Faecal near-infrared reflectance spectroscopy profiling for the prediction of dietary nutritional characteristics for equines

Abstract

Faecal near infrared reflectance spectroscopy (NIRS) profiling involves the use of NIRS for the prediction of nutrient components of animal diets from their faecal samples. This study evaluated the potential use of faecal NIRS profiling for the prediction of minerals, fibre fractions, and organic matter (OM) in horses’ diet. Two feeding trials were conducted consisting of 29 horses housed in individual stables and fed unique diets (D) to obtain diet reference data and faecal (F) NIR spectra (D:F) pairs. The D:F pairs were used for the development of faecal NIRS predictive equations. The trials ran for a total of five weeks (7 days adaptation and 4 weeks of measurement). Diets were adjusted weekly by changing the forage and concentrate supplied to include variation in the diet. Faecal and diet samples were pooled per week per horse resulting in 116 D:F pairs (29 horses x 4 weeks). Samples from weeks 1–3 (n = 87) were used to develop calibration and week 4 samples were used as an independent validation set (n = 29). Modified partial least square regression was used for calibration development. Results obtained showed that faecal NIRS prediction of dietary OM was excellent [coefficient of determination of validation (R2val) = 0.92 and ratio performance deviation (RPD) = 3.54], while that of neutral detergent fibre (aNDF) was very good (R2val and RPD: 0.83 and 2.12). The validation statistics (R2val: 0.77 and RPD: 1.56) showed the prediction model for acid detergent fibre (ADF) was acceptable but not as precise as those of OM and aNDF. Prediction of acid detergent lignin (ADL) was the poorest (R2val = 0.41and RPD = 1.05). For the mineral components, prediction of copper (Cu) was very good (R2val = 0.81, RPD= 2.11) and better than those of Phosphorus (P), Sulphur (S) and Manganese (Mn) which all had R2val of 0.60–0.67 and RPD of 1.60–1.64. Overall, this study showed that faecal NIRS profiling can be used to successfully predict OM, aNDF and Cu concentration in horses’ diet. Although, the predictions of other fibre fractions were less satisfactory, faecal NIRS can still be used for monitoring purposes and improved for future application to independent samples.