Animal movements are typically driven by areas of supply and demand for animal products and by the seasonality of production and demand. As animals can potentially spread infectious diseases, disease prevention can benefit from a better understanding of the factors influencing movements patterns in space and time. In Mauritania, an important cultural event, called the Tabaski (Aïd el Kebir) strongly affects timing and structure of movements, and due to the arid and semi-arid climatic conditions, the season can also influence movement patterns. In order to better characterize the animal movements patterns, a survey was carried out in 2014, and those data were analysed here using social network analysis (SNA) metrics and used to train predictive gravity models. More specifically, we aimed to contrast the movements structure by ruminant species, season (Tabaski vs. Non-Tabaski) and mode of transport (truck vs. foot). The networks differed according to the species, and to the season, with a changed proportion of truck vs. foot movements. The gravity models were able to predict the probability of a movement link between two locations with moderate to good accuracy (AUC ranging from 0.76 to 0.97), according to species, seasons, and mode of transport, but we failed to predict the traded volume of those trade links. The significant predictor variables of a movement link were the human and sheep population at the source and origin, and the distance separating the locations. Though some improvements would be needed to predict traded volumes and better account for the barriers to mobility, the results provide useful predictions to inform epidemiological models in space and time, and, upon external validation, could be useful to predict movements at a larger regional scale.