Decoding Acoustic Signals of Herbage Intake

SANTIAGO A. UTSUMI; DIEGO HUMBERTO MILONE; JULIO GALLI; JOSÉ OMAR CHELOTTI; EMILIO LACA; HUGO LEONARDO RUFINER; SEBASTIÁN RODRIGO VANRELL; LEONARDO GIOVANINI

Michigan

Congreso; International Society of Ecoacoustics Ecoacoustics Congress; 2016

International Society of Ecoacustics

Easy and accurate measurement of grazing behavior and herbage intake is critical to promote progress in science and management of grazing systems. Hence, we seek to improve the efficacy of automatically detecting and classifying digestive events of grazing ruminants by means of more effective acoustic instrumentation and analytical procedures. Rigorous testing of a system called Chew‐Bite Real‐Time Analysis (CBRTA) was conducted with dairy cows of the MSU and UNR grazing herds. The CBRTA processing algorithm was capable to evaluate complex grazing soundtracks with a speed of 50 mes faster than real-time and without affecting accuracy in event detection and classification or sound proper es quantification. Furthermore, when 24 h sound tracks were sampled, and digitally and automatically processed, 96% of the digestive events were correctly detected and up to 83% of them were effectively classified as being bites, chews or composite chew‐bites (i.e. compound chewing and biting). The ability of CBRTA to integrate both, behavior and acoustic variables for real‐time assessment of herbage dry matter intake (DMI) was also tested. Energy flux density (EFD) of chewing sounds was linearly related to DMI and up to 74% of the total variation in EFD was due to variation in DMI alone. The best predictive model explained 91% of the observed DMI (CV 17%) and included the predictors: total chewing EFD, number of chew‐bites and plant height (tall vs. short). The present research confirmed that digestive sounds contain valuable information to remotely monitor feeding behavior and to predict herbage intake.