A network representation of the exchange of animals among the holdings can act as the backbone of the spatial dependence analysis of disease spreading. Therefore, the analysis of this network at different scales of time and space is an important part of livestock epidemiological studies. Here, the objective is three fold: (1) to characterize the network; (2) to investigate its evolution over time; and (3) to identify the modifications in its structure due to restrictions. Based on the French National Cattle Database (BDNI) from 2005-2011, we construct networks at different spatiotemporal scales, analyze them with methods of graph theory (indicators such as, clustering, reciprocity, assortativity, distributions of degree and strength, centrality, activity, entropy, disparity, preference index). Comparisons of means and distributions of these indicators (e.g. Kolmogorov-Smirnov test) lead to identification of patterns in cattle movement in terms of seasonality, demographics and dependence on geographical and administrative units. The structure of the network at different spatiotemporal scales and its response to interventions (restrictions imposed due to bluetongue outbreak in France in 2007) reveals the scales (e.g. communes as nodes and monthly time window), where interventions affect network descriptors, which may be used in redesigning control policies. We observed that although most of the average indicators do not show statistically significant variations, centrality distributions related to capacity of bridging among different nodes show significant differences for the period of intervention (2007). We also study the network in terms of measures used for competition in market (e.g. preference index) and note that with interventions imposed, the network reorganized itself (through changes, for a given holding, in the composition of subnetworks of its neighbors and/or the intensity of corresponding trade) to sustain its structure.