Introduction
Proper management of nitrogen fertilization is essential for high grain yield in cereal crops such as maize. At harvest, nitrogen deficiency can reduce grain yield by 14 to 80% 8,28. Maize is socio-economically important as it is the most grown crop species worldwide 21,23.
Among the soil nutrient fertilizers available, nitrogen (N) is the most expensive fertil ization cost and can reach about 30% of total production cost 10. However, when applied to the soil, it can cause environmental damages since a is usually lost by leaching and vola tilization 19. In addition, nitrogen fertilizer manufacturing consumes much oil, which is a non-renewable energy source. Therefore, new alternatives need to be sought to streamline use of nitrogen fertilizers 5,13, as one of the major agricultural challenges for the coming years is to produce sustainable food and optimize existing resources 17.
Studies have indicated plant breeding and Azospirillum genus bacteria inoculation as alternatives to partially supply maize nitrogen demands. These bacteria are highly capable of performing biological nitrogen fixation (BNF) 25, which can reduce use of nitrogen fertilizers.
These bacteria also play a role in plant hormone production such as auxins, cytokines, and gibberellins, which stimulate plant shoot and root growth 26, and hence plant light interception and dry mass yield 22. Such stimulus also increases plant water and nutrient uptakes by increasing root system branching 20, thereby improving corn grain yield 4,15.
The results of Azospirillum brasilense inoculation depend on biotic and abiotic factors such as genotype, soil microbial community, and climate variations 14. Genotypes may have different physiological performances in terms of N uptake 9. Thus, identifying and selecting maize genotypes responsive to inoculation and nitrogen fertilizations is a forth coming approach to improve crops and reduce nitrogen fertilizer use and respective envi ronmental contamination 18.
The hypothesis for the present study is that the agronomic performance of maize genotypes diverges in their potential when subjected to nitrogen topdressing fertilization and inoculation with A. brasilense. This study aims to evaluate and characterize the agronomic performance of second-season maize genotypes as a function of topdressing nitrogen fertilization and A. brasilense inoculation via soil and select the best genotypes within each management.
Material and methods
An experiment was conducted in the second season of 2017, at the Fazenda de Ensino, Pesquisa e Extensão (FEPE) of the São Paulo State University, School of Agricultural and Veterinarian Sciences, Jaboticabal-SP (Brazil). The study area is at an average altitude of 615 meters, near the coordinates of 21° 14’ 05’’ S latitude and 48° 17´ 09’’ W longitude. According to Köppen’s classification, the climate is classified as Aw, which stands for humid tropical with rainy season in summer and dry winter. The soil was classified as eutrophic Red Latosol (Oxisol) 6. The region has average precipitation of 1.425 mm.
Soil chemical attributes and particle size were determined in the 0.00-0.20 and prior to common maize sowing. The results were pH (CaCl2) 5.45; organic matter (g dm-3) 24.37; N 0.18%; P (mg dm-3) 8.26; K (mmolc dm-3) 3.08; Ca (mmolc dm-3) 38.73; Mg (mmolc dm-3) 16.98; S (mg dm-3) 7.07; B (mg dm-3) 0.21; Fe (mg dm-3) 25; H+AL (mmolc dm-3) 20.13; cation exchange capacity (mmolc dm-3) 78.93; base saturation 74.17%; clay 540 g kg-1; silt 230 g kg-1; sand 230 g kg-1.
Forty-eight maize (Zea mays L.) genotypes were used in the experiment. These corre spond to 46 synthetic maize populations from random crossbreeding, which were developed by Phoenix Agricultural Ltda, as well as two commercial cultivars AL Bandeirantes (Check A) and hybrid DKB 390 VT PRO2 (Check B).
Sowing was performed on February 17, 2017. Plots were delineated in randomized blocks, arranged in strips, devised by a plot seeder. Sowing fertilization was performed using 350 kg ha-1 of the 8-28-16 formulation. Each plot consisted of four 5-m long rows spaced 0.50 m between rows and 0.33 m between plants, with a population of 60,000 plants ha-1. As a useful area for evaluation and harvesting, only the two central lines were used.
All the genotypes were supplied with nitrogen by: 1) biological fixation with A. brasilense inoculation in the soil; and 2) chemical fertilization with urea as source. Both topdressing and inoculation were performed on March 23, 2017, when plants were at V4 stage.
We used the commercial inoculant Qualyfix Gramíneas® (A. brasilense, strain AbV5 and AbV6, at concentration of 5 x 108 cells mL-1), which was applied at the manufacturer’s recom mended rate (600 mL ha-1 for soil spraying). The inoculant was applied to the soil with the aid of a costal sprayer at 10 cm from maize row. Nitrogen topdressing was performed at a rate of 140 kg ha-1 N, using urea as source. It was applied in continuous fillet at 10 cm from crop row. After fertilization and A. brasilense application, irrigation was performed by a sprinkler system. Weed and pest controls were performed according to recommendations for maize cropping 7.
The following variables were evaluated: number of vegetative days until male (MF) and female (FF) flowerings, plant height (PH), ear height (EH), lodging and breaking rates (L + B), stunting rate (STU), Fusarium spp. (FUS) and grain yield (GY).
We also evaluated number of plants tilted more than 45° from the vertical or lodged on the soil at harvest time. This result was added to the breaking parameter, which is the number of plants with broken stem below the main ear. The data were transformed to for data normality, and later converted to percentage values.
Before threshing, five maize ear samples were collected randomly from each plot to be evaluated for Fusarium spp. (FUS) symptoms. The number of ears presenting FUS symptom was counted and transformed to for data normality and then converted to percentage.
Grain yield (GY) was obtained by threshing ears harvested within the useful area of each plot, weighing the grains and correcting to 13% moisture, and then converting the data to tons per hectare (t ha-1).
The stuting rate (STU) was analyzed according to percentage of Daubulus maidis incidence in each plot, obtained by equation 1.
where:
NPS = number of plants showing symptoms of generalized redness or yellowing and whitish streaks caused by Dalbulus maidis
NTP = total number of plants in the plot
The data were transformed to , aiming at normality, and later converting them to percentage.
During the experiment, the averages of temperature maximum and minimum and rainfall were 29.1 °C, 16,6 °C and 358 mm, respectively (Figure 1).
Meteorological data were provided by the agroclimatological station at FCAV/UNESP (College of Agricultural and Veterinary Sciences - University of São Paulo State). Water was provided via sprinkler irri gation when there were longer periods with no rain. Harvesting was performed manually on 07/03/2017, at 136 days after emergence (DAE).
Data from each variable was recorded separately, and genotypes and the two N supplies were considered as fixed factors. A factorial variance analysis (2 nitrogen supplies x 48 genotypes) was performed, considering all genotypes and agronomic traits. Means were compared by the Scott-Knott test at 5% probability, using AgroEstat® software 2.
After standardization, all data were submitted to principal component analysis (PCA). The number of components was chosen according to Kaiser’s criterion 11, wherein eigen values > 1.00 are selected. The results of principal component analysis (PC) were presented in biplot graphs, with genotypes plotted in the X-axis and matrices in the Y-axis, being repre sented by points. A scatterplot of the principal component analysis was used to have a clear view of the performance of each genotype in each variable. Genotypes were identified by a -2 to 2 ellipse in PC1 and PC2, wherein genotypes within the ellipse have no specific prop erties, while those located outside the ellipse are characterized as promising for each N supply. The analyses were performed using the Statistica 7.0® software 24.
Results and Discussion
The results of the analysis of variance (Table 1, page 72) showed that N source had a significant effect on grain yield (GY) and on days for female flowering (FF), when exchanging the N supply via inoculation with A. brasilense for topdressing.
Regarding the source of genotype variation, all parameters differed significantly, except for male flowering (MF) and lodging + breaking rates (L + B), These results corroborates the hypothesis that the agro nomic performance of maize genotypes diverges in their potential when subjected to nitrogen topdressing fertilization and inoculation with A. brasilense.
The interaction GEN x NSS (genotype x N supply source) showed a significant difference only for Fusarium spp., therefore, genotypes and N supply vary differently for incidence of this disease (Table 1, page 72). Similar results were observed by Khan and Doohan (2009), who found decreases between 12 and 21% in the disease incidence for inoculated plants.
These authors pointed out that favorable conditions for Fusarium spp. development may have been due to an increase in fertilization rates, besides being directly related to leaf disease intensity.
In both N supply systems, the genotypes presented overall GY means of 2.14 t ha-1 (Table 1), among which Check A and Check B stood out with averages of 3.23 and 3.55 t ha-1, respectively (Table 1). These averages may reflect the low L + B rates, STU and FUS of these genotypes. In this same context, genotype 22 had the lowest yield, with an average of 1.21 t ha-1, which may be linked to other to other factors such as the low PH and high staging value presented by this genotype, in addition to the above average L + B rates (Table 2, page 73-74).
In comparing the performance of genotypes with respect to the two N supply sources, only genotypes 27 and 34 showed significant differences in GY. These genotypes presented higher GY by A. brasilense application, with increments of 1.08 and 1.19 t ha-1, respectively, in relation of mineral fertilizantion (Table 2, page 73-74). These results are in agreement with those found by Müller et al. (2016), who obtained increases of up to 28% in grain yield after inoculation of A. brasilense in relation to genotypes with topdressing nitrogen fertilization.
Superior grain yield was also reported by Araújo et al. (2014) after inoculating maize with the same diazotrophic bacteria. When topdressed (Table 2, page 73-74), 31.25% of the genotypes were superior to the others for GY. The same was observed in A. brasilense inocu lation, in which 45.83% of the genotypes differed from the others regarding their response to inoculation. For both N supply sources, each genotype showed significant GY differences, therefore, this trait is greatly influenced either by chemical or biological sources.
Genotypes showed significant differences in PH and EH when comparing both N supplies. Genotype 18 increased plant height and ear insertion height by 0.20 m when inoculated with A. brasilense, that is, it was responsive to diazotrophic bacteria application (Table 2, page 73-74). Favorable effects provided by bacteria may have occurred due to changes in the root system of inoculated maize plants, leading to improvements in plant growth, water absorption, and nutrient intake 3, highlighting the height gains.
Genotype 32 presented the same increase (0.20 m) in plant and EH when urea was supplied as topdressing (Table 2, page 73-74); therefore, this genotype was more responsive to urea fertilization in terms of plant height. This result corroborates Vogel et al. (2013), who observed that nitrogen-deficient maize plants are more capable of synthesizing carbo hydrates during photosynthesis, thus growing more.
Both A. brasilense inoculated and urea topdressed plants obtained similar increase in PH and EH, being thus responsive to both N supply sources. These responses can reduce final production cost and environmental impact.
Regarding the incidence of L + B, all genotypes behaved similarly with no significant differences between inoculation and topdressing. When comparing the 48 genotypes for each N supply, 68.75 and 62.50% of the genotypes were more responsive to topdressing and inoculation, respectively (Table 2, page 73-74). Moreover, no disease infestation was observed in genotype Check B in both environments, Therefore, this genotype may be tolerant to vector transmission (leafhopper - Dalbulus maidis), thus benefiting plant devel opment and health in these plots, as well as improving productivity.
By comparing Fusarium spp. incidence in both N supply sources, significant differences were observed in genotypes 17, 18, and 30, which showed higher tolerance when inoculated with A. brasilense. This benefit may be due to a better responsiveness to plant inoculation, possibly due to an improved root system with higher nutrient absorption capacity, which is also reflected in healthier and better nourished plants. However, when topdressed, geno types showed no significant differences, therefore, all plants behaved similarly. On the other hand, when N is supplied by biological means, a significant difference was shown on 54.16% of the genotypes, thus highlighting a possible response of inoculation with the bacteria.
By analysis of principal components (PC), genotype distribution was evaluated for each N supply source, using a biplot graph method (Figure 2).
When fertilized with urea topdressing, the first two PCs explained 55.58% of the original data variability (31.64% and 23.94%, respectively) (Figure 2A, page 75). As for the biological supply with A. brasilense inoculation, 54.89% of the original data variability was explained by the first two PCs (29.12% and 25.77% respectively) (Figure 2B, page 75).
For nitrogen fertilization in topdressing (Figure 2A, page 75), the genotypes Check A, Check B, 12, and 44 were discriminated by the parameters grain yield, plant height, and ear height. Check B was the most productive genotype, with high values of plant height and ear height, and the lowest values of incidence of lodging + breaking, stunting, and Fusarium spp.
Genotype 44 had the second-highest yield but high values for STU and FUS, which indicates problems in plant development and grain quality. In turn, genotypes 43 and 48 showed high values of male and female flowerings, and stunting but low values for GY, PH, and EH. Because of such characteristics, these genotypes should be discarded from breeding programs. The same was observed in genotypes 1, 18, and 30; yet, they had grain yields considered above the general average and with high values for L + B, STU, and FUS as well. Genotype 32 was discrim inated by PH and EH, presenting a significant difference between the sources of N supply, in addition to values above the general average for grain yield and Fusarium spp.
On the positive side of the X-axis (Figure 2A, page 75), it can be noted that genotypes 21, 22, 23, 24, and 25 were discriminated by lower GY, which may be explained by high percentage of L + B, stunting and FUS, which directly affect grain yield, besides presenting the longest flowering cycles.
For inoculation with A. brasilense (Figure 2B, page 75), it can be noted on the negative side of the X-axis that the genotypes Check A, Check B, 27, and 35 were characterized by the variables GY, PH, and EH, corroborating results shown in Table 2 (page 73-74). Check B was the most productive and with above-average values for PH and EH, as well as lower rates of L + B, STU and FUS.
Genotype Check A presented the second highest GY, the highest values for PH and EH, besides good plant health. Genotype 35 had the fourth highest yield but low PH and EH, with above average values for incidence of FUS. Despite the high GY, Genotype 27 also had higher and undesirable indices for lodging, breaking, stunting, and FUS Genotype 43 showed lower value for GY, PH and EH but high rates for L + B, STU and FUS. when compared to the geno types located on the upper-left side of the graph (Figure 2, page 75).
Genotype 2 presented low grain yield in relation to the above genotypes, presenting late flowering and above-average values for L + B. The same was seen in genotype 47, which also showed above-average values for incidence of STU and FUS. Genotype 22 presented high EH and low presence of FUS. However, it presented high values for L + B and STU but later flowering and the lowest grain yield (Figure 2B, page 75).
On the positive side of the X-axis (Figure 2, page 75), it can be noted that the genotypes 21, 23, and 26 were in the opposite direction to the others, which indicates low GY and high L + B values, as well as high incidence of STU and FUS, which directly affect plant devel opment and yield. Therefore, these genotypes may have presented the lowest responses to A. brasilense inoculation.
The hypothesis tested in the present study was confirmed, as the maize genotypes showed different agronomic performance when submitted to topdressing mineral fertil ization and inoculation with A. brasilense via soil. It was observed that genotypes 27, 34 and 35 showed superior agronomic performance when inoculated with A. brasilense, while genotypes 12 and 44 obtained higher agronomic performance under nitrogen mineral fertilization.
Conclusions
The maize genotypes showed contrasting agronomic performance in relation to the nitrogen supplies evaluated. Inoculation with A. brasilense via soil has a significant effect on increasing maize grain yield, making it a viable and more sustainable alternative in the supply of N. However, the responses of agronomic attributes vary with the genotype. The choice of the genotype is an essential factor for the successful use of mineral nitrogen fertil ization or A. brasiliense inoculation.