Coefficient Of Determination In Validation Research Articles

Use of Near Infrared Reflectance Spectroscopy (NIRS) for the rapid compositional analysis of di-, tri-, and tetraploid eastern oysters (Crassostrea virginica)

The oyster aquaculture industry in the USA, especially the Chesapeake Bay, has shown rapid growth over the last decade. Breeding has played a large role in this growth in the Chesapeake Bay, with emphasis on the selection for disease resistance, growth and the development of triploid Crassostrea virginica. With major gains realized in both of these traits, we are attempting to add more, such as, compositional traits. A specific interest for the Aquaculture Genetics and Breeding Technology Center is to derive compositional data for use with our selective breeding program. Near Infrared Reflectance Spectroscopy (NIRS) estimates of moisture and glycogen may lead to the definition of these new traits for selection to accompany those traditional measures of survival, meat yield, and growth rate. This study describes the development of NIRS models to estimate moisture and glycogen in the eastern oyster C. virginica.NIR spectra of 138 homogenized oyster meats were modeled against physicochemically measured moisture and glycogen data. The robustness of resulting models was assessed by an independent validation that showed a high coefficient of determination for validation (R2val=0.95) and high residual predictive deviation (RPD≥4.1) for both parameters, making them suitable for quantification. In a second step, these models have also been validated to obtain compositional data on different levels of ploidy in C. virginica (diploid, triploid, and tetraploid). These parameters will be applied to our family based selection program where heritabilities will be calculated. Statement of relevanceFor the first time we have developed NIRS models to determine compositional analysis in C. virginica. NIRS estimates of moisture and glycogen may lead to the definition of new traits for selection to accompany other traditional measures (survival, meat yield, and growth rate). NIRS measurements will be applied to our family based selection program where heritabilities will be calculated.

A Wavelet-Based Area Parameter for Indirectly Estimating Copper Concentration in Carex Leaves from Canopy Reflectance

Due to the absence of evident absorption features and low concentrations, the copper (Cu) concentration in plant leaves has rarely been estimated from hyperspectral remote sensing data. The capability of remotely-sensed estimation of foliar Cu concentrations largely depends on its close relation to foliar chlorophyll concentration. To enhance the subtle spectral changes related to chlorophyll concentration under Cu stress, this study described a wavelet-based area parameter (SWT (605−720), the sum of reconstructed detail reflectance at fourth decomposition level over 605−720 nm using discrete wavelet transform) from the canopy hyperspectral reflectance (350−2500 nm, N = 71) of Carex (C. cinerascens). The results showed that Cu concentrations had negative and strong correlation with chlorophyll concentrations (r = －0.719, p < 0.001). Based on 1000 random dataset partitioning experiments, the 1000 linear calibration models provided a mean R2Val (determination coefficient of validation) value of 0.706 and an RPD (residual prediction deviation) value of 1.75 for Cu estimation. The bootstrapping and ANOVA test results showed that SWT (605−720) significantly (p < 0.05) outperformed published chlorophyll-related and wavelet-based spectral parameters. It was concluded here that the wavelet-based area parameter (i.e., SWT (605−720)) has potential ability to indirectly estimate Cu concentrations in Carex leaves through the strong correlation between Cu and chlorophyll. The method presented in this pilot study may be used to estimate the concentrations of other heavy metals. However, further research is needed to test its transferability and robustness for estimating Cu concentrations on other plant species in different biological and environmental conditions.

Quantification of whey proteins by reversed phase-HPLC and effectiveness of mid-infrared spectroscopy for their rapid prediction in sweet whey

In the dairy industry, membrane filtration is used to reduce the amount of whey waste and, simultaneously, to recover whey proteins (WP). The composition of WP can strongly affect the filtration treatment of whey, and rapid determination of WP fractions would be of interest for dairy producers to monitor WP recovery. This study aimed to develop mid-infrared spectroscopy (MIRS) prediction models for the rapid quantification of protein in sweet whey, using a validated rapid reversed phase (RP)-HPLC as a reference method. Quantified WP included α-lactalbumin (α-LA), β-lactoglobulin (β-LG) A and B, bovine serum albumin, caseinomacropeptides, and proteose peptone. Validation of RP-HPLC was performed by calculating the relative standard deviation (RSD) in repeatability and reproducibility tests for WP retention time and peak areas. Samples of liquid whey (n=187) were analyzed by RP-HPLC and scanned through MIRS to collect spectral information (900 to 4,000cm−1); statistical analysis was carried out through partial least squares regression and random cross-validation procedure. Retention times in RP-HPLC method were stable (RSD between 0.03 and 0.80%), whereas the RSD of peak area (from 0.25 to 8.48%) was affected by WP relative abundance. Higher coefficients of determination in validation for MIRS model were obtained for protein fractions present in whey in large amounts, such as β-LG (0.58), total identified WP (0.58), and α-LA (0.56). Results of this study suggest that MIRS is an easy method for rapid quantification of detail protein in sweet whey, even if better resolution was achieved with the method based on RP-HPLC. The prediction of WP in sweet whey by MIRS might be used for screening and for classifying sweet whey according to its total and individual WP contents.

Design and Fabrication of a Real-Time Measurement System for the Capsaicinoid Content of Korean Red Pepper (Capsicum annuum L.) Powder by Visible and Near-Infrared Spectroscopy.

This research aims to design and fabricate a system to measure the capsaicinoid content of red pepper powder in a non-destructive and rapid method using visible and near infrared spectroscopy (VNIR). The developed system scans a well-leveled powder surface continuously to minimize the influence of the placenta distribution, thus acquiring stable and representative reflectance spectra. The system incorporates flat belts driven by a sample input hopper and stepping motor, a powder surface leveler, charge-coupled device (CCD) image sensor-embedded VNIR spectrometer, fiber optic probe, and tungsten halogen lamp, and an automated reference measuring unit with a reference panel to measure the standard spectrum. The operation program includes device interface, standard reflectivity measurement, and a graphical user interface to measure the capsaicinoid content. A partial least square regression (PLSR) model was developed to predict the capsaicinoid content; 44 red pepper powder samples whose measured capsaicinoid content ranged 13.45–159.48 mg/100 g by per high-performance liquid chromatography (HPLC) and 1242 VNIR absorbance spectra acquired by the pungency measurement system were used. The determination coefficient of validation (RV2) and standard error of prediction (SEP) for the model with the first-order derivative pretreatment method for Korean red pepper powder were 0.8484 and ±13.6388 mg/100 g, respectively.

Prediction of naturally-occurring, industrially-induced and total trans fatty acids in butter, dairy spreads and Cheddar cheese using vibrational spectroscopy and multivariate data analysis

This study investigated the use of vibrational spectroscopy [near infrared (NIR), Fourier-transform mid-infrared (FT-MIR), Raman] and multivariate data analysis for (1) quantifying total trans fatty acids (TT), and (2) separately predicting naturally-occurring (NT; i.e., C16:1 t9; C18:1 trans-n, n = 6 … 9, 10, 11; C18:2 trans) and industrially-induced trans fatty acids (IT = TT – NT) in Irish dairy products, i.e., butter (n = 60), Cheddar cheese (n = 44), and dairy spreads (n = 54). Partial least squares regression models for predicting NT, IT and TT in each type of dairy product were developed using FT-MIR, NIR and Raman spectral data. Models based on NIR, FT-MIR and Raman spectra were used for the prediction of NT and TT content in butter; best prediction performance achieved a coefficient of determination in validation (R2V) ∼ 0.91–0.95, root mean square error of prediction (RMSEP) ∼ 0.07–0.30 for NT; R2V ∼ 0.92–0.95, RMSEP ∼ 0.23–0.29 for TT.

Development of Models for the Prediction of Domestic Red Pepper (Capsicum annuum L.) Powder Capsaicinoid Content using Visible and Near-infrared Spectroscopy

Purpose: The purpose of this study was to non-destructively and quickly predict the capsaicinoid content of domestic red pepper powders from various areas of Korea using a pungency measurement system in combination with visible and near-infrared (VNIR) spectroscopic techniques. Methods: The reflectance spectra of 149 red pepper powder samples from14 areas of Korea were obtained in the wavelength range of 450-950 nm and partial least squares regression (PLSR) models for the prediction of capsaicinoid content were developed using area models. Results: The determination coefficient of validation (RV2), standard error of prediction (SEP), and residual prediction deviation (RPD) for the capsaicinoid content prediction model for the Namyoungyang area were 0.985, ±2.17 mg/100g, and 7.94, respectively. Conclusions: These results show the possibility of VNIR spectroscopy combined with PLSR models in the non-destructive and facile prediction of capsaicinoid content of red pepper powders from Korea.

Prediction of ozone concentration in tropospheric levels using artificial neural networks and support vector machine at Rio de Janeiro, Brazil

It is well known that air quality is a complex function of emissions, meteorology and topography, and statistical tools provide a sound framework for relating these variables. The observed data were contents of nitrogen dioxide (NO2), nitrogen monoxide (NO), nitrogen oxides (NOx), carbon monoxide (CO), ozone (O3), scalar wind speed (SWS), global solar radiation (GSR), temperature (TEM), moisture content in the air (HUM), collected by a mobile automatic monitoring station at Rio de Janeiro City in two places of the metropolitan area during 2011 and 2012. The aims of this study were: (1) to analyze the behavior of the variables, using the method of PCA for exploratory data analysis; (2) to propose forecasts of O3 levels from primary pollutants and meteorological factors, using nonlinear regression methods like ANN and SVM, from primary pollutants and meteorological factors. The PCA technique showed that for first dataset, variables NO, NOx and SWS have a greater impact on the concentration of O3 and the other data set had the TEM and GSR as the most influential variables. The obtained results from the nonlinear regression techniques ANN and SVM were remarkably closely and acceptable to one dataset presenting coefficient of determination for validation respectively 0.9122 and 0.9152, and root mean square error of 7.66 and 7.85, respectively. For these datasets, the PCA, SVM and ANN had demonstrated their robustness as useful tools for evaluation, and forecast scenarios for air quality.

Direct determination of minerals in human diets by infrared spectroscopy and X-ray fluorescence

The use of near and mid infrared spectroscopy and X-ray fluorescence (XRF) to determine the concentration of mineral elements in Spanish human diets was investigated. Thirty-five commercial baby foods, 6 children fast food menus and 13 university canteen menu samples were analysed by infrared and XRF spectroscopy and spectra evaluated by using reference data obtained by inductively coupled plasma optical emission spectroscopy (ICP-OES). Models for calcium, potassium, iron, magnesium, sodium and zinc determination were built and validated. Spectra were pre-treated by using different pre-processing algorithms (multiplicative scatter correction, standard normal variation, first derivate, orthogonal signal correction, smoothing and mean centre) prior to developing calibration models using partial least squares and were evaluated by cross-validation and external validation. The highest coefficients of determination in validation (R2val) and the lowest relative root-mean-square error of prediction were obtained for potassium determination by MIR (0.86 and 11%), NIR (0.9 and 11%) and XRF (0.9 and 10%).

Rapid determination of soil organic matter quality indicators using visible near infrared reflectance spectroscopy

Assessment and monitoring of soil organic matter (SOM) quality are important for determining and developing management practices that will enhance and maintain the productivity of agricultural soils. This requires routine analysis of multiple soil parameters, which can be time-consuming and expensive. Research has suggested that visible near infrared reflectance spectroscopy (VNIRS) may be used as a rapid and cost-efficient tool for SOM quality assessment. In this study, VNIRS (400–2498nm) was used for the first time to simultaneously predict microbial biomass nitrogen (MBN), water-extractable organic N (WEON), light fraction organic matter N (LFOMN), particulate organic matter N (POMN), soil total N (TN), soil organic carbon (SOC) and soil C/N ratio as soil SOM quality indicators in Chernozemic soils of western Canada. The soil samples (n=200) were collected at the 0–15cm depth from a crop rotation experiment conducted at 6 sites in 2010 and 2011. After removal of outliers (five samples) identified by principal components analysis (PCA), 75% of the sample set was randomly selected for calibration (n=146) and the remainder used for validation (n=49). Modified partial least squares regression with cross-validation was used to develop prediction models. The reliability of the models was assessed using the coefficient of determination in validation (R2V) and the ratio of standard deviation of the reference data in the validation set to the standard error of prediction (RPDV). The VNIRS predictions were considered reliable for LFOMN, POMN, TN, and SOC (R2V>0.80, RPDV>2.4), as well as for MBN (R2V=0.74, RPDV=1.93), but less reliable for WEON (R2V=0.67, RPDV=1.70) and soil C/N ratio (R2V=0.54, RPDV=1.45). This study showed that VNIRS has the potential as a non-destructive and cost-efficient tool for rapid determination of SOM quality indicators.

Hyperspectral measurements of severity of stripe rust on individual wheat leaves

The objective of this study was to assess the effect of severity of stripe rust (Puccinia striiformis) on the hyperspectral reflectance of wheat. A total of 110 leaf samples with a range of disease severity were collected at the heading stage (Stage І, 29 April) and grain filling stage (Stage II, 21 May). The spectra of the adaxial and abaxial surfaces of the leaf samples were taken using an ASD Leaf Clip, and the spectral characteristics were analysed. The photochemical reflectance index (PRI) was used to build two linear regression functions from the two growth stages using 70 leaves, and the remaining 40 leaves were used to validate their effectiveness. The results indicated that P. striiformis caused changes in foliar water and chlorophyll, and those changes made it feasible to assess disease severity using in situ hyperspectral measurements. In general, the reflectance values from the adaxial surfaces of the leaf samples were smaller than the abaxial surfaces. In comparison to Stage І, the spectral contrast of four different disease severities was greater at Stage II. By comparing the regression functions, the coefficient of determination using the set of leaves for validation for Stage І (R 2 = 0.74) was smaller than that for Stage II (R 2 = 0.83). However, the coefficient of determination for validation for Stage І (R 2 = 0.91) was slightly larger than that of Stage II (R 2 = 0.90). The results suggest that the ASD Leaf Clip is an ideal tool to collect in situ hyperspectral measurements of wheat leaves showing symptoms of stripe rust, and Stage II is more appropriate to assess severity compared to Stage І.

The use of near infrared spectroscopy (NIRS) to predict the chemical composition of feed samples used in ostrich total mixed rations

The wet chemical analysis of feed samples is time consuming and expensive. Near infrared spectroscopy (NIRS) was developed as a rapid technique to predict the chemical composition of feeds. The prediction of accuracy of NIRS relies heavily on obtaining a calibration set which represents the variation in the main population, accurate laboratory analyses and the application of the best mathematical procedures. In this study NIRS was used to determine the chemical composition of total mixed rations (TMRs) used in ostrich diets. A sample population of 479 ostrich feed samples was used in the calibration and 94 samples were used in the independent validation of dry matter (DM), ash, crude protein (CP), ether extract (EE), crude fibre (CF), acid detergent fibre (ADF), neutral detergent fibre (NDF), gross energy (GE), calcium (Ca) and phosphorus (P). Coefficient of determination in validation (r 2 v) and standard error of prediction (SEP) was satisfactory (r 2 v values higher than 0.80). Coefficient of determination and SEP values for CP, EE, CF, ADF, NDF and GE were 0.97% and 0.74%, 0.89% and 0.50%, 0.94% and 1.41%, 0.89% and 2.67%, 0.95% and 2.81% and 0.80% and 0.28 MJ/kg, respectively. Less accurate values (r2v below 0.80) were obtained for DM, ash, Ca and P being 0.57% and 0.28%, 0.67% and 1.29%, 0.43% and 0.59% and 0.49% and 0.11%, respectively. The study indicated that NIRS is a suitable tool for a rapid, non-destructive and reliable prediction of the chemical composition of ostrich TMRs. Keywords : NIRS, ostrich TMR, chemical composition, nutritive value

Rapid predictive models for minimally destructive Kappa number and pulp yield of Acacia spp. with near infrared reflectance (NIR) spectroscopy

Kraft pulp and wood powder rom Acacia Spp. were selected for the development of rapid, minimally-destructive, and predictions of kappa number and pulp yield, by means of near infrared reflectance (NIR) spectra. The models, based on Partial Least Squares Regression (PLS-R), were established with fifty-four calibration samples selected by Principle Component Analysis (PCA), while the validation models resulted from nineteen samples that were not included in the calibration set. The and stability of calibration models were evaluated by coefficient of determination for calibration (R2cal) and root mean square error of cross-validation (RMSECV). The coefficient of determination for validation (R2val) and root mean square error of prediction (RMSEP) were used for validation models. The main results showed that: (1) the predictive models from pulp were more credible in terms of the R2cal and R2val values than those from wood powder by 25 to 70%; and (2) a validation model for kappa number from pulp showed a better stability than the corresponding calibration model, since RMSEP was 23.5% less than RMSECV, while calibration models for pulp yield were more steady than validation models. This study provided reliable models for predicting kappa number and pulp yield rapidly and with a minimal need for physical sampling.

Near-Infrared Reflectance Spectroscopy (NIRS) for Protein, Tryptophan, and Lysine Evaluation in Quality Protein Maize (QPM) Breeding Programs

Quality protein maize (QPM) has approximately twice the tryptophan (Trp) and lysine (Lys) concentrations in protein compared to normal maize. Because several genetic systems control the protein quality of QPM, it is essential to regularly monitor Trp and/or Lys in breeding programs. Our objective was to examine the potential of near-infrared reflectance spectroscopy (NIRS) to enhance the efficiency of QPM research efforts by partially replacing more expensive and time-consuming wet chemistry analysis. More than 276 maize samples were used to develop NIRS models for protein content (PC), Trp, and Lys. The standard error of prediction (SEP) for the calibration and the coefficient of determination for validation (R(2)(v)) were 0.26 and 0.96 for PC, 0.005 and 0.85 for Trp, and 0.02 and 0.75 for Lys. When the NIRS models were used to evaluate 266 S2 lines from five QPM breeding populations, the coefficients of determination between NIRS and the chemical data were 0.94, 0.76, and 0.80 for PC, Trp, and Lys, respectively. Therefore, the NIRS models can be used to support the QPM breeding efforts.

Prediction of fat quality in pig carcasses by near-infrared spectroscopy

This study was conducted to evaluate the potential of near-infrared (NIR) spectroscopy (NIRS) technology for prediction of the chemical composition (moisture content and fatty acid composition) of fat from fast-growing, lean slaughter pig samples coming from breeding programmes. NIRS method I: a total of 77 samples of intact subcutaneous fat from pigs were analysed with the FOSS FoodScan NIR spectrophotometer (850 to 1050 nm) and then used to predict the moisture content by using partial least squares (PLS) regression methods. The best equation obtained has a coefficient of determination for cross-validation (CV; R2cv) and a root mean square error of a CV (RMSECV) of 0.88 and 1.18%, respectively. The equation was further validated with (n = 15) providing values of 0.83 and 0.42% for the coefficient of determination for validation (R2val) and root mean square error of prediction (RMSEP), respectively. NIRS method II: in this case, samples of melted subcutaneous fat were analysed in an FOSS XDS NIR rapid content analyser (400 to 2500 nm). Equations based on modified PLS regression methods showed that NIRS technology could predict the fatty acid groups, the main fatty acids and the iodine value accurately with R2cv, RMSECV, R2val and RMSEP of 0.98, 0.38%, 0.95 and 0.49%, respectively (saturated fatty acids), 0.94, 0.45%, 0.97 and 0.65%, respectively (monounsaturated fatty acids), 0.97, 0.28%, 0.99 and 0.34%, respectively (polyunsaturated fatty acids), 0.76, 0.61%, 0.84 and 0.87%, respectively (palmitic acid, C16:0), 0.75, 0.16%, 0.89 and 0.10%, respectively (palmitoleic acid, C16:1n-7), 0.93, 0.41%, 0.96 and 0.64%, respectively (steric acid, C18:0), 0.90, 0.51%, 0.94 and 0.44%, respectively (oleic acid, C18:1n-9), 0.97, 0.25%, 0.98 and 0.29% (linoleic acid, C18:2n-6), 0.68, 0.09%, 0.57 and 0.16% (α-linolenic acid, C18:3n-3) and 0.97, 0.57, 0.97 and 1.22, respectively (iodine value, calculated). The magnitude of this error showed quite good accuracy using these rapid methods in prediction of the moisture and fatty acid composition of fat from pigs involved in breeding schemes.

The Rapid Estimation of Cellulose, Hemicellulose, and Lignin Contents in Rice Straw by Near Infrared Spectroscopy

Rice straw can be used for biochemical conversion for energy production. The use of near infrared spectroscopy was explored to estimate the contents of cellulose, hemicellulose, and lignin in this study. A total of 172 samples from 17 provinces of China were collected and scanned using NIRSystem 6500 spectrophotometer. Different spectra pretreatments and regression methods were adopted to optimize the models. When evaluated on independent validations, the standard errors of validation and determination coefficient of validation were 1.49%, 1.59%, 0.65% and 0.82, 0.71, 0.78, respectively. It was concluded that near infrared spectroscopy can be used to roughly analyze the contents of cellulose and lignin, and can only distinguish high contents and low contents of hemicellulose in rice straw.

Near‐infrared spectroscopy on combine harvesters to measure maize grain dry matter content and quality parameters

AbstractDry matter (DM) content and quality parameters of maize grain are important traits in breeding of maize hybrids. New developments in near‐infrared spectroscopy (NIRS) allow to determine DM content and product quality parameters directly on plot combine harvesters. The main objective of our study was to examine the potential of NIRS on combine harvester (NOCH) for determination of DM, crude protein (CP) and starch (ST) contents in maize grain. Plot combine harvesters equipped with spectrometers were used. Eleven locations were harvested in Europe during 2003 and 2004. The NOCH calibration models were developed with standard error of prediction (SEP) and coefficient of determination for validation (R) of 1.2% and 0.95 for DM, 0.3% and 0.88 for CP, and 1.0% and 0.79 for ST, respectively. NOCH is a promising technology for determination of DM, CP and ST contents of maize grain for breeding purposes, because it automates the assessment of each harvested plot, and the information for making selection decision is rapidly available to plan the next generation.

Quality Determination of Chinese Rice Wine Based on Fourier Transform near Infrared Spectroscopy

To evaluate the applicability of near infrared (NIR) spectroscopy for the determination of five enological parameters (alcoholic degree, pH value, total acid, amino acid nitrogen and °Brix) in Chinese rice wine, transmission spectra were collected in the spectral range from 800 nm to 2500 nm in a 1 mm path length rectangular quartz cuvette with air as the reference at room temperature. Five calibration equations for the enological parameters were established between the reference data and NIR spectra by partial least squares (PLS) regression, separately. The best calibration results were achieved for the determination of alcoholic degree and °Brix. The coefficients of determination in calibration ( R2cal) for alcoholic degree and °Brix were 0.93 and 0.96, respectively. The predictive deviation ratio ( RPD) value of the calibration for alcoholic degree was higher than 3 (3.95), which demonstrated the robustness of the calibration model. The RPD value for °Brix was 2.34. The performance of the calibration models for pH, total acid and amino acid nitrogen were not as good as that of alcoholic degree and °Brix. The RPD values for the three parameters were 1.41, 1.38 and 1.27, respectively and R2cal values were 0.97, 0.83 and 0.90, respectively. In the validation step, r2val values for alcoholic degree and °Brix were all higher than 0.9. The performance of pH and total acid were acceptable [the coefficients of determination in validation ( r2val) for pH and total acid were 0.82 and 0.77, respectively]. The performance of amino acid nitrogen model was the worst, with an r2val value of 0.62. The results demonstrated that the NIR spectroscopic technique could be used to predict the concentrations of these five enological parameters in Chinese rice wine.

Prediction of Colour and pH in Grapes Using a Diode Array Spectrophotometer (400–1100 nm)

One hundred and twenty thawed samples of homogenised and whole frozen grape berries were analysed using a diode array spectrophotometer (400–1100 nm) (CORONA 45VISNIR, Carl Zeiss, Germany). The spectra and the analytical data were used to develop partial least squares calibrations to predict colour and pH in both presentation modes to the instrument. The visible (vis: 400–700 nm), near infrared (NIR: 700–1100 nm) and vis + NIR (400–1100 nm) regions were used to perform the calibrations. Cross validation models for colour and pH on homogenised samples gave a coefficient of determination in validation ( R2val) and the root mean square error of cross-validation ( RMSECV) of 0.92 and 0.07 mg g−1 for colour, and 0.90 and 0.04 for pH, respectively, using the vis region. Presentation of intact whole grape berries gave R2val and RMSECV values of 0.50 and 0.14 mg g−1 for colour, 0.60 and 0.08 for pH using the NIR region. It was concluded that homogenised samples gave the best calibration statistics. More research needs to be done to improve calibration on whole samples if the technology is to be used for rapid analysis for either on-farm or on-harvester applications.

Use of Near-Infrared Reflectance Spectroscopy to Analyze Vitamin Content

Near-infrared reflectance spectroscopy (NIRS) has been used to determine the composition of many products used in animal nutrition. This study sought to determine how best to apply NIRS to analyze the concentration of 11 vitamins in a standard premix. Two methods of regression, stepwise and modified partial least-squares, were used to analyze the results. Stepwise regression showed a confounding effect among various wavelengths and was problematic for use at extreme wavelengths. Modified partial least-squares regression gave more accurate results, as indicated by values of the coefficient of determination in calibration (1.0 for all vitamins) and the coefficient of determination in validation (0.96 to 1.0). Application of NIRS in conjunction with modified partial least-squares regression shows promise as a quick and inexpensive procedure for analysis of vitamin content.

Determination of the fatty acid composition of the oil in intact‐seed mustard by near‐infrared reflectance spectroscopy

AbstractNear‐infrared reflectance spectroscopy (NIRS) was used to estimate the fatty acid composition of the oil in intact‐seed samples of Ethiopian mustard (Brassica carinata Braun) within a mutation breeding program that produced seeds with variable fatty acid compositions. Five populations, from 1992 to 1996 crops, were included in this study; and NIRS calibration equations for major fatty acids (palmitic, stearic, oleic, linoleic, linolenic, eicosenoic, and erucic) were developed within each single population. Furthermore, global calibration equations, including samples from the five populations, were developed. After external validation, the NIRS technique permitted us to obtain a reliable and accurate nondestructive estimation of the fatty acid composition of the oil, especially for the major acids—oleic, linoleic, linolenic, and erucic. For these, the r2 in external validation was higher than 0.95 by using both single‐and multipopulation equations, and higher than 0.85 for the remaining fatty acids. Moreover, the multipopulation equations provided an accurate estimation of samples from a population not represented in the calibration data set, with values of coefficient of determination in validation (r2) from 0.80 (palmitic and eicosenoic acids) to 0.97 (erucic acid). The ability of NIRS to discriminate among different fatty acid profiles was mainly due to changes within six spectral regions, 1140–1240, 1350–1400, 1650–1800, 1880–1920, 2140–2200, and 2240–2380 nm, all of them associated with fatty acid absorbers. Thus, NIRS can be used to estimate the fatty acid composition of Ethiopian mustard seeds with a high degree of accuracy, provided that calibration equations be developed from calibration sets that include large variability for the fatty acid composition of the oil.