Page 172 - BOOK OF PROGRAM ICAGRI 2

#4060

Comparison single and mix species NIRS database accuracies
of dairy fibre feed value detection

3
2*
1
2
2
3
I Agustiyani , Despal , L A Sari , R Chandra , R Zahera , I G Permana
1 Study Program Nutrition and Feed Science, Department of Nutrition and Feed
Technology, Faculty of Animal Science, IPB University. Jl. Agatis Kampus IPB
Darmaga 16168, Bogor-Indonesia
2 Department of Nutrition and Feed Technology, Faculty of Animal Science, IPB
University, Jl. Agatis Kampus IPB Darmaga 16168, Bogor-Indonesia
3 Study Program Nutrition and Feed Technology, Department of Nutrition and Feed
Technology, Faculty of Animal Science, IPB University, Jl. Agatis Kampus IPB
Darmaga 16168, Bogor-Indonesia

E-mail: despaltk@gmail.com

Abstract. Near Infrared Reflectance Spectroscopy (NIRS) accuracy is determined by its
database. However, our previous developed database for dairy cattle dietary fibre feed (DFF)
showed low accuracy for complex organic substance detection such as ADF and NDF due to
mix species used in the database. Effort to improve DFF prediction accuracy have been done
using single species fibre feed database (SSFFD). This paper aimed to compare single and mix
species NIRS database accuracy in predicting DFF nutrient contents. In mix database, five feeds
(Napier grass, natural grass, corn leaves, cornhusk, and rice straw) from three areas of dairy
cattle farming were sampled (two farms from each area). In single database, thirty Napier grass
from 30 areas were collected. Samples were analysed chemo-metrically for its dry matter (DM),
ash, crude protein (CP), crude fibre (CF), neutral detergent fibre (NDF), acid detergent fibre
(ADF). Spectra of each sample were collected thrice (two spectra for calibration and a spectrum
for validation) using FT-NIRS Spectrometer Solid Cell. Calibration and validation models were
carried out using the Partial regression. For external validation, seven independent Napier grass
samples were tested. The result shows that calibration using mix species produced residual
predictive deviation (RPD) > 2 on all parameters. Meanwhile, calibration using single species
with RPD more than 2 were only for CP, CF, and NDF. In external validation, the ratios of the
standard error of prediction (SEP) to standard error of laboratory (SEL) using the mix species
database were less than 2 for DM, ash, CF, and NDF. Meanwhile, SEP/SEL using single species
were > 2 for all parameters, except for DM and ash. Single species NIRS database develop using
Napier grass were less accurate than mixed species DFF due to large nutrient content variation
between varieties of grass. It is concluded that database accuracy develop rom mixed dietary
fibre feed were more accurate in compare to single species but different varieties of Napier
grass.

© 2020 ICAGRI - The 2nd International Conference on Agriculture and Bioindustry (ICAGRI), 169
Agriculture Faculty, Universitas Syiah Kuala. All Rights Reserved