Pet Food Applications
Introduction
The pet food industry is constantly examining new methods for the analysis of nutritional and chemical composition of products. While analytical needs are similar to those of animal feed and forage, pet food analysis presents different challenges because of the complexity of the products. One parallel between pet food and traditional animal food is using feed formulation to design a well-balanced and complete diet with proper nutritional content. Pet food is specifically designed to keep the animal healthy if the food is the only food consumed by the animal. In the case of treats, they are designed to have high flavor content and appealing texture but are not meant to be the sole source of proper nutrition for the animal. Unlike animal feed and forage products, pet food can be manufactured as wet food or dry food. Both types offer different advantages for pets and owners. Wet food requires a range of 60% to 70% moisture content while dry food must contain a maximum of 10% moisture. Nutritional content and proper labeling are strictly regulated by the FDA. Typical nutrients include protein, carbohydrates, fats, vitamins, minerals, and water. Primary ingredients include meat, meat by-products, cereals, grains, and supplemental vitamins and minerals. Meat byproducts used in pet food include liver, kidney, and lungs as well as leftover poultry and fish parts.
The animal feed industry is constantly examining new methods for the analysis of nutritional and chemical composition of products. Traditional methods are often expensive and time-consuming. They are ill-suited to measure large numbers of samples and especially for determining variation within a batch of animal feed because most methods can only measure a small portion of sample. Other disadvantages include the inability of most techniques to perform real-time, on-line measurements during the manufacturing process and being unable to determine more than one parameter from a single measurement. In recent years, NIR spectroscopy has emerged as a technique that offers many advantages over traditional analytical methods. It is fast, non-invasive, and able to determine multiple parameters of interest from a single measurement. It requires no chemical or toxic solvents and can not only measure large amounts of samples but can measure a large sample amount in one reading. This is very important when accounting for variation within a batch, as animal feed and other natural products can often show large amounts of variation.
The ability of NIR spectroscopy for use in measuring parameters in animal feed has been proven for some time. However, recent advances in technology have helped facilitate its use as a complete process analytical tool from the analysis of raw materials in the field to a final quality control check. Handheld spectrometers can be used for field measurements and to check incoming batches of raw materials before entering the manufacturing process. On-line instruments can be used as real-time process control tools to optimize the manufacturing process, resulting in improved product quality, optimization of resources, and potentially significant cost savings. It can also be used to both detect and quantify adulterants in animal feed, which is a significant problem with potential health and economic consequences. An examination and review of applications using NIR spectroscopy in the pet food industry is presented here. Reviewed topics include the use of the technique to determine nutritive value, energy parameters, proper labeling of the finished product, and detection of adulteration in various animal feed products.
Analytes
- Total Starch
- Gelatinized Starch
- Neutral Detergent Fiber (NDF)
- Acid Detergent Fiber (ADF)
- Acid Detergent Lignin (ADL)
- Organic Matter (OM)
- Crude Fiber (CF)
- Crude Protein (CP)
- Ether Extract (EE)
- Nitrogen Free Extracts (NFE)
- Gross Energy (GE)
- Digestible Energy (DE)
- Apparent Digestibility Coefficients
- Digestible Nutrients
- Cellulose
- Hemicellulose
- Minerals
Scientific References and Statistics
At-line Prediction of Gelatinized Starch and Fiber Fractions in Extruded Dry Dog Food Using Different Near-Infrared Spectroscopy Technologies
Extruded dog food is the most common type of dry dog food. Extrusion uses heat and very high pressure to create an “air-popped” kibble. Upon entering the extruder, the temperature of the food mixture is between 200°F and 250°F and the high pressure creates steam with an intense burst of heat. The time inside the extruder is short and can range from fifteen seconds to two minutes. There is much debate between the nutritional value and digestibility of extruded foods vs. baked foods. Proponents of baking argue that baking makes the food more digestible and subjects it to less breaking down of the nutrients. However, the baking process is much slower and takes place at a much higher temperature (typically around 500°F). Regardless of the food type, cooking in some form is required to break down starches to get the food to the desired digestibility level. Starch is a non-fibrous carbohydrate that represents an important percentage of pet food composition. The degree of gelatinization from the cooking process is a very useful indicator of starch digestibility in animal diets. Fiber fractions are also important for determining animal health and proper nutrition. There is a need in the pet food industry for fast and cost-effective methods to determine starch and fiber parameters in extruded dry dog food. In this study, NIR spectroscopy was examined to determine the feasibility of measuring total starch, gelatinized starch, neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) in extruded dry dog food.
Two separate NIR spectrometers with different sampling methods were used in this study for comparative purposes: a NIR reflectance spectrometer and a NIR transmittance spectrometer. Eighty-one sealed commercial packages of extruded dog food for both puppies and adults of several breed sizes were procured for the study. The diversity of the products was considered representative of the available market products and there was high variability in the nutritional composition. Chemical composition of dry matter, crude protein, crude fat, crude fiber, and crude ash was provided by the manufacturer. Main carbohydrates sources in the different foods were oats, corn, potato, pea, rice, and sorghum. Main protein sources (many of the foods had multiple protein sources) were chicken, duck, fish, lamb, rabbit, horse, pork, and venison. All samples were ground before analysis. For reflectance measurements, samples were placed in a spinning cup and scanned from 400 nm to 2500 nm at 0.5 nm intervals. Thirty-two scans were collected per reading and averaged into one spectrum. For the transmittance instrument, samples were scanned from 850 nm to 1050 nm at 2 nm intervals. Sixteen scans were collected per reading and averaged into one spectrum. The same samples were scanned consecutively with both instruments. Standard methods were performed to determine reference values for the parameters of interest. Various pre-processing algorithms were applied to the NIR spectra before chemometric analysis. The processed NIR spectra and reference values were used to create Partial Least Squares (PLS) calibration models that correlate the spectra to the parameters of interest. Results for both the NIR reflectance and transmittance spectra are shown below.
| Reflectance: | ||
| Total Starch (g/100 g DM) | R2 = 0.97 | SEV = 1.16 |
| Gelantinized Starch (g/100 g DM) | R2 = 0.95 | SEV = 1.32 |
| NDF (g/100 g DM) | R2 = 0.93 | SEV = 1.52 |
| ADF (g/100 g DM) | R2 = 0.97 | SEV = 0.46 |
| ADL (g/100 g DM) | R2 = 0.93 | SEV = 0.22 |
| Transmittance: | ||
| Total Starch (g/100 g DM) | R2 = 0.84 | SEV = 2.77 |
| Gelantinized Starch (g/100 g DM) | R2 = 0.77 | SEV = 2.94 |
| NDF (g/100 g DM) | R2 = 0.80 | SEV = 2.90 |
| ADF (g/100 g DM) | R2 = 0.67 | SEV = 0.69 |
| ADL (g/100 g DM) | R2 = 0.85 | SEV = 0.31 |
All models were validated using cross validation, which removes data points from the models, creates new models, and gives predicted values for the parameters of interest. Results for the reflectance spectra models demonstrated that both the starch and fiber parameters can be accurately predicted using NIR spectroscopy. Results for the transmittance spectra demonstrated low accuracy and suggest that this instrument cannot adequately predict these parameters in extruded dog food. The sampling method used here was at-line and the results warrant confirming the feasibility of using this application as an on-line quality control method in dog food manufacturing. Real-time feedback that monitors the proportion of gelatinized starch to total starch during the manufacturing process could be incredibly beneficial in the dog food industry.
Energy Evaluation of Extruded Compound Foods for Dogs by Near-Infrared Spectroscopy
Determining the digestible or metabolizable energy content of dog food is crucial for good health and proper quality control. There are traditional methods for estimating energy content in dog food. These include using the Atwater method to estimate the metabolizable energy content from their proximate analysis and estimation of digestible energy content by the digestibility coefficient estimated from crude fiber or other indigestible carbohydrate fractions of foods. Obtaining the data required to implement these methods is traditionally done by time-consuming and expensive wet chemistry methods. NIR spectroscopy is a proven method for fast, non-invasive analysis for the determination of chemical composition and nutritive evaluation of animal feeds and mixed diets. In this study, the potential of NIR spectroscopy for the prediction of the chemical composition, apparent digestibility, and content of digestible nutrients and energy of commercial extruded compound dog food was examined.
Fifty-six commercial extruded compound foods of known chemical composition and in vivo apparent digestibility were procured for the study. Only fifty-one of these samples were used for gross energy digestibility and digestible energy content. Before scanning with a NIR spectrometer, all samples were air-dried and milled through a 1 mm sieve. Each sample was scanned twice in reflectance mode from 1100 nm to 2500 nm at 2 nm intervals. Various pre-treatment algorithms were applied to the spectral data before chemometric modeling. Partial Least Squares (PLS) chemometric models were created to correlate the NIR spectra to the parameters of interest. Models were developed for organic matter (OM), crude protein (CP), ether extract (EE), crude fibre (CF), nitrogen free extracts (NFE) and gross energy (GE) content, apparent digestibility coefficients (OMD, CPD, EED, NFED and GED), and digestible nutrient and energy content (DOM, DCP, DEE, DNFE and DE) of the commercial dog food samples. Apparent digestibility was determined using six adult Beagles ranging from two to six years of twelve to fifteen kg live weight. After a ten-day adaptation period of food being given at a fixed level of intake, food refusals and feces were collected from the beagles for ten days. Ash, CP, EE, and CF were determined using standard AOAC methods and gross energy was determined using an adiabatic calorimetric bomb. In the table below, statistics from the chemometric models for chemical composition and gross energy content, apparent digestibility, and digestible nutrients and energy content are listed.
| Chemical composition (g/kg DM) and gross energy content (MJ/kg DM) | |
| OM | R2 = 0.779 |
| CP | R2 = 0.995 |
| CF | R2 = 0.977 |
| EE | R2 = 0.927 |
| NFE | R2 = 0.940 |
| GE | R2 = 0.952 |
| Apparent digestibility (%) | |
| DMD | R2 = 0.967 |
| OMD | R2 = 0.945 |
| CPD | R2 = 0.903 |
| EED | R2 = 0.847 |
| NFED | R2 = 0.945 |
| GED | R2 = 0.942 |
| Digestible nutrients (g/kg DM) and energy content (MJ/kg DM) | |
| DOM | R2 = 0.960 |
| DCP | R2 = 0.980 |
| DEE | R2 = 0.942 |
| DNFE | R2 = 0.922 |
| DE | R2 = 0.961 |
The results for most parameters in the calibration models were excellent and proved the feasibility of using NIR spectroscopy for energy evaluation of extruded compound foods for dogs. The R2 correlation coefficients were above 0.9 for all chemical constituents and gross energy (GE) with the exception of organic matter (OM), which was below 0.8. The likely reasons for this are a small range of reference values and the fact that a portion of OM concentration contains minerals, which do not absorb NIR wavelengths. One reason for the high correlation is the broad ranges of values that covered most commercial compound extruded foods that are usually fed to healthy dogs. While the correlation coefficients were lower for apparent digestibility than the chemical constituents, most were still above 0.9. Results were also excellent for digestible nutrients and energy content. Overall, results obtained from this study for predicting energy value of compound extruded dog foods using NIR spectroscopy were comparable to those obtained using proximate chemical analysis and standard reference methods. It is likely that results could be improved by using more samples that fully cover the complete range of commercial dog foods available on the market.
Use of Near-Infrared Spectroscopy to Predict Energy Content of Commercial Dog Food
NIR spectroscopy is a well-established and proven method for feed evaluation. It offers the advantages of being fast, non-invasive, and does not use toxic chemicals or solvents that are often used for traditional reference methods. Little to no sample preparation is required and once chemometric models are created that correlate the NIR spectra to chemical and physical parameters of interest, multiple measurements can be made from a single light reading. Gross energy (GE) and digestible energy (DE) are crucial energy parameters in pet food and standard reference techniques to determine these are based on proximate analysis of pet food, requiring chemical analysis. The standard reference methods for these energy measurements also have limited accuracy in food with high crude fiber (CF) content. In this study, NIR spectroscopy was assessed to determine the feasibility of using it to predict energy content of commercial extruded dry dog food.
Seventy-one commercial extruded compound dog foods were procured for the study. In order to determine the apparent digestibility of energy from each sample, in vivo feeding trials were conducted according to a standard protocol suggested by the European Pet Food Industry Federation (FEDIAF). Measured portions of food were given to six healthy female beagles. Food refusals and feces samples were collected for five to seven days. Standard methods were used to determine measured GE, DE, and Gross Energy Digestibility (GED). Five samples with CF content over 8% were removed from the data as such samples are recommended to be removed from the equations used to determine the energy parameters.
Once the reference values were obtained, all samples were scanned using an NIR spectrometer. Before scanning, fresh samples were milled through a 1 mm sieve. All samples were scanned twice in reflectance mode from 1100 nm to 2500 nm at 2 nm intervals. Before chemometric modeling, various pre-processing algorithms were applied to the spectral data. Chemometric models were created using the Partial Least Squares (PLS) method that correlates the NIR spectra to GE, DE, and GED. Modeling results are shown below.
| GE, MJ/kg of Food on DM Basis | R2 = 0.95 | SEC = 0.25 |
| GE Digestibility, % | R2 = 0.89 | SEC = 1.62 |
| DE, MJ/kg of Food on DM Basis | R2 = 0.94 | SEC = 0.49 |
The results of this study were excellent and proved the feasibility of using NIR spectroscopy to predict energy content of commercial dog food. Correlation coefficients were well above 0.9 for GE and DE. In the case of GE Digestibility, the correlation coefficient value was slightly below 0.9. There are a number of possible reasons for the less precise digestibility estimation compared to other parameters, although results are still considered good enough for accurate predictions. The prediction of energy availability when estimating digestibility is more complex than simply estimating energy parameters. Digestibility estimation is complex and depends on feed characteristics and animal response to the feed. There was also a limited variability in the digestibility of the available commercial dog food samples. 84% of the samples had a digestibility range between 80% and 91%. A larger number of samples with a wider range of values should improve results. The advantages of using NIR spectroscopy over a traditional feeding trial to estimate energy parameters in dog food are vast. A feeding trial requires a minimum of one week of feces collection plus time for animal adaptation. Chemical analysis of the food is required as well. A comparison between using NIR spectroscopy and estimations obtained from NRC equations (the prediction method recommended by FEDIAF and the Association of American Feed Control Officials) showed that the NIR method was more accurate. The potential was demonstrated to implement NIR spectroscopy as a method to replace traditional time-consuming and expensive methods for the evaluation of energy parameters in commercial dog food.
Application of a Handheld Near-Infrared Spectrometer to Predict Gelatinized Starch, Fiber Fractions, and Mineral Content of Ground and Intact Extruded Dry Dog Food
Gelatinized starch, fiber fractions, and mineral content are critical quality control parameters in the pet food industry. Traditional methods for measuring these parameters are time-consuming and expensive. NIR spectroscopy has emerged as an alternative to traditional methods and offers many advantages. The method is fast, non-invasive, and requires little to no sample preparation. It can measure large amounts of samples in a reasonable amount of time and can determine multiple chemical and physical parameters of interest from a single light reading. Implementing NIR spectroscopy often requires a modification in the production process or a laboratory that requires sample transport. The use of handheld devices for screening can potentially expand the industrial applicability of this technology. In this study, the feasibility of predicting total starch, gelatinized starch, insoluble fibrous fractions, and the mineral content of both intact and ground extruded dry dog food using NIR spectroscopy was examined.
Ninety-nine commercial packages of extruded dog food were procured for the study. They were representative of the Italian pet food market in terms of composition and intended for consumption by both puppy and adult dogs of small, medium, and large breed sizes. A portion of each sample was ground and both an intact and ground portion of each sample was scanned using a handheld NIR spectrometer from 740 nm to 1070 nm at 1 nm intervals. Replicate spectra were collected a total of ten times for the intact samples and five times for the ground samples. After all the samples were scanned, the replicate spectra were averaged into a single spectrum for each sample. The purpose of scanning both types of samples was to determine if sample homogenization could improve the prediction performance of the instrument. Reference tests were conducted for the parameters of interest. Concentration of macrominerals (Ca, P, Mg, Na, K, and S) and trace minerals (Al, B, Ba, Cr, Cu, Fe, Li, Mn, Mo, Ni, Sr, V, and Zn) were determined. Neutral Detergent Fiber (NDF), Acid Detergent Fiber (ADF), Acid Detergent Lignin (ADL), Total Starch, and Gelatinized Starch were also determined. Before chemometric modeling, various pre-processing algorithms were applied to the spectral data. Partial Least Squares (PLS) calibration models were created that correlate the NIR spectra to each parameter of interest. Results are shown below.
| Ground Kibbles | |
| Total Starch | R2 = 0.91 |
| Gelatinized Starch | R2 = 0.91 |
| NDF | R2 = 0.91 |
| ADF | R2 = 0.57 |
| ADL | R2 = 0.76 |
| Cellulose | R2 = 0.56 |
| Hemicellulose | R2 = 0.73 |
| K | R2 = 0.70 |
| Intact Kibbles | |
| Total Starch | R2 = 0.89 |
| Gelatinized Starch | R2 = 0.89 |
| NDF | R2 = 0.78 |
| ADF | R2 = 0.69 |
| ADL | R2 = 0.78 |
| Cellulose | R2 = 0.59 |
| Hemicellulose | R2 = 0.83 |
| K | R2 = 0.80 |
Models were validated using cross-validation and the results were mixed for both types of samples. In the case of total and gelatinized starch, correlation coefficients were high enough and prediction error during cross-validation low enough to justify using this instrument and method for accurate quality control prediction. There was no large distinction between results for intact and ground samples for the starch parameters. In the case of fiber fractions, the results were poor and insufficient for any kind of quality control. The only mineral that showed reasonable results that could be used for screening purposes was K. Correlation was so low for the other minerals that results are not shown here. Minerals do not absorb light in the near-infrared range of the infrared spectrum because they are inorganic. In some cases, mineral concentration can be determined using NIR spectroscopy through an indirect correlation. However, such models must be carefully examined and validated. Past studies have proven the feasibility of NIR spectroscopy to accurately determine fiber fractions in different types of food products. Limited wavelength range and low light penetrations are likely the reasons for worse results using a handheld spectrometer. Despite the mixed results, the potential was demonstrated to using NIR spectroscopy and a handheld spectrometer for quality control analysis of extruded dry dog food.
Prediction of Mineral Composition in Commercial Extruded Dry Dog Food by Near-Infrared Reflectance Spectroscopy
Mineral content of dog food is essential for proper development and good continued health. Quality control analysis of minerals is time-consuming and expensive. Many companies are not equipped with the technology to perform mineral analysis. NIR spectroscopy is a fast, non-invasive method that is widely used in the food industry for the prediction of chemical and physical quality control parameters. It is a proven method for the analysis of organic compounds like fat, protein, and moisture. There are limited studies that analyze the potential of using NIR spectroscopy to predict mineral concentration. It is understood that minerals are inorganic compounds and that the NIR portion of the infrared spectrum does not absorb light from inorganic components. However, there has been some success using NIR spectroscopy to measure mineral concentration in products like cheese and processed meat. An indirect correlation is acceptable using NIR spectroscopy if the models are carefully examined and validated. In the case of minerals, a measurement is possible if the minerals are part of organic molecules or if they produce some alteration in the water absorbing region of the NIR spectrum. In this study, the feasibility of using NIR spectroscopy to determine mineral concentration in extruded dry dog food was examined.
One hundred nineteen sealed packages of extruded commercial dry dog food were procured for the study. Different compositions and the number of samples for each main protein source represent the availability of these products in the Italian pet food market. Varieties were included for puppy and adult dogs of small, medium, and large dog breed sizes. All samples were ground before scanning with a NIR spectrometer. Each sample was scanned from 850 nm to 2500 nm. Thirty-two scans were collected per reading and averaged into one spectrum. Reference tests were conducted for various major minerals and trace minerals. Major minerals included Ca, P, K, Na, S, and Mg. Minor minerals included Fe, Zn, Al, Mn, Cu, Sr, Ba, B, Cr, Ni, Mo, V, and Li. Before chemometric analysis, various pre-processing algorithms were performed on the spectral data. Partial Least Squares (PLS) were created correlating the NIR spectra to concentration of the different minerals. Results were poor for most minerals and the results that were considered good enough for screening purposes are shown below.
| K | R2 = 0.85 | RPD = 2.58 |
| S | R2 = 0.89 | RPD = 3.04 |
Models were validated using cross-validation and most showed poor predictions. In the case of Li, P, B, and Sr, the correlation coefficients were adequate but cross-validation results were poor. This indicates that the model correlation is not valid and that the models are likely correlating to something else besides the mineral concentration. RPD is a measurement of prediction accuracy. Generally, a RPD between two and three indicates that a model is considered good enough for screening purposes. K and S were the only models to show an RPD above two. Some of the minerals that showed poor results here have shown decent results in studies with other types of food products. The likely reason for this is the limited range of reference values in the samples. The study did reiterate the well-known fact that measuring mineral concentration using NIR spectroscopy is difficult and often impossible due to the lack of direct absorption of NIR light by minerals.
Prediction of Chemical Composition, Nutritive Value, and Ingredient Composition of European Compound Feed for Rabbits by Near Infrared Reflectance Spectroscopy
NIR spectroscopy is a widely used proven method for predicting the chemical composition and nutritive value of forage and raw materials used in compound animal feel. The process of formulating compound diets for both ruminant and non-ruminant animals can be complex. Traditional methods require equations based on chemical composition, digestible nutrients, or in vitro digestibility to determine digestible energy concentration. Determining the values for the physical and chemical parameters of interest that are needed to implement these methods often requires the use of wet chemistry tests that are expensive, time-consuming, and ill-suited to test large numbers of samples. NIR spectroscopy offers numerous advantages over traditional methods and some studies have shown that this method can be more accurate than using equations to predict nutritive value of compound diets. The study summarized here was conducted among six European laboratories for the purpose of comparing the efficacy of various techniques for predicting the nutritive value of compound diets for rabbits. Methods including NIR spectroscopy, in vitro methods, chemical constituents, and digestible nutrients. The aim was to reduce the utilization of animals for in vivo trials and to offer the industry a fast and cost-efficient evaluation tool. NIR spectroscopy is fast and non-invasive while requiring little to no sample preparation. It does not use toxic chemicals or solvents and can determine multiple chemical and physical parameters of interest from a single light measurement. In this study, the efficacy of NIR spectroscopy in predicting the chemical constituents, nutritive value, and ingredient composition of compound rabbit feeds was examined.
A common sample bank was created over the course of three years. The six participating laboratories collected experimental compound diets that were tested in vivo. In total, one hundred seventy-seven compound feeds for weaning, growing, and reproducing rabbits were formulated for the study. The formulations were specifically designed to create variation in different nutrient concentrations and the nutritive value of specific raw materials. Chemical analysis and in vivo digestibility trials were conducted according to standard methods. In total, reference values for fourteen different chemical, energy, and digestibility parameters were determined. A portion of each sample was ground, sealed in a plastic bottle, and transported to one of the laboratories for scanning with an NIR spectrometer. Samples were scanned from 1100 nm to 2500 nm at 2 nm intervals. Each sample was scanned on two separate days and the two spectra per sample were averaged into a single spectrum. Before chemometric analysis, various pre-processing algorithms were performed on the spectral data. Initial examination led to the removal of thirteen samples due to their out-of-range chemical composition. One hundred and eleven samples were used to create Partial Least Squares (PLS) regression models that correlate the NIR spectra to the parameters of interest. The remaining samples were used as an independent validation set. Results are shown below.
| Dry Matter (g/kg) | R2 = 0.79 | SEP = 4.8 |
| Organic Matter (g/kg DM) | R2 = 0.35 | SEP = 8.6 |
| Crude Protein (g/kg DM) | R2 = 0.84 | SEP = 5.6 |
| Ether Extract (g/kg DM) | R2 = 0.90 | SEP = 4.2 |
| Crude Fiber (g/kg DM) | R2 = 0.63 | SEP = 16 |
| NDF (g/kg DM) | R2 = 0.56 | SEP = 32 |
| ADF (g/kg DM) | R2 = 0.84 | SEP = 14 |
| ADL (g/kg DM) | R2 = 0.74 | SEP = 11 |
| Starch (g/kg DM) | R2 = 0.94 | SEP = 16 |
| Gross Energy (MJ/kg DM) | R2 = 0.61 | SEP = 0.25 |
| Digestible Energy (MJ/kg DM) | R2 = 0.80 | SEP = 0.39 |
| Dry Matter Digestibility | R2 = 0.84 | SEP = 0.019 |
| Crude Protein Digestibility | R2 = 0.71 | SEP = 0.026 |
| Gross Energy Digestibility | R2 = 0.82 | SEP = 0.019 |
Analysis of the results were mixed but most parameters showed good enough correlation and low enough prediction error to demonstrate the viability of NIR spectroscopy to predict them. Dry matter had a low correlation coefficient for a moisture measurement but prediction error was sufficiently low as well. Water is very absorbing of NIR light and the lower correlation could be due to a change in the moisture content of the samples. Organic matter contains minerals which do not absorb in the NIR region and unless an indirect correlation is obtained based on the effect of minerals on organic molecules, poor correlation can be expected. Crude protein and ether extract (a fat measurement) showed excellent results. In the case of crude fiber, NDF, ADF, and ADL, prediction error was high and this could be due to a number of factors, including reference error and a small range of reference values in the samples. Starch showed high correlation and low prediction error. Gross energy and digestible energy showed decent results compared to other studies. There is likely introduced reference error because animal response to feeding is involved in these parameters. The same could be said for the digestibility parameters. Overall, while the results were mixed for this study, it did show that NIR spectroscopy has the ability to be used as a screening tool for determination of chemical composition and nutritive value of compound feed for rabbits.