CORRELATIONS AND PATH ANALYSIS UNDER MULTICOLINEARITY IN THE ASSOCIATION OF VEGETATIVE CHARACTERISTICS WITH THE YIELD OF SNAP BEANS

Authors

Rodrigo Anschau, Univerisidade Estadual de Londrina
Gustavo H. Freiria, Univerisidade Estadual de Londrina
Felipe F. Furlan, Univerisidade Estadual de Londrina
Douglas J. Bertoncelli, Univerisidade Estadual de Londrina
Guilherme R. Gomes, Univerisidade Estadual de Londrina
Verônica Pellizzaro, Univerisidade Estadual de Londrina
Mônica S. Omura, Univerisidade Estadual de Londrina
Nicolas A.S. Vengrus, Univerisidade Estadual de Londrina
Luiz H.C. Almeida, Univerisidade Estadual de Londrina
Lúcia S.A. Takahashi, Univerisidade Estadual de Londrina

Document Type

Article

Date of this Version

3-2017

Citation

ANNUAL REPORT OF THE BEAN IMPROVEMENT COOPERATIVE, No. 60, March 2017. Published by USDA.

Comments

U.S. government work.

Abstract

INTRODUCTION: The comprehension of the plant vegetative attributes in the yield is of a fundamental importance to the management, as well as to the genetic improvement. The main tool used in this study are the estimation of correlations, however, because they do not provide the relative importance of the direct and indirect influence of other characters in the yield, they do not determine the cause effect relation among them (Furtado et al., 2002).

The best comprehension of the evolved causes in this association can be obtained by the path analysis. On the other hand, multicolinearity conditions could produce inconsistent values or with any congruence with the biological phenomenon studied (Moreira et al., 2013). With the intention to study strategies to contour this effect it was aimed to evaluate different methods to sidetrack the multicolinearity, as well as study the simple correlation among the studied characters.

MATERIAL & METHODS: The experiment was conducted in the city of Londrina, Parana state, Brazil (23º19’41.00”S, 51º12’18.19”W and altitude 590m), in the period of March 18th until June 10th of 2016, in field conditions. Were evaluated three cultivars (Feltrin Vicenza Amarelo Baixo, UEL 1 e UEL 2) and seven accesses of snap beans from the germplasm bank of State University of Londrina (T1, T3, T13, T24, T25, T39 and T41). The evaluations were divided at the development stages R1 (plant´s height, total dry mass, foliar area, specific leaf area index and leaves dry mass) and R7 (plant´s height, total dry mass and yield of pods).

Were estimated the Pearson correlations and path analysis, in which the yield of pods was considered the basic variable and the other characters were considered the explicative variables. In conditions of high multicollinearity (Cruz; Carneiro, 2003) were proceeded the disposal of variables of high interrelation, as well as the crest path analysis.