Random amplified polymorphic DNA (RAPD) and isozyme electrophoresis (IE) techniques were used to estimate the level of genetic diversity in a sample of cacao germplasm existing at the International Cocoa Genebank, Trinidad. Twenty-six cocoa populations represented by 459 cocoa genotypes were analysed using IE and 22 populations represented by 353 cocoa genotypes were analysed using RAPD. Despite few differences in the classification of the populations, both techniques revealed three major groups: the indigenous trees, the cultivated Trinitario and the cultivated trees from Ecuador. Two-thirds of the partitioned diversity were found within populations and one-third between the populations, with both techniques.

Field observations were conducted from 1998 to 2001 at the International Cocoa Genebank, Trinidad, to evaluate 57 cacao clones for resistance to black pod (BP) and witches'-broom (WB) diseases (caused by Phytophthora sp. and Crinipellis perniciosa, respectively). Each month ripe pods were harvested and the number of healthy and diseased was recorded. The number of brooms on vegetative shoots was recorded three times a year on selected branches. Twenty-three clones showed less than 10% of infection for both BP and WB on pods. Among those, eight clones showed an absence of brooms on the observed branches: IMC 6, MAN 15/60 [BRA], PA 67 [PER], PA 195 [PER], PA 218 [PER], PA 296 [PER], PA 303 [PER], and POUND 32/A [POU]. Broad-sense heritability was estimated at 0.38 and 0.57 for WB disease on pods and shoots, respectively, and at 0.51 for BP disease. Genetic correlation between WB disease on pods and on shoots was low and estimated at 0.39, whereas the correlation between WB and BP diseases on pods was 0.48. To choose putative parents for breeding schemes, it is suggested that clones are first assessed for their level of resistance to WB on shoots, and the most promising individuals are screened for BP with a detached pods test. Further studies are needed to confirm whether the level of resistance to WB on pods can be predicted using an early test on seedlings.