Methods
Experimental Design
Using a cross-over replicated Latin square design controlling for ‘day’ and ‘animal’ effects, 13 dry cows and 5 heifers were each subjected to 3 treatments: positive (POS; large pen, deep straw bedding, brush access, fence-line companion heifer), neutral (NEU; midsized pen, minimal bedding, visual of companion heifer), and negative (NEG; small pen, no bedding, no visual contact with other animals, pail of inaccessible pellets). Animals were moved from their group-housed home pens to individual treatments pens, where treatments were applied in random order, with one treatment applied each day for 45 min and 1 washout day between treatments. During each treatment period, we collected heart rate (HR), vocalizations, and continuous video data. Animals were then moved to a test pen, passing by a thermal camera before completing a novel object test (pool toy, hamper, airplane) and attention bias test (dog vs. pellets) with continuous RGB video recording. A single trained observer who was blind to treatments performed a QBA and assessed NO and AB reactivity scores in the test pen.
Treatments
Tarp
Gate
Feed/ Water
Bucket of inaccessible pellets
Positive Pen
Neutral Pen
Negative Pen
Figure 3. Treatment pens and test arena layout
Positive Treatment
Neutral Treatment
Negative Treatment
The positive treatment pen was the largest, had deep straw bedding, fence line contact with a companion heifer (allowing for physical contact), brush access, and nutritional enrichment (natural vegetation during summer/fall months; salt lick during winter months). Free access to feed and water.
The neutral treatment pen was mid-sized with some straw bedding, visual contact with the companion heifer (not allowing physical contact), no brush or nutritional enrichment. Free access to feed and water.
The negative treatment pen was the smallest, had no bedding, and allowed no visual contact with other animals (tarps were placed to isolate the pen and obstruct the view). There was free access to feed and water as well as a bucket of inaccessible pellets (frustration bucket).
Data Collection
Heartrate monitor band
Image of cow wearing heart rate monitor​
Image of FLIR camera (summer)
Image of FLIR camera (winter)
Image of airplane (novel object)
In Treatment Pen
-
Heart rate
-
Vocalizations​
Between Treatment Pen and Test Arena
-
Thermal video recording
In Test Arena
-
Qualitative Behaviour assessment
-
Reactivity Scores (of behaviour tests)
Before entering the treatment pen, the animal was outfitted with a heart rate monitor (pictured) which recorded continuously while in the treatment pen. Vocalizations were recorded manually from live observations.
While being moved from the treatment pen to the test arena, animals passed by a FLIR camera (pictured) which recorded thermal video. Temperature data, of the animal's eye area, were extracted from the video by a trained observer blind to treatments.
Behaviour and reactivity scores were assessed by a trained observer who was blind to treatments. Possible reactivity scores were:
0 - approaches stimuli
1 - no reaction
2 - stares
3 - moves away (stays in arena)
4 - leaves arena to get away ​
The novel object was different each day to maintain novelty. The novel objects included a hamper, a pool float, and an airplane (pictured). The attention bias test was the same each day. The test always started once the animal was eating pellets from the designated bucket and ended within 20 seconds of the dog appearing and a barking sound being played over a speaker.
​
​Behaviour was scored based on video predetermined descriptors and rated between 0 - 100 (e.g. 12/100 for calm would mean the animal is not very calm whereas, 86/100 for calm would mean the animal is quite calm).
Predictor Variables
-
Treatment
-
Cow
-
Group
-
Day
Response Variables
-
Heart rate
-
Number of Vocalizations
-
Eye-area temperature
-
Reactivity scores
-
​Qualitative behaviour assessment scores
Analysis Type
-
Multivariate
Controlled for during study design by randomization and included in the model for statistical analyses.
Data were collected throughout the experiment by a trained observer and specific measurement devices (i.e. video cameras and heart rate monitors).
Discriminant analysis to maximize variance between treatment groups.