Seed size is correlated with a number of other plant attributes, and such correlations are commonly interpreted as adaptations to different life histories. However, seed size is also strongly related to species phylogeny. In the current comparative literature, correlations between trait variation and taxonomic membership are often interpreted as evidence for phylogenetic constraints, and such phylogenetic explanations of variation are presented as alternatives to adaptive explanations. However, phylogenetic patterns of trait variation should not be discussed exclusively in terms of constraints. Neither should phylogenetic and adaptive explanations for variation be automatically presented as alternatives. Another explanation for similarity among related taxa is phylogenetic niche conservatism (PNC). This process is well known to exist-indeed, it is an inevitable consequence of evolution by natural selection-and is explicitly adaptive. Under PNC, radiation within a lineage is shaped by phylogeny, by the environment, and by competition from other lineages present in the environment. These factors may inhibit divergence within a lineage (hence "conservatism"), but there are no insuperable obstacles (constraints) to divergence. We have included multiple data sets within one analysis, permitting an assessment of how a given lineage has diverged under different physical conditions or in the presence of different competing lineages. We found that taxonomic membership statistically accounted for the majority of seed size variation within data sets. Differences among data sets were also largely associated with shifts in taxonomic composition. These two findings do not distinguish between phylogenetic constraints and PNC; however, three other features of the data indicate that phylogeny has its effect through niche conservatism at least as much as through constraint. First, within many lineages, divergence of taxa from their lineage means was greater than divergence between different lineages, showing that phylogeny does not absolutely constrain variation. Second, much seed size variation correlated with taxonomic membership was also correlated with growth form and dispersal mode, suggesting a role of niche adaptation in the conservation of sets of traits within lineages. Third, patterns of divergence within some lineages were not always consistent across data sets, pointing to the role of niche divergence in trait divergence, which is the logical complement of PNC. These results indicate that a PNC model of trait evolution provides a better explanation for seed size variation than phylogenetic constraint.