1st Page

Journal of Animal Science - Animal Production

Relationships of a novel objective chute score and exit velocity with growth performance of receiving cattle12


This article in JAS

  1. Vol. 94 No. 11, p. 4819-4831
    Received: Mar 03, 2016
    Accepted: Aug 31, 2016
    Published: October 13, 2016

    4 Corresponding author(s):

  1. K. A. Brunoa33,
  2. E. S. Vanzant 4a,
  3. K. A. Vanzanta and
  4. K. R. McLeoda
  1. a Department of Animal and Food Sciences, University of Kentucky, Lexington 40546


Animals with excitable temperaments often have decreased gains that have been associated with decreased intake and efficiency. Different temperament measures probably measure different specific underlying traits. Commonly used temperament measures include both objective and subjective measures. Subjective measures present potential difficulties for making across-study comparisons and thus for generalizing quantitative relationships. One objective of this experiment was to evaluate 2 related, but different, measures associated with temperament, where 1 measure is a new, objective measurement based on the common subjective chute score measures. Also, there is reason to believe that RDP requirements of animals may vary with temperament. To examine the relationships between temperament measures and nutrient use, 192 crossbred steers were used in a 58-d randomized complete block design experiment. Temperament treatments (assigned prior to d 1) were chute exit velocity (EV; slow vs. fast) and objective chute score (WSD; low vs. high), a novel temperament measure that was the SD of weights collected at 5 Hz for 10 s while an animal was restrained in a chute with its head caught. Both were measured on d −8, 1, 2, 16, 30, 56, and 58, where d 1 was the day that animals were allotted to treatment groups and began receiving experimental diets. Steers were fed a diet with 1 of 3 RDP levels (75%, 105%, and 120% of RDP requirements). There were no main effects or interactions with RDP (P ≥ 0.12); thus, it was removed from the statistical model for subsequent analyses. There were no interactions between EV and WSD (P ≥ 0.11). Slow EV animals had greater ADG (P = 0.02) and DMI (P ≤ 0.09) than fast EV animals, but there was no effect of EV on G:F (P > 0.14). For d 0 to 58, high WSD animals had greater DMI (P ≤ 0.09) than low WSD animals but no difference in ADG (P = 0.23), whereas low WSD animals tended to have increased G:F (P = 0.11). Results of this study give additional confirmation that EV is associated with DMI and growth and provide evidence that a novel measure of behavior, WSD, is also related to growth, independently of EV. Because WSD and EV appear to measure different underlying behavioral traits, use of both measures may improve our ability to discriminate among temperament categories for growing cattle.


Growth rates and immunocompetence have been shown to be depressed in cattle with excitable temperaments (Voisinet et al., 1997; Burdick et al., 2011). Changes in both intake (Café et al., 2011b) and efficiency (Petherick et al., 2002) have been implicated as contributors to such responses. Receiving stress is associated with decreased intake and is likely to compromise ruminal function (Loerch and Fluharty, 1999). As a result, it is important to ensure adequate RDP intake in receiving cattle. If receiving stress is differentially expressed in animals from different temperament groups, it is possible that these groups will respond differently to RDP supply. Furthermore, animals’ behavior may be related to gut microbial populations (Bercik et al., 2011). In ruminants, passage rate is a large determinant of the microbial population in the gut. Varied intake among temperament groups would be expected to correlate to varied passage rate and thus to potential variation in gut microbial populations. Because of the potential for differences in intake, growth rate, and gut microbial populations, it is reasonable to assume that differences in temperament among receiving cattle may be associated with differences in RDP requirements.

Both exit velocity (EV; Burrow et al., 1988; Curley et al., 2006) and subjective chute score (SCS; Grandin, 1993; MacKay et al., 2013) are commonly used measures of temperament in cattle research. As a subjective measure, SCS should be viewed with some caution (Curley et al., 2006; Core et al., 2009). Objective measures that attempt to assess behaviors similar to SCS may also be related to production traits and avoid some of the concerns associated with subjective measures (Manteca and Deag, 1993; Curley et al., 2006; Sebastian et al., 2011). One approach for an objective chute score for cattle is to measure variation in scale readings obtained during a defined holding period in a squeeze chute.

Additionally, although several studies have demonstrated relationships between temperament and growth (e.g., Reinhardt et al., 2009; Cafe et al., 2011b; Bates et al., 2014), relatively few studies have looked at the implications of sorting animals into like-temperament groups. Because at least part of the temperament effect may be expressed through influences on social interactions among individuals, this scenario is different from those in which animals of various temperament classes are penned together. If differential management of animals with different temperaments proves warranted (e.g., different rations, vaccination schedules, etc.), an understanding of the effects of grouping animals by temperament class is essential.

Thus, the objectives of this study were to evaluate a novel objective chute score measurement and EV as independent measures by which to group animals to determine if temperament interacts with RDP to affect growth, intake, and morbidity of newly received feeder cattle.


All procedures were approved by the University of Kentucky Institutional Animal Care and Use Committee.

Animals and Behavioral Measures

One hundred ninety-two mixed breed beef steers (243 ± 1.75 kg) in 48 pens (4 steers/pen) were used in a randomized complete block design with a 2 × 2 × 3 factorial treatment structure, using initial weight as a blocking factor. Steers were purchased from central Kentucky livestock auctions by an order buyer and housed at the C. Oran Little Beef Research Unit in Woodford County, KY. Steers comprised several mixed breeds dominated by British influence. Within 48 h of arrival cattle were weighed and ear tagged for individual identification, and temperament assessments were obtained for treatment assignment.

Three temperament assessments were obtained on each animal: 1) SCS (Grandin, 1993), 2) objective chute score, measured as the SD of weights recorded across a set time interval (WSD), and 3) EV (Burrow et al., 1988). Subjective chute score was completed by observing behavior in the chute (Silencer hydraulic squeeze chute, Moly Manufacturing, Lorraine, KS, hanging on load cells) for 10 s after the head was caught, with no squeeze applied, and was represented as the average score across 4 observers. The 1 to 5 scoring scale used was that of Grandin (1993): 1) calm, no movement, 2) restless shifting, 3) squirming, occasional shaking of device (squeeze chute or scale), 4) continuous vigorous movement and shaking of device, and 5) rearing, twisting, or violently struggling.

Measures for calculation of WSD were obtained during the same 10-s interval. A 10-s time interval was chosen on the basis of preliminary studies that indicated that this time interval was sufficient to delineate differences between animals while also accommodating the need to routinely process 192 animals in a reasonable amount of time. The scale head (Lynx, Mettler Toledo, Columbus, OH) was programmed to export weights at 5 Hz via RS232 to a laptop computer (Acer Aspire V5, Acer America, San Jose, CA). During this period, animals were restrained in the head catch but otherwise undisturbed, and WSD was determined as the SD of the 50 recorded weight values across this 10-s interval. At the end of this 10-s period, squeeze was applied. Each animal’s weight (for determination of ADG) was calculated as the average of weights measured at 200-ms intervals during the most stable 2-s interval subsequent to squeeze being applied.

Upon being released from the chute, flight time was measured over 1.68 m using an infrared sensor (FarmTek Inc., North Wylie, TX) and converted to EV (m/s).

Cattle received ear tags for individual identification on arrival, and subsequent processing (d 0) included viral and bacterial vaccinations (Bovi-Shield Gold 5, Zoetis, Florham Park, NJ; Once PMH, Merck Animal Health, Summit, NJ; Somubac, Zoetis; Ultrachoice 7, Zoetis) and an injection of anthelmintic (Dectomax, Zoetis). Cattle were reimmunized on d 14 with Ultrachoice 7 and Somubac.

Animals were assigned to treatments within weight blocks. Treatments included 2 levels of EV (fast or slow), 2 levels of WSD (high or low), and 3 levels of dietary protein supply (formulated to provide 75%, 105%, and 120% of NRC (2000) RDP requirements) arranged in a 2 × 2 × 3 factorial. Behavioral treatments were established on the basis of the EV and WSD measured for each animal during its first exposure to the handling facility (within 48 h of arrival for each of 4 groups of cattle; ranging from 2 to 15 d before allotment to treatment groups and initiation of experimental diets). Steers were assigned to pens such that each pen had 4 animals, all with similar behavioral treatment. The experimental unit was the pen, consisting of 48 pens total and 4 pens per treatment.

Because the 2 behavior-associated treatment factors were inherent characteristics of the animals (in contrast to exogenously applied treatments), the ability to divide the animals into these treatment groups depended on the independence of and the relationships between the distributions of these treatment variables. Thus, prior to developing an allotment strategy, the distributions of the 2 factors were evaluated using SAS JMP (SAS Inst. Inc., Cary, NC), and the selected allotment strategy (division into WSD groups preceding division into EV groups) was the one that provided the least overlap among treatments. Graphical comparison of the distributions of the treatment variables is provided in Fig. 1 and 2. Of 11 distribution types compared within SAS JMP, the Johnson SL distribution provided the best fits (based on Akaike’s information criterion) for both WSD and EV and was thus used for illustrative purposes in Fig. 1 and 2. Because animals were first divided into high and low WSD groups, there was essentially no overlap in the distribution of WSD for either of the high and low WSD treatment groups (Fig. 1). Additionally, across both EV groups, there was substantial similarity in the WSD distributions of each of the WSD groups. Thus, the allotment strategy provided for a clear delineation between high and low WSD treatments. Alternatively, EV allotments were determined subsequent to the establishment of the 2 WSD groups. Thus, EV delineations (Fig. 2) were not as distinct as those for WSD. However, within each level of WSD (high and low), there was essentially no overlap between the EV distributions of the fast and slow groups. A high degree of overlap existed between the slow EV, high WSD group and the fast EV, low WSD group. More importantly, the EV distributions differed somewhat for the high and low WSD groups. This situation would be expected to increase the likelihood of detecting interactions between EV and WSD and would need to be considered in any interpretation of interaction effects. However, the distributions depicted in Fig. 2 indicate sufficient delineation in EV between the fast and slow groups as a whole to suggest confidence in main effects attributed to this factor. Overall, this approach increased the power of our design for detecting differences between WSD treatments while compromising on the ability to detect EV effects. The degree of confounding would have been greater (greater overlap between distributions) had the alternate strategy been chosen.

Figure 1.
Figure 1.

Fitted Johnson SL distribution curves for weight standard deviation (WSD) for each treatment in the 2 × 2 factorial. These distributions were constructed from all animals (n = 192) that were assigned to the various treatments. These curves provide a graphical description of the degree of overlap in distribution of WSD among treatments. EV = exit velocity.

Figure 2.
Figure 2.

Fitted Johnson SL distribution curves for exit velocity (EV) for each treatment in the 2 × 2 factorial. These distributions were constructed from all animals (n = 192) that were assigned to the various treatments. These curves provide a graphical description of the degree of overlap in distribution of EV among treatments. WSD = weight standard deviation.


Steers were randomly assigned to pen and dietary treatment within weight blocks. Steers were housed, 4 to a group, in 2.44 × 14.63 m pens partially covered by a 3-sided, concrete floored barn. Each pen had 2.44 linear meters of bunk along the fence line (0.61-m bunk space per animal). Pens were scraped clean and bedded with sawdust routinely.

Dietary Treatments

Each pen group was fed a corn silage–based total mixed ration (Table 1) with 1 of 3 different RDP levels (denoted in Table 1 as RDP levels 1, 2, and 3) once daily at approximately 0700 h. Diets were prepared and adjusted daily to provide ad libitum intake with minimal amounts of feed refusals. To this end, feed bunks were observed twice daily at 0700 and 1500 h when the remaining amount of feed in the bunk of each pen was estimated to determine the next day’s feeding amount for each pen. Ingredient DM were determined once a week by drying samples for 24 h in a forced-air oven (100°C, model 1690, VWR Scientific Products, Cornelius, OR), and rations were adjusted accordingly. Steers had free access to water; adjacent pens shared a water source.

View Full Table | Close Full ViewTable 1.

Ingredient composition of experimental diets

Ingredient, % diet DM RDP level 11 RDP level 2 RDP level 3
Cool season grass hay 20.00 20.00 20.00
Switchgrass hay 20.00 20.00 20.00
Cracked corn 29.90 29.90 29.90
High-moisture corn 12.82 12.82 12.82
AminoPlus2 14.30 4.84 5.53
Soybean meal 0.00 9.38 8.15
Urea 0.00 0.08 0.62
Limestone 1.30 1.30 1.30
Potassium chloride 0.50 0.50 0.50
Trace mineral salt3 0.75 0.75 0.75
Vitamin ADE Premix4 0.05 0.05 0.05
Choice white grease 0.38 0.38 0.38
1Levels refer to the amount of RDP in the diet, nominally described as 75% (level 1), 105% (level 2), and 120% (level 3) of NRC (2000) RDP requirements.
2AminoPlus is a product of Ag Processing Inc. (AGP, Omaha, NE)
3Trace mineralized salt provided 92.9% salt, 68 mg/kg Co, 1,838 mg/kg Cu, 120 mg/kg I, 9,290 mg/kg Mn, 19 mg/kg Se, and 5,520 mg/kg Zn.
4Vitamin premix supplied 1,820 IU/kg vitamin A, 363 IU/kg vitamin D, and 227 IU/kg vitamin E.

The RDP treatments were established to represent 75%, 105%, and 120% of requirements, calculated according to NRC (2000), using a RDP requirement equivalent to 11% of TDN. Intakes were estimated at 2.8% of BW, and diets were formulated to be isocaloric and to meet metabolizable protein requirements for 1.36 kg/d ADG. Using RDP requirements of 13% of TDN, these treatments would have provided 62%, 89%, and 102% of animal requirements. Protein degradability values for forages were determined at a commercial laboratory (Dairy One, Ithaca, NY) using a Streptomyces griseus protease assay according to Coblentz et al. (1999). Degradability values for other protein containing feedstuffs were determined as described by Kenney et al. (2015).

Once a week, feed refusals were collected, weighed, recorded, and combined within treatments. Treatment composites were subsampled, and DM was determined (duplicate 250- to 500-g samples dried at 100°C for 24 h or until constant weight) and recorded. Any feed refusals on the floor outside of the bunk were weighed but not included in the orts sample for DM determination.

Animals were not withheld from feed or water prior to weighing, although weights were obtained prior to feeding. In addition to the initial measures used for treatment assignment (described above), animal weights, EV, and WSD were recorded on d 0, 1, 15, 29, 56, and 58. An average of the d 0 and 1 weights was used as the initial weight. Average weight from d 56 and 58 was used as the final BW.

Medication Protocol.

Steers were examined daily and treated for sickness if required. In order for animals to qualify for treatment, the animals must have displayed clinical signs (e.g., lethargy, emaciation, coughing, runny nose) and have had a rectal temperature exceeding 39.7°C. The treatment regimen for respiratory disease consisted of initial treatment with a single s.c. injection of Draxxin (2.5 mg/kg BW; Zoetis); a second treatment, if necessary, with a single s.c. injection of Nuflor (40 mg/kg BW; Merck Animal Health); and third treatment, if necessary, with a single s.c. injection of Baytril (7.5 mg/kg BW; Bayer HealthCare Animal Health Division, Shawnee Mission, KS). One animal was treated with PenJect (6 mL/100 kg BW; Butler Schein Animal Health, Dublin, OH) for an injured leg. Animals were rechecked on d 1, 2, and 5 subsequent to treatment. Second and third treatments were only warranted if animals failed to respond to the initial treatment per label instructions. Two animals were treated more than 3 times, 1 for a leg injury and 1 for coccidiosis. Data for the animal with coccidiosis were retained, as the animal responded to treatment. The animal with the leg injury did not recover and lost weight, and it and 8 other animals were removed for various health reasons, coupled with weight loss. Weight gain data for those animals were removed prior to statistical analysis, and the mean of the remaining animals in each pen was used, such that 12 pens (experimental units) per treatment were used. For intake data, which was acquired per pen, data for the entire pen were removed prior to analysis, resulting in 8 to 11 pens per temperament treatment for intake and efficiency data. Diagnosis of illness was based on clinical signs and at the discretion of the Beef Research Unit manager. No diagnostic testing was done.

Growth Performance

Performance measures were collected for 3 periods: d 0 to 29, 29 to 58, and 0 to 58. Dry matter intake was calculated for each pen by subtracting the orts from the total amount of feed offered during each period. Average daily gain was calculated for each animal as the total BW gain per period divided by the total number of days per period. Gain-to-feed ratio was calculated as BW gain divided by DMI.


Four classification categories were used for measures of morbidity. Animals were grouped according to 1) display of clinical signs of respiratory disease coupled with elevated rectal temperature (≥39.7°C), 2) display of clinical signs without elevated rectal temperature (pulls, not treated), 3) weight loss across a 29 to 30 d period not coupled with either of the above, or 4) other illness not included in the above categories. Clinical signs of respiratory disease included incidence of lethargy, unusual breathing, and quantity and character of nasal discharge.

Statistical Analysis

Distributions of Studentized residuals of each response variable were determined using JMP 10 (SAS Inst. Inc.) to confirm normality assumptions and to identify potential outliers. Normality assumptions were confirmed, and no outlier data points were found using a criterion of greater than 1.5 times the interquartile range below the first quartile or above the third quartile.

For descriptive purposes, WSD was regressed on SCS using JMP 10. For this descriptive analysis, the ordinal variable SCS was treated as a continuous variable. According to Tabachnick and Fidell (2013), this is an appropriate strategy when the variable can be considered to arise from a continuously distributed underlying scale, and there are more than 7 categories in the scale. As the mean of rankings on a 1 to 5 scale from 4 independent raters, our SCS contained 12 (of a potential 17) categories.

Interrater reliability of SCS was calculated using Krippendorff’s α (Hayes and Krippendorff, 2007). This statistic was used as a reliability index because it can be used with any number of observers, accommodates missing data, and satisfies all of the important criteria for a good measure of reliability (Hayes and Krippendorff, 2007). Krippendorff’s α values and confidence intervals were determined using a SAS macro ( The measurement type was specified as “ordinal,” and a bootstrap sample size of 10,000 was used for confidence interval construction.

All performance data (ADG, DMI, G:F) were analyzed with the GLM procedure of SAS (SAS Inst. Inc.) using pen as the experimental unit. Initially, data were analyzed using a model that included main effects of EV, WSD, RDP, all interactions among these 3, and block. However, there were no main effects of RDP (P ≥ 0.15) or any interactions with RDP (P ≥ 0.12) for any of the response variables. Thus, for analysis of behavioral treatments, RDP was removed from the model. The model statement for analysis of morbidity contained only the effects of behavioral treatment using CATMOD procedures of SAS.

Time effects on EV and WSD were analyzed using the mixed procedure of SAS allowing for repeated measures analysis. The model statement included EV, WSD, day, and their interactions. Denominator degrees of freedom were calculated using the Kenward-Roger method. Day was specified as the repeated term, and a first-order autoregressive structure was used for the error variance/covariance matrix. Time effects were characterized using linear, quadratic, and cubic contrasts. Main effects and interactions were considered significant at P < 0.10.


Main effects of dietary RDP treatments are presented in Table 2. Level of RDP in the diet did not influence (P ≥ 0.15) any of the response variables in this study. Likewise, no interactions between RDP, EV, and WSD were detected (P ≥ 0.12) for the variables analyzed and reported here.

View Full Table | Close Full ViewTable 2.

Effect of RDP treatments on ADG, DMI, and G:F conversion1

RDP level2
Item 1 2 3 SEM3 P-value
Initial wt, kg 244 244 242 1.2
ADG, kg/d
    d 0 to 29 1.02 1.03 0.96 0.045 0.57
    D 29 to 58 1.54 1.53 1.55 0.051 0.97
    d 0 to 58 1.27 1.28 1.26 0.032 0.85
DMI, kg/d
    d 0 to 29 6.70 6.94 6.65 0.146 0.34
    D 29 to 58 7.93 8.16 7.73 0.156 0.18
    d 0 to 58 7.32 7.55 7.19 0.131 0.17
    d 0 to 29 2.60 2.70 2.61 0.053 0.38
    D 29 to 58 2.70 2.79 2.67 0.045 0.20
    d 0 to 58 2.62 2.71 2.60 0.039 0.15
    d 0 to 29 0.150 0.148 0.143 0.0055 0.61
    D 29 to 58 0.199 0.191 0.199 0.0056 0.51
    d 0 to 58 0.177 0.171 0.173 0.0035 0.50
1There were no 2- or 3-way interactions involving RDP level (P ≥ 0.12).
2Levels refer to the amount of RDP in the diet, nominally described as 75% (level 1), 105% (level 2), and 120% (level 3) of NRC (2000) RDP requirements.
3Here n = 16 for initial weight and ADG; n = 14, 13, and 13 for RDP levels 1, 2, and 3 for intake and efficiency variables. With unequal n among treatments, SEM is the average of the values for the 3 treatments.

Repeatability among raters for SCS (Krippendorff’s α) was 0.75 (95% CI, 0.70 to 0.80), where an α value of 0 indicates lack of repeatability and a value of 1 corresponds to perfect agreement among raters. Measures of WSD and SCS, collected simultaneously at the time of first processing, were positively, although weakly, correlated (P < 0.01, R2 = 0.247; Fig. 3).

Figure 3.
Figure 3.

Linear regression analysis of initial weight standard deviation (WSD) on subjective chute score (SCS) at d 0. WSD = (1.12 ± 0.499) + (1.90 ± 0.239) × SCS, P < 0.01, R2 = 0.247, n = 192.


There were no interactions between EV and WSD for ADG, DMI, or G:F (P ≥ 0.11; Table 3). Average daily gain was greater in slow compared with fast EV animals for d 0 to 29 (P < 0.01) and d 0 to 58 (P = 0.02), although ADG was not influenced (P = 0.91) by EV during the second 29 d of the study. Dry matter intake, both in absolute terms and as a percentage of BW, was greater in slow EV animals for both 29-d periods and across the entire study (P ≤ 0.09). There were no significant effects of EV on G:F (P ≥ 0.14).

View Full Table | Close Full ViewTable 3.

Effect of behavioral treatments on ADG, DMI, and feed conversion efficiency

Item Slow Fast Low High SEM3 EV × WSD EV WSD
EV,4 m/s 1.74 3.56 0.081
WSD,4 % 1.17 2.79 0.001
Initial wt, kg 245 241 240 245 1.0
ADG, kg/d
    d 0 to 29 1.09 0.91 0.97 1.04 0.035 0.89 <0.01 0.13
    D 29 to 58 1.54 1.54 1.53 1.54 0.039 0.34 0.91 0.86
    d 0 to 58 1.31 1.23 1.25 1.29 0.025 0.40 0.02 0.23
DMI, kg/d
    d 0 to 29 7.0 6.5 6.6 6.9 0.12 0.14 0.01 0.06
    D 29 to 58 8.2 7.6 7.7 8.2 0.12 0.30 <0.01 0.01
    d 0 to 58 7.6 7.1 7.1 7.5 0.11 0.15 <0.01 0.01
    d 0 to 29 2.68 2.58 2.60 2.66 0.043 0.11 0.09 0.34
    D 29 to 58 2.78 2.65 2.67 2.76 0.034 0.27 0.01 0.08
    d 0 to 58 2.70 2.58 2.60 2.68 0.032 0.11 0.01 0.09
    d 0 to 29 0.153 0.143 0.148 0.148 0.0046 0.64 0.14 1.00
    D 29 to 58 0.193 0.201 0.203 0.191 0.0041 0.95 0.16 0.05
    d 0 to 58 0.175 0.175 0.178 0.171 0.0028 0.61 0.99 0.11
1Exit velocity (EV) is the time taken for steers to travel 1.68 m on exiting the chute; assignment to slow and fast treatments is described in text.
2Weight standard deviation (WSD) is the SD of 50 recorded weight values across a 10-s interval while the animal was restrained by the head in a chute; assignment to low and high treatments is described in text.
3Here n = 24 for initial weight and ADG; n = 18, 22, 21, and 19 for slow EV, fast EV, low WSD, and high WSD for intake and efficiency variables. With unequal n among treatments, SEM is the average of the values for the 4 treatments.
4Average initial (d 0) measures of EV and WSD by treatment.

There was no effect of WSD on ADG (P ≥ 0.13). High WSD animals had a higher absolute DMI than low WSD animals for all periods (P ≤ 0.06). As a percentage of BW, high WSD animals had higher DMI than low WSD animals during d 29 to 58 (P = 0.08) and d 0 to 58 (P = 0.09). Low WSD animals had a higher G:F than high WSD animals during d 29 to 58 (P = 0.05) along with a similar trend for d 0 to 58 (P = 0.11).

There were no interactions between EV and WSD (P ≥ 0.28) or significant effects of EV on morbidity (P ≥ 0.12; Table 4). Compared with low WSD animals, more high WSD animals were pulled from pens for displaying clinical signs of illness, without a corresponding increase in rectal temperatures (P = 0.06).

View Full Table | Close Full ViewTable 4.

Effect of behavioral treatments on morbidity

Item Slow Fast Low High EV × WSD EV WSD
Respiratory3 13 19 17 15 0.47 0.25 0.70
Pulls, not treated4 5 3 1 7 0.56 0.49 0.06
Weight loss5 6 0 1 5 0.72 0.12 0.13
Other6 5 1 4 2 0.28 0.14 0.39
1Exit velocity (EV) is the time taken for steers to travel 1.68 m on exiting the chute; assignment to slow and fast treatments is described in the text.
2Weight standard deviation (WSD) is the SD of 50 recorded weight values across a 10-s interval while the animal is restrained by the head in a chute; assignment to low and high treatments is described in the text.
3Number of animals displaying respiratory illness as described in the Materials and Methods section.
4Number of animals that were displaying clinical signs of illness without a corresponding increase in rectal temperature and thus were not treated.
5Number of animals that were observed to be losing weight but were not displaying clinical signs of illness.
6Number of animals that were treated for illness other than respiratory (leg injury, coccidiosis, etc.).

Exit velocity and WSD means within treatment groups changed over time (Fig. 4 and 5). There was a significant day × EV interaction (P < 0.01) and day × WSD interaction (P < 0.01). For the slow EV treatment, change in EV across time was characterized by linear (P < 0.01) and cubic (P < 0.01) effects; time effects for the fast EV treatment were characterized by linear (P < 0.01), quadratic (P < 0.01), and cubic (P < 0.01) effects. However, EV for the slow EV treatment groups remained below (P < 0.01) that of the fast EV groups at all measurement times. Change in WSD across time for the low WSD group were characterized by a quadratic (P < 0.01) effect, whereas time effects for the high WSD treatment were characterized by linear (P < 0.01), quadratic (P < 0.05), and cubic (P < 0.01) effects. At all measurement times except for d 58, WSD for the low WSD group was less (P < 0.05) than that for the high WSD group.

Figure 4.
Figure 4.

Least squares means of exit velocity (EV) on each day of temperament assessment (d −8, 0, 1, 15, 29, 56, and 58) by EV treatment (slow or fast). Treatment was based on initial measures of EV (average = d −8). The initial average EV for the slow treatment was 1.74, and initial average EV for the fast treatment was 3.56 (n = 192). There was a significant EV by day interaction (P < 0.01). Points with different letters on a given day are significantly different (P < 0.01). The time effects of the slow EV treatment were characterized by linear (P < 0.01) and cubic (P < 0.01) effects. The time effects of the fast EV treatment were characterized by linear (P < 0.01), quadratic (P < 0.01), and cubic (P < 0.01) effects.

Figure 5.
Figure 5.

Least squares means of weight standard deviation (WSD) on each day of temperament assessment (d −8, 0, 1, 15, 29, 56, and 58) by WSD treatment (low or high). Treatment was based on initial measures of WSD (average = d −8). The initial average WSD for the low treatment was 2.84 kg, and initial average WSD for the high treatment was 6.91 kg (n = 192). There was a significant WSD by day interaction (P < 0.01). Points with different letters on a given day are significantly different (P < 0.05). The time effects of the low WSD treatment were characterized by a quadratic (P < 0.05) effect. The time effects of the high WSD treatment were characterized by linear (P < 0.01), quadratic (P < 0.05), and cubic (P < 0.01) effects.



There were 2 reasons RDP was chosen as the dietary treatment to be analyzed with temperament in this study. The first was to evaluate whether differences exist in nutrient requirements between temperament groups, where temperament was defined by EV and WSD. Previous research has found that animals with fast EV typically have lower ADG and DMI than slow EV animals (Cafe et al., 2011b; Bates et al., 2014). Sufficient dietary RDP is important to help minimize intake depression during the stress of the receiving period. If the intake of temperamental animals is further depressed through accentuated stress responses, it is possible that higher than normal levels of RDP could help alleviate some of that intake depression. The second was that differences in rumen microbial populations might be linked to animal temperament. There is a growing body of evidence that gut microbes play an important role in animal behavior in general (e.g., Bercik et al., 2011). In ruminants, Naglaa and Ghada (2014) found that administering a Saccharomyces cerevisiae probiotic to ewes for 1 mo altered SCS, indicating that behavior of ruminants may be related to gut flora. Differences in gut microbial populations could insinuate differences in microbial protein requirements and thus in RDP requirements. However, no significant effects of RDP were detected in the present study. In preliminary studies from our laboratory (unpublished), we found differences in growth performance with similar cattle receiving similar dietary treatments. However, in Kenney et al. (2015), as in the present study, we did not detect differences in growth response to similar differences in RDP in the absence of an added direct-fed microbial. Reasons for differences in response to RDP among our studies are unclear. Regardless, the lack of any significant effects of RDP in the present study indicates that our RDP treatments did not provide an adequate test of our hypothesis. Thus, we cannot draw any conclusions from these data regarding potential relationships between RDP provision and temperament.

Exit velocity has commonly been used as a measure of temperament and has been suggested to be indicative of physiological stress related to animals’ encounters with humans (Curley et al., 2006). The general nature of the physiology underlying the relationship between temperament and growth has been described. Cortisol concentrations have been positively correlated to EV (Curley et al., 2006; Cafe et al., 2011a) and SCS (Cafe et al., 2011a). Carroll and Forsberg (2007) summarized research linking increased glucocorticoids to suboptimal growth in livestock. Differences in performance between EV treatments are thought to be due to the elevated stress response in high EV animals, which is associated with increased stress hormone concentrations and decreases in activities such as eating and sleeping (Burdick et al., 2011; Cafe et al., 2011a), coupled with compromised immune function (Fell et al., 1999).

Differences in ADG between fast and slow EV were confined to the first 29 d, even though slow EV animals had higher intake across d 0 to 29 and 29 to 58. This may indicate an attenuation of EV effects on ADG across time. However, it is also possible the difference in significance between the 2 periods is a consequence of end point errors associated with differences in gut fill. Although steps were taken to minimize differences in shrink among periods, such differences can comprise a substantial part of gain over short periods of time. More important are the effects across the entire study period, indicating that slow EV animals had a higher ADG than fast EV animals, which is consistent with previous research (Petherick et al., 2002; Cafe et al., 2011b; Bates et al., 2014). The present results build on the existing literature, which has predominantly focused on finishing cattle. In the present study, we have demonstrated that EV can also be related to gain in receiving animals consuming diets with comparatively high forage:concentrate ratios. Müller and von Keyserlingk (2006) also demonstrated a relationship between EV and growth of animals on a forage-based diet across a similar length of time. In their study, animals were group housed, and regression of ADG on EV indicated a quadratic relationship, accounting for 14% of the variation in ADG. The present study indicates that although the broad nature of this relationship may account for a relatively small proportion of variation in gain, separation of animals based on temperament can result in management groups with biologically meaningful differences in growth rate. This has potential implications for increasing uniformity of growth rates of cattle within a given pen and for differential management of such groups.

In addition to the potential for managing cattle according to temperament, our experimental approach has other important implications. First, it is possible that at least part of the relationship between temperament and gain could be a consequence of social interactions among animals (MacKay et al., 2013). For example, animals exhibiting different EV could be more or less aggressive at the feed bunk than their pen mates. By grouping like animals together, we have minimized opportunities for such social factors to operate among animals in different temperament categories. Although not ruling out the possibility of social interactions as a temperament-related factor, the present results suggest that if they do play a role, such a role is not necessary for the expression of ADG differences between temperament groups. Second, grouping animals in this manner had a direct influence on the type of statistical analysis that was appropriate. Specifically, this grouping of animals called for use of ANOVA techniques, as opposed to the variety of regression approaches that have been used in many studies. Discretization of continuous variables for analysis unfortunately appears to be a somewhat common occurrence in much of the temperament literature and should be discouraged (Streiner, 2002). However, in the present study, EV was not discretized for the purpose of data analysis. Rather, we were evaluating a management system based on separating animals into temperament categories. Thus, experimental groups were necessarily established by dichotomizing EV and WSD, thus making them classification variables appropriate for use in ANOVA. Third, our approach necessitated that animal temperament was categorized on the basis of initial measures, prior to the experimental period. Although some researchers have indicated that EV and other measures of temperament are more robust when averaged across several times during an experiment (e.g., Cafe et al., 2011b), in practice, if animals are to be managed in temperament groups, the decision must be made prior to the feeding period. Thus, we chose to evaluate a system in which temperament categories were determined from a single measurement, obtained shortly after arrival of the animals at the facility. Ultimately, an important observation from our experimental approach was that even when groups were established in such a fashion, effects of EV on ADG were still apparent.

The differences in DMI, as an absolute value and as a percentage of BW, between fast and slow EV during both periods were consistent with previous studies where slow EV animals had higher DMI than fast EV animals (Nkrumah et al., 2007; Cafe et al., 2011b; MacKay et al., 2013). Alteration in time spent feeding has been proposed as the cause for differences in intakes between animals with different EV (MacKay et al., 2013). However, alterations in feeding behavior are possibly the result, rather than the cause, of differences in intake. The stress response has been well documented. Animals with a more reactive stress response have increased concentrations of neurohormones and glucocorticoids, such as epinephrine and circulating cortisol, released by stimulation of the HPA axis and the sympathetic nervous system (Burdick et al., 2011). Stimulation of these stress-related hormones can inhibit feed intake (Burdick et al., 2011); therefore, it seems likely that intake is directly affected by these endocrine actions and that reductions in feeding time are a consequence of a reduced intake. Relationships among simple measures of behavior, like EV, social or feeding behaviors, intake, and gain are worthy of further study.

The cause and effect relationship between temperament measures and growth performance is not yet clear. Some researchers have speculated that differences in the stress response of excitable animals imply that these animals are less efficient in maintaining and gaining weight (Petherick et al., 2003), although there were no differences in G:F between EV treatments in the present study. Nkrumah et al. (2007) and Fox et al. (2004) found phenotypic correlations between EV and DMI, but not with measures of efficiency. Alternatively, Cafe et al. (2011b) found that temperament was related to DMI and time spent eating, with lesser effects on efficiency of feed utilization. As argued above, it seems likely that depressions in time spent feeding are a consequence, rather than a driver, of the lower intakes experienced by high EV animals that are caused by activation of the HPA axis (Burdick et al., 2011). However, MacKay et al. (2013) suggested that differences in intake between temperament levels are due to differences in activity and home pen behavior such that animals with a fast EV might have higher levels of extraneous activity and spend less time eating. Results of the present study add additional verification that differences in gain between EV treatments derive heavily from differences in intake. Future research aimed at better understanding the mechanistic relationships between behavior and intake could help establish management strategies to fully capitalize on behavior-related effects.

The tendency for higher gains during the first 29 d with high WSD animals and the positive relationship between WSD and DMI contrast with previous studies, which have used SCS and have generally reported negative correlations between chute score and growth (Hoppe et al., 2010; Cafe et al., 2011b; Sebastian et al., 2011). However, Graham et al. (2001) found a positive relationship between SCS and ADG in 2 out of 4 sire groups (with no effect of SCS on ADG in the other 2). Studies evaluating the relationship between SCS and intake are quite limited. Cafe et al. (2011b) found a negative relationship between DMI and SCS, whereas MacKay et al. (2013) found no relationship between these measures. In the present study, initial SCS and WSD were weakly, but positively, correlated. The positive correlation was not surprising as these measures were obtained simultaneously and both were intended to measure level of agitation associated with confinement in the squeeze chute. However, objective and subjective measures do not measure precisely the same things. In this case, the objective measure is an integrated measure (across 10 s) of the magnitude of the changes in force on the load cells as induced by the animal’s movements. It does not account for factors other than body movement (e.g., vocalization, exposed eye white, etc.) that are accounted for in subjective measures. Despite the moderate degree of agreement among raters for SCS (Krippendorff’s α of 0.75), there is substantial variation in this measure simply because it is subjective. Thus, part of the low correlation between WSD and SCS is a consequence of random variation in SCS. Nevertheless, because WSD and SCS are positively related, even if weakly, the positive relationship between WSD and DMI in the present study is not likely due to some unique attribute of WSD compared with SCS. It is possible that in the present study, where chute scores were obtained at initial exposure to the facilities, animals with a higher stress response displayed freezing as opposed to aggressive behavior in response to being restrained in the squeeze chute (Burrow and Corbet, 2000). Thus, we suggest that the present results are consistent with earlier work showing relationships between chute scores and intake in that intake is lower in more temperamental animals. However, we further suggest that the conditions under which objective chute scores are obtained may influence whether “high” scores are associated with calm or excitable temperaments.

Importantly, others have reported on objective methods for assessing chute scores. Specifically, Sebastian et al. (2011) evaluated various objective measures of animal movement while animals were confined in a squeeze chute. Their measures included force exerted on the head gate, measured through strain gauges, and animal movement, using a movement measurement device. Our approach was most similar to their use of the movement measuring device, although our approach differed from theirs in that we obtained an integrated measure (i.e., the SD) of the magnitude of recorded weight changes at a frequency of 5 Hz, whereas they focused on the number of fluctuations in direction of voltage changes, collected at a frequency of 122 Hz. There is little reason to suspect that these measures are influenced similarly by animal activity, and continued evaluation of these and other approaches to objectively quantify animal movement is warranted. Nonetheless, both Sebastian et al. (2011) and the present work demonstrated general agreement between objective chute score and SCS measures.

Although DMI was greater for high WSD animals, low WSD animals were more efficient during the last 28 d and tended to be during the entire study period, such that ADG did not differ between these groups. The lack of a significant relationship between WSD and ADG seen in this study is consistent with some previous findings with SCS (Francisco et al., 2012) but inconsistent with others in which cattle with a high SCS had lower ADG (Voisinet et al., 1997; Reinhardt et al., 2009; Cafe et al., 2011b). This inconsistency points to a need for a more mechanistic understanding of the relationship between chute scores and ADG. Use of an objective chute score may facilitate standardization in such evaluations but may not be related to production measures in the same way as a subjective score.

Weight SD was not significantly related to incidence of respiratory disease in cattle, but it was related to the number of cattle that were pulled from their pens for displaying clinical signs of illness without a corresponding increase in rectal temperature, where high WSD animals were pulled more often than low WSD animals. Results suggest that high WSD animals may show more symptoms of illness than low WSD animals. Previous relationships between temperament and external display of illness have been reported. For example, Hulbert et al. (2011) suggested that temperamental animals are more difficult to treat for illness because they do not show clinical signs, but that study used an average of EV and SCS to define temperament scores. Thus, it is unclear how these measures might have related individually with observation of sickness.

Cattle can habituate to handling over time with repeated handling events (Burrow and Dillon, 1997; Cafe et al., 2011b), which means that their temperament measures should change with time. With the exception of substantial changes on the second day of consecutive day handling, the largest changes across time for EV were for the fast EV group between initial processing and the start of the experiment. Likewise, with the additional exception of an anomalous point for the low WSD group on d 1 of the experiment, the largest change in WSD occurred just subsequent to the first time the animals were processed at this facility. These observations are in agreement with previous work (Curley et al., 2006; Cafe et al., 2011b) and should be expected in that more excitable animals would be likely to calm somewhat with repeated handling, whereas calmer animals would experience little change. Importantly, despite some convergence after the initial measurements, both EV groups and both WSD groups remained distinct for those respective measures for the duration of the study. Perhaps not surprisingly, measurements obtained on the second day of consecutive processing are of questionable value. Although the length of time between handling events to avoid such issues remains unclear, these results suggest that biweekly processing can be accommodated without causing convergence of measures for discrete groups. Others have observed different patterns in temperament measures across time (Petherick et al., 2002; Petherick et al., 2003). Various patterns should be expected, as there are likely a number of external factors (climatological, etc.) that would be expected to influence average animal responses on a given measurement day. The important consideration is that general rankings remain intact across time, and the present data indicate that for short-term (i.e., about 2 mo) growing studies, this is the case for both EV and WSD.


The results of this study give additional confirmation of a relationship between EV and growth performance and extend the existing knowledge by demonstrating these effects in receiving cattle sorted by temperament classification. Additionally, we have demonstrated the use of a novel objective chute score (WSD), which overcomes concerns of the subjectivity generally associated with chute scores. This objective score did not interact with EV for any response variable and was positively, but weakly, correlated with subjective scores, indicating that the behavior measured by this approach is not identical to that determined when using SCS. Furthermore, in this study, animals with high WSD had characteristics often associated with less temperamental animals (higher intakes, greater display of sickness, and tendency for higher gains during the first 30 d of the study). Although EV relationships appear to be generally consistent across the literature (higher EV associated with lower ADG), it appears that the relationship between chute scores and production characteristics may be more complex than is often assumed. At a minimum, researchers and practitioners should take care to use EV and chute scores as independent measures and should not presume that they measure similar characteristics.




Be the first to comment.

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