Theory has long predicted allocation patterns for plant defense against herbivory, but only recently have both above- and belowground plant defenses been considered simultaneously. Milkweeds in the genus Asclepias are a classic chemically defended clade of plants with toxic cardenolides (cardiac glycosides) and pressurized latex employed as anti-herbivore weapons. Here we combine a comparative approach to investigate broadscale patterns in allocation to root vs. shoot defenses across species with a species-specific experimental approach to identify the consequences of defense allocational shifts on a specialist herbivore. Our results show phylogenetic conservatism for inducibility of shoot cardenolides by an aboveground herbivore, with only four closely related tropical species showing significant induction; the eight temperate species examined were not inducible. Allocation to root and shoot cardenolides was positively correlated across species, and this relationship was maintained after accounting for phylogenetic nonindependence. In contrast to long-standing theoretical predictions, we found no evidence for a trade-off between constitutive and induced cardenolides; indeed the two were positively correlated across species in both roots and shoots. Finally, specialist root and shoot herbivores of common milkweed (A. syriaca) had opposing effects on latex production, and these effects had consequences for caterpillar growth consistent with latex providing resistance. Although cardenolides were not affected by our treatments, A. syriaca allocated 40% more cardenolides to shoots over roots. We conclude that constitutive and inducible defenses are not trading off across plant species, and shoots of Asclepias are more inducible than roots. Phylogenetic conservatism cannot explain the observed patterns of cardenolide levels across species, but inducibility per se was conserved in a tropical clade. Finally, given that above- and belowground herbivores can systemically alter the defensive phenotype of plants, we concur with recent calls for a whole-plant perspective in testing models of plant defense allocation.