Rapid differentiation of South African game meat using portable near-infrared (NIR) spectroscopy

Payne, Kiah (2019-04)

Thesis (MScFoodSc)--Stellenbosch University, 2019.

Thesis

ENGLISH ABSTRACT: The South African game meat industry operates as a free-market enterprise; however, this can create certain problems for producers and consumers. For example, in South Africa no standardised meat cuts exist and there are no quality standards in place for game. Therefore, allowing the legal sale of inferior quality game meat. Due to a general lack of regulations as well as varying carcass dressings, the chance that a species may be mislabelled or substituted is greatly increased. In recent years, meat authenticity awareness has increased, as there have been incidences where meat has been fraudulently mislabelled. Typical cases involve the intentional substitution of high value raw ingredients with inferior species or materials, the addition of non-declared proteins from several origins, or the marketing of frozen-thawed meat as fresh. This type of food fraud concerns consumers in terms of economic loss, food allergies, religious compliance, and food safety. This study aimed to investigate a feasible alternative to the manual, tedious and time-consuming conventional analytical methods used for meat differentiation and authentication that could provide the meat industry with a rapid, non-destructive, accurate and reliable solution in the near future. Near-infrared (NIR) spectroscopy combined with multivariate data analysis (MDA) techniques were used to rapidly differentiate between South African game species, irrespective of the treatment (fresh or previously frozen) or the muscle type as well as determine these individual classes (fresh; previously frozen; frozen period; muscle type) per species. Meat samples of four game species [black wildebeest (Connochaetes gnou), zebra (Equus quagga burchelli), springbok (Antidorcas marsupialis), ostrich (Struthio camelus)] were scanned at ca. 23° C. Spectra were collected with a portable MicroNIR OnSite spectrophotometer (Viavi Solutions Inc., Milpitas, USA), in the range of 908 – 1676 nm, after which the data was analysed using MDA. It was possible to differentiate between game species, irrespective of the treatment (fresh or previously frozen), the frozen period or the muscle type. The partial least squares discriminant analysis (PLS-DA) model was successful and achieved accuracies ranging 89.8 – 93.2%. It was also possible to distinguish between fresh and previously frozen meat, and with low accuracies determine the frozen period. The principal component analysis (PCA) score plots illustrated good separation between the fresh and frozen-thawed samples, however, the frozen-thawed samples exhibited an overlap between the individual frozen periods. Therefore, lacking separation and distinct clustering for the different frozen periods. Thus, the MDA models were not effective when trying to classify the different frozen periods. The PLS-DA models could however discriminate between the fresh and previously frozen meat, irrespective of the frozen period or muscle type. The black wildebeest (99.2%), zebra (94.4 and 99.3%), springbok (100%) and ostrich (90 and 98.3%) models achieved good overall accuracies. Lastly, this study found that the ostrich muscles could be distinguished from each other with a 100% accuracy. Furthermore, the results suggested that discrimination of the different muscle types for both zebra and springbok was less sufficient due to the lack of separation between the different muscles. Misclassification mostly occurred between muscles that are anatomically located near to one another. Therefore, the samples with spectral similarities were grouped to form a two-group class discrimination model. The PLS-DA results showed that it was possible to differentiate between the forequarters and hindquarters of the zebra (90.3%) and springbok (97.9%) muscles. The results showed NIR spectroscopy’s potential as a rapid and non-destructive method for species identification, fresh and previously frozen meat differentiation as well as muscle type determination. Furthermore, this technique has the potential of providing the South African game meat industry with an alternative technique to the current manual, destructive and time-consuming authentication methods.

AFRIKAANSE OPSOMMING: Die Suid-Afrikaanse wildsvleisbedryf word as ‘n vryemarkonderneming bestuur; dit kan egter sekere probleme vir produsente en verbruikers voortbring. Byvoorbeeld, in Suid-Afrika is daar geen gestandardiseerde vleissnitte nie en daar is geen kwaliteitstandaarde in plek vir wild nie. Daarom, word die wettige verkoop van miderwaardige kwaliteit wildsvlies toegelaat. As gevolg van ‘n algemene gebrek aan regulasies sowel as verskeie karkasse, word die kans dat ‘n spesie dalk verkeerdelik geetiketteer of vervang word, aansienlik verhoog. In onlangse jare het die bewustheid van vleis-egtheid toegeneem, aangeien daar voorvalle was waar vleis bedrieglig geetiketteer was. Tipiese gevalle behels die opsetlike vervanging van hoё-waarde rou bestanddele met minderwaardige spesies of materiale, die toevoeging van nie-verklaarde proteïene vanaf verskeie oorspronge, of the bemarking van bevrore-ontdooide vleis as vars. Hierdie soort vodeselbedrog het betrekking op verbruikers in terme van ekonomiese verlies, voedselallergieё, godsdienstige nakoming en voedselveiligheid. Hierdie studie beoog om ‘n haalbare alternatief vir die huidige met-die-hand, vervelige en tydrowende konvensionele analitiese metodes vir vleisdifferensiasie en verifikasie te ondersoek, wat die vleisbedryf in die nabye toekoms ‘n vinnige, nie-vernietigende, akkutate en betroubare oplossing sal bied. Naby-infrarooi (NIR) spektroskopie gekombineer met meerveranderlike data analise (MDA) tegnieke is gebruik om vinnig tussen Suid-Afrikaanse wildspesies te onderskei, ongeag die behandelling (vars of voorheen bevrore) of die spier tipe sowel as om hierdie individuele klasse te bepaal (vars; voorheen bevrore; bevrore tydperk; spier tipe) per spesie. Vleismonsters van vier wildspesies [swartwildebees (Connochaetes gnou), sebra (Equus quagga burchelli), springbok (Antidorcas marsupialis), volstruis (Struthio camelus)] is geskandeer teen ca. 23° C. Spekta is versamel met ‘n draagbare MicroNIR OnSite spektrofotometer (Viavi Solutions Inc., Milpitas, USA), in die golflengte reeks van 908 – 1676 nm, waarna die data met MDA geanaliseer is. Dit was moontlik om tussen wildspesies te onderkei, ongeag die behandelling (vars of voorheen bevrore), die bevrore tydperk of die spier tipe. Die parsiёle kleinste kwadrate diskriminantanalise (PLS-DA) model was suksesvol en het akkuraatheid bereik was 89.8 – 93.2% dek. Dit was ook moontlik om tuseen vars en voorheen bevrore vleis te onderskei, en met lae akkuraatheid die bevrore periode te bepaal. Die hoofkomponentanalise (PCA) telling-beelde het goeie skeiding tussen die vars en bevrore ontdooide monsters geïllustreer, maar die bevrore-ontdooide monsters het ‘n oorvleueling tussen die individuele bevrore tydperke getoon. Daarom ontbreek skeiding en afsonderlike groepering vir die verskillende bevrore tydperke. Die MDA modelle was dus nie effektief om die verskillende bevrore tydperke te probeer klassifiseer nie. Die PLS-DA modelle kon egter onderskei tussen die vars en voorheen bevrore vleis, ongeag die bevrore tydperk of die spier tipe. Die swartwildebees (99.2%), sebra (94.4 en 99.3%), springbok (100%) en volstruis (90 en 98.3%) modelle het goeie algehele akkuraatheid behaal. Laastens het hierdie studie bevind dat die volstruisspiere van mekaar onderskei kan word met ‘n 100% akkuraatheid. Verder het die resultate voorgestel dat die diskriminasie van die verskillende spier tipes vir beide sebra en springbok minder voldoende was weens die gebrek aan skeiding tussen die verskillende spiere. Misklassifikasie het meestal plaasgevind tussen spiere wat anatomies naby mekaar geleё is. Daarom is die monsters met spektrale ooreenkomste gegroepeer om ‘n tweegroep-klas diskriminasie model te vorm. Die PLS-DA resultate het getoon dat dit moontlik was om tussen die voor- en agterkwarte van die sebra (90.3%) en springbok (97.9%) spiere te onderskei. Die resultate het NIR spektroskopie se potensiaal getoon as ‘n vinnige en nie-vernietigende metode vir die identifikasie van spesies, vars en vorheen bevrore vleis differensiasie sowel as spier-tipe bepaling. Verder het hierdie tegniek die potensiaal om die Suid-Afrikaanse wildsvleisbedryf te voorsien met ‘n alternatiewe tegniek vir die huidige met-die-hand, destruktiewe en tydrowende verifikasiemetodes.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/106069
This item appears in the following collections: