From innovation to sustainability: precision methods transform livestock farming

Cutting-edge technologies and systems promise to contribute to more sustainable practices but need academics, scientists, processors and farmers to join forces, according to speakers at the São Paulo School of Advanced Science held on the Jaboticabal campus of São Paulo State University.

Agência FAPESP* – The use of advanced technologies to monitor and manage livestock production more efficiently and sustainably is higher than ever on the wish list of cattle, pig, poultry and fish farmers. With tools such as sensors, cameras, artificial intelligence and data analysis, it is possible to obtain information in real time on the livestock’s behavior, health, genetics and nutrition, optimizing productivity and reducing environmental impact.

Precision livestock farming (PLF), a broad term referring to all kinds of novel technologies for automated monitoring of animal production, is rapidly becoming a fundamental driver of change in the agricultural sector, offering opportunities for various segments to benefit but requiring close collaboration among researchers, breeders and processors, according to the speakers at the São Paulo School of Advanced Science on Precision Livestock Farming.

The ten-day event was organized by São Paulo State University (UNESP) via the Graduate Program in Animal Science of its School of Agrarian and Veterinary Sciences (FCAV) and the Graduate Program in Aquaculture of its Center for Aquaculture (CAUNESP).

The event was funded by FAPESP and took place in October on UNESP’s Jaboticabal campus. It was attended by 122 graduate students and early-career researchers from 14 countries. “The aim was to foster strong interest in research on PLF, train young scientists to play a global role in the advancement of science, and reduce the distance between basic and applied research in the field of PLF,” said Luciano Hauschild, a professor at UNESP and coordinator of the event.

“As one of the world’s leading producers of beef, pork and chicken, Brazil is strategically positioned to adopt PLF. Some of its largest farms are already implementing PLF solutions for animal health monitoring, feed control and meat traceability,” Hauschild added.

According to the United States Department of Agriculture (USDA), Brazil and Australia are the leading exporters of beef. Global beef exports are expected to total 11.9 million tons in 2024, for annual growth of 1%. Brazil is the leading chicken exporter, shipping to 172 countries, and the third-largest producer of poultry, as well as being the fourth-largest producer and exporter of pork, with exports of about 590,000 tons in the first half of this year, up 2% compared with first-half 2023. Santa Catarina is the Brazilian state with the most pork exports.

Novel technologies for PLF can be advanced only if academics, scientists, industrial firms and farmers join forces, Tomas Norton (Catholic University of Leuven, Belgium) said in his plenary lecture opening the second week. “Together we can develop effective models and integrate knowledge for practical application. This development requires collaboration among researchers, producers and key players such as breeders," Norton stressed. “We’re approaching a future in which the digital and physical worlds will be even more integrated. As a result, we can create innovative and effective solutions for the agricultural sector.”

Monitoring and decision-making support

Artificial intelligence (AI) is already in widespread use for decision-making support in farm management, with the aim of increasing the efficiency of production, reducing costs, and promoting more sustainable agricultural practices, including programs to safeguard animal welfare.

An example is the use of a machine learning algorithm to diagnose laminitis, a painful condition involving inflammation of the sensitive tissue (laminae) in the hoof, hindering mobility and potentially causing permanent damage to the foot. “Computer vision software has detected 25 points on the cow’s body for the purpose of measuring variables such as stride duration and length, gait speed and rump angle,” said João Dorea (University of Wisconsin-Madison, USA).

Another application of PLF technology is in monitoring of animal behavior. Shuwen Hu (Commonwealth Scientific and Industrial Research Organization, Australia) studies how triaxial accelerometers (sensors that capture movement along three axes) can measure and analyze the behavior and daily activity of cattle. “We want to understand better how types of animal behavior, such as feeding, moving, lying and rising, can be monitored with precision by means of these devices, providing valuable insights for herd management,” she said, adding that this can make livestock farming more efficient, sustainable, and focused on animal well-being.

Growing concern

Animal welfare is a growing concern, for both ethical and economic reasons. Gustavo Venâncio da Silva (UNESP Botucatu) investigates behavioral signs that can be used to diagnose acute pain in cattle after castration, a common procedure in livestock farming. “A better understanding of how animals manifest pain can lead to more responsible management practices and increase production efficiency by reducing stress,” he said.

Big data analytics using digital technology can be applied to improve beef cattle selection and production, noted Guilherme Rosa (University of Wisconsin-Madison), who studies ways of combining genomics and field data to support decision-making. “Novel technologies for automated farm monitoring using sensors and cameras are generating large amounts of data on cattle, but much of this data is still underutilized. Furthermore, falling costs of genetic sequencing and other ‘omics’ technologies permit detailed analysis of cattle at the molecular level,” he said.

Prediction of animal growth is another possible application of PLF. Ariana Negreiro (University of Wisconsin-Madison) studies different methods of predicting characteristics of growth and reproduction in Holstein heifers (Holstein is one of the most popular and productive breeds of dairy cattle), including “transformers” – machine learning models that use complex neural networks to identify patterns in large datasets. Her findings can help improve heifer management, raising growth and reproduction rates with enhanced accuracy. 

Precision aquaculture

In aquaculture, the appropriate term is precision fish farming (PFF), which is based on the same principles as PLF but takes other key conditions into account owing to the fact that fish are typically farmed in tanks or ponds. “Fish and shrimp are highly sensitive to environmental factors, such as variations in water temperature, pH and dissolved oxygen,” said Diogo Hashimoto, a researcher at the Center for Aquaculture (UNESP Jaboticabal) and one of the organizers of the SPSAS. 

Most advances in PFF therefore focus on monitoring and automation of environmental conditions. Monitoring is the main challenge. “The transparency of the water can vary significantly, interfering directly with the capture of animal performance data. The animals are often bred in ponds with plenty of sediment and can’t even be seen. Furthermore, they move fast, horizontally, vertically and in all three dimensions, with the angle of motion constantly changing,” Hashimoto said.

Nevertheless, technological advances have been achieved to automate fish biometry monitoring using computer vision systems. “We’re able to take measurements automatically and even predict fillet yield by means of imaging, without having to sacrifice fish. We’re currently working on ways of detecting stress and stress response using video,” he explained.

Another example mentioned by researchers at UNESP’s Center for Aquaculture is their innovative software that automates accurate real-time biometric measurement of pacus (Piaractus mesopotamicus). The system uses deep learning to analyze photographs of the pacus and identify specific body parts such as the head, pelvic girdle and fins. This makes the genetic selection process less stressful for the fish, which do not have to be handled for the purpose of taking measurements, and reduces the risk of disease. In addition, the software can be integrated with genomic data to accelerate genetic improvement without the need to euthanize any fish, keeping them alive for use as reproducers during the selection process and raising yields sustainably (read more at: agencia.fapesp.br/40284). 

* With information from the press office of UNESP Jaboticabal’s Graduate Program in Aquaculture.