Abstract
Brown seaweeds are the most studied and exploited algae type for ruminant nutrition due to their biomass availability, ease of harvest and content of bioactive compounds. Infrared spectroscopy represents a rapid, non-invasive and chemical-free technique that is widely applied for the chemical characterization and digestible quality of many terrestrial forages. However, there is limited information regarding its application to seaweeds. This study compared the effectiveness of Near-Infrared (NIR: 9000–4000 cm-1) and Mid-Infrared (MIR: 4000–400 cm-1) spectroscopy to measure the nutritional value and in vitro dry matter rumen digestibility of brown seaweeds. Due to the small number of seaweed samples available, 40 samples were analysed in triplicate with a total dataset of 120 samples. For partial least-squares regression model development and evaluation purposes, the dataset (n = 120) was divided into two subsets, the first one for training and model development purposes (70% of data, n = 84), and the second one for model testing and evaluation (internal evaluation) purposes (30% of data, n = 36). Partial least-squares regression was employed to develop multivariate calibration models which were internally and externally validated. The samples were analysed using established wet chemistry methods which were regarded as the reference methods. NIR showed high accuracy for the quantitative prediction of crude protein (R2P = 0.99; RMSEP = 0.51; RER = 26.9; RPD = 6.9) and total polyphenolic content (R2P = 0.94; RMSEP = 0.20; RER= 10; RPD= 3.2), whereas MIR could only accurately predict crude protein (R2P = 0.96; RMSEP = 1.12; RER = 11.64; RPD = 3.14). Ash, neutral and acid detergent fiber, lignin (sa) and in vitro dry matter rumen digestibility models showed limited applicability for quantitative measurements (R2P < 0.85; RPD < 2). Overall, NIR and MIR could be used to rapidly evaluate the nutritional composition and digestibility of brown seaweeds in their dried form but further evaluation on an external database would be required to assess the robustness of these models on unrelated data. Furthermore, the use of these spectroscopic methods showed lower accuracy and precision compared to wet chemistry methods, which better qualifies them for screening rather than confirmatory analysis.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have