1. Premise of the study: Determining how species perform in novel climatic environments is essential for understanding responses to climate change, and evolutionary consequences of biological invasions. For the vast majority of species, the population characteristics that will predict performance and patterns of natural selection in novel locations in the wild remains limited. 2. Methods: We evaluated phenological, vegetative, architectural and fitness-related traits in experimental gardens in contrasting climates (ON and SC) in the North American non-native distribution of Arabidopsis thaliana. We assessed the effects of climatic distance, geographic distance, and genetic features of history on performance and patterns of natural selection in the novel garden settings. 3. Key results: We found plants had greater survivorship, flowered earlier, were larger, and produced more fruit in the south, and that genotype by environment interactions were significant between gardens. However, our analyses revealed similar patterns of natural selection between gardens in distinct climate zones. After accounting for genetic ancestry, we also detected that population climatic distance best predicted performance within gardens. 4. Conclusion: These data suggest that colonization success in novel, non-native environments is determined by a combination of climate and genetic history. When performance at novel sites was assessed with seed sources from geographically and genetically disparate established non-native populations, proximity to the garden alone was insufficient to predict performance. Our study highlights the need to evaluate seed sources from diverse origins to describe comprehensively phenotypic responses to novel environments, particularly for taxa where many source populations may contribute to colonization.