ABSTRACT Fish meal and meat & bone meal from avian sources are the available animal protein sources mostly used in Bangladesh while formulating poultry diet. Due to the strong link between feed and food, the quality and safety of animal originated feed protein materials are the crucial public concern. The aim of this study was to explore the nutritional evaluation of fish meal and meat & bone meal by using near-infrared reflectance spectroscopy (NIRS). Samples were collected from different locations and feed mills in Bangladesh and were scanned in duplicate (scanning number 32, resolution 8) with an FT-NIRS systems monochromator (700–2400 nm) using a Qurtz cup sampling device. Multivariate analyses were performed for the development of calibration equations and data were centred using Partial Least Squares algorithm where spectral outliers were identified from each calibration. The accuracy of the calibration models was validated by root mean square error cross-validation (RMSECV), and correlation coefficient (R 2) between the measured values determined by analytical laboratory vs. predicted values of NIRS. High correlation between laboratory values and predicted values (R 2 > 90.00) were observed in predicting moisture, CP and total ash contents in both of the feed protein meals although the correlation (R 2) is relatively lower in predicting EE (88.45) and P (75.41) contents in fish meal, and Ca (85.84) and P (82.23) contents in meat & bone meal. The highest RMSECV was observed in CP and total ash contents of fish meal (1.15% and 1.01%) and meat and bone meal (1.27% and 1.32%), but the RMSECV in predicting other nutrients of feed protein meals were in acceptable ranges. Therefore, it is revealed that the NIRS could potentially be used to predict the nutrient contents in fish meal and meat & bone meal in Bangladesh.

中文翻译：

---

使用近红外反射光谱开发鱼粉和肉骨粉营养评价的局部校准

摘要 来自禽类的鱼粉和肉骨粉是孟加拉国主要用于制定家禽日粮的可用动物蛋白来源。由于饲料和食品之间的紧密联系，动物源性饲料蛋白材料的质量和安全是公众最关心的问题。本研究的目的是利用近红外反射光谱（NIRS）探索鱼粉和肉骨粉的营养评价。从孟加拉国的不同地点和饲料厂收集样品，并使用 Qurtz 杯采样装置使用 FT-NIRS 系统单色器（700-2400 nm）一式两份扫描（扫描次数 32，分辨率 8）。对校准方程的开发进行了多变量分析，并使用偏最小二乘算法对数据进行了集中，其中从每次校准中识别出光谱异常值。校准模型的准确性通过均方根误差交叉验证 (RMSECV) 和分析实验室确定的测量值与 NIRS 预测值之间的相关系数 (R 2) 进行验证。尽管在预测 EE (88.45) 和 P 时相关性 (R 2) 相对较低，但在预测两种饲料蛋白粉中的水分、CP 和总灰分含量时，观察到实验室值和预测值之间的高度相关性 (R 2 > 90.00)鱼粉中的 (75.41) 含量，以及肉骨粉中 Ca (85.84) 和 P (82.23) 的含量。RMSECV 在鱼粉（1.15% 和 1.01%）和肉骨粉（1.27% 和 1.32%）的 CP 和总灰分含量中观察到最高，但 RMSECV 用于预测饲料蛋白粉的其他营养成分在可接受的范围内。因此，研究表明近红外光谱技术可用于预测孟加拉国鱼粉和肉骨粉中的营养成分。