Modelling mould spoilage in cold-¢lled ready-to-drink beverages by Aspergillus niger and Penicillium spinulosum Alyce Stiles Battey, Siobain Du¡y and Donald W. Scha¡ner* Mathematical models have been developed to predict the probability of growth of spoilage moulds in response to various preservative systems in ready to drink beverages. A Box-Behnken experimental de- sign included ¢ve variables, each at three levels: pH (2?8, 3?3, 3?8), titratable acidity (0?20%, 0?40%, 0?60%), sugar content (8?0, 12?0, 16?0 8Brix), and preservative concentrations (sodium benzoate and potassium sorbate, each 100, 225, 350 ppm). Duplicate samples were inoculated with a mould cocktail consisting of equal proportions of Aspergillus niger and Penicillium spinulosum spores (5?0610 4 spores/ml).The inoculated samples were plated on malt extract agar after 0,1, 2,4, 6, and 8 weeks. Lo- gistic regression was used to create predictive models. The pH, titratable acidity, sugar content, sodium benzoate, and potassium sorbate levels were all found to be signi¢cant factors in predicting the prob- ability of mould growth over time. Interactions between pH and sodium benzoate, pH and potassium sorbate, and pH and sugar content were also statistically signi¢cant. This logistic model was validated against 14 new conditions and predicted the growth of mould after 8 weeks with over 96% accuracy. Product developers can use these models to predict mould growth in ready to drink beverages. # 2001Academic Press Introduction Ready-to-drink beverages have high water ac- tivities (A w ) that permit microbial growth. Combinations of hurdles, such as pH, sugar content and chemical preservatives, prevent the growth of most organisms in ready to drink beverages (Leistner 1995). Aspergillus niger and Penicillium spinulosum are highly resistant to chemical preservatives such as sorbic and ben- zoic acids, and can tolerate both high acid and lower A w environments (Banwart 1979, De Boer and Nielsen 1995). The food industry conducts challenge stu- dies to assess the ability of organisms to grow in a particular foodstu¡, but these studies re- quire considerable labour, time, and materials, and the number of parameters that can be tested is often limited. Validated predictive models can provide rapid information about the microbial stability of a product that can re- duce the time and e¡ort needed for challenge studies. The ¢eld of predictive microbiology has been focussed on creating polynomial regression equations to model foodborne pathogens, re- sulting in such models as the Food MicroModel ORIGINAL ARTICLE *Corresponding author. Fax: +001-732-932-6776. E-mail: scha¡ner@aesop.rutgers.edu Received: 28 March 2001 Food Risk Analysis Initiative, 65 Dudley Road, Rutgers,The State University of New Jersey, New Brunswick, New Jersey 08901-8520, USA 0740-0020/01/100521 +09 $35.00/0 # 2001 Academic Press Food Microbiology, 2001, 18, 521^529 doi:10.1006/fmic.2001.0438 Available online at http://www.idealibrary.com on