RESUMEN
The mule duck accounts for over 90% of French foie gras production, a sector where feed represents two-thirds of production costs. This study focuses on analysing the feeding behaviours of the mule duck and its parental populations (Pekin and Muscovy) using automated feeders. To assess feed efficiency, feed conversion ratio and residual feed intake were analysed, along with six traits derived at the daily and meal levels. Genetic parameters were estimated separately in purebred populations, as well as with a joint crossbred model that estimated the parental contributions to the hybrid crossbred performances. In relation to higher feed intakes and much-reduced feeding times (P < 0.001), the feeding rate in the Pekin population was twice as high as in the Muscovy population (19 g/min vs 9 g/min), while the mule duck exhibited a large heterosis for this trait (29 g/min). Feeding traits exhibited moderate (0.38 ± 0.11) to high (0.65 ± 0.11) heritabilities. Similar correlation patterns were observed between feeding traits in the two parental populations. In the Pekin line, the feed conversion ratio did not significantly correlate with feeding traits except for daily feed intake. However, in the Muscovy population, it was negatively correlated with the number of meals (-0.51 ± 0.21) and positively with meal feed intake and meal duration (+0.79 ± 0.17 and + 0.71 ± 0.26, respectively). The contributions of the two parental species to the hybrid's performance differed, with the Pekin contributing more to feeding and meat traits compared to the Muscovy. They were similar only for liver weight. Additionally, unfavourable correlations between meat traits and liver traits were estimated in both pathways. Genetic relationships between feeding traits and slaughter traits varied by parental origin, suggesting different strategies for improving hybrid performance in the two parental species. However, in both pathways, genetic correlations between feed conversion ratio and meat traits (breast muscle and thigh weights) were favourable (<-0.42 ± 0.18), whereas they were unfavourable (>0.41 ± 0.20) for fatty liver weight. Altogether, improving liver traits and feed efficiency in the hybrid through selection in the parental populations could be enhanced by considering feeding traits recorded with electronic feeders, provided that adverse correlations are properly accounted for in a multitrait index.