Detection of counterfeiting rice bran using near-infrared reflectance spectroscopy (NIRS)

This study aims to compare rice bran quality from different origins and to differentiate rice bran from rice husk based on the spectrum produced by Near Infrared Reflectance Spectroscopy (NIRS) technology. The rice bran used in this study comes from mills and poultry shops in West Java and Sumatra. This study used husks as an adulteration mixing material in rice bran with mixing levels of 25% rice bran: 75% husk, 50% rice bran: 50% husk, and 75% rice bran: 25% husk. Proximate compositions (water content, ash content, ether extract (EE), crude protein (CP), and crude fiber (CF)) were observed. NIRS spectra were collected using the Buchi NIRCal 5.2 application. The results showed an interaction between provider and origin on rice bran EE, CP, and CF contents. Rice bran provided by the miller had better quality with higher CP (17.3% vs. 12.2%) and lower CF (12.0% vs. 16.6%) content in the West Java origin case and higher EE (16.5% vs. 11.1%) in West Sumatera case. The broad spectra distribution showed the variety of the rice bran quality from the poultry shop. The artificial adulteration spectra confirm the impure rice bran quality from the poultry shop.


Introduction
Feeds play a significant role in dairy cattle productivity improvement [1].Lack of feed quality influences dairy cattle performance significantly [2].Rice bran is one of the most frequent feeds used in the dairy ration.Rice bran is a byproduct of rice milling.It is used as an energy source for dairy cattle.Rice bran contained 85.56% dry matter (DM), 11.59% ash, 11.21% crude protein (CP) [3], 4 -15% crude fibre (CF) [4].Rice bran was seasonally available [5].Besides the availability problem, rice bran qualities also varied due to the rice variety, milling process [6] and adulteration [7].During harvest time, rice bran availability increased, and the price was low.However, during the offseason, the price was high, which increased the risk of adulteration.Rice is commonly adulterated by material with similar physical characteristics, such as milled rice husk [7].Such adulteration lowered the rice bran quality, reduced dairy cow performance, and harmed animals [8].Although rice husk is also the byproduct of the rice milling industry, its nutritional value is very low.It contained 20% ash, 2 -2.8% CP, 0.3 -0.8% extract ether (EE) and 35 -45% CF [5].The fibre fractions comprised 15 -20% hemicellulose, 35 -40% cellulose and 20 -25% lignin [9].The high content of fibre and low CP and EE resulted in low nutrient density and digestibility of the rice husk adulterated rice bran [7].Regarding the importance of rice bran in dairy feed and the negative effect of rice husk contamination in the rice bran, a detection method to prevent adulteration should be provided.Feed quality can be analysed using physical characteristics, chemical or biological.The physical characteristic of rice bran can be measured using physical properties such as bulk density, water absorption capacity and fat absorption capacity [10].According to Sairam et al. [10], rice bran has 0.34 g/cc bulk density, 240 ml/100 g water absorption capacity and 210 ml/100 g fat absorption capacity.While Cheong dan Xu [5] reported the rice husk dimensions about 8-10mm in length, 2-3mm in width, and 0.2mm in thickness, with bulk density ranging from 100 to 160kg/m 3 .The true density of rice husk ranges from 670 to 740kg/m 3 and can be only compressed to 400kg/m 3 .Although the rice bran and rice husk were physically different, some modifications, such as milling or grinding, might change the physical characteristic of the rice husk to imitate the rice bran.Therefore, physical characteristics alone were difficult to determine adulteration.The chemical and biological assessment might be more accurate.However, both of these methods also have their constraints, such as being time-consuming, expensive, laborious, and requiring chemical reagents which were not environmentally friendly [11].Therefore, an alternative method that overcomes the limitation of the existing method should be offered.
Near Infrared Reflectance Spectroscopy (NIRS) is one of the alternative technology suggested.The technology has been used in many aspects of assessment, such as chemical composition [12], [13], [11], feed digestibility [14], milk fatty acid health index [15] and differentiating morning from afternoon milk [16].The technology has the advantage of its reagent-free, fast, non-destructive sample, and no sample preparation required [17]; therefore, it is helpful for timely intervention [18].The NIRS technology was based on the reflection of specific chemical bonds featured after being irradiated by NIR light [19].Each feedstuff produces a specific spectrum that measures the unique optical and characteristics of the sample.The spectrum contains information on a specific molecular vibrational aspect of the sample, its physical properties and its unique interaction with the measuring instrument, which is interpreted into structural information representing the measured sample's molecular detail [20].The accuracy of NIRS detection on samples is based on the NIRS database and was improved in line with the local database used [21].Therefore, this study is aimed to compare rice bran quality from different origin and to differentiate rice bran from rice husk based on the spectrum produced by NIRS instrument.

Rice bran and rice husk collecting sample
Rice bran and rice husk sample were collected from rice millers and poultry shops in Bogor regency, West Java Province and Damasraya, Solok, Agam and Padang regencies, West Sumatera Province of Indonesia.Forty samples of 500 g each were collected.Rice bran from millers was used as the pure sample, which was then adulterated with rice husk at (0, 25, 50, 75 and 100%) for the NIRS spectrum of rice bran adulteration detection.

Chemical analysis
Proximate analysis to measure water content, ash, CP, and EE was conducted according to AOAC [22].At the same time, CF was measured according to AOCS [23].Water content was measured after heating the sample in a 105 o C Eyela NDO 400 (made in Japan) oven for 24 hours (until the sample weight was constant).The water evaporated were calculated from the different weight of the sample before and after heating.Ash content was measured after combustion of the sample in a 500 o C muffle furnace for six h.The remaining residue was compared with the DM weight of the sample to calculate the ash percentage.The CP and EE were determined using Kjehdal and shoxlet system from Gerhart Instrument (made in Germany).While CF was measured using Ankom 200 (made in the USA) technology according to procedure Ba 6a-05 AOCS (2005).
3 2.3.Spectrum analysis using NIRS Spectrum analyses were conducted using pre-warmed, and system suitability tested NIRFlex N-500, modular FT-NIR Spectrometer Solids Cell from BUCHI (made in Switzerland).About 50 g samples were placed in a 100 mm diameter petri dish.The sample was distributed evenly to cover the dish surface thoroughly.The dish was put into a petri dish holder, and infrared light (800 -2500 nm wavelength) was sent into the sample.Each scanning was conducted three times.The adulteration analysis using NIRS spectrum tested 5 types of artificial adulterated rice bran, namely A = pure rice bran from miller, B = pure milled rice husk, C = 75% A + 25% B, D = 50% A + 50 B, and E = 25%A + 75% B.

Statistical analysis
This study used 2 x 2 factorial completely randomized design with 10 replications.The first factor was provinces of origin (P1 = West Java, P2 = West Sumatera).At the same time, the second factor was the rice bran provider (M1 = miller, M2 = poultry shop).Parameters observed include water content, ash, crude protein, crude fibre, ether extract and NIRS spectrum.Data analysis was conducted using ANOVA using the statistical software SPSS version 21.The spectrum produced by NIRS from the original and adulterated rice bran was compared using an excel plot.

Chemical composition of rice bran from the different provinces of origins and providers
The chemical composition of rice bran from different city and provider are shown in Table 1.The table shows that rice bran's nutrient content varies greatly.The moisture content of rice bran from West Java was significantly higher than that from West Sumatra.According to [24], moisture below 14% is safe for dairy cattle trading concentrate.Therefore, rice bran from both provinces of origin was still in the moisture content range save from trading.High moisture feed increases the risk of mould and fungal growth during storage, which might produce toxins and cause adverse effects on cows [25].The higher moisture content of rice bran from West Java Province compared to West Sumatera Province might be caused by the more humid area of West Java Province, especially Bogor regency from where the sample was collected.Annually, Bogor regency rainfall was 3,992.7 mm [26] was higher than the average West Sumatera Province annual rainfall (3709 mm in Padang city [27], 4285 mm in Solok regency, 2482 mm in Damasraya regency and 2467 -3611 mm in Agam regency ).
Ash contents in rice bran from both provinces of origin were not significantly different.In contrast, crude protein, crude fibre and ether extract were influenced by the interaction between the province of origin and provider.The differences were significantly visible for the rice bran from miller between west Java and West Sumatera.In contrast, the rice bran from the poultry shop was not significantly different among the origins.In general, rice bran from miller has better quality than poultry shops because it contains higher protein and extracts ether but lower fibre than poultry shops.The data suggested that rice bran in poultry shops might have been adulterated with low-quality fibrous feed but similar physical performance, such as milled rice husk.
The significant interaction between source and provider was shown in rice bran from miller.Ether extract and crude fibre contents in rice bran from the West Java miller were lower, while the crude protein was higher than in rice bran from the West Sumatera miller.The difference might be caused by the variation in rice variety, and the milling technique used [28].West Sumatera rice farmers proudly use local varieties such as Mundam, Kuniang Kulik, Simeru, Rahmat, Suntiang Ameh, Srikandi, Padi Putiah, Cintaku, Salih Berganti, Bakwan, Suntiang, Makwan, and Anak Daro which tastier according to the local preferences [29].On the contrary, farmers in West Java were more open to the national variety introduction, such as Inpari-1, Inpari-12, Logawa, Silugonggo, Dodokan, IR-66, and Tukad Unda.According to Kalpanadevi et al. [6], different milling techniques also produced different rice bran quality.They reported that the abrasive milling technique produced higher protein and crude fibre than the friction milling technique.

The NIRS spectrum of rice bran from the different provinces of origins and providers
The spectrum produced by the NIRS after penetrating rice bran samples of the different provinces of origin and providers with near-infrared light is shown in figure 1.The figures show that the spectrum of rice bran from millers (a and b) was more solid, while the poultry shop was more diverse (c and d).The spectrum reflected from West Java miller rice bran was more solid than West Sumatera.It might be caused by the more similar variety and milling technique used to produce the rice bran [28].Unlike the West Java miller rice bran, which was only collected from the Bogor regency, the West Sumatera miller rice bran was collected from several regencies such as Padang, Solok, Damasraya and Agam regencies.West Sumatra farmers' frequent use of local rice varieties [29] might also cause the spectrum variety.
After comparing the rice bran spectrum between miller and poultry shops, it can be seen that the spectrum from poultry shops was more diverse than from miller.In contrast to the more solid spectrum of rice bran from West Java Miller, the spectrum from the West Java poultry shop was more diverse than the West Sumatera poultry shop.The spectrum suggested that the risk of adulteration in West Java poultry was higher than in West Sumatera poultry shop.It might be caused by the higher demand for rice bran in west Java and less rice bran source, leading to higher prices during the off-season.The figure suggested that rice bran from the poultry shop might have been added with other materials different from the pure rice bran from the miller.The most common adulterated used was milled rice husk.NIRS has been used widely for adulteration detection.According to Cen and He [30], an adulteration by a concentration of 0.1% could be detected and evaluated by NIRS.It was reviewed that the beverage industry has used NIRS to detect the adulteration of beverages.In dairy farms, NIRS used to detect adulteration, such as addition of water, the addition of whey to milk, and fraudulent addition of melamine, urea and glucose, were also common [18].
Figure 1 suggested two different sides of the spectrum in which figures 1a and 1b were higher on specific wave numbers but lower in others.They are in line with the data in table 1, in which rice bran from millers was higher in ether extract and crude protein contents but lower in crude fibre than the rice bran from poultry shops.Asaduzzaman et al. [31] used spectra fingerprints to monitor raw milk quality.Based on PCA scores, the resulting spectral data were processed and reduced to focus on the most relevant wavelengths.The spectrums were used later to build a multivariate control chart.The approach has only three wavelengths to monitor milk quality and test its authenticity.In this study, chemical parameters such as moisture, ash, crude protein, crude fibre and ether extract have been used to detect the adulteration of rice bran.

The NIRS spectrum of rice bran with different degrees of artificial adulteration
The NIRS spectrum of rice bran with different degrees of artificial milled rice husk adulteration is shown in figure 2. The figure shows that the 100% rice bran spectrum produced reflectance lower at wavelength > 1400 nm but higher at wavelength < 1200 nm compared to the artificial adulterated rice bran.As reflectance has an inverse relation with substrate concentration [31], the spectrum suggested that the pure rice brans have a higher substrate concentration that overtone at the wavelength > 1400 nm and lower at the wavelength < 1200 nm.According to Stuth et al. [32], lipid or ether extract was measured by NIRS because of the characteristic aliphatic -CH absorptions seen at around 2310 and 1725 nm, with weaker overtone bands at 1400 and 1210 nm.Digestibility-related absorption in the region of 2270 nm was also reported to be associated with the crude fibre, which consisted of cellulose and lignin, and at around 1670 nm, which was unique to the lignin aromatic -CH first overtone band.Crude protein have been observed at spectral regions of 1,254-1,348 nm, 1,410-1,502 nm, 1,901-1,502 nm, 1,901-1,928 nm, 1,956-2,070 nm, 2,083-2,132 nm, 2,150-2,288 nm, and 2,340-2,500 nm) [33].

Original Spectra
All Spectra According to Stuth et al. [32], lipid or ether extract was measured by NIRS because of the characteristic aliphatic -CH absorptions seen at around 2310 and 1725 nm, with weaker overtone bands at 1400 and 1210 nm.In agreement with Stuth, Chen and Yang [33] reported that the spectral region of 1,738-1,744, 2,298-2,306, and 2,322-2,328 nm have been observed for lipids.Digestibility-related absorption in the region of 2270 nm was also reported to be associated with the crude fibre, which consisted of cellulose and lignin, and at around 1670 nm, which was unique to the lignin aromatic -CH first overtone band.Crude protein have been observed at spectral regions of 1,254-1,348 nm, 1,410-1,502 nm, 1,901-1,502 nm, 1,901-1,928 nm, 1,956-2,070 nm, 2,083-2,132 nm, 2,150-2,288 nm, and 2,340-2,500 nm) [33].By knowing the specific region of the nutrient composition in the spectral, we might detect the abnormality in the content due to adulteration.

Conclusions
It is concluded that rice bran quality varied among provinces of origin and providers.The rice bran from West Java contained more moisture than rice bran from West Sumatra.There were no significant differences in ash content between the provinces or providers.The crude protein and Ether extract content of rice bran from miller was higher than in poultry shops, but the crude fibre was lower than in poultry shops.NIRS spectrum differentiates rice bran from miller and poultry shop from its divers' spectrum.The artificial adulteration of rice bran with milled rice husk showed a different spectrum between pure rice bran and adulterated ones.It confirms the adulteration of rice bran in the poultry shop.

Figure 1 .
Spectrum of rice bran from different province and provider (a = West Java miller, b =West Sumatera miller, c = West Java Poultry Shop and d = West Sumatera Poultry Shop

Table 1
Chemical composition of rice bran from different origins and providers.