From pixels to breeding values: genetic analysis of detection and movement traits in laying hens using automated tracking data of ArUco-marked birds

BACKGROUND: Automated tracking technologies make it possible to perform large-scale genetic analyses of behavioural traits in livestock. However, behavioural phenotypes derived from automated systems can be noisy, incomplete, or difficult to interpret in terms of specific behaviours or welfare outcomes. This study evaluated the feasibility of using ArUco marker-based tracking to derive three behavioural phenotypes expressed in the litter area (detected or not, minutes detected, walking speed) in a population of 1,132 crossbred laying hens kept under semi-commercial conditions.

RESULTS: Despite the fact that 63% of the time, individuals were not detected due to occlusion or system limitations, the tracking method yielded sufficient data (on average 5,326 detections per individual per hour) to estimate genetic parameters with genomic prediction. All traits exhibited significant additive genetic variance, with heritability estimates ranging from 0.05 to 0.08 for hourly measurements and from 0.13 to 0.26 for daily measurements. Genetic correlations revealed shared architecture between detection traits (detected or not and minutes detected, [Formula: see text]= 0.71), but an unexpected strong negative correlation between walking speed and minutes detected ([Formula: see text]= -0.73) probably because faster individuals sooner disappeared from the camera's field of view. Cross-validation accuracies were modest (0.25-0.32), while model-based accuracies were nearly twice as high. Multi-trait modelling did not improve the accuracy of estimated breeding values.

CONCLUSIONS: Our results demonstrate that automated tracking can generate useful phenotypes for the genetic evaluation of complex behaviours. The current methodology provides individual-level behavioural data suitable for genetic analyses, though its application at scale still requires substantial effort in data collection and processing. Additionally, further ethological understanding is required to confirm which specific behaviours these traits reflect.