A Study on Light Preference in Gilts via Behavioral Pattern Analysis

The rational design of artificial lighting systems in pig housing can enhance animal welfare, thereby boosting gilt health and reproductive performance while improving economic metrics for swine farms. To identify the optimal light environments for gilts under artificial illumination, we conducted self-selection-based photic preference testing, ultimately providing actionable insights for welfare-centric precision lighting protocols in modern pig production. In this study, a dynamic multi-chromatic self-selection system was developed, integrating programmable RGBW-LED arrays for spectral control, inter-compartment access channels for autonomous gilt movement, and real-time image recognition technology to investigate light color preferences. Twenty-four gilts (nulliparous female pigs) were housed for five weeks in pens with white, yellow, green, blue, or red light (100 lux), and they were given free access to all of the chromatic zones through inter-compartment channels. A YOLOv8n-based deep learning framework was used to quantify their spatiotemporal distribution, activity levels, and eating behavior. The key findings were the following: (1) a significant preference for green light environments (21.29 ± 3.77% distribution proportion) (p < 0.05), peaking at 6:00-13:00 and 18:00-20:00; (2) the average activity was the highest in a white light environment (25.49 ± 0.77%), significantly exceeding yellow (22.69 ± 0.63%) and green light (21.55 ± 0.61%) (p < 0.05); and (3) the daily feed consumption under green light was the lowest, significantly lower than that under white, blue, and red light (p < 0.05). The findings from this study offer insights into the light environment preferences of gilts, which could improve animal welfare.