Very little is known about how spatial effects influence invasive species throughout the invasion sequence. We propose here two mechanisms to explain the changes in spatial effects throughout the stages of invasion, using the soybean aphid (Aphis glycines) as a model. First, the “hierarchical spatial effect” hypothesis, based on a change in the relative importance of the spatial scales throughout the invasion process, with main effect at broad scale during the first years of invasion, and main effect at local scale during the subsequent years. Second, the “host-switching spatial effect” hypothesis, stating that the spatial effect is driven by a switch in the effect of the host/habitat throughout the invasion process, from effect of main summer host/habitat during the first years of invasion to effect of overwintering host/habitat during the subsequent years. Data from governmental archives and field samplings enabled to investigate the spatial effects on aphid density at three scales (regional, landscape, local) during a 7 year period (2006–2012). Our results demonstrate that the hierarchical spatial effect hypothesis is not an adequate model for the soybean aphid, aphid density being more affected by landscape-scale factors irrespective of years. In contrast, our results are in accordance with the host-switching spatial hypothesis, with positive effect of the main summer host/habitat (soybean) during the first steps of invasion (2006–2008), followed by a positive effect of overwintering habitats (buckthorn, woodland) during the subsequent years (2010–2012). Overall, investigating these hypotheses in other systems would determine whether the same tendency is observed for other invasive species.