Métodos estatísticos de análise da adaptabilidade e estabilidade de cultivares de soja na região centro sul do estado do Tocantins, visando a produção de biodiesel

Abstract

Aiming to evaluate eleven soybean cultivars on the productivity of oil in the municipalities of Gurupi and Palmas - TO 04 through analysis methods adaptability and stability were conducted 05 trials of soybean cultivars in the years 2008/2009 and 2009/2010, where each test environment was a: (Palmas I (30/11/2008); Palmas II (16/12/2008); Palmas III (04/12/2009), Gurupi I (03/12/2008); Gurupi II (05/01/2009)). The experiments were conducted in randomized complete block design with four replications. Treatments consisted of eleven soybean cultivars: (P98Y70; M-8766RR; M-9144RR; BR/EMGOPA314; P98R91; P98Y51; P99R01; M-8867RR; M-9056; M-8527RR e M-9350). This work was done in the form of chapters: Chapter 1 deals with the theoretical basis. In chapter 2 we studied the adaptability and stability by Eberhart and Russell (1966), Lin and Binns (1988) modified by Carneiro (1998), Annicchiarico (1992) and Centroid (ROCHA et al., 2005). The oil yield ranged from 337.88 kg ha-1 (Gurupi II) to 968.49 kg ha-1 (I Palmas), averaging between environments 634.58 kg ha-1. For the methodology of Eberhart & Russell (1966) the ideal cultivar was not found, however cultivars M-9144, M-8527RR, had the highest average PO, respectively 717.99 and 700kg/há, wide adaptability (*1i=1), but relative stability, little predictable (*2d#0). The 9144RR cultivar M-, M-8527RR, and BR/EMGOPA-314 can be recommended for broad environmental conditions by the other three methods. There was partial agreement in the ranking of cultivars for general conditions, favorable and unfavorable, so the methodologies used to complement and enhance credibility in the classification and names of cultivars. In chapter 2 we studied the Spearman correlation between four methods of adaptability and stability: methods based on analysis of variance: Annicchiarico (1992), in linear regression: Eberhart and Russell (1966), in non-parametric statistics: Lin and Binns (1988) modified by Carneiro (1998) and in multivariate analysis: Centroid (ROCHA et al . 2005). Were found significant correlations between Eberhart and Russell, Lin and Binns modified by Carneiro, Annicchiarico and Centroid. The correlations of greater magnitude occurred between these latter three methods. The use of methods of Eberhart and Russel should preferably be used simultaneously with one of the other three methods, given that these correlations have significant and of greater magnitude among its parameters.