Plants demonstrate some of the greatest variation in genome size among eukaryotes, and their genome sizes can vary dramatically across individuals and populations within species. This genetic variation can have consequences for traits and fitness, but few studies have been able to attribute genome size differentiation to ecological and evolutionary processes. Biological invasions present particularly useful natural laboratories to infer selective agents that might drive genome size shifts across environments and population histories. Here, we test hypotheses for the evolutionary causes of genome size variation across 14 invading populations of yellow starthistle, Centaurea solstitialis, in California, USA. We use a survey of genome sizes and trait variation to ask: (1) Is variation in genome size associated with developmental trait variation? (2) Are genome sizes smaller toward the leading edge of the expansion, consistent with selection for 'colonizer' traits? Or alternatively, does genome size increase toward the leading edge of the expansion, consistent with predicted consequences of founder effects and drift? (3) Finally, are genome sizes smaller at higher elevations, consistent with selection for shorter development times? We found that 2C DNA content varied 1.21-fold among all samples, and was associated with flowering time variation, such that plants with larger genomes reproduced later, with lower lifetime capitula production. Genome sizes increased toward the leading edge of the invasion, but tended to decrease at higher elevations, consistent with genetic drift during range expansion but potentially strong selection for smaller genomes and faster development time at higher elevations.