Previsão de massa seca de Brachiaria brizantha e ganho de peso por bovinos
Carregando...
Arquivos
Data
2018-05-04
Autores
Sousa, Clayson Correia de [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
o objetivo desta tese foi ajustar modelos de previsão da massa seca (MS) desta forrageira relacionada a variáveis explicativas de clima, do solo, do pasto e dos animais, a partir de metadados de experimentos feitos em 8 localidades diferentes da região Centro-Sul. As análises primeiramente foram feitas para os dados agregados de experimentos de não irrigados e irrigados e na sequência para dados de experimentos de não irrigados. Quanto aos dados agregados as variáveis que mais influenciaram a MS foram Excedente Hídrico (EXC = precipitação mensal menos a evapotranspiração mensal); temperatura média mensal (T) e aplicação de fertilizantes (K2O, N). O modelo com melhor ajuste foi MS0,5=0,18K2O+5,56T9+ 0,14ETR8-103,63 (subscritos 8 e 9 representam respectivamente os meses de agosto e setembro), o qual pode ser utilizado para prever a MS de pasto com um ano de antecedência. Os dados de pastagem irrigada demonstraram aumento da média de MS a partir de setembro, sobretudo em função do aumento da temperatura média reduzindo o efeito sazonal e antecipando a MS. O erro dos modelos foi elevado (MAPE ≥ 29%), contudo na sequência das análises, a inclusão de variáveis morfoestruturais do pasto (% folhas) e de manejo (SPE, suplementação concentrada protéico/energética, altura do pasto) melhorou significativamente a acurácia (MAPE < 2%). A MS de folhas (variável dependente do modelo) é explicada pelo GMD (ganho médio diário) (variável independente) sendo -59,3 kg ha-1 por kg de GMD. Observam-se efeitos da TL (taxa de lotação) sobre a MS total, altura do pasto, frequência de corte sendo os efeitos respectivamente de -368,2kgUA-1ha-1, 254kg ha-1m-1 e 313,3kg ha-1dia-1. Dentre as variáveis climáticas que mais influenciaram a MS e a MSf foram temperatura e variáveis de balanços hídricos e observou-se que GMD dos animais sofreu maior efeito da SPE.
the objective of this thesis was to adjust forecast models of the dry mass (DM) of this forage related to explanatory variables of climate, soil, grass and animals, from metadata of experiments done in 8 different locations in the CenterSouth region. First the analyzes were made for the aggregate data of irrigated and rainfed experiments and in the sequence for data from not irrigated experiments. Regarding the aggregated data, the variables that most influenced the DM were: Water Surplus (EXC = monthly precipitation minus monthly evapotranspiration); monthly mean temperature (T) and fertilizer application (K2O, N). The best fit model was DM0,5=0,18K2O+5,56T9+0,14ETR8-103,63 (subscripts the 8 and 9 respectively represent the months of August and September), which can be used to forecast the average DM with a year in advance. Irrigated pasture data increased the average of DM from September, mainly due to the increase of the average temperature reducing the seasonal effect and anticipating the DM. The error of the models was high (MAPE ≥ 29%), however in the analysis, the inclusion of morphostructural variables of pasture (% leaves) and management (SPE, protein / energetic supplementation, pasture height) significantly improved the accuracy (MAPE <2%). Leaf DM (MSf) (model dependent variable) is explained by GMD (independent variable) being -59.3 kg ha-1 per kg of GMD. The effects of TL (total stocking rate) on total DM, pasture height, cutting age, and effects respectively of -368.2kg UA-1 ha-1 , 254kg ha-1m-1 and 313.3kg ha-1 day -1 . Among the climatic variables that influenced MS and MSF were temperature and water balance variables, and it was observed that GMD of the animals was most influenced by supplementation.
the objective of this thesis was to adjust forecast models of the dry mass (DM) of this forage related to explanatory variables of climate, soil, grass and animals, from metadata of experiments done in 8 different locations in the CenterSouth region. First the analyzes were made for the aggregate data of irrigated and rainfed experiments and in the sequence for data from not irrigated experiments. Regarding the aggregated data, the variables that most influenced the DM were: Water Surplus (EXC = monthly precipitation minus monthly evapotranspiration); monthly mean temperature (T) and fertilizer application (K2O, N). The best fit model was DM0,5=0,18K2O+5,56T9+0,14ETR8-103,63 (subscripts the 8 and 9 respectively represent the months of August and September), which can be used to forecast the average DM with a year in advance. Irrigated pasture data increased the average of DM from September, mainly due to the increase of the average temperature reducing the seasonal effect and anticipating the DM. The error of the models was high (MAPE ≥ 29%), however in the analysis, the inclusion of morphostructural variables of pasture (% leaves) and management (SPE, protein / energetic supplementation, pasture height) significantly improved the accuracy (MAPE <2%). Leaf DM (MSf) (model dependent variable) is explained by GMD (independent variable) being -59.3 kg ha-1 per kg of GMD. The effects of TL (total stocking rate) on total DM, pasture height, cutting age, and effects respectively of -368.2kg UA-1 ha-1 , 254kg ha-1m-1 and 313.3kg ha-1 day -1 . Among the climatic variables that influenced MS and MSF were temperature and water balance variables, and it was observed that GMD of the animals was most influenced by supplementation.
Descrição
Palavras-chave
Agrometeorologia, modelagem, meta-análise, pastagens, forragens, regressão, Agrometeorology, modeling, meta-analysis, pastures, forages, regression