1st Page

Journal of Animal Science - Animal Nutrition

Technical Note: Validation of rumination collars for beef cattle1


This article in JAS

  1. Vol. 91 No. 6, p. 2858-2862
    Received: Sept 27, 2012
    Accepted: Feb 27, 2013
    Published: November 25, 2014

    2 Corresponding author(s):

  1. C. Goldhawk*†,
  2. K. Schwartzkopf-Genswein* and
  3. K. A. Beauchemin 2
  1. Agriculture and Agri-Food Canada, Research Centre, Lethbridge, AB, Canada
    University of Calgary Veterinary Medicine, Calgary, AB, Canada


Two studies were conducted to evaluate a wireless rumination monitoring system (Hi-Tag; SCR Engineers Ltd., Netanya, Israel) for recording rumination in beef cattle. The system operates based on acoustics of rumination and has been used previously for monitoring rumination in both young and adult dairy cattle. Study 1 consisted of beef cattle in tie-stall housing with 72 observations for 4 cattle fed a high forage backgrounding diet and 47 observations for 3 cattle fed a high grain finishing diet. Study 2 consisted of 44 observations for 6 beef cattle fed a high forage diet in a loose-housing feedlot pen. Each observation consisted of the rumination time during a 2-h period as estimated by visual observations made by trained observers (inter-observer correlation r = 0.97) and the Hi-Tag system. The mean difference between visual and Hi-Tag rumination times was 9.8 ± 18.7 min/2 h. The rumination times from the Hi-Tag system were only moderately correlated with visual observations (r = 0.41, P < 0.001). The difference between visual and Hi-Tag rumination times was not correlated with head posture during the 2 h period (r = –0.02, P = 0.89). Physical and dietary differences between dairy and beef cattle may have resulted in the inaccuracy of the Hi-Tag system when used in beef cattle fed typical backgrounding and finishing diets. More work is required to improve the accuracy of this automated system for rumination monitoring before it can be used reliably with beef cattle.


Rumination is a process of regurgitating ingesta from the reticulorumen into the mouth where the bolus is masticated and mixed with saliva for 30 to 60 s and reswallowed followed by a short pause before repeating the process (Beauchemin, 1991). Rumination enables microbial digestion, increases saliva secretion (Beauchemin, 1991), and can be influenced by changes in diet, management, and health status in cattle (Metz, 1975; Beauchemin et al., 1994; Borderas et al., 2008). Therefore, monitoring rumination provides insight into the health and well-being of cattle.

Automating the monitoring of behaviors such as rumination is beneficial as it removes the influence of observers, may reduce the costs of obtaining the information, and may identify nuances of behavior not readily visible to human observers (Magnusson, 2000; Rushen et al., 2012). Many automated systems have been developed for monitoring feeding and rumination behavior of grazing or tethered cattle using acoustics (Laca and Wallis DeVries, 2001; Clapham et al., 2011), jaw motion (Beauchemin et al., 1989; Matsui, 1994), or position of the body (Ungar et al., 2011). These systems, however, are not suitable for loose-housed or nongrazing cattle, such as beef cattle in feedlots.

One system, the Hi-Tag rumination monitor (SCR Engineers Ltd., Netanya, Israel), monitors rumination behavior in loose-housed cattle using collars to collect and store data and a reader that downloads data when cattle pass by (Schirmann et al., 2009). Rumination is recorded by this system using a microphone on the collar of the animal and is summarized as total time spent ruminating during 2-h periods. The Hi-Tag system has been successfully validated for adult dairy cattle (Schirmann et al., 2009). The system performed less reliably when used with dairy heifers, potentially due to dietary differences or differences between the acoustics of rumination of young and adult cattle (Burfeind et al., 2011). The Hi-Tag system has not been evaluated for growing beef cattle in feedlots, which are often fed diets containing more concentrates and less forage than the diets fed to adult dairy cattle. Beef cattle also have anatomical differences such as dewlaps, musculature of the neck, and skin thickness, which may influence the acoustics of rumination in a way that interferes with the function of the Hi-Tag system. The objective of this study was to evaluate the Hi-Tag system for beef cattle by determining 1) the accuracy of the Hi-Tag system for 2 typical feedlot diets relative to visual observations and 2) the effect of head posture on the agreement between estimates of rumination time from the Hi-Tag system and visual observations.


Data were collected during 2 studies conducted with approval from the Lethbridge Research Center Animal Care Committee. All animals were managed according to the Canadian Council on Animal Care (Olfert et al., 1993). Hi-Tag rumination collars were placed on the cattle according to the instructions of the company 1 to 2 d before beginning the observations.

The definition of rumination and method for recording bouts was defined according to Schirmann et al. (2009, P. 6053) as follows: “the onset of rumination was defined as the time when regurgitation took place, namely when a bolus came up the esophagus and reached the mouth. Observers recorded the start time when a bolus reached the mouth and the cow began to chew rhythmically and recorded the ending time when the bolus was swallowed.” The Hi-Tag system outputs a summary of total time spent ruminating during 2-h periods, with no access to raw data on timing or duration of individual rumination bouts within a period. Therefore, the start and end time for each bolus was recorded and used to calculate length of each rumination bout, which was then summarized as total duration of rumination per 2-h period to match with the output from the Hi-Tag system.

Observers were trained using both live observations and video recordings. First, 2 observers discussed live observations until an agreement could be reached on the application of the aforementioned definition of rumination. To determine inter-observer reliability, 2 observers recorded rumination for twenty-four 2-h live observation periods using 4 cattle simultaneously in tie-stall housing (6, 6, 4, and 5 periods from each animal). Observers were situated below camera mountings and were unable to see the recording sheets of each other.

For video observations, 2 independent observers recorded rumination during thirty 2-h periods from 4 cattle in the same tie-stall housing described below for Study 1 (6, 6, 8, and 10 periods from each animal) to determine inter-observer reliability. To determine intra-observer reliability of video observations, 1 of the trained observers watched forty 2-h observation periods from 4 cattle (8, 8, 12, and 12 periods from each animal) using video recorded in tie-stall housing and repeated the observations from the same video after a 1-wk interval.

Due to equipment failure during live observations, only 6 periods of live observations from 2 cattle (3 per animal) were available that also had video observations. The 6 periods with both live and video recorded observations were used to compare live vs. video observations of rumination. The Pearson’s correlation coefficient was calculated between observers for both live and video observations to determine inter-observer reliability for each method of observing rumination and within observer to determine intra-observer reliability of video observations and between video and live observations to compare methods.

Study 1 consisted of 7 Angus or Angus-Hereford yearling heifers kept in tie-stall housing, tethered by a halter during observations so as not to interfere with the Hi-Tag collar. All heifers had been acclimated to handling and tie-stalls from the time they were weaned at 5 to 6 mo of age. Cattle had been fed a high grain feedlot finishing ration (HG; Table 1) when they started the trial, which was continued for 1 wk before switching to a high forage backgrounding diet (HF; Table 1) for the remaining observations. Cattle were continuously video recorded to observe rumination behavior and posture. Video equipment was placed 1 m away from the head of each animal to ensure that rumination behavior could be accurately identified. There were a total of 72 unobstructed 2-h periods of 4 cattle during feeding of the HG diet (16, 17, 18, and 21 observation periods per animal) and 47 unobstructed 2-h periods for 3 cattle during feeding of the HF diet (14, 16, and 17 observation periods per animal). In Study 2, 6 Angus or Angus-Hereford yearling steers were loose housed for 3 wk in a typical feedlot pen with a bunk feeder and fed a diet similar to the HF diet in Study 1 (Table 1). A live observer positioned at the feed bunk of the pen recorded rumination behavior for animals in Study 2 and had an unobstructed view of the animals at all times. There was a total of forty-four 2-h periods across all 6 animals in study 2 (4 cattle each provided 8 observation periods, with the remaining 2 cattle providing 5 and 7 observation periods). Video cameras were placed on the pen fencing to record animal posture.

View Full Table | Close Full ViewTable 1.

Diet composition

Study 1
Study 2
Item Backgrounding (HF)1 Finishing (HG)2 Backgrounding (HF)1
Ingredient (% of the dietary DM)
    Barley silage 60 9 60
    Barley grain, dry rolled 30 81 35
    Canola meal 3.3 3.3 0.48
    Beet pulp 5.0 5.0
    Ground barley 2.73
    Calcium carbonate 1.20 1.20 1.29
    Urea 0.25 0.25 0.10
    Salt 0.16 0.16 0.16
    Dried molasses 0.01 0.01 0.13
    Mineral and vitamin premix3 0.05 0.05 0.05
    Melengestrol acetate premix4 0.03 0.03
    Flavoring agent5 0.003 0.003
    Vitamin E premix6 0.003
    Canola oil 0.05
1HF = high forage diet.
2HG = high grain diet.
3Supplied per kilogram of dietary DM: 51 mg of Zn, 24 mg of Mn, 13 mg of Cu, 0.6 mg of I, 0.2 mg of Co, 0.25 mg of Se, 2,200 IU of vitamin A, 275 IU of vitamin D, and 10 IU of vitamin E.
4Contained 220 mg/kg.
5Anise 422 powder containing ground cumin, fennel, fenugreek, silicon dioxide, and wheat bran (Canadian Bio-Systems Inc., Calgary, AB, Canada).
6Contained 500,000 IU/kg.

A subset of observation periods was selected from video recordings to evaluate the effect of posture (head up vs. head down) on the agreement between the Hi-Tag equipment and observers. The percentage of time during the 2-h period that each animal spent with its head up and head down was recorded for fifteen 2-h periods for the HG diet in the tie-stalls, fifteen 2-h periods for the HF diet in the tie-stalls, and fifteen 2-h periods for the HF diet in the feedlot pen.

The concordance correlation coefficient (CCC) and Pearson’s correlation coefficient were calculated for pooled data as well as by diet and location to determine the relationship between visual observation and Hi-Tag output for rumination time in minutes and the deviation from a 1:1 relationship in both slope and intercept of the actual data (SAS Inst. Inc., Cary, NC). To evaluate the effect of posture on the performance of the Hi-Tag system, the Pearson’s correlation coefficient was calculated for the percentage of the 2-h period spent in the head-up position and the difference between visual observations of rumination time and the output of rumination time from the Hi-Tag system.


There was a high correlation between the rumination times recorded by 2 independent observers for both live (r = 0.98, P < 0.001) and video recorded (r = 0.97, P < 0.001) observations. There was also a high intra-observer correlation for video observations (r = 0.99, P < 0.001). Despite small sample size, the correlation between video and live observations of rumination was also high (r = 0.97, P = 0.001). The high correlation between methods of observation and both within and between observers indicates that rumination was consistently evaluated by observers, similar to other research (Schirmann et al., 2009; Burfeind et al., 2011).

Across all diets and housing systems, the Hi-Tag system estimated rumination time to be 16.7 ± 11.2 min/2 h whereas visual observations estimated 26.6 ± 20.3 min/2 h (mean ± SD). On average, the visual observations were 9.8 ± 18.7 min/2 h longer than the rumination time estimated by the Hi-Tag system. The differences in average estimates of rumination time per 2 h indicate that although there was considerable variability in the estimates of rumination time, the Hi-Tag systems generally underestimated rumination relative to visual observations. Table 2 presents the Pearson’s correlation coefficient, CCC, location, and scale shift between the rumination time estimated by the Hi-Tag rumination monitors and the rumination time estimated by visual observations. The overall Pearson’s correlation coefficient was low when data from all diets and housing systems were pooled. When analyzed by diet and location, the lowest correlation was in the feedlot with the HF diet (Table 1). As indicated by the CCC and variation in location and scale shifts, there was no consistent deviation from a 1:1 relationship, which is apparent in the spread of the data displayed in Fig. 1.

View Full Table | Close Full ViewTable 2.

Pearson’s correlation coefficient, concordance correlation coefficient, location, and scale shift between Hi-Tag1 rumination collar and visual observations of rumination by beef cattle as affected by housing system and diet

Housing Diet r P-value2 CCC3 P-value4 Location shift Scale shift
Tie-stall High forage 0.46 0.001 0.24 ± 0.07 0.001 1.02 2.35
High grain 0.39 <0.001 0.29 ± 0.08 <0.001 0.67 1.63
Feedlot High forage 0.08 0.595 0.07 ± 0.14 0.280 0.33 1.45
Overall 0.41 <0.001 0.30 ± 0.05 <0.001 0.65 1.82
1SCR Engineers Ltd., Netanya, Israel.
2P-value for Pearson’s correlation coefficient.
3CCC = concordance correlation coefficient.
4P-value for concordance correlation coefficient.
Figure 1.
Figure 1.

Number of minutes that beef cattle spent ruminating during a 2 h period estimated by visual observation vs. Hi-Tag rumination collars for tie-stall and feedlot housing and high forage and high grain diets.


It was noted during Study 1 that although the collars were fitted when cattle were in a head-up position, the collars appeared to loosen and move from their initial position when the cattle lowered their head. There was, however, no significant correlation between the percentage of time during an observation period spent in the head-up position and the disagreement between visual observation and Hi-Tag collar estimations of rumination time in either housing system or diet (Table 3).

View Full Table | Close Full ViewTable 3.

Pearson’s correlation coefficient between the differences in rumination minutes estimated by visual observation and Hi-Tag1 rumination collars and the percent of observation periods that cattle spent with their head in the raised position as affected by housing system and diet

Housing Diet r P-value
Tie-stall High forage 0.34 0.22
High grain –0.14 0.62
Feedlot High forage 0.13 0.63
Overall –0.02 0.90
1SCR Engineers Ltd., Netanya, Israel.
2P-value for significance of Pearson’s correlation coefficient.

Research with dairy cattle found very high correlations (r-values ranged from 0.86 to 0.96) between rumination times estimated by the Hi-Tag system and visual observations (Schirmann et al., 2009). Cattle in the study by Schirmann et al. (2009) were loose-housed dairy cows fed diets with much greater forage content than the diets fed in this study. Burfeind et al. (2011) evaluated the Hi-Tag system vs. visual observations of rumination time for Holstein calves (46 ± 14 d) and heifers (208 ± 78 d) and found reasonable correlation with animals >9 mo of age (r = 0.88) and high variation in the agreement between live observations and Hi-Tag measurement of rumination between groups of calves that were <9 mo old (r-values from 0.47 to 0.89). Burfeind et al. (2011) suggest that there may be differences in rumination acoustics of different dietary components and differences between young and adult dairy cattle that resulted in the lower correlation between visual observation and Hi-Tag estimates of rumination found in their study compared with the results of Schirmann et al. (2009). Cattle used in this study were older than cattle used by Burfeind et al. (2011) but likely younger than the cows used by Schirmann et al. (2009). It is plausible that age-related factors played a role in the low correlation found in our studies. Dietary and physical differences, such as dewlaps, musculature of the neck, and skin thickness, may also have contributed to the low correlation between visual observations and Hi-Tag rumination collars found with the beef cattle in our studies.

In conclusion, more research is required to refine the functioning of the Hi-Tag equipment for estimation of rumination time when used with beef cattle fed backgrounding and finishing diets in feedlots. Because the Hi-Tag system compiles the rumination data in 2-h summaries with no access to the raw data, it was not possible to further investigate the nature of the disagreement between Hi-Tag and live observation estimates of rumination behavior in beef cattle. From the research presented here, different postures do not appear to be associated with the error in estimating rumination times. Given the extensive nature of beef production, automating the monitoring of rumination behavior would be an excellent development worth pursuit.




Be the first to comment.

Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.