Evaluating near infrared spectroscopy as a rapid screening tool for microbiological safety of rooibos (Aspalathus linearis)

Thesis (MScFoodSc)--Stellenbosch University, 2024.

ENGLISH ABSTRACT: Rooibos, or Aspalathus linearis, is a popular South African beverage, known worldwide for its wide array of health benefits. While rooibos is marketed as a health product, very limited information is available on the microbiological safety of rooibos. Microbiological testing methods currently available in the industry is time- consuming, costly and potentially inaccurate. Given rooibos’ known association with microbiological contaminants such as Salmonella, E. coli and coliforms, this study aimed to assess near infrared (NIR) spectroscopy as a feasible alternative for rapid screening of the microbiological safety of rooibos. Rooibos samples (n=30) were collected from the final stages of processing, to represent the samples used in industry for microbiological screening. Both control (n=15) and contaminated (n=15) samples from various regions and harvesting seasons were chosen to increase the variation within the sample set. All samples had known moisture contents (in the range of 8.8 % to 9.8 %) and microbiological profiles. All unsafe samples had only coliforms present as no samples containing E. coli or Salmonella could be obtained at the time of the study. Through principal component analysis (PCA), a distinctive separation emerged when plotting PC 1 (82.28%) against PC 4 (0.45%). This separation, observed in the direction of PC 4, suggested a significant distinction, between safe and unsafe samples. To investigate potential correlations with moisture content, samples were colour-coded accordingly. However, no discernible patterns related to moisture content emerged in the direction of PC 4. As a result, the conclusion was drawn that the observed separation along PC 4 was indicative of the presence of coliforms, serving as a marker for microbiological safety. The study demonstrated that employing NIR spectroscopy and partial least squares discriminant analysis (PLS-DA) enabled the development of a calibration and prediction model with 100% classification accuracy, sensitivity and specificity. With no instances of misclassification, this model emerges as a feasible option for use in industry. The study not only established the potential of NIR spectroscopy for microbiological safety screening, it also employed multivariate data analysis techniques. These included the examination of PC loadings and PLS-DA regression vectors, identifying critical wavelengths (1720 nm, 1930 nm and 2310 nm). These wavelengths, associated with moisture content and CH stretching, played a crucial role in discriminating between safe and unsafe samples. The presence of coliforms were inferred from these wavelengths. While addressing the current gap in microbiological safety assessments of rooibos, this research shows the potential of using NIR spectroscopy as a rapid and accurate screening tool for rooibos classification. The findings pave the way for improved food safety control practices and highlight the need for future research in the application of spectroscopic techniques to ensure the safety of rooibos.

AFRIKAANSE OPSOMMING: Rooibos, of Aspalathus linearis, is 'n gewilde Suid-Afrikaanse drankie, wat wêreldwyd bekendheid geniet vir ‘n wye verskeidenheid gesondheidsvoordele. Terwyl rooibos bemark word as 'n gesondheidsproduk, is daar baie beperkte inligting beskikbaar oor die mikrobiologiese veiligheid van rooibos. Mikrobiologiese toetsmetodes wat tans in die bedryf beskikbaar is, is tydrowend, duur en potensieël onakkuraat. Gegewe rooibos se bekende assosiasie met mikrobiologiese kontaminante soos Salmonella, E. coli en koliforme, het hierdie studie ten doel gestel om naby-infrarooi (NIR) spektroskopie te assesseer as 'n bruikbare alternatief vir bepaling van die mikrobiologiese veiligheid van rooibos. Rooibos monsters (n=30) is versamel vanaf ‘n finale verwerkingsaanleg om die monsters te verteenwoordig wat in die bedryf gebruik word vir mikrobiologiese toetsing. Beheer (n=15) en gekontamineerde (n=15) monsters is uit verskeie streke en oesseisoene gekies om die variasie binne die monsterstel te verhoog. Alle monsters het bekende voginhoud (in die reeks van 8.8 % tot 9.8 %) en mikrobiologiese profiele gehad. Alle onveilige monsters was slegs met koliforme gekontamineer aangesien geen monsters met E. coli of Salmonella verkry kon word tydens die studie nie. Deur middel van hoofkomponent analise (HA) het 'n duidelike skeiding na vore gekom wanneer PC 1 (82.28%) teenoor PC 4 (0.45%) geplot is. Hierdie skeiding, waargeneem in die rigting van PC 4, het 'n beduidende onderskeid tussen veilige en onveilige monsters voorgestel. Om potensiële korrelasies met voginhoud te ondersoek, is monsters ooreenkomstig gekleur. Geen waarneembare patrone wat met voginhoud verband hou, het egter na vore gekom in die rigting van PC 4 nie. As gevolg hiervan is die gevolgtrekking gemaak dat die waargenome skeiding in die rigting van PC 4 'n aanduiding was van die teenwoordigheid van koliforme, wat as 'n merker vir mikrobiologiese veiligheid dien. Die studie het gedemonstreer dat die gebruik van NIR spektroskopie en gedeeltelike kleinste kwadrate diskriminantanalise (PLS-DA) die ontwikkeling van 'n kalibrasie- en voorspellingsmodel met 100% klassifikasie akkuraatheid, sensitiwiteit en spesifisiteit moontlik gemaak het. Met geen gevalle van misklassifikasie nie, blyk hierdie model 'n uitvoerbare opsie vir gebruik in die bedryf te wees. Die studie het nie net die potensiaal van NIR spektroskopie vir mikrobiologiese veiligheid klassifisering gevestig nie, maar het ook kritiese golflengtes (1720 nm, 1930 nm en 2310 nm) geïdentifiseer. Hierdie golflengtes, wat geassosieer word met voginhoud en CH-bindings, het 'n belangrike rol gespeel in die onderskeiding tussen veilige en onveilige monsters. Die teenwoordigheid van koliforme is afgelei uit hierdie golflengtes. Terwyl die huidige gaping in mikrobiologiese veiligheid analises van rooibos aangespreek word, toon hierdie navorsing die potensiaal van die gebruik van NIR spektroskopie as 'n vinnige en akkurate hulpmiddel vir rooibos klassifikasie. Die bevindinge baan die weg vir verbeterde voedselveiligheidsbeheerpraktyke en beklemtoon die behoefte aan toekomstige navorsing in die toepassing van spektroskopiese tegnieke om die veiligheid van rooibos te verseker.

Masters