Malou Van Der Sluis

One-third of the human population’s protein intake is provided by livestock products and there is a rapidly increasing demand for these products. To meet this demand, animal agriculture has intensified. With the resulting large numbers of animals per farm, keeping track of individual animals is challenging. However, individual records, for example of activity, are of great relevance for assessing animal welfare and for implementation of traits in breeding programmes. The main aims of this thesis were 1) to study whether, and how, detailed activity data on individual broilers can be collected in an automated manner throughout life, while they are housed in a group, and 2) to study the relationship between individual activity, leg health and body weight, with the ultimate goal of examining whether selecting on activity to improve health and welfare of broilers is feasible. In chapters 2 and 3, we implemented an ultra-wideband (UWB) tracking system to collect data on individual activity levels of broilers from approximately two-weeks-old onwards. In chapter 2, we compared the UWB recordings to video recordings and observed a moderately strong agreement between the two approaches. Furthermore, we observed that, with the UWB system, we could detect decreases in activity as the broilers grew older, as well as differences in activity levels between body weight categories. On average, lightweight broilers were observed to move longer distances than heavyweight broilers. Overall, this chapter indicated that the relative activity of broilers can be tracked well with the UWB system, from approximately two-weeks-old onwards. In chapter 3, we used the UWB data to assess the relationship between activity and gait in broilers. We found indications for relationships between gait classification and different measures of activity, with lower activity levels for birds with a suboptimal gait. Furthermore, we observed a tendency for an interaction between gait classification and weight category. In this interaction, a difference in level of activity was observed between gait classifications in lightweight birds, with lower activity levels for birds with a suboptimal gait, but not in heavyweight birds. It has to be further investigated if this is a consequence of higher body weight already limiting activity levels. However, in general the differences in activity levels of birds with different gait classifications were not very clear and therefore it remained difficult to distinguish gait classifications based on distances moved during the period from 16 to 32 days old. The UWB system was not suitable for collecting activity data early in life, due to the relatively large and heavy tags that had to be fitted to the birds. In chapters 4-6, we therefore implemented a passive radio frequency identification (RFID) tracking system to collect data on individual activity levels of broilers from hatching onwards. In chapter 4, we validated the RFID system and observed that in 62.5% of the cases the RFID system was in full agreement with video in terms of the location of the animals. In 99.2% of the cases, near matches (allowing for a deviation of one antenna grid cell) were observed. Furthermore, we observed that there were moderately strong correlations in terms of distances moved between RFID, video and UWB. We concluded that the RFID system could accurately register among-individual variation in activity throughout life and continued collecting activity data with this system. In chapter 5, we used the resulting RFID data to further elaborate on the relationship between activity patterns early in life and body weight, or growth, over time. In this chapter, we not only looked at average activity levels, but also at dynamic descriptors of activity. We found a negative correlation between average daily gain (ADG) and the root mean square error of activity, indicating that broilers with more deviations from the expected linear trend in activity had a lower ADG. In chapter 6, we used RFID data to estimate the heritability of activity in broilers. We observed an estimated heritability of 0.31 ± 0.11 across the full production period and a decrease in heritability of activity as the birds aged. Furthermore, we observed positive and mostly moderate to strong genetic and phenotypic correlations between activity levels recorded in different weeks, with the correlations between adjacent weeks being strongest. Overall, our results suggested that genetic selection on activity is feasible, which could in the future potentially help to improve broiler welfare, through the hypothesized positive effects of increased activity on leg health. In chapter 7, the general discussion, I brought together the results from the different chapters. I discussed the differences between, and advantages and limitations of, UWB and passive RFID tracking systems for individual activity tracking of group-housed broilers. I concluded that passive RFID systems are better suitable for activity tracking throughout life, due to their small and lightweight tags that can be fitted to broilers from hatching onwards. I furthermore addressed the relationship between activity, leg health and body weight in more detail and showed that these are all related. Finally, I discussed potential directions for future implementation of RFID tracking systems in larger scale broiler systems, including less detailed RFID tracking and a sensor-fusion approach of RFID and computer vision.