Sensory, chemical and consumer analysis of Brettanomyces spoilage in South African wines Janita Botha Thesis defence 17 February 2009 Research Team : M Muller, W J du Toit, A J de Villiers & A G J Tredoux
Found further strong links between 4-EP and isovaleric acid
Investigated effect of isovaleric acid on detection thresholds of 4-EP and 4-EG
Isovaleric acid? 4-ethylcatechol?
Research Aims Determine the sensory detection thresholds of 8 Brett-related compounds To determine the sensory profiles of 4 Brett-related compounds To determine the sensory interactions of 4 Brett-related compounds To determine the consumer preference of 4 Brett-related compounds Chapter 1 Chapter 2 Chapter 3 To investigate realationships between 8 Brett-related compounds in selected South African wines Chapter 4
Overall results 4-EP 4-EG 4-EC Isovaleric acid Elastoplast Leather Medicinal Smoky Savoury Pungent Sick-sweet Berry-like Not in literature Clove-like?? More detail obtained about the contributions of the respective compounds
All four compounds affect intensity of Elastoplast descriptor Compounds have different effects at different levels Intensity affected by other compounds present “ Sweaty leather” aroma attributed to Brett?
PCA vs PARAFAC: sensory data Judges Samples Attributes Samples Attributes Judge 1 Judge 2 Judge n PCA: Analysis of (unfolded) 2-way data PARAFAC: Analysis of data cube Effects of judges not ignored Often simplified by finding mean over judges
PCA vs PARAFAC 1 st factor calculated Subsequent orthogonal factors calculated 1 st factor calculated Subsequent factors calculated in sucessive directions of variance Factors recalculated untill they converge PCA PARAFAC Less noise modelled