Microsoft Word - 13-Agra_42292_communication


1432 
Bioscience Journal  Communication 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

NITROGEN DOSES IN TOPDRESSING AFFECT VEGETATION INDICES 
AND CORN YIELD 

 
DESES DE NITROGÊNIO EM COBERTURA AFETAM ÍNDICES DE VEGETAÇÃO E 

PRODUTIVIDADE DE MILHO 
 

Fabio Henrique Rojo BAIO¹; Eder Eujácio da SILVA; Igor Mendes SCARPIM; 
 Cid Naudi da Silva CAMPOS¹; Paulo Eduardo TEODORO¹ 

1. Universidade Federal de Mato Grosso do Sul, Campus de Chapadão do Sul (UFMS/CPCS), 
Chapadão do Sul, MS, Brasil. eduteodoro@hotmail.com  

 
ABSTRACT: Nitrogen is the main nutrient required by corn crop, especially in Cerrado soils. Remote 

sensing techniques can be used to generate additional information now of nitrogen fertilization 
recommendation. This work investigated the association of plant height and dry matter phenological variables 
together with NDVI, REDEDGE, SAVI, and IV 760/550 vegetation indices (VIs) with corn grain yield, under 
different N doses. Sowing occurred in November 2016, at a spacing of 0.45 m between rows and a 60,000 ha-1 
plant population. Four N doses (0, 80, 160, and 240 kg of N ha-1) were applied at phenological stage V4. The 
experimental design consisted of randomized blocks, containing four N doses in topdressing and 16 
replications. The active optical sensor Crop Circle ACS-470 was used to obtain the VIs. The NDVI, SAVI, and 
RE indices have a high positive association with each other and with the variables plant height and dry matter. 
Polynomial regression equations were adjusted between the variables in response as doses of N. Afterwards, 
they were estimated as correlations between variables and results expressed through the network of 
correlations. Finally, a multivariate analysis of canonical variables was performed to understand the 
interrelationship between the variables and each dose of N applied. NDVI and RE have a positive relationship 
of moderate magnitude with grain yield in corn crops. 
 

KEYWORDS: Multivariate analysis. Remote sensing. Correlation network. Zea mays. 
 
INTRODUCTION 

 
Brazil is among the world's three largest 

corn producers, with a production of 79.9 million 
tons of grain in the 2016/2017 harvest, and an 
average yield of 6.5 t ha-1 (CONAB, 2018). This 
yield can be increased, and the nitrogen (N) 
management is one of the main factors that need to 
be improved in this crop. N is one of the most 
difficult nutrients to manage in the soils of 
subtropical regions due to its high soil instability. 
This fact is because it is subject to a large number of 
interactions (ERNANI, 2003). 

Remote sensing techniques can be used to 
generate additional information at the moment of 
nitrogen fertilization recommendation. Data from 
the spectral response of the crop canopy during its 
development, characterized by the process of 
reflectance of incident electromagnetic waves, can 
be used as indirect indicators of nutritional status 
and productive potential of a culture (SERRANO et 
al., 2000). Sensors that measure spectral properties, 
reflectance, and transmittance of plant leaves are 
affected by the Nitrogen concentration (N) since this 
nutrient is essential to the chlorophyll molecule, 

which is the pigment that first absorbs the light 
energy necessary for photosynthesis (BLACKMER; 
SCHEPERS, 1995; TARPLEY et al., 2000). 

Two wavelengths are directly related to 
plant variables, being indicators of the productive 
potential: red spectrum waves, whose reflectance is 
lower under a greater amount of chlorophyll 
(TUCKER, 1979); and infrared spectrum waves, 
whose reflectance is higher under a greater plant dry 
matter accumulation (DM). Leaf chlorophyll content 
and DM production positively correlate to N doses 
and yield (SCHADCHINA; DMITRIEVA, 1995). 
Vegetation indices are used to minimize the 
variability caused by external factors in the data 
obtained from reflectance, which are sensitive to the 
green biomass of the plants and, therefore, to the 
amount of chlorophyll per unit area (PONZONI, 
2001). 

Shaver et al. (2017) studied the efficiency of 
nitrogen use in corn crops and observed that canopy 
sensors could be an effective tool to determine the N 
doses in topdressing, mainly in irrigated fields, 
where fertirrigation can be used, and the several 
split applications can be conducted. Fang et al. 
(2014) studied the interaction between different 

Received: 15/05/18 
Accepted: 05/12/18 



1433 
Nitrogen doses in topdressing…  BAIO, F. H. R. et al. 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

vegetation indices (VIs) to estimate yield, obtaining 
errors of 3.5% in the estimate of field yield when 
using interaction, and 13%, when using only one VI. 

However, in-field data collection is cost-
effective, in which remote sensing helps guide 
sampling (MLADENOVA et al. 2017). The 
objective of this work was to study the phenological 
variables and different vegetation indices to estimate 
corn yield potential. 
 
CONTENTS 

 
The experiment carried out in the 

experimental area of the Federal University of Mato 
Grosso do Sul (lat. 18°46'17.9"S; long. 
52°37'25.0"W, alt. 810 m asl) during the 2016/17 
harvest, in the municipality of Chapadão do Sul - 
MS. The experimental design was randomized 
blocks with four urea doses and 16 replications. The 
experimental plots consisted of nine 4 x 6 m rows, 
totaling 64 plots. The hybrid DKB 310 VT PRO 3® 

was used. At sowing, 30 kg ha-1 of nitrogen (urea, 
43% N), 120 kg ha-1 of phosphorus (super simple 
phosphate 20% P2O5), and 60 kg ha- 1 potassium 
(Potassium chloride, 58% K2O) were applied. 
Sowing was carried out on November 12th, 2016, at 
a spacing of 0.45 m and a 60,000 ha-1 population. 
When corn plants reached stage V4, at 22 days after 
emergence (DAE), the nitrogen fertilizer urea (43% 
N) was applied at doses of 0, 80, 160, and 240 kg of 
N ha-1. 

The three central rows were used for 
analyses, discarding 1 m of each end to avoid 
possible interference of other treatments, totaling a 
useful area of 5.4 m2. At 64 DAE, the following 
variables were evaluated: plant height (PH), using a 
graduated ruler; vegetation indices (VI) [NDVI 
(Normalized Difference Vegetation Index), RE 
(REDEDGE), SAVI (Soil-Adjusted Vegetation 
Index), and VI 760/550], with a Crop Circle sensor, 
model ACS-470 (Holland Scientific, Lincoln, 
EUA), using the filters 550, 670, 730, and 760 nm; 
shoot dry matter (SDM), by randomly collecting 
three plants in the plot, which were cut at the stem. 
The material was stored in paper bags, dehydrated 
in an oven at 105 °C for 48 hours, and weighed. The 
means of the three evaluated plants were processed. 

For the N doses applied in topdressing and 
each evaluated variable, maps were generated by the 
ESRI ArcGis 10.5 Geographic Information Systems 
(GIS) software. Correlation network was performed 
to study the interrelation between the studied 

variables and the N application in topdressing. 
Positive correlations are highlighted in green and 
negative correlations are represented in red. The line 
thickness is proportional to the correlation 
magnitude. Canonical variables were analyzed to 
study the association between variables and each 
treatment evaluated. These analyses were performed 
using the Rbio program (BHERING, 2017). 

Figure 1 shows the maps for the variables of 
the experimental area containing the N doses 
applied in topdressing, plant height, dry matter, 
grain yield, and the vegetation indices NDVI, RE, 
SAVI, and IV 760/550 IVs. The distribution of the 
classes of values of vegetation indices and dry 
matter are very similar in this figure. This 
association was more evident when the Pearson’s 
correlation network was used between the evaluated 
variables (Figure 2). All the vegetation indices, 
except for VI 760/550, are inter-correlated with each 
other and present a high correlation with shoot dry 
matter of corn plants. 

The VIs NDVI and RE were the most correlated 
to grain yield and N doses since they have greater 
proximity to these variables in the correlation network 
generated. Similar results were reported by Baio et al. 
(2018), who investigated the association between 
vegetation indices and relative deposition of different 
application rates in corn, which generated higher 
correlations of these indices with plant dry matter. In 
addition, the N doses applied are directly related to corn 
grain yield. Moreover, linear correlations can be easily 
interpreted using the graphical network correlation 
technique. The efficiency of this innovative technique has 
previously been reported by Silva et al. (2016). 

The graphs presented in Figure 3 show the 
regression equations for the studied variables in 
function of the N doses applied in topdressing to 
corn plants. With the increase of N doses, the values 
in a quadratic equation increased for the VIs NDVI 
(A), REDEDGE (B), and SAVI (C), that is, these 
VIs presented a saturation close to the maximum N 
dose (2240 kg N ha-1). Conversely, IV 760/550 has a 
linear equation, showing that this VI can respond to 
doses greater than 240 kg de N ha-1. For plant 
height, the quadratic equation showed a good 
relation to the N doses in topdressing and to dry 
matter; however, the latter did not present a good 
correlation. The observed yield shows that the 
hybrid used in this study has a high response 
capacity to N in topdressing. The highest N dose 
also presented the highest yield, and therefore, the 
equation for this variable is linear. 



1434 
Nitrogen doses in topdressing…  BAIO, F. H. R. et al. 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

 
Figure 1. Maps of the experimental area containing N doses applied in topdressing (A), plant height (B), dry 

matter - DM (C), yield (D), and VIs NDVI (E), REDEDGE (F), SAVI (G), and VI 760/550 (H). 
 

 
Figure 2. Correlations network between N doses and the variables plant height (PH), dry matter (DM), the VIs 

VI 760/550, NDVI, REDEDGE, and SAVI, and yield (Y). 
 



1435 
Nitrogen doses in topdressing…  BAIO, F. H. R. et al. 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

 
Figure 3. Regression graphs for the variables NDVI (A), REDEDGE (B), SAVI (C), VI 760/550 (D), plant 

height (E) and yield (F) in N doses applied in topdressing. 
 

The canonical variables analysis was used to 
verify the contribution of each variable to the 
difference between N doses (Figure 4). This 
technique is analogous to the principal components 
analysis but should be preferred when the research 
has an experimental design (replications). Each 
canonical variable is a linear combination of the 
variables used. To represent the scores in a two-
dimensional graph, the percentage of variance 
retained in the first two canonical variables must be 
greater than 80% (Mingoti 2005). In this work, the 
variance accumulated in the first two canonical 
variables was 97.4%, allowing its precise 
interpretation 

The control (dose of 0 kg ha-1 of N) did not 
associate with any variable, which was already 
expected since the genetic material available in the 
market is highly demanding in N. The vectors grain 
yield (Y) and NDVI, SAVI, and IV 750/560 are 
close to the highest N dose. These results indicate 
that the highest dose was the most efficient to 
increase these variables, corroborating the results 
obtained by Raper and Varco (2015), who observed 
that NDVI is more associated with cotton plant dry 
matter when subject to high N doses. Conversely, 
the variables DM and PH are more associated with 
the intermediate doses of N used, which indicates 
the quadratic behavior of these variables in response 
to the increasing N doses applied in topdressing, as 
also reported by Borges et al. (2006). 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 



1436 
Nitrogen doses in topdressing…  BAIO, F. H. R. et al. 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

 
Figure 4. Analysis of canonical variables between the variables PH, DM, VIs (IV760 / 550, NDVI, 

REDEDGE, and SAVI), Y, and N doses in topdressing. 
 
CONCLUSION 

 
The indices NDVI, SAVI, and RE have a 

high positive association with each other and with 

the variables plant height and dry matter. NDVI and 
RE have a positive relationship of moderate 
magnitude with maize grain yield. 

 
 

RESUMO: O nitrogênio (N) é o principal nutriente requerido pela cultura do milho, principalmente em 
solos do Cerrado. Técnicas de sensoriamento remoto podem ser usadas para gerar informações adicionais agora 
sobre a recomendação de fertilização nitrogenada. Este trabalho investigou a associação de variáveis 
fenológicas de altura de plantas e matéria seca com os índices de vegetação (IVs) NDVI, REDEDGE, SAVI e 
IV 760/550 com a produtividade de grãos de milho, sob diferentes doses de N. A semeadura ocorreu em 
novembro de 2016, com espaçamento de 0,45 m entre linhas e 60.000 ha-1 de população de plantas. Quatro 
doses de N (0, 80, 160 e 240 kg de N ha-1) foram aplicadas no estádio fenológico V4. O delineamento 
experimental foi o de blocos casualizados contendo quatro doses de N em cobertura e 16 repetições. O sensor 
óptico ativo Crop Circle ACS-470 foi usado para obter os IVs. Equações de regressão polinomial foram 
ajustadas entre as variáveis em resposta como doses de N. Posteriormente, foram estimadas como correlações 
entre variáveis e resultados expressos através da rede de correlações. Por fim, foi realizada uma análise 
multivariada de variáveis canônicas para entender a inter-relação entre as variáveis e cada dose de N aplicada. 
Os índices NDVI, SAVI e RE apresentam alta associação positiva entre si e com as variáveis altura de planta e 
matéria seca. NDVI e REDEDGE têm uma relação positiva de magnitude moderada com a produtividade de 
grãos na cultura do milho. 
 

PALAVRAS-CHAVE: Análise multivariada. Sensoriamento remoto. Rede de correlação. Zea mays. 
 

 
REFERENCES 
 
BHERING, L. L. Rbio: a tool for biometric and statistical analysis using the R platform. Crop Breeding and 
Applied Biotechnology, Viçosa, v. 17, p. 187-190, 2017. https://doi.org/10.1590/1984-70332017v17n2s29 
 
BLACKMER, T. M., SCHEPERS, J. S. Use of chorophyll meter to monitor nitrogen status and schedule 
fertigation for corn. Journal of Production Agriculture, Madison, v. 8, n. 1, p.56-60, 1995. 
https://doi.org/10.2134/jpa1995.0056 



1437 
Nitrogen doses in topdressing…  BAIO, F. H. R. et al. 

Biosci. J., Uberlândia, v. 35, n. 5, p. 1432-1437, Sep./Oct. 2019 
http://dx.doi.org/10.14393/BJ-v35n5a2019-42292 

BORGES, E. A., FERNADES, M. S., LOSS, A., SILVA, E. V., SOUZA, S. R. Acúmulo e remobilização de 
nitrogênio em variedades de milho. Caatinga, Mossoró, v. 19, n. 3, p. 278-286, 2006. 
 
CONAB. Companhia Nacional de Abastecimento. Acompanhamento da safra brasileira: grãos, décimo 
levantamento, fevereiro, 2018. Disponível em: <https://www.conab.gov.br/index.php/info-agro/safras>. Acesso 
em 15 de abril de 2018. 
 
ERNANI, P. R. Disponibilidade de nitrogênio e adubação nitrogenada para macieira. Lages: Graphel; 
2003. 
 
FANG, H., LIANG, S., HOOGENBOOM, G. Integration of MODIS LAI and vegetation index products with 
the CSM–CERES–Maize model for corn yield estimation. International Journal of Remote Sensing, v. 32, n. 
4, p. 1039-1065, 2011. https://doi.org/10.1080/01431160903505310 
 
MINGOTI, S. A. Análise de dados através de métodos de estatística multivariada. Belo Horizonte: Editora 
UFNG; 2005. 
 
MLADENOVA, I. E., BOLTEN, J. D., CROW, W. T., ANDERSON, M. C., HAIN, C. R., JOHNSON, D. M., 
MUELLER, R. Intercomparison of Soil Moisture, Evaporative Stress, and Vegetation Indices for Estimating 
Corn and Soybean Yields Over the U.S. IEEE Journal of Selected Topics in Applied Earth Observations 
and Remote Sensing, v. 10, n. 4, p. 1328 – 1343, 2017. https://doi.org/10.1109/JSTARS.2016.2639338 
 
PONZONI, F. J. Comportamento espectral da vegetação. In. Meneses PR, Madeira Netto JS, editores. 
Sensoriamento remoto - reflectância dos alvos naturais. Brasília, DF: Universidade de Brasília; 2001. p.157-99. 
 
RAPER, T. B., VARCO, J. J. Canopy-scale wavelength and vegetative index sensitivities to cotton growth 
parameters and nitrogen status. Precision Agriculture, v. 16, n. 1, p. 62-76, 2015. DOI:10.1007/s11119-014-
9383-4 https://doi.org/10.1007/s11119-014-9383-4 
 
SCHADCHINA, T. M., DMITRIEVA, V. V. Leaf chlorophyll content as a possible diagnostic mean for the 
evaluation of plant nitrogen uptake from the soil. Journal Plant Nutricion, v. 18, n. 14, p. 27-37, 1995. 
https://doi.org/10.1080/01904169509364992 
 
SERRANO, L., FILELLA I., PEÑUELAS, J. Remote sensing of biomass and yield of winter wheat under 
different nitrogen supplies. Crop Science, v. 40, n. 7, p. 23-31, 2000. 
https://doi.org/10.2135/cropsci2000.403723x 
 
SHAVER, T. M., KRUGER, G. R., RUDNICK, D.R. Crop Canopy Sensor Orientation for Late Season 
Nitrogen Determination in Corn. Journal of Plant Nutrition, 2017. DOI: 10.1080/01904167.2017.1346681. 
https://doi.org/10.1080/01904167.2017.1346681 
 
SILVA, A. R.; RÊGO, E. R.; PESSOA, A. M. S.; RÊGO, M. M. Correlation network analysis between phenotypic and 
genotypic traits of chili pepper. Correlation network analysis between phenotypic and genotypic traits of chili pepper. 
Pesquisa Agropecuária Brasileira, Brasília v.51, n.4, p.372-377, 2016. https://doi.org/10.1590/S0100-
204X2016000400010 
 
TARPLEY, L., REDDY, K. R., Sassenrath-Cole, G. F. Reflectance indices with precision and accuracy in 
predicting cotton leaf nitrogen concentration. Crop Science, Madson, v. 40, p. 1814–1819, 2000. 
https://doi.org/10.2135/cropsci2000.4061814x 
 
TUCKER, C. J. Red and photographic infrared linear combination for monitoring vegetation. Remote Sensing 
Environment, v. 8, n. 1, p. 27-50, 1979. https://doi.org/10.1016/0034-4257(79)90013-0