Simultaneous determination of goat milk adulteration with cow milk and their fat and protein contents using NIR spectroscopy and PLS algorithms
Introduction
World goat milk production surpassed 18 million tonnes per year in 2017, accumulating an increase of approximately 20% in a decade (2007–2017) (FAOSTAT, 2018). This is due in part to the growing consumer interest in this product and its derivatives, motivated by the unique characteristics of goat milk when compared to cow milk, such as low allergenicity, high digestibility and functional properties (Ahmed, El-Bassiony, Elmalt, & Ibrahim, 2015; Hodgkinson et al., 2012).
From the economic point of view, goat milk is more expensive than cow milk and, consequently, its derivatives have a higher added value. Thus, driven by this economic representativeness, adulteration of this food type is a reality. The most fraud is in the replacement category, where there is a partial or total substitution of a food constituent by one or more substitutes (i.e. one adulterant or a mixture of adulterants) of lower cost without the knowledge of consumers. The main adulterant found in goat milk and its derivatives is cow milk, due to its lower cost and greater abundance (Azad & Ahmed, 2016; Dabrowska et al., 2010; Di Pinto et al., 2017).
Considering the similarity of these matrices and, consequently, the difficulty of identifying them in mixtures, chromatographic methods, immunological tests, electrophoretic techniques and DNA identification are currently used to determine this kind of adulteration (Di Pinto et al., 2017; Kamal & Karoui, 2015; Pesic et al., 2011; Song, Xue, & Han, 2011). However, such techniques have some drawbacks such as: high cost, time-consuming, laborious and requiring various steps of sample pretreatment.
In this scenario, non-destructive methodologies like Near Infrared Spectroscopy (NIRS) have represented an important, interesting and promising alternative for the detection of adulteration with cheaper substituents in different types of foods (Shao, Xuan, Hu, & Wang, 2019), especially for cow milk and its derivatives (Azad & Ahmed, 2016). The increase in the popularity of this technique can be attributed to its main characteristics: nondestructive, absence of sample pretreatment, rapidity and accuracy. In comparison with other methodologies based on vibrational spectroscopy (FTIR and RAMAN, for example), the recent advances in the NIR instrumentation, including portable, low-cost and easy-to-use devices, coupled with the construction of the qualitative and quantitative libraries of chemometric models is becoming this technique a reliable, efficient and accessible tool that can be widely employed for different applications in scientific research and industries (Lohumi, Lee, Lee, & Cho, 2015; Pasquini, 2018).
Despite all the methodological advantages, NIR spectroscopy has its limitations, mainly with regard to the sensibility of the technique, which also depends on certain operational and environmental factors, such as the variability of the matrix and the physical state of the sample (Dupont, Croguennec, & Pochet, 2018; Pasquini, 2018). The main sources of variability of milk from various species that affect its chemical composition include race, feed, breeding system (extensive and intensive), lactation period, environmental conditions, age and animal health status (Nagy et al., 2019; McDermott et al., 2017; Manousidis et al., 2018; Miloradovic, Miocinovic, Kljajevic, Tomasevic, & Pudja, 2018).
Classification models based on NIR spectroscopy developed in different studies have already demonstrated the high predictive capacity for the determination of different adulterants in cow milk, and addition of milk from other species in camel milk too (Chen et al., 2017; Mabood et al., 2017a, 2017b; Karunathilaka, Yakes, He, Chung, & Massoba, 2018). However, to the best of our knowledge, there are no reports on the application of NIR spectroscopy for the determination of goat milk adulteration by adding cow milk.
Since in typical NIR spectra exhibit a redundancy of the recorded data due to the strong correlation over the different analytical channels, it has been very useful the use the noise-reduction properties of Partial Least Squares (PLS) combined with the discard of non-informative or redundant variables. In such scenario, Interval PLS (iPLS) and the Successive Projections Algorithm for interval selection in PLS regression (iSPA-PLS) have been successfully employed to improve the performance of the regression models, generating more stable models with superior interpretability, and fewer prediction errors. The iSPA-PLS algorithm is an extension of SPA that can select multiple intervals of variables, instead the selection of one interval alone as in iPLS, which can eventually not contain the required analytical information to solve the problem under study (Diniz, Pistonesi, & Araújo, 2015; Gomes, Galvão, Araújo, Véras, & Silva, 2013; Krepper et al., 2018). It is important to highlight that NIR and iSPA-PLS have already been successfully employed in food analysis for determining protein content in wheat and extract concentration in beer (Gomes, Galvão, Araújo, Véras, & Silva, 2013), moisture and total polyphenols in tea (Diniz, Pistonesi, & Araújo, 2015), total anthocyanins in jaboticaba fruit (Mariani, Teixeira, Lima, Morgensten, Nardini, & Cunha Júnior, 2015) and fat content in chicken hamburgers (Krepper et al., 2018).
This study is focused on the identification and quantification of goat milk adulteration by adding cow milk, besides the determination of their fat and protein contents employing NIR spectroscopy and PLS algorithms. For classification aims, considering both the economic losses and the potential health risk to allergic consumers regardless of the amount of cow milk ingested, Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA) were tested to differentiate adulterated samples from the pure ones (i.e. non-adulterated goat and cow milk). For quantification, full-spectrum PLS, iPLS and iSPA-PLS have been employed and then compared.
Section snippets
Samples
In order to construct a representative model containing different variability sources of the raw materials, 7 lots with 18 samples of goat milk and 7 lots with 18 samples of bovine milk were acquired between November 2017 to June 2018 from different farmers located in the micro-regions Agreste and Sertão of the Paraíba and Pernambuco states, Brazil. The flocks were composed of pure and mestizo animals of different ages with different farrowing number and lactation stages maintained in extensive
Spectra investigation and pre-processing procedures
In order to eliminate noise and systematic variations on the baseline, NIR spectra (Fig. 1) were initially preprocessed with 13-point moving mean and then with different preprocessing methods (SNV; MSC; LBC; LBC-BO; and BO) before the construction of PLS, iPLS and iSPA-PLS models.
As can be seen, a prominent and broad band around 6900 cm−1 (the first overtone of O–H stretching vibration) has been attributed to water. The band at 5000 cm−1 has been associated with protein as amide, besides
Conclusions
In this work, we demonstrate the feasibility of NIR spectroscopy in the identification and quantification of adulteration of goat milk by the addition of cow milk, despite the high complexity in determining one dairy matrix into another. The use of pre-processed NIR spectra coupled with interval selection by iSPA-PLS provided the best results for the determination of both adulteration and fat contents, while PLS gave better results for the protein quantification. The good performances obtained
CRediT authorship contribution statement
Elainy Virginia dos Santos Pereira: Conceptualization, Methodology, Investigation, Formal analysis, Writing - original draft, Writing - review & editing. David Douglas de Sousa Fernandes: Software, Formal analysis, Validation, Data curation, Writing - original draft. Mário César Ugulino de Araújo: Resources, Funding acquisition. Paulo Henrique Gonçalves Dias Diniz: Writing - original draft, Writing - review & editing, Data curation, Visualization, Supervision, Project administration. Maria Inês
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
The authors are grateful to Prof. Dr. Mário Araújo for using the NIR spectrometry facilities and the NUPEA-UEPB laboratory for supporting all reference analysis. The authors also thank to CAPRIBOV cooperative and all farmers that provided the milk samples. David Fernandes and Mário Araújo gratefully acknowledge CNPq Brazil for their research fellowships.
References (39)
- et al.
Identification of potent antioxidant bioactive peptides from goat milk proteins
Food Research International
(2015) - et al.
Composition of goat and cow milk produced under similar conditions and analyzed by identical methodology
Journal of Food Composition and Analysis
(2009) - et al.
Effect of seasonal variation on the composition and properties of raw milk destined for processing in the UK
Food Chemistry
(2014) - et al.
Detection of melamine adulteration in milk by near-infrared spectroscopy and one class partial least squares
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
(2017) - et al.
Quality of UHT goat's milk in Poland evaluated by real-time PCR
Small Ruminant Research
(2010) - et al.
DNA-based approach for species identification of goat-milk products
Food Chemistry
(2017) - et al.
Quantification and identification of adulteration in the fat content of chicken hamburgers using digital images and chemometric tools
Lebensmittel-Wissenschaft und -Technologie- Food Science and Technology
(2019) - et al.
Allergic responses induced by goat milk αs1-casein in a murine model of gastrointestinal atopy
Journal of Dairy Science
(2012) - et al.
Analytical methods coupled with chemometric tools for determining the authenticity and detecting the adulteration of dairy products: A review
Trends in Food Science & Technology
(2015) - et al.
Non-targeted NIR spectroscopy and SIMCA classification for commercial milk powder authentication: A study using eleven potential adulterants
Heliyon
(2018)
Suitability of somatic cell count, electrical conductivity, and lactate dehydrogenase activity in foremilk before versus after alveolar milk ejection for mastititis detection
Journal of Dairy Science
Determination of fat content in chicken hamburgers using NIR spectroscopy and the Successive Projections Algorithm for interval selection in PLS regression (iSPA-PLS)
Spectrochimica Acta
A review of vibrational spectroscopy techniques for the detection of food authenticity and adulteration
Trends in Food Science & Technology
Development of new NIR-spectroscopy method combined with multivariate analysis for detection of adulteration in camel milk with goat milk
Food Chemistry
FT-NIRS coupled with chemometric methods as a rapid alternative tool for the detection & quantification of cow milk adulteration in camel milk samples
Vibrational Spectroscopy
Relationships among nutritive value of selected forages, diet composition and Milk quality in goats grazing in a Mediterranean Woody rangeland
Livestock Science
NIRS and iSPA-PLS for predicting total anthocyanin content in jaboticaba fruit
Food Chemistry
Cow and environmental factors associated with protein fractions and free amino acids predicted using mid-infrared spectroscopy in bovine milk
Journal of Dairy Science
The influence of Milk heat treatment on composition, texture, colour and sensory characteristics of cows' and goats' Quark-type cheeses
Small Ruminant Research
Cited by (75)
Dapsone determination in tablets to leprosy treatment using a portable NIR spectrometer
2024, Journal of Molecular StructureDetermination of moisture and total protein and phosphorus contents in powdered chicken egg samples using digital images, NIR spectra, data fusion, and multivariate calibration
2024, Journal of Food Composition and AnalysisClassification and quantification of multiple adulterants simultaneously in black tea using spectral data coupled with chemometric analysis
2024, Journal of Food Composition and AnalysisBR-Net: Band reweighted network for quantitative analysis of rapeseed protein spectroscopy
2023, Spectrochimica Acta - Part A: Molecular and Biomolecular Spectroscopy