detected for any parameters of morphology structure of maize grains except thickness of superior grains The effect of various N levels on inferior grains was generally higher than on superior grains and the application of 450 kg N ha −1 decreased the variation between superior and inferior grains and increased grain length and width, thus improving maize sink and yield. Figure 2 and Table 3 show the effects of N treatments on starch granules in superior and inferior grains. The effects on big starch granules were much greater than those on small starch granules. No significant differences in the diameter of starch granules <5 μm

Effect of N on Starch Granules of Superior and Inferior Grains.
Year Hybrid N Treatment Grain density 1000-grain weight geometrical morphology of grain (cm)   Table 2. Effect of nitrogen levels on the morphology structure of inferior grains. Means in the same column followed by the same letter do not differ statistically at the 0.05 probability level by an ANOVA protected Duncan's test; * , **NS Significantly different at the 0.05 and 0. 01 probability levels and no significant difference, respectively. were observed among superior and inferior grains. The application of 450 kg N ha −1 resulted in a marked increase in the number of big starch granules. In the superior grains of Suyu 29, for example, the percentages starch granules with a diameter >15 μm using 0, 360, 450, and 540 kg N ha −1 were 49.13%, 57.93%, 64.16%, and 53.20% (Figs 2C and D and 3E and F). Figures 2 and 3 illustrate that the distribution of starch granules of superior grains was different from that of inferior grains; the starch granules of superior grains showed smaller peak diameters, but a wider distribution than that in inferior grains. In terms of F-value (Table 3), various N levels imparted significant differences in surface area and volume between superior and inferior grains from hybrids. And, significant hybrid × N treatment interactions were detected for surface area and volume of superior and inferior grains of maize. Overall, the application of 450 kg N ha −1 increased surface area and volume of starch granules, also, increased the number of big starch granules Furthermore, big starch granules improved maize yield compared to small starch granules. This may explain the observed improvement in grain density and 1,000-grain weight, which resulted in a higher maize yield.
Correlation analysis. Table 4 illustrate that grain yield of maize positively and significantly correlated with any parameters of grain and starch granule of grain except surface area and volume of inferior grains. Compared with 1,000-grain weight of superior grains, there were more significant correlation relations with inferior grains. Based on the correlation coefficients, maize yield was most closely related to any morphology parameters of both superior and inferior grains, followed by 1000-grain weight and grain density.

Starch Granule Morphology of Superior and Inferior Grains.
The starch granule morphology of the three maize hybrids using 0 and 450 kg N ha −1 was examined using SEM. The granules showed polygonal and spherical shapes (Fig. 4), similar to the finding of Wei et al. 15 using rice starch. However, several striking differences were observed between the superior and inferior grains. For example, the polygonal starch granules of superior grains (Fig. 4A,B,E,F,I and J) were bigger than inferior grains (Fig. 4C,D,G,H,K and L). Conversely, at 0 kg N ha −1 , isolated starch granules from inferior grains with more smooth surfaces were spherical in shape (Fig. 4C,G and K). The SEM figures (10,000×) of the starch granules also showed mechanical damage (Fig. 5), 'Hollow' and 'solid' inner structures were also observed in the starch granules. The wall of spherical starch granules was thin and its chambers were large. The walls of the polyhedral starch granules were thick and layered and its chambers were small. These features (except starch granule damage) may explain the observed improvement in grain density and 1,000-grain weight, which improved maize production grain density.

Discussion
Grain and Starch Granule Morphology and its Relationship with Maize Yield. Improved sink activity, sources, and assimilate and nutrient flow are essential in increasing yield. Increasing the grain number currently serves as the main approach in improving sink activity in maize, although evidence on the relationship between morphology structure of grain and yield is limited. Previous studies have shown that grain morphology (length, width, and surface area) could be regulated by moisture, as evidenced by studies using rice 9 or starch granule morphology of rice could be regulate by N level 13 . In cereals, maximum seed volume is established before reaching its maximum dry weight and this is generally estimated as the developmental stage when the kernel maximum water content is achieved 24 . The 1,000-grain weight of maize and length of grain are quantitative traits that are controlled by multiple genes and are often influenced by environmental factors 25 . Because the ovaries at the tip of the ear have not reached their maximum size at the silking stage 7 , the apical kernels of maize ears have smaller size and dry matter accumulation rates than basal kernels. However, this difference could be regulated by changing the temperature during growth 26 and N levels 12 . Our research indicated that under high-density conditions (82,500 plants ha −1 ), the application of 450 kg N ha −1 also mitigated this difference in grain density and 1,000-grain weight between superior and inferior grains and significantly increased grain density, length, width, and volume of grains, with minimal effects on grain thickness. It appears that smaller differences in superior and inferior grains improved maize yield. N largely influenced grain morphology in inferior grains than that in superior grains. Our this result is very similar to one of previous reports 12 .
The starch content was positively correlated with the maize starch granule volume ranges of 0.8 to 2 μm, 2 to 10 μm, and 10 to 15 μm, but negatively correlated with other size ranges 18 . Starch granules with a diameter of ≥15 μm were predominant 19 . Temperature could also regulate starch granule size. Interestingly, temperature affected granule size, pasting temperature, and transition temperature, in which a lower temperature prevailed during the filling stage, resulting in larger starch granules and lower pasting and transition temperatures 20 . Our results demonstrated a positive correlation between large starch granules and maize yield. Furthermore, the predominance of large starch granules could be strongly influenced by N levels, although several striking differences  Table 3. Effects of nitrogen levels on weighted average of surface area and volume of superior and inferior grains of maize. Means in the same column followed by the same letter do not differ statistically at the 0.05 probability level by an ANOVA protected Duncan's test; * , **NS Significantly different at the 0.05 and 0. 01 probability levels and no significant difference, respectively.  15 and Nordmark et al. 21 . However, some starch granules have small cavities with a diameter range of 0.07 to 0.10 μm, as well as channels and pores randomly distributed across their surfaces, often in clusters and showing variations 22,23 . Evidences from SEM and chemical analyses have shown that the pores are randomly distributed, varying in number per granule, and the channels were irregularly bent 27 . These granules provide the architecture of blocklets, and a normal blocklet is mainly formed by crystalline and amorphous lamellae 28,29 previously reported that the semi-crystalline blocklets generally consist of two types, 'normal' and 'defective' , which occur within the same starch molecule and are the basic units that construct starch granules. The normal blocklets comprise the hard shells, whereas soft shells consist of the defective ones; both  32,33 . We observed 'hollow' and 'solid' inner structures in the starch granules (Fig. 5); the wall of the spherical starch granules was thin and the chamber was big. The wall of the polyhedral starch granules was thick and layered and the chamber was small. Li et al. 29 suggested that the hydrolysis of α-amylase or other chemicals that flowed into starch granules along channels were responsible for this 'hollow' formation. However, Wei et al. 13 and our current study suspect that the physicochemical characteristics of granule walls, as well as the 'hollow' and 'solid' inner structures of starch granules, may generally occur in nature and are controlled by the conditions of the environment (e.g., N and temperature) during the filling stage, thus affecting maize yield. More extensive research on this area is therefore warranted. Experimental Design. A split-plot experimental design consisting of three replicates was used in the study, with hybrids planted in the main plots and N treatments conducted in the sub-plots. Seeds were sowed on June 27 each year at a density of 82,500 plants ha −1 in test plot area of 110 m 2 . Planting followed a 0.3 m by 0.7 m row spacing pattern. Four N levels (Granule urea, N concentration = 46%) were utilized (0, 360, 450, and 540 kg ha −1 ) based on the previous experimental results in which the yield of maize reached its peak at 450 kg N ha −1 . Vegetative and reproductive N treatments (Granule urea, N concentration = 46%) was incorporated to the soil at the 10-leaf and anthesis stages. Table 5 shows the N application scheme in detail. Base fertilizers of 150 kg ha −1 P 2 O 5 and 225 kg ha −1 K 2 O were applied once upon planting. Figure 6 shows the temperature condition during the period from sowing to final harvest for the 2010 and 2012 growing seasons. Other field management measurements were carried out according to the maize requirements for high yield.   Grain Morphology. Ten ears from each treatment were harvested 50 days after anthesis and fully dried.

Plant
The grains of four rows at the base of the ear were removed and divided into 4 equal portions. The grains from the lower one-fourth portion were designated as superior, whereas those from the top one-fourth portion were defined as inferior. The length (a), width (b), thickness (c), and 1,000-grain weight of the superior and inferior grains were measured. The volume of the grains (approximately 100 g) was assessed using the 95% alcohol displacement method. The following calculations were performed: geometric volume of the grain = a × b × c; and grain density = weight of the grain (approximately 100 g)/volume of the grains.   . Approximately 18 h later, the starch was washed with 0.2% NaOH and centrifuged (4,000 g for 10 min) until no biuret reaction occurred (or protein content was less than 0.5%), and finally washed with water. The starch was further treated with anhydrous ethanol and a mixture of methanol and chloroform (v/v = 1:1) three times. Finally, the starch was collected and dried under atmosphere, ground into powder, and then passed through a 200-mesh sieve.
Starch Granule Size. The analysis of starch granule size was conducted using a laser particle size analyzer (Mastersizer 2000, Malvern, England) as described elsewhere 34 , with minor modifications. Granule sizing experiments were carried out through laser diffractometry using a Mastersizer 2000 instrument, equipped with HydroMu dispersing unit (Malvern). Measurements were taken under the following conditions: anhydrous alcohol mobile phase, particle refractive index of 0.54, particle absorption coefficient of 4, anhydrous alcohol refractive index of 1.366, and a general calculation model for irregular particles. Ten measurement cycles of 10 s each were taken, and the data obtained were averaged using the equipment software (Mastersizer 2000, ver. 5.20 from Malvern).
Granule Morphology. Granule morphology was examined using the method described by Lu et al. 35 , with minor modifications. Starch granules that were steeped in 95% ethanol for 4 h were mounted on circular aluminum stubs and air-dried. Subsequently, these were coated with gold, examined under a scanning electron microscope (SEM, XL-30 ESEM, Philips, Netherlands), and then photographed at an accelerating potential of 20 kV.
Grain Yield and Other Components. The harvested plot size was 22 m 2 (four 11-m rows at the center of each plot). Mean grain yield and other components were estimated for each treatment at each location. Statistical Analysis. Analysis of variance (ANOVA) using a linear model was performed using the software Statistical Product and Service Solutions (SPSS Inc., Chicago, IL, USA) to detect the main effects of hybrids, nitrogen, years and their interaction. Treatments were compared by Duncan's test and differences were declared as statistically significant if P < 0.05. Pearson's correlations were calculated to determine the relationship between different parameters of maize. All graphics were drawn using Sigmaplot 12.5. The results from 2010 and 2012 followed the same trend. So, all of the calculations in the present study based on the averages of 2010 and 2012.