3.4.9
USE OF MULTIVARIATE ANALYSIS TO SELECT CACAO (THEOBROMA CACAO) RESISTANT TO CRINIPELLIS PERNICIOSA

SDVM SILVA1, EDMN LUZ1 and K MATSUOKA2

1CEPLAC/CEPEC/SEFIT, Cx Postal 07, 45600-000 Itabuna, Bahia, Brazil; 2UFV, Departamento de Fitopatologia, 35670-000 Višosa, MG, Brazil

Background and objectives
This work aimed to choose the best variables to assess cacao genotypes resistant to witches' broom disease caused by Crinipellis perniciosa.

Materials and methods
Open pollinated progenies of 70 clones were inoculated with 75,000 basidiospores/ml using an automated inoculation and incubation system. One hundred and twelve plants of each progeny were inoculated (4 replications of 28 plants each). Symptoms were assessed 60 days after inoculation. As several symptoms were observed it was neccessary to select the variables which best expressed the disease by means of multivariate statistical analysis.

Results and conclusions
The following variables were selected: terminal, axillary and cotyledonary brooms; lenght of the terminal broom; stem, petiole and pulvinus swellings; and stem hypertrophy. Progenies were grouped by genetic similarities according to canonic variables, Mahalanobis genetic distance, Tocher method and Anderson discriminating analysis. AB 1; CC 34; CEPEC 519; EEG 13, 50, 64; ICS 75; IMC 27; MA 15; OC 66, 67; P 7; SC 5; SCA 6, 12; SIAL 207, 407, 495, 644; SIC 21, 662; TSH 774, 1188 and UF 296 were resistant while CEPEC 26, 44; SIC 18, 24 and SPA 17 were more susceptible than Catongo, used as a standard susceptible progeny.