Multivariate prediction of odor from pig production based on in-situ measurement of odorants. (June 2016)

• Record Type: Journal Article
• Authors: Hansen, Michael J.; Jonassen, Kristoffer E.N.; Løkke, Mette Marie; Adamsen, Anders Peter S.; Feilberg, Anders
• Is Part Of: Atmospheric environment. Volume 135(2016)
• Page Start: 50
• Page End: 58
• Publication Date: 2016-06
• Subjects: Principal component analysis -- Partial least squares regression -- Proton-transfer-reaction mass spectrometry -- Olfactometry; Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2016.03.060
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: The aim of the present study was to estimate a prediction model for odor from pig production facilities based on measurements of odorants by Proton-Transfer-Reaction Mass spectrometry (PTR-MS). Odor measurements were performed at four different pig production facilities with and without odor abatement technologies using a newly developed mobile odor laboratory equipped with a PTR-MS for measuring odorants and an olfactometer for measuring the odor concentration by human panelists. A total of 115 odor measurements were carried out in the mobile laboratory and simultaneously air samples were collected in Nalophan bags and analyzed at accredited laboratories after 24 h. The dataset was divided into a calibration dataset containing 94 samples and a validation dataset containing 21 samples. The prediction model based on the measurements in the mobile laboratory was able to explain 74% of the variation in the odor concentration based on odorants, whereas the prediction models based on odor measurements with bag samples explained only 46–57%. This study is the first application of direct field olfactometry to livestock odor and emphasizes the importance of avoiding any bias from sample storage in studies of odor-odorant relationships. Application of the model on the validation dataset gave a high correlation between predicted and measured odor concentration (R 2 = 0.77). Significant odorants in the prediction models include phenols and indoles. In conclusion, measurements of odorants on-site in pig production facilities is an alternative to dynamic olfactometry that can be applied for measuring odor from pig houses and the effects of odor abatement technologies. Highlights: Analytical and sensory measurements were carried out in parallel. Sampling in bags was avoided by using direct on-site measurements. A partial-least-squares regression model using on-site data was achieved. On-site measurements of odor clearly improved the model validation (R 2 = 0.77).