Precision phenotyping as a tool to automatically monitor health and welfare of individual animals housed in groups

Farm animals are increasingly housed in large group housing systems. Monitoring health and welfare in these large groups can be challenging. In current welfare assessment schemes, attention for animal-based welfare indicators is increasing, resulting in a shi in focus from environment-based to animal-based indicators. However, in group settings monitoring these animal-based welfare indicators is challenging, especially at the level of the individual animal. Focusing on the individual level is relevant in many dierent settings, e.g. for precision feeding and management, or targeted veterinary care. Also, in the context of animal breeding, focus on the individual level is pivotal; this is frequently done by housing animals of potential interest for selection individually to measure (or evaluate) their phenotypes. In this way, however, no information about the performance of the individual in a group setting can be included in the breeding program, while social interactions can have profound eects on group performance. New genetic methodology now allows the modelling of these social interactions using direct and indirect genetic eects models. is type of methodology can for instance provide information on the propensity of an animal to become a victim of damaging behaviour (direct genetic eect), but also on the propensity to perform damaging behaviour (indirect genetic eect). To utilise this methodology, the ability to measure accurate individual phenotypes in a group setting becomes very important. Breeding companies currently have a strong interest in developing methods for precision phenotyping, relying on new technology that enables tracking of individuals using combinations of dierent sensors. is should allow the use of group housing in future breeding operations and should allow more accurate phenotyping. In the PhenoLab project, we investigated possibilities for tracking of location, activity and proximity of individual laying hens. To meet that aim, we tracked individual hens during a ve-minute Open Field test using two dierent tracking systems: Ultra-wideband tracking using TrackLab and automatic video tracking using EthoVision. Ultra-wideband tracking consists of an active RFID tag that is placed on the bird in a backpack. is tag is then located by triangulation by four beacons. Comparing distance moved between TrackLab and Ethovision yielded 96% similar results (sample of 24 hens). In a second step, the ultra-wideband tracking was also used to measure dierences in activity between high (HFP; n=45) and low (LFP; n=41) feather pecking lines of laying hens. e system was well able to detect the higher activity level in the HFP line compared with the LFP line (10 vs 5 m moving distance), that was also found in previous studies. Interestingly, within the HFP line, birds phenotyped as feather peckers using traditional video observations were found to be the most active individuals compared with the other phenotypes. Precision phenotyping technology could become an important tool to automatically monitor health and welfare of individual animals housed in groups.