Attempts over the past 50 years to explain variation in the abundance, distribution and diversity of plant secondary compounds gave rise to theories of plant defense. Remarkably, few phylogenetically robust tests of these long-standing theories have been conducted. Using >50 species of milkweed (Asclepias spp.), we show that variation among plant species in the induction of toxic cardenolides is explained by latitude, with higher inducibility evolving more frequently at lower latitudes. We also found that: (1) the production of cardenolides showed positive-correlated evolution with the diversity of cardenolides, (2) greater cardenolide investment by a species is accompanied by an increase in an estimate of toxicity (measured as chemical polarity) and (3) instead of trading off, constitutive and induced cardenolides were positively correlated. Analyses of root and shoot cardenolides showed concordant patterns. Thus, milkweed species from lower latitudes are better defended with higher inducibility, greater diversity and added toxicity of cardenolides.