Dairy Summit highlights on-farm research

ARLINGTON, Wis. — The Dairy Innovation Hub held its fifth annual Dairy Summit Nov. 20 on the University of Wisconsin-Madison campus. Through poster sessions, research presentations and panel discussions, the summit highlighted examples of the latest hub-funded research taking place at UW-Madison, UW-Platteville and UW-River Falls to help meet the challenges facing today’s dairy community.

Approximately 200 legislators, researchers, campus affiliates, students, dairy professionals and farmers attended. In addition, 100 people attended virtually. University researchers, graduate students and farmers presented preliminary findings on nine research projects during the opening panel and research introductions.

After lunch, the summit transitioned to two tour opportunities showcasing hub investments at the Arlington Agricultural Research Station and in UW–Madison campus research facilities.

At Arlington, visitors experienced several tour stops, starting with the Emmons Blaine Dairy Cattle Center. Here, researchers discussed studies they are conducting using artificial intelligence and barn camera footage to determine body condition and locomotion scores; specialized gates for measuring feed intake and feed efficiency; and equipment to measure me-thane emissions from cows’ breath.

In the first barn, Joao Dorea explained how artificial intelligence can provide real-time animal monitoring through a computer vision system. More than 100 cameras are installed in the barn, some of which measure the body shape and body weight of cows to detect disease.

One project is a study of the transition period, with a focus on metabolic diseases. The majority of metabolic diseases in dairy cows occur during the transition period, which spans from three weeks before calving to three weeks after calving, Dorea said.

“One way to monitor for potential health problems during this period is to look to the body condition score,” he said. “The severity of negative energy balance can increase the risk of various peripartum disorders, including ketosis, hypocalcemia, retained placenta, metritis, endometritis and displaced abomasum.”

Dorea said economic losses for these diseases can be substantial, as investments in first-lactation animals may not be fully recouped if the animals are culled, in addition to treatment costs. 

Body condition score is a subjective measurement on a 5-point scale that is difficult to measure consistently and systematically in large dairy operations, Dorea said.

“The problem with subjectively assigning a score is that people may not have time on a dairy to watch cows and score them for body condition,” he said. “The other problem with this method is the inability to physically see the subtle changes in body shape. This is what we dis-covered when using cameras.”

The team collects data 21 days, 14 days and seven days before calving to determine if there is a change in body shape. The cameras map the whole shape of the cow, with images providing a 3-D map of the animal from above. For example, a cow that was visually scored a 4 at 21 and 14 days before calving revealed differences in shape on the computer image, illustrating the significance of the technology.

“The difference was minimal and not possible to observe through human vision, but it may be relevant to detect a cow that might get sick,” Dorea said.

In his experiment, Dorea utilized pre-partum data to detect sub-clinical ketosis in dairy cows at least seven days before its onset. 

“It would be helpful if we could accurately classify cows so we would know at least seven days before calving which cows would and would not have sub-clinical ketosis after calving,” Dorea said.

Cameras also track feed bunk behaviors. They compute time spent eating, the number of visits and the time between visits.

“We take this information and combine it with the body shape to make predictions,” Dorea said. “Every time the animal accesses feed, we know they are there eating, and we compute that behavior and use the info to detect diseases.”

The barn also includes cameras that measure mobility, focusing on walking speed, stride length and duration, and head movement. All angles of the animal are examined to as-sign mobility scores and detect animals with problems. How the cow’s spine behaves as she walks is also analyzed.

“Maybe a cow that is healthy will walk straight, while a cow that is not healthy will have rump movement and will not walk straight,” Dorea said. “We can capture that. And that’s what we use to classify locomotion scores very accurately.”

In another barn, feed efficiency trials are being conducted on one pen using specialized feeding equipment and gates. Data is generated from 64 cows that eat from 32 feeders.

Luiz Ferraretto and Francisco Peñagaricano explained how the equipment monitors individual intake and feeding behaviors. Ferraretto said they can control which cow goes through which feeding gate, and the system records which cow ate from which gate and how much she ate.

“Every time a cow enters the system, it’s recording,” he said. “We can collect the data and start calculating intake. We care about intake because there is a linear relationship between intake and production. We want cows to produce more milk with lower intake for better efficiency.”

Their study is also looking at how different diets and ingredients affect emissions as well as energy and metabolism based on body weight.

“We’re trying to select cows that produce the same amount of milk and consume less feed,” Peñagaricano said. “When a farmer produces more milk but uses less feed, it affects the profitability of the farm and will also affect the environmental impact of that farm.”

Peñagaricano said a new challenge they are working on is to reduce methane emissions.

“As a geneticist, my goal is to reduce methane emissions through breeding,” he said. “To do that, we first need to measure emissions, which we are doing with the GreenFeed system.”

The cow puts her head inside the gate which tracks air and measures methane emissions, including carbon dioxide, oxygen and hydrogen. The farm is running 8-week trials to attain adequate measurements for each cow.

“We need her to stay in at least two minutes to have a good record,” Peñagaricano said. “She can stay in for 4-5 minutes six times a day and must wait two hours in between visits. Methane emissions change during the day, and we want to make sure we are measuring her throughout the day and getting precise data measurement.”

The system reports that cows are producing an average of 450 grams of methane per day. There are also cows producing more than 600 grams and less than 300 grams.

“It’s a beautiful variation,” Peñagaricano said. “Twenty-five percent of that variation is due to genetic factors. We are confident that in a couple of years, we will have genetic evaluations for methane emissions. We need to introduce methane as an economic selection.”

Peñagaricano and Ferraretto are finding that feed efficiency and methane are favorably related.

“Cows that are more feed efficient tend to emit less methane,” Peñagaricano said. “We already have feed efficiency into the economic selection index. If we get methane too, we could push both traits in the same direction at the same time.”