The basidiomycete fungus Crinipellis perniciosa (Stahel) Singer is the causative agent of witches' broom disease of cocoa (Theobroma cacao). Witches' broom disease is endemic to large areas of the Americas and is a major constraining factor on cocoa production in such areas. The disease cycle includes a primary monokaryotic biotrophic phase and a secondary dikaryotic saprotrophic phase. In order to study the early stages of infection the possibility of quantifying the relative concentrations of the primary and secondary phases using changes in the ratio of mannan to chitin was investigated. Both chitin and mannan are fungal cell wall polymers. Separate gas chromatographic based mannan and chitin assays were developed. The new chitin estimation procedure was significantly more sensitive than the commonly used colorimetric based methods and was also selective for chitin rather than a range of similar components. The new method required only about 2oo ug of material whilst the previous and colorimetric based procedures required approximately 25 mg. Thus, the new procedure was sufficiently sensitive to detect extremely low levels of infection in clearly defined plant regions, such as meristems. The mannan assay was not significantly more sensitive than the previous assays, but produced cleaner samples, and thus interpretation of results was simplified and mass spectrometer maintenance reduced. To capitalise on the new chitin and mannan assays, the work was extended to encompass neutral and amino sugars. No useful biomarkers or ratios between components were discovered. Ratios between sugars may have been capable of apportioning the relative concentration of primary and secondary phase C. perniciosa in infected cocoa. All extractable lipids were also evaluated as potential biomarkers, and the lipid profiles assessed for potentially useful ratios. No useful lipid ratios were discovered, but ergosterol was found to be a useful biomarker. The newly developed chitin assay was used to study C. perniciosa infection in green and brown brooms. The assay was also used to investigate the distribution of C. perniciosa in sections taken throughout the length of an entire broom. The results indicated C. perniciosa was highly localised at the broom base and at the growing points. The use of random amplified polymorphic DNA polymerase chain reaction as a guide to pathogenicity was also evaluated. The method was adapted for use with C. perniciosa at the Institut Fur Genbiologische Forschung, Berlin. The method indicated the presence of fungal plasmids in the primary phase of C. perniciosa but not in the secondary.