Most evolutionary innovations—power-enhancing phenotypes previously absent in a lineage—have arisen multiple times within major clades. This repetition permits a comparative approach to ask how, where, when, in which clades, and under which circumstances adaptive innovations are acquired and secondarily lost. I use new and literature-based data on the phylogeny, functional morphology, and fossil record of gastropods to explore the acquisition and loss of the siphonal canal and its variations in gastropods. The siphonal indentation, canal, notch, or tube at the front end of the shell is associated in living gastropods with organs that detect chemical signals directionally and at a distance in an anteriorly restricted inhalant stream of water.

Conservative estimates indicate that the siphonate condition arose 23 times and was secondarily lost 17 times. Four siphonate clades have undergone prodigious diversification. All siphonate gastropods have a shell whose axis of coiling lies at a low angle above the plane of the aperture (retroaxial condition). In early gastropods, the siphonal canal was short and more or less confined to the apertural plane. Later (mainly Cretaceous and Cenozoic) variations include a dorsally deflected canal, a long canal, and a closed canal. The closed siphonal canal, in which the edges join to form a tube, arose 15 times, all in the adult stages of caenogastropods with determinate growth.

Gastropods in which the siphonate condition arose were mobile, bottom-dwelling, microphagous animals. Active predaceous habits became associated with the siphonate condition in the Mesozoic and Cenozoic Purpurinidae-Latrogastropoda clade. Loss of the siphonate condition is associated with nonmarine habits, miniaturization, and especially with a sedentary or slow-moving mode of life.

The siphonate condition arose seven times each during the early to middle Paleozoic, the late Paleozoic, and the early to middle Mesozoic, and only once each during the Late Cretaceous and Cenozoic. Well-adapted incumbents prevented most post-Jurassic clades from evolving a siphonal indentation and its associated organs. Dorsally deflected, long, and closed canals are known only from Cretaceous and Cenozoic marine gastropods, and represent improvements in sensation and passive armor.

In a discussion of contrasting ecologies of clades that gained and lost the siphonate condition, I argue that macroevolutionary trends in the comings and goings of innovations and clades must incorporate ecological and functional data. Acquisitions of energy-intensive, complex innovations that yield greater power have a greater effect on ecosystems, communities, and their resident clades than do reversals, which generally reflect energy savings.