Predicting which species will become invasive when introduced to a novel range is one of the main goals of invasion biology. However, many studies of the factors influencing invasion either use the wrong comparison group, fail to control for relatedness, or measure traits in only a single environment. I compared invasive and noninvasive species in the Commelinaceae (dayflower family) across multiple environments in the greenhouse and developed a molecular phylogeny of invasive and noninvasive Commelinaceae for comparative analyses. Invasive Commelinaceae had faster growth rates, higher fecundity, thinner leaves, and greater vegetative reproduction than noninvasive relatives. However, whether some traits, such as thin leaves, were associated with invasiveness is environment-dependent, suggesting that predictive models need to take the trait-environment relationship into account to more accurately predict invasiveness. Further, invasive Commelinaceae exhibited opportunistic responses to high nutrient environments, and noninvasive species did not, suggesting that a lack of opportunism may be useful for predicting noninvasive species. Demographic models parameterized with greenhouse data suggest that fecundity (for self-compatible species), time to first reproduction, and
vegetative reproduction may be particularly important for explaining the greater performance of invasive taxa in high nutrient environments. Phylogenetic analyses based on trnL-trnF (a non-coding chloroplast region) and 5S NTS (a non-coding nuclear ribosomal repeat) were mostly consistent with tribal and subtribal relationships in the Commelinaceae, and with a previous rbcL phylogeny at the genus level and relationships within genera were mostly well-resolved and consistent with current taxonomy, with the exception of Callisia, which is not monophyletic in these analyses. Comparative tests incorporating phylogeny suggest that weediness, self-compatibility, annual life history, vegetative reproduction, and possibly seed heteromorphism all increase the probability of becoming invasive. In phylogenetically naive tests, capsule dimorphism, autogamous selfing proportion, propagule pressure, and date of first introduction were also associated with invasiveness, although those relationships were not significant in tests incorporating phylogeny. Overall, many traits, including several novel ones presented here (e.g. seed heteromorphism) were associated with invasiveness in the Commelinaceae. Also, environment and phylogeny both influence trait associations, suggesting that taking these factors into account may improve the ability to predict and prevent invasions.