As human influence reshapes communities worldwide, many species expand or shift their ranges as a result, with extensive consequences across levels of biological organization. Range expansions can be ranked on a continuum going from pulled dynamics, in which low-density edge populations provide the “fuel” for the advance, to pushed dynamics in which high-density rear populations “push” the expansion forward. While theory suggests that evolution during range expansions could lead pushed expansions to become pulled with time, empirical comparisons of phenotypic divergence in pushed vs. pulled contexts are lacking. In a previous experiment using Trichogramma brassicae wasps as a model, we showed that expansions were more pushed when connectivity was lower. Here we used descendants from these experimental landscapes to look at how the range expansion process and connectivity interact to shape phenotypic evolution. Interestingly, we found no clear and consistent phenotypic shifts, whether along expansion gradients or between reference and low connectivity replicates, when we focused on low-density trait expression. However, we found evidence of changes in density-dependence, in particular regarding dispersal: populations went from positive to negative density-dependent dispersal at the expansion edge, but only when connectivity was high. As positive density-dependent dispersal leads to pushed expansions, our results confirm predictions that evolution during range expansions may lead pushed expansions to become pulled, but add nuance by showing landscape conditions may slow down or cancel this process. This shows we need to jointly consider evolution and landscape context to accurately predict range expansion dynamics and their consequences.