Qualidade dos prognósticos do estado nutricional de boro na cultura da soja
Carregando...
Data
2016-11-22
Autores
Traspadini, Edilaine Istéfani Franklin [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A interpretação do estado nutricional da soja pode ser realizada por métodos uni, bi e multivariados, mas necessita de estudos sobre a acurácia dos diagnósticos produzidos. Diante disso, objetiva-se com esse estudo avaliar a qualidade dos prognósticos a partir da interpretação foliar pelo Nível Crítico, pelo DRIS e pelo CND, comparando o diagnóstico de boro com a resposta na produtividade da soja. O estudo foi realizado em área experimental e comercial na região de Chapadão do Sul-MS, Brasil. No experimento de calibração, foram usadas as doses 0; 300; 600; 1.200 e 1.800 g ha-1 de B, em três pulverizações e cinco repetições. Na área experimental e em 140 lavouras comerciais, foi realizada amostragem foliar, determinando os teores nutricionais e estimado a produtividade. Apenas nas parcelas experimentais foi realizada a análise dos diagnósticos, através do confronto do diagnóstico com a resposta da planta na adição do nutriente. Cinco medidas de acurácia foram usadas para testar a eficiência dos métodos do NC (calibrado, da literatura distribuição normal reduzida (NCDNR)); DRIS (com três critérios de seleção das relações entre nutrientes), e o CND, que foram a acurácia total, acurácia para deficiência e suficiência, razão de deficiência e razão de eficiência, e o incremento líquido da produtividade. De maneira geral, os métodos testados obtiveram resultados abaixo do esperado ou na linha limite, sendo que somente os métodos NC (literatura e DNR), DRIS (pela transformação logarítmica das relações) e CND atingiram boa acurácia (70 a 80%). Ao considerar o incremento líquido da produtividade, somente dois métodos foram considerados eficientes: o DRIS (pela transformação logarítmica das relações) e o NC (calibrado). Sendo que o segundo obteve incremento na produtividade cinco vezes superior ao primeiro (22 kg ha-1). Os critérios de seleção das relações bivariadas a compor o índice DRIS não apresentaram o mesmo desempenho nos prognósticos, desses apenas a transformação logarítmica resultou em ganhos produtivos devido a seus prognósticos. Os métodos DRIS e CND não apresentaram prognósticos iguais.
The interpretation of the nutritional state of soybean can be performed by uni, bi and multivariate methods, but needs to study the accuracy of the diagnoses produced. Therefore, the objective of this study is to evaluate the quality of the prognoses from the foliar interpretation by the Critical Level, the DRIS and the CND comparing the diagnosis of boron with the response in soybean yield. The study was conducted in an experimental and commercial area in the region of Chapadão do Sul, MS, Brazil. In the calibration experiment were used doses 0, 100, 200, 400 and 600 g ha-1 B, working with three sprayings and five repetitions. In the experimental area and in the 140 commercial fildes was carried out foliar sampling to determine the nutritional content, and estimated productivity. Only in the experimental plots was carried out the analysis of diagnoses by comparing the diagnostic with the plant response in the addition of the nutrient. Five accuracy measures were used to test the efficiency of NC methods (calibrated, from reduced normal distribution literature (NCDNR)); DRIS (with three selection criteria of the relation between nutrients) and; the CND, which were the overall accuracy, accuracy and sufficiency for disability, rate of disability and rate of efficiency and the net increase of productivity. In general, the tested methods had results beneath expects or on the outline. Only the NC methods (literature and DNR), DRIS (by logarithmic transformation of relations) and CND achieved good accuracy (70-80%). When we considered the net increase, only two methods were efficient, the DRIS (by logarithmic transformation of relations) and NC (calibrated). The second one obtained a increment in productivity five times greater than the first one (22 kg ha-1). The selection criteria of the relationships are compared with the results obtained by DRIS did not present the same performance in the prognoses, and only by logarithmic transformation resulted in productive gains due to their prognoses. The DRIS and CND methods did not present equal prognoses.
The interpretation of the nutritional state of soybean can be performed by uni, bi and multivariate methods, but needs to study the accuracy of the diagnoses produced. Therefore, the objective of this study is to evaluate the quality of the prognoses from the foliar interpretation by the Critical Level, the DRIS and the CND comparing the diagnosis of boron with the response in soybean yield. The study was conducted in an experimental and commercial area in the region of Chapadão do Sul, MS, Brazil. In the calibration experiment were used doses 0, 100, 200, 400 and 600 g ha-1 B, working with three sprayings and five repetitions. In the experimental area and in the 140 commercial fildes was carried out foliar sampling to determine the nutritional content, and estimated productivity. Only in the experimental plots was carried out the analysis of diagnoses by comparing the diagnostic with the plant response in the addition of the nutrient. Five accuracy measures were used to test the efficiency of NC methods (calibrated, from reduced normal distribution literature (NCDNR)); DRIS (with three selection criteria of the relation between nutrients) and; the CND, which were the overall accuracy, accuracy and sufficiency for disability, rate of disability and rate of efficiency and the net increase of productivity. In general, the tested methods had results beneath expects or on the outline. Only the NC methods (literature and DNR), DRIS (by logarithmic transformation of relations) and CND achieved good accuracy (70-80%). When we considered the net increase, only two methods were efficient, the DRIS (by logarithmic transformation of relations) and NC (calibrated). The second one obtained a increment in productivity five times greater than the first one (22 kg ha-1). The selection criteria of the relationships are compared with the results obtained by DRIS did not present the same performance in the prognoses, and only by logarithmic transformation resulted in productive gains due to their prognoses. The DRIS and CND methods did not present equal prognoses.
Descrição
Palavras-chave
Acurácia, Diagnose foliar, Glycine max L., Accuracy, Leaf analysis