Modelling the catabolite and microbiological profile of cheddar cheese manufactured from ayrshire milk

Abstract

Branded dairy products have lately become a global trend. As a result of this, the origin of the milk used in the manufacturing of branded cheeses must be declared by the producer, since it is known that these products are highly adulterated with foreign milk. In South Africa, branded Ayrshire Cheddar cheese has become highly popular due to its unique organoleptic properties and in light of claims that it ripens much faster than cheese made from other milk (not including Ayrshire).

This study was therefore directed to investigate the unique properties of branded Ayrshire Cheddar cheese versus Cheddar cheese manufactured from a mixture of other breeds’ milk (not including Ayrshire milk) and to establish a catabolite profile for each cheese type. The outlay of the thesis was constructed into six chapters each with its own outcomes. The first chapter focused on the variations between the two Cheddar cheese batches (produced from Ayrshire and other breeds’ milk) with regards to organic acid, selected chemical parameters and starter microbiotia. In the following three chapters mathematical models were developed that would predict organic-; fatty and amino acid fluxtuations respectively in the cheese made from Ayrshire and other milk. In the last chapter two artificial neural networks were designed with the two starter organisms, Lactococcus lactis and Streptococcus thermophilus as variable indicator respectively.

Thirty-two cheese samples of each batch (pure Ayrshire (4) / mix breed with no Ayrshire (4)) were ripened and samples were analysed under the same conditions on the following days after production: 2, 10, 22, 36, 50, 64, 78, and 92. In the subsequent chapters, the following analysis were done on each day of analysis: organic acid by means of high performance liquid chromatography (HPLC); fatty acids by means of Gas Chromatography Mass Spectometry (GCMS); amino acids by means of GC-MS; microbial analysis by means of traditional methods, total DNA extraction and polymerase chain reaction (PCR); and standard chemical analysis for moisture, NaCl and pH.

In the first research chapter, the minimum and maximum (min/max) values, standard deviations and proposed rel X values of organic acids were evaluated in Ayrshire and the mixed-breed Cheddar cheese, and showed that isovaleric acid is the organic acid with the least variation relative to concentration in both cheeses and it was assumed that this organic acid is the most effective indicator of cheese uniformity. Clear differences in organic acids, chemical variables and starter micro-organisms were also evident in the two cheese batches.

Results obtained from the regression models which was defined for each organic -; amino - and fatty acid by means of mathematical equations can be used by the manufacturer to achieve i.e. the selection of cheese for specialist lines, the early exclusion of defective cheeses, and the establishment of brand origin (Ayrshire vs. mixed-breed Cheddar cheeses). The regression graphs also illustrate unique flux patterns in Ayrshire and the mixed-breed in terms of organic -, fatty -, and amino acid content.

In the last chapter, the discrimination between the two batches was respectively done via artificial neural network (ANN) modelling of Lactococcus lactis and Streptococcus thermophilus as indicator organisms. The ANN consisted of a multilayered network with supervised training arranged into an ordered hierarchy of layers, in which connections were allowed only between nodes in immediately adjacent layers. The construction thereof allowed for two output nodes, connected to an input layer consisting of two nodes to which the inputs were connected. In both cheeses the results from the ANN showed acceptable classification of the cheeses based on the counts of L. lactis and S. thermophilus.