Accelerating genetic progress.
Genetic improvement now plays an essential role in achieving a modern, efficient, and sustainable pig production system. In recent years, the use of technological advances applied to breeding programmes has made it possible to accelerate the genetic progress in breeding populations, thereby increasing the profitability of producers.
The application of technological advances in breeding programmes is nothing new. Since its foundation in 1962, PIC is committed to technological innovation, progressively implementing new technologies to generate continuous genetic improvement in all traits of the breeding programme. Investment in genomic selection coupled with extensive commercial data collection, elite population expansion and high selection intensity has resulted in the highest rates of genetic improvement that PIC has ever recorded. This improvement translates commercially into an increased number of pig and better quality of the pigs, higher robustness, faster and more efficient growth and high carcass value. All in all, PIC’s genomic programme has generated an estimated total additional value of €3.45 / £ 2.94 per slaughter pig – just in 2020 alone.
Despite the high rates of improvement achieved, we are aware that we still do not have the perfect pig. For this reason, and following the motto “Never stop improving”, PIC is investing in the latest innovative technologies, looking at their practical applications in animal genetic improvement. Among the available technologies, those directly related to genetic technology stand out, such as gene editing and total genome sequencing, which will most probably mark a before and after in pig breeding in the coming years. On the other hand, and in parallel to the above, investment in data collection has increased exponentially in recent years to the point where the application of new developments in artificial intelligence and its branch known as machine learning are already a reality. This article focuses on phenotyping technologies, showing concrete examples of how artificial intelligence applied to pig production will generate a revolution in the way we measure phenotypes and select our breeders.
Applied breeding technology
In current selection programmes, the selection objectives in terminal sire lines focus on:
- Efficient growth
- Robustness
- Carcass and meat quality
Of all these, traits related to robustness have traditionally been the most difficult to address genetically. There are several reasons for this: A greater difficulty to identify the specific trait to be selected for and/or to measure the trait objectively, with sufficient volumes, and/or because of the low heritability of the trait.
A typical example where these difficulties come together would be attempts to select for aspects of animal behaviour. Therefore, while it is expected that technological advances in data collection will impact to a greater or lesser extent on all selection objectives, the initial impact could be greater for robustness traits, as they allow new traits to be described and measured objectively, so that their use for genetic improvement can be assessed.
Applications of artificial intelligence in pig breeding – Real world examples
Artificial intelligence refers to the study, development and application of computer techniques with the aim of enabling machines to acquire certain skills similar to human intelligence. The concept of machine learning consists of machine learning by exposing the computer to an enormous amount of data so that it can process, analyse and learn from it. With all this, we can literally train machines to help us select pigs.
One of the most widely used lines of research today is the combination of cameras recording animals 24 hours/seven days a week and machine learning. Indeed, the volume of data generated by the cameras is so large that one person, or even a group of people, would have to spend enormous amounts of time analysing the material generated. And, even if it were possible, their measurements would tend to have a degree of subjectivity that would make the use of such data for breeding less efficient. In contrast, by combining the collection of large volumes of data with artificial intelligence we can generate valuable information for breeder selection as shown in the following examples.
Automation of assessment of leg soundness
Within the group of robustness traits, leg and feet problems represent one of the main causes for unvoluntary culling, both for males and for breeding females. Traditionally, the assessment of leg structure is carried out by specialised professionals who generate an assessment useful for phenotypical selection and genetic improvement. However, their assessments are inevitably linked to the human eye performing the assessment. The image analysis and machine learning technology that we are developing allows us to generate automated leg structure assessments to produce a much higher volume of highly objective leg scorings on elite farms, with the added benefits of facilitating on-farm work and increasing data accuracy. (Figure 2)
Behavioural studies.
Much more complex is the collection of animal behavioural data. The 24-hour-a-day recording of pigs reared under commercial conditions is a unique opportunity to generate data to analyse the behaviour of our pigs. In the experimental units of PIC (picture 3) we are teaching a machine to analyse these recordings. The result makes it possible to individually record aspects such as resting and activity times, access and times at feeders and drinkers, study their behaviour and social interaction in different environmental conditions, etc. By developing specific algorithms, this technology is making it possible to describe new traits that can be genetically improved. In addition, the information generated provides valuable know-how for technical support to pig producers. Think for example of the analysis of behavioural changes associated with different environmental conditions such as feed, densities, or ventilation.
FertiBoar
While the above examples are still in a process of development prior to commercial application, the FertiBoar technology is already a reality and therefore an example worth describing in more detail.
The FertiBoar technology allows early selection of males in relation to their semen quality. Using testicular ultrasound scans (Figure 4) and automated image analysis through artificial intelligence (Figure 5), we have developed a specific algorithm that allows us to reliably predict the semen quality of young boars already at the farm of origin, and therefore prior to their entry into Gene Transfer Centres/AI Studs. The benefits are multiple, both for the AI-Stud by greatly reducing the cost of quarantine and increasing the quality and predictability in the production of semen doses, and for the farmer thanks to a higher quality of the semen used to mate the females and the stability in the supply of doses due to the lower replacement of young males.
This technology represents a successful collaboration between academic institutions (IFN Schönow, Germany) and private companies (PIC). In the joint research project (2017-2020), ultrasound images and semen quality data from more than 1,000 animals were analysed. The high quality of this research has resulted in scientific publications and recognition as an innovation at industry events (Innov’SPACE and Technical Innovation at FIGAN), as well as its rapid implementation in the PIC boar supply chain.
The first PIC®408 boars tested with this technology are already at an AI Stud, in Riufred (Spain). Underlining this project as a successful example for practical application of state-of-the-art technology. Thanks to the progressive implementation of FertiBoar at all sire line multiplication farms, in the short term all PIC customers will benefit from this innovation.
Conclusions
- Innovation is key to maintain a competitive advantage. For PIC this is part of the company’s founding values and since 1962 its R&D has been focused on delivering tangible benefits to the producer.
- Continued investment in technology is required, with the scale, volume, and scope necessary to generate further genetic progress for the benefit of the producer.
- The impact of terminal boars for the commercial level is rapid (responsible for 50% of the genes) and decisive, representing a great opportunity for innovation application (considering the moderate relative cost of investment in paternal genes, approximately 0.06-0.07% of the final cost in relation to their impact on the porcine value chain).
To learn more, please contact your PIC team.
Author: Juan Manuel Herrero, Genetic Services PIC Southern Europe
