Vibrational spectroscopy to predict in vitro digestibility and the maturity index of different forage crops during the growing cycle and after freeze- or oven-drying treatment

The aims of the study were to utilize vibrational spectroscopy as a rapid predictive tool of forage quality; to compare two preparation methods, freeze- (FD) vs. oven-dried (OD); to focus on the progression of intra- and inter-family maturity by adopting a multivariate crop maturity index (CMI) based on composition, digestibility and tillage traits. A panel of forages (n = 158) composed of 12 crops (borage, chia, false flax, flax, galega, hemp, perilla, quinoa, ravizzone, safflower, sunflower and white lupin) derived from 8 botanic families, sampled at different vegetative stages, and which were FD or OD, were examined. Two spectrometers were used at different spectral regions: a Perkin Elmer IdentiCheck™ (PE, B-band, 714–1025 nm; C-band, 1026–2500 nm, D-band, 2501–3333 nm) and a portable Analytical Spectral Device (ASD, A-band, 350–713 nm, UV–Vis; B-band, as above). The absorption spectra were constantly higher in the OD samples and showed very high discriminability. The average prediction response (RPD, defined as the performance-deviation ratio) was better with the PE instrument, because of its enhanced band capabilities. However, the response over the spectral regions differed on the basis of which instrument was used and according to the preparations. The ASD instrument was more efficient in the B-band, for the OD preparation and better than PE in the pooled calibration (RPD:1.63 vs. 1.20; P=0.0005). A significant superiority in the NIR C-band for the FD preparation was observed (RPD: 2.46 vs. 1.95; P=0.004), while, unexpectedly, the MIR D-band was 25% more performing (RPD: 2.78 vs. 2.21; P=0.0005). The ash, the neutral detergent fiber (NDFom) and its indigestible part (INDF) were placed at the highest prediction rank in both instruments, albeit at different precision levels, caused by the different instrumental capabilities, with an overall 1-VR avg. of 0.81 (1-VR, defined as the r-square of internal cross-validation). In a composite FD-OD equation, the best prediction was made by the INDF (1-VR of 0.91 and 0.88 for the PE and ASD instruments, respectively). The worst performances were observed for the digestible neutral detergent fiber (DNDF) prediction. The CMI was influenced by the INDF (R2 0.91) and was accurately predicted by vibrational spectroscopy (RPD 5.2 and 2.9 for PE and ASD, respectively). CMI was able to summarize the botanical differences and highlight a rank between the eight families from the less mature pole: Boraginaceae and Chenopodiaceae < Lamiaceae < Asteraceae and Fabaceae < Cannabaceae < Brassicaceae < Linaceae (the most mature type). Four particular wavelengths have been identified as they are related to the CMI, namely: 701 (red), 905, 2451 and 2799 nm, respectively, in the A, B, C and D bands.