%0 Journal Article
%T Statistical modeling of sugar cane potential bud sprouting within the context of a 3x3x2 factorial experiment Modelado estad赤stico de la brotaci車n potencial de ca a de az迆car en un experimento factorial 3x3x2
%A Osvaldo E. A. Arce
%A Patricia A. Digonzelli
%A Eduardo R. Romero
%J Revista Industrial y Agr赤cola de Tucum芍n
%D 2010
%I Estaci車n Experimental Agroindustrial Obispo Colombres
%X The aim of this study was to model sugarcane potential bud sprouting temporal evolution within the context of a 3x3x2 factorial experiment. Statistical modeling was accomplished by means of a three-parameter logistic model using nonlinear mixed models. The trial was carried out with seed cane from three varieties (LCP 85-384, CP 65-357 and CP 48-103), considering three harvesting dates (May, August, and October) and two seed cane origins (micropropagation and hot water treatment). Experimental units were distributed in a 3x3x2 completely randomized factorial arrangement with two replicates. The number of sprouts was recorded from the day after plantation up to the 21st day. The selection of effects associated with a particular parameter was made by means of backward selection. Once the significant effects were selected, the need for the corresponding random effect was tested. The same procedure was followed with the rest of the parameters. Heteroscedasticity was corrected using a power variance function and autocorrelation of residuals was included as an order 1 autoregressive model. After statistical modeling, the conclusion to be drawn is that the methodology of nonlinear mixed models is an adequate and powerful tool when analyzing data obtained from factorial experiments, where repeated measurements over a period of time have been recorded. El objetivo del trabajo fue modelar la evoluci車n temporal de la brotaci車n potencial en ca a de az迆car, en el contexto de un experimento factorial 3x3x2. El modelado se hizo mediante el modelo log赤stico de tres par芍metros, utiliz芍ndose modelos mixtos no lineales. Se trabaj車 con ca a semilla de tres variedades (LCP 85-384, CP 65-357 y CP 48-103), tres 谷pocas de cosecha (mayo, agosto y octubre) y dos or赤genes de la semilla (micropropagaci車n y propagaci車n convencional por estacas termotratadas). Las unidades experimentales se ubicaron en un arreglo factorial 3x3x2 completamente aleatorizado con dos repeticiones. Se registr車 diariamente el n迆mero de yemas brotadas, desde el d赤a siguiente al de plantaci車n hasta los 21 d赤as. En modelos mixtos, la selecci車n de los efectos asociados con un par芍metro particular se hizo mediante una selecci車n hacia atr芍s. Una vez seleccionados los efectos significativos, se evalu車 si el efecto aleatorio correspondiente era necesario. El mismo procedimiento se sigui車 con el resto de los par芍metros. La heterocedasticidad se corrigi車 mediante una funci車n de potencia de varianza, y para la autocorrelaci車n de residuales se us車 un modelo autoregresivo de orden 1. Luego del an芍lisis, se concluye qu
%K Modelos mixtos
%K Factorial con medidas repetidas
%K Modelos no lineales
%K Curvas de crecimiento
%K R
%K Mixed models
%K Repeated measurements
%K Nonlinear models
%K Growth curves
%K R
%U http://www.scielo.org.ar/scielo.php?script=sci_arttext&pid=S1851-30182010000200003