Adubação nitrogenada à taxa variada em Brachiaria decumbens usando sensor multiespectral de baixo custo
Loading...
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Federal de Viçosa
Abstract
A alimentação é o principal custo na produção pecuária, sendo as pastagens a fonte mais econômica. Porém, grande parte das pastagens estão degradadas, reduzindo a produtividade. A deficiência de nitrogênio afeta a qualidade das forragens, e a utilização de sensores de baixo custo auxiliam na gestão do nitrogênio, otimizando a produção e reduzindo custos na agricultura. Por isso, este trabalho buscou avaliar um espectrofotômetro de baixo custo com base em adubação nitrogenada com taxa variada, utilizando características espectrais da Brachiaria Decumbens. Foi adotado um delineamento de blocos casualizados com quatro repetições e cinco tratamentos cada, totalizando 20 tratamentos. Estes incluíram um tratamento testemunha (sem nitrogênio), um tratamento de referência (150 kg ha?¹ de N em dose única) e três tratamentos com doses variáveis de nitrogênio baseadas em índices espectrais (NDRE, CGI, NDVI). O Índice de Suficiência de Nitrogênio (NSI), calculado a partir de leituras espectrais em diferentes alturas das plantas, determinou a necessidade de doses corretivas. O experimento contou com três ciclos (ocorrendo de março a maio de 2023), em que variáveis espectrais foram medidas usando sensores como SPAD, GreenSeeker, câmera Micasense Red Edge – MX acoplada ao drone Matrice 100 e ReflecSense. As coletas de forragem foram realizadas ao final de cada ciclo para determinar massa seca (MS), produtividade e teor de proteína bruta. Amostras foram cortadas a 5 cm do solo e processadas para estimar MS e MSA, utilizando estufas de circulação forçada. O solo foi previamente analisado e corrigido com fósforo. Os dados meteorológicos no período dos três ciclos foram analisados, no qual o primeiro ciclo apresentou temperaturas mais altas e precipitação moderada, condições favoráveis ao crescimento da cultura estudada. Já os ciclos subsequentes foram marcados por temperaturas mais amenas e menor precipitação, limitando a absorção de nitrogênio e o desenvolvimento da Brachiaria. A análise de correlação de Pearson revelou a capacidade do ReflecSense de detectar variações nutricionais e de biomassa. A avaliação entre a câmera Micasense RedEdge-MX (NDRE) com o ReflecSense apresentou correlações variáveis: moderada a forte em condições de estresse nutricional e nula em situações de deficiência extrema. Já ao avaliar o ReflecSense com o sensor comercial GreenSeeker (utilizado para NDVI), o ReflecSense demonstrou forte correlação nas condições de maior variabilidade nutricional, destacando-se em doses elevadas de nitrogênio. No entanto, enfrentou limitações em detectar variações em cenários de estresse severo. Na avaliação do índice CGI (clorofila), o ReflecSense exibiu correlações moderadas a fortes, com resultados mais consistentes no último ciclo. A correlação negativa em algumas condições sugere desafios relacionados ao comportamento espectral dos sensores e interferências externas, como biomassa e luminosidade. Observou-se que o coeficiente de variação (CV%) apresentou uma redução consistente ao longo dos ciclos, principalmente no teor de proteína bruta, em que reduziu entre o ciclo 1 e 2 mais de 10%. Com o teste de médias (Tukey a 5% de probabilidade), concluiu-se que todos os tratamentos com adubação nitrogenada alcançaram produtividades superiores ao tratamento testemunha (sem adição de nitrogênio). Além disso, observou-se que, em todos os três ciclos e em todos os índices de vegetação, bem como nas características nutricionais, não houve diferença estatística entre o tratamento referência (T1) e os tratamentos com adubação em taxa variável (T2, T3 e T4). Em síntese, o ReflecSense apresentou desempenho promissor, especialmente em condições de manejo variado, comprovando sua viabilidade como alternativa acessível e eficiente para monitoramento nutricional e de produtividade em pastagens. Entretanto, limitações em cenários de estresse hídrico apontam para possíveis melhorias no sensor para otimizar sua aplicação em diferentes contextos ambientais. Palavras-chave: índices de vegetação; estresse nutricional; adubação nitrogenada; zootecnia de precisão
Feed is the main cost in livestock production, with pastures being the most economical source. However, a large portion of pastures are degraded, reducing productivity. Nitrogen deficiency affects forage quality, and the use of low-cost sensors assists in nitrogen management, optimizing production and reducing costs in agriculture. Therefore, this study aimed to evaluate a low-cost spectrophotometer based on variable-rate nitrogen fertilization, using spectral characteristics of Brachiaria decumbens. A randomized block design was adopted, with four replications and five treatments each, totaling 20 treatments. These included a control treatment (no nitrogen), a reference treatment (150 kg ha?¹ of N applied in a single dose), and three treatments with variable nitrogen rates based on spectral indices (NDRE, CGI, NDVI). The Nitrogen Sufficiency Index (NSI), calculated from spectral readings at different plant heights, determined the need for corrective doses. The experiment comprised three cycles (from March to May 2023), during which spectral variables were measured using sensors such as SPAD, GreenSeeker, the Micasense RedEdge–MX camera attached to a Matrice 100 drone, and the ReflecSense sensor. Forage samples were collected at the end of each cycle to determine dry matter (DM), yield, and crude protein content. Samples were cut 5 cm above the soil and processed to estimate DM and DMS, using forced-air circulation ovens. The soil was previously analyzed and corrected with phosphorus. Meteorological data from the three cycles were analyzed, showing that the first cycle had higher temperatures and moderate rainfall, conditions favorable for the growth of the studied crop. The subsequent cycles were characterized by milder temperatures and lower rainfall, limiting nitrogen uptake and the development of Brachiaria. Pearson’s correlation analysis revealed the ReflecSense’s ability to detect nutritional and biomass variations. The comparison between the Micasense RedEdge-MX camera (NDRE) and the ReflecSense showed variable correlations: moderate to strong under nutritional stress and null under extreme deficiency. When comparing the ReflecSense with the commercial GreenSeeker sensor (used for NDVI), the ReflecSense demonstrated a strong correlation under conditions of greater nutritional variability, standing out at higher nitrogen doses. However, it faced limitations in detecting variations under severe stress conditions. In the evaluation of the CGI (chlorophyll index), the ReflecSense exhibited moderate to strong correlations, with more consistent results in the final cycle. The negative correlation observed under certain conditions suggests challenges related to the spectral behavior of the sensors and external interferences such as biomass and light intensity. It was observed that the coefficient of variation (CV%) consistently decreased over the cycles, especially in crude protein content, which dropped by more than 10% between the first and second cycles. Using Tukey’s test (at a 5% probability level), it was concluded that all nitrogen-fertilized treatments achieved higher yields than the control treatment (without nitrogen). Moreover, in all three cycles and across all vegetation indices as well as nutritional characteristics, there was no statistical difference between the reference treatment (T1) and the variable- rate fertilization treatments (T2, T3, and T4). In summary, the ReflecSense showed promising performance, especially under variable management conditions, confirming its viability as an accessible and efficient alternative for monitoring pasture nutrition and productivity. However, its limitations under water stress conditions highlight potential improvements to optimize the sensor’s performance across different environmental contexts. Keywords: vegetation indices; nutritional stress; nitrogen fertilization; precision animal husbandry.
Feed is the main cost in livestock production, with pastures being the most economical source. However, a large portion of pastures are degraded, reducing productivity. Nitrogen deficiency affects forage quality, and the use of low-cost sensors assists in nitrogen management, optimizing production and reducing costs in agriculture. Therefore, this study aimed to evaluate a low-cost spectrophotometer based on variable-rate nitrogen fertilization, using spectral characteristics of Brachiaria decumbens. A randomized block design was adopted, with four replications and five treatments each, totaling 20 treatments. These included a control treatment (no nitrogen), a reference treatment (150 kg ha?¹ of N applied in a single dose), and three treatments with variable nitrogen rates based on spectral indices (NDRE, CGI, NDVI). The Nitrogen Sufficiency Index (NSI), calculated from spectral readings at different plant heights, determined the need for corrective doses. The experiment comprised three cycles (from March to May 2023), during which spectral variables were measured using sensors such as SPAD, GreenSeeker, the Micasense RedEdge–MX camera attached to a Matrice 100 drone, and the ReflecSense sensor. Forage samples were collected at the end of each cycle to determine dry matter (DM), yield, and crude protein content. Samples were cut 5 cm above the soil and processed to estimate DM and DMS, using forced-air circulation ovens. The soil was previously analyzed and corrected with phosphorus. Meteorological data from the three cycles were analyzed, showing that the first cycle had higher temperatures and moderate rainfall, conditions favorable for the growth of the studied crop. The subsequent cycles were characterized by milder temperatures and lower rainfall, limiting nitrogen uptake and the development of Brachiaria. Pearson’s correlation analysis revealed the ReflecSense’s ability to detect nutritional and biomass variations. The comparison between the Micasense RedEdge-MX camera (NDRE) and the ReflecSense showed variable correlations: moderate to strong under nutritional stress and null under extreme deficiency. When comparing the ReflecSense with the commercial GreenSeeker sensor (used for NDVI), the ReflecSense demonstrated a strong correlation under conditions of greater nutritional variability, standing out at higher nitrogen doses. However, it faced limitations in detecting variations under severe stress conditions. In the evaluation of the CGI (chlorophyll index), the ReflecSense exhibited moderate to strong correlations, with more consistent results in the final cycle. The negative correlation observed under certain conditions suggests challenges related to the spectral behavior of the sensors and external interferences such as biomass and light intensity. It was observed that the coefficient of variation (CV%) consistently decreased over the cycles, especially in crude protein content, which dropped by more than 10% between the first and second cycles. Using Tukey’s test (at a 5% probability level), it was concluded that all nitrogen-fertilized treatments achieved higher yields than the control treatment (without nitrogen). Moreover, in all three cycles and across all vegetation indices as well as nutritional characteristics, there was no statistical difference between the reference treatment (T1) and the variable- rate fertilization treatments (T2, T3, and T4). In summary, the ReflecSense showed promising performance, especially under variable management conditions, confirming its viability as an accessible and efficient alternative for monitoring pasture nutrition and productivity. However, its limitations under water stress conditions highlight potential improvements to optimize the sensor’s performance across different environmental contexts. Keywords: vegetation indices; nutritional stress; nitrogen fertilization; precision animal husbandry.
Description
Citation
BANDEIRA, Priscila Pascali da Costa. Adubação nitrogenada à taxa variada em Brachiaria decumbens usando sensor multiespectral de baixo custo. 2025. 49 f. Dissertação (Mestrado em Engenharia Agrícola) - Universidade Federal de Viçosa, Viçosa. 2025.
