Modelagem não linear da extração de zinco de um solo tratado com lodo de esgoto

• Autor(a) principal: Souza, Edcarlos Miranda de
• Data de Publicação: 2010
• Outros Autores: Muniz, Joel Augusto, Marchi, Giuliano, Guilherme, Luiz Roberto Guimarães
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/42132
• topic: Regressão não-linear; Elementos-traço; Lodo de esgoto; Bootstrap; Curvatura de Bates e Watts; Non-linear regression; Trace element; Sewage sludge; Bates and Watts curvature

The prediction of trace element availability to plants using chemical extractants is hard to achieve. For this reason, a successive extraction method which simulated a rhizosphere solution employing synthetic organic acids was previously proposed. The method was developed in order to access the long term potential availability of trace elements to plants. Data from successive zinc extractions with organic acid mixtures, as well as with routine extractants (Mehlich I, DTPA, and NH4OAc) adjusted to a non-linear model based in first order kinetic mechanisms. The aim of the present study was to verify whether the estimator asymptotic properties for that adopted non-linear model were valid. Classical methodology for sampling and likelihood was used, with the bootstrap methodology as an alternative. Zinc extractants composed of organic acid mixtures presented similar inferential results; however, these results were different from those obtained with routine extractants. Data adjustment with routine extractants presented narrow confidence bands, suggesting that predictions with these extractants were considered precise within the studied interval, except for data obtained with NH4OAc. In general, zinc availability prediction inferential results were precisely adjusted to the non-linear model as set by the methodologies used: linear approximation, bootstrap, and curvature measure.