The quantification of feed composition parameters, as well as feed digestibility, requires time-consuming and expensive laboratory reference methodologies. In the context of precision livestock feeding, fast, accurate and on-line methods to quantify at a cost-effective the composition and characteristics of total mixed ration (TMR) are desired tools. Therefore, the aim of the present study was to develop prediction models using near infrared spectroscopy (NIRS) for a plethora of TMR compositional traits and digestibility. A total of 205 TMR samples were collected in herds located in the Parmigiano Reggiano area in different experimental trials. Samples were analysed, using reference methodologies, for crude protein (CP), starch, amylase-treated ash-corrected neutral detergent fiber with addition of sodium sulfite (aNDFom), acid detergent fiber (ADF), acid detergent lignin (ADL) and ash. Moreover, the fiber undegradable fractions (uNDF) of TMR was quantified in vitro at different timepoints (24 h, 30 h, 120 h, 240 h). Spectral data in the range from 900 to 2500 nm were collected for all these samples using a TANGO FT-NIR spectrometer. Partial least squares regression was employed to calibrate NIRS prediction models through a cross-validation, but also to validate such models on a subset including a random 30% of the total observations which were excluded from the calibration set. Different mathematical pre-treatments were also applied to the spectra data in order to identify the transformation which provided the most accurate NIRS prediction model (in terms of maximised explained variance and minimised root mean square error of prediction). The coefficient of determination in external validation (R2P) was >0.80 only for starch, between 0.60 and 0.80 for CP, aNDFom, ADF and ADL, while <0.50 for ash content. The R2P for uNDF prediction models at different timepoints ranged between 0.56 (uNDF30) and uNDF240 (0.68). The residual prediction deviation (RPD) in external validation was <2 for all NIRS prediction models, with the exception of ADF model (RPD =2.20). In conclusion, NIRS can be a feasible and rapid method for the prediction of different TMR chemical components and digestibility measures, although large samples size and variability would be desired and could contribute to increase models accuracies.

NIRs calibration model for chemical composition and digestibility of total mixed rations for Parmigiano Reggiano ration

Giovanni Buonaiuto;Francesca Ghiaccio;Damiano Cavallini;Alberto Palmonari;Ludovica Mammi;Luca Campidonico;Giorgia Canestrari;Giulio Visentin;Andrea Formigoni
2021

Abstract

The quantification of feed composition parameters, as well as feed digestibility, requires time-consuming and expensive laboratory reference methodologies. In the context of precision livestock feeding, fast, accurate and on-line methods to quantify at a cost-effective the composition and characteristics of total mixed ration (TMR) are desired tools. Therefore, the aim of the present study was to develop prediction models using near infrared spectroscopy (NIRS) for a plethora of TMR compositional traits and digestibility. A total of 205 TMR samples were collected in herds located in the Parmigiano Reggiano area in different experimental trials. Samples were analysed, using reference methodologies, for crude protein (CP), starch, amylase-treated ash-corrected neutral detergent fiber with addition of sodium sulfite (aNDFom), acid detergent fiber (ADF), acid detergent lignin (ADL) and ash. Moreover, the fiber undegradable fractions (uNDF) of TMR was quantified in vitro at different timepoints (24 h, 30 h, 120 h, 240 h). Spectral data in the range from 900 to 2500 nm were collected for all these samples using a TANGO FT-NIR spectrometer. Partial least squares regression was employed to calibrate NIRS prediction models through a cross-validation, but also to validate such models on a subset including a random 30% of the total observations which were excluded from the calibration set. Different mathematical pre-treatments were also applied to the spectra data in order to identify the transformation which provided the most accurate NIRS prediction model (in terms of maximised explained variance and minimised root mean square error of prediction). The coefficient of determination in external validation (R2P) was >0.80 only for starch, between 0.60 and 0.80 for CP, aNDFom, ADF and ADL, while <0.50 for ash content. The R2P for uNDF prediction models at different timepoints ranged between 0.56 (uNDF30) and uNDF240 (0.68). The residual prediction deviation (RPD) in external validation was <2 for all NIRS prediction models, with the exception of ADF model (RPD =2.20). In conclusion, NIRS can be a feasible and rapid method for the prediction of different TMR chemical components and digestibility measures, although large samples size and variability would be desired and could contribute to increase models accuracies.
ASPA 24th Congress Book of Abstract
170
170
Giovanni Buonaiuto, Francesca Ghiaccio, Damiano Cavallini, Alberto Palmonari, Ludovica Mammi, Luca Campidonico, Giorgia Canestrari, Giulio Visentin, Andrea Formigoni
File in questo prodotto:
Eventuali allegati, non sono esposti

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11585/834025
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact