0 avis
Influence of measurement method of methane on the performances of faecal NIRS models in cattle
Archive ouverte
Edité par CCSD ; Wageningen Academic Publishers -
Session 72. Methods and technologies for research and smart nutrition management in dairy and beef cattle. International audience. Enteric methane (CH4) emissions in cattle is one of the largest greenhouse gas contributor from agriculture. MeasuringCH4 emissions is important in order to assess mitigation strategies but existing methods cannot be used on largenumber of animals on farm. Spectroscopic methods using milk have been suggested as an alternative for predictingemissions on farm but are limited to lactating dairy cows. Faeces could be a more universal matrix but few studies areavailable regarding the use of spectroscopic methods on the prediction of CH4 emissions. The objective of this studywas to asses faecal near-infrared spectroscopic (NIRS) prediction models, based on CH4 emissions measured withdifferent methods (respiration chambers (RC), sulphur hexafluoride tracer gas (SF6) and GreenFeed® (GF)). In thiswork based on two in vivo studies, we used SF6 measurements as the reference. In study 1, RC and SF6 measurementswere performed simultaneously leading to two NIRS models based on 32 RC and SF6 methane values in non-lactatingcows. In study 2, SF6 and GF values were collected simultaneously and two NIRS models were developed using 24values from dairy cows. In both studies, bias between measured and predicted values was non-significant comparedto the standard error of cross-validation corrected by the bias (SECV(C)), when values issued from models based onSF6 method were compared to SF6 values (bias=0.54 g/d and SECV(C)=45.1 g/d for study 1 and bias=3.00 g/d andSECV(C)=95.4 g/d for study 2). However, if the cross-validation is done on RC or GF and compared to SF6 values,bias greatly increased (-38.0 g/d and 101.2 g/d in study 1 and 2, respectively whereas SECV(C) in each study were 50.3and 100.4 g/d). These results suggest that SECV(C) values were similar between methods within each study but therewas a bias. This bias was more important for GF than for RC when compared to SF6 method. These results could helpto minimize errors due to the combination of CH4 emission databases obtained from several measurement methods.
Consulter en ligne
Suggestions
Du même auteur
