Use este identificador para citar ou linkar para este item: https://repositorio.unimontes.br/handle/1/1082
Título: Diagnose nutricional da mangueira 'Palmer' e modelagem da produção no semiárido
Autor(es): Sales, Rafael Pereira
Orientador(ra): Kondo, Marcos Koiti
Membro(s) Banca: Maia, Victor Martins
Megda, Michele Xavier Vieira
Megda, Marcio Mahmoud
Pegoraro, Rodinei Facco
Portugal, Arley Figueiredo
Palavras-chave: Mangueira;Plantas Floração;Redes neurais artificiais
Área: Ciencias Agrarias
Subárea: Agronomia
Data do documento: 18-Out-2019
Resumo: The first chapter aimed to determine reference values DRIS and CND to the interpretation of the results of ‘Palmer’ mango tree leaf tissue analysis, establishing adequate nutrient ranges for the cultivation. The second chapter was measured the artificial neural networks (ANN) and multiple linear regression (MLR) potential to predict the ‘Palmer’ mango yield and determine the variables with the greatest impact on fruit yield. The work was accomplished on the irrigation district of the Gorutuba Project (Janaúba and Nova Porteirinha counties) and Jaíba Projects (Jaíba and Matias Cardoso counties), in the semi-arid region of Minas Gerais, Brazil. Ninety-eight plots of 23 ‘Palmer’ mango farm were selected, being, 49 plots in three harvests (2015, 2016, 2017) and the other 49 plots in two harvests (2016, 2017), totalizing 245 samples. Each sample was composed by leaf tissue analysis (N, P, K, S, Ca, Mg, Cu, Zn, Fe, Mn and B) used on the first chapter and more 36 variables related to weather, soil, handling and nutritional status, besides the flowering percentage of 40 plots, used on the second chapter. In the reference standards reference development (chapter 1), the samples were divided as to yield into two subpopulations, high yield (>30 Mg ha-1) and low yield (≤ 30 Mg ha-1). The DRIS and CND indexes were calculated, determining the critical level by the normal distribution. From these values, the Nutritional Balance Index (NBI) and the Mean Nutritional Balance Index (NBIm) were calculated, determined the Beaufils ranges and classified the nutrients as to limitation and fertilization response. The DRIS and CND methods are equally efficient to ‘Palmer’ mango tree nutritional diagnosis. The adequate ranges to cultivate it on the semi-arid from DRIS indexes are: : 1,50-1,77 (N); 0,10-0,12 (P); 0,93-1,19 (K); 2,34-3,73 (Ca); 0,20-0,27 (Mg); 0,13-0,17 (S) dag kg-1 to macronutrients; and 15-46 (Cu); 77-154 (Fe); 24-48 (Zn); 931-2985 (Mn) and 49-83 (B) mg kg-1 to micronutrients. The B, followed by Cu was the main yield limiting nutrients by deficiency and, the Ca and N, were the main yield limiters by excess, on low yield population, identified by the DRIS method as CND. On chapter 2 the 36 variables were: climatic data at the time of floral induction (maximum temperature, average temperature, minimum temperature, hourly solar radiation, evapotranspiration, relative humidity and wind velocity); soil chemicals (pH, MO, P, K, Ca, Mg, SB, Al, H+Al, t, T, V, m, Cu, Fe, Zn, Mn, B and Prem); handling related attributes: technological level (1 – low, 2 – medium and 3 - high), uniformity coefficient of irrigation distribution, irrigation water (1 – well, 2 – irrigation, 3 - mixed), number of plants by hectare, mean canopy diameter in meters, canopy area by hectare in m-2 ha, last growth regulator dose used in grams of active ingredient per canopy diameter, floral induction month (1 to 10, initiating count in October – start of harvest year induction in the region), area nutritional status (mean nutritional balance index calculated by the DRIS – dNBI method and mean nutritional balance index calculated by the CND – cNBI method), and yield (Mg ha-1). The data were submitted to analysis by ANN and MLR, for prediction models determination, and were selected the most relative variables importance by the Garson test. For the flowering percentage, a mathematical model in function of average temperature were adjusted. ANNs with 84.9% determination coefficient (R2 ) were adjusted using all of the variables to prediction and 87.02% using just the seven variables of most relative importance. Now, for the MLR, the R2 was 81.40%. The most relative variables importance were: mean temperature, maximum temperature, canopy diameter, CND mean nutritional balance index, canopy area, paclobutrazol dose, and technological level. The quadratic model for flowering percentage in the function of the average temperature showed 94.12% of R2 , where the upper limit of temperature for floral induction with 70% of the flowering response was 29.4°C.
URI: https://repositorio.unimontes.br/handle/1/1082
Aparece nas coleções:Teses e Dissertações

Arquivos associados a este item:
Arquivo Descrição TamanhoFormato 
Sales, Rafael Pereira_Diagnose nutricional da mangueira 'Palmer' e modelagem da_2019.pdf1,45 MBAdobe PDFVisualizar/Abrir


Este item está licenciada sob uma Licença Creative Commons Creative Commons