species and despite apparent absence of phytoplanktonfeeding larvae in this genus since its origin 20 million years ago. Retention of a larval esophagus and a full complement of velar ciliary tracts needed for particle capture and ingestion in encapsulated larvae of some muricids may help explain how larval planktotrophy re-emerged within this clade.

Comparative data on the developing gastropod foregut suggest that this multicomponent feeding complex consists of two developmental modules. Modularity is revealed by delayed development of the buccal cavity and radular sac (“ventral module”) relative to the dorsal food channel (“dorsal module”) in gastropods with feeding larvae compared with those that may have never had a feeding larval stage. If nonfeeding larvae like those of extant patellogastropods and vetigastropods are ancestral for gastropods, then the uncoupling and heterochronic offset of dorsal and ventral foregut modules allowed the post-metamorphic dorsal food channel to be co-opted as a simple but functional esophagus for feeding larvae. Furthermore, by reducing energy cost per ovum, the heterochronic offset may have given mothers the evolutionary option of increasing fecundity or investing in protective egg encapsulation material. A second developmental innovation was spatial separation of the dorsal and ventral foregut modules, as illustrated by distal foregut development in buccinid neogastropods and venom gland development in cone snails. Spatial uncoupling may have enhanced the evolvability of gastropod foreguts by allowing phenotypic variants of ventral module components to be selected within post-metamorphic ecological settings, without needing to be first tested for compatibility with larval feeding. Finally, we describe a case in which foregut modularity has helped facilitate a highly derived life history in which encapsulated embryos ingest nurse eggs.

Evolution of two novel feeding strategies among caenogastropod molluscs, suspension feeding in calyptraeids such as Crepidula fornicata and predatory feeding with a pleurembolic proboscis among neogastropods, may have both involved elongation of the anterior esophagus. Emergence of predatory feeding with a proboscis is particularly significant because it correlates with the rapid adaptive radiation of buccinoidean and muricoidean neogastropods during the Cretaceous. However, the notion that this important evolutionary transition involved elongation of the anterior esophagus to extend down a long proboscis has been disputed by evidence that it may have been the wall of the buccal cavity that elongated. We undertook a comparative study on foregut morphogenesis during larval and metamorphic development in C. fornicata and in three species of neogastropods with a pleurembolic proboscis to examine the hypothesis that the same region of foregut has elongated in all. We approached this by identifying a conserved marker for the boundary between buccal cavity and anterior esophagus, which was recognizable before the developing foregut showed regional differences in length. A survey of four species of littorinimorph caenogastropods suggested that the site of neurogenic placodes for the buccal ganglia could serve as this marker. Results showed that foregut lengthening in C. fornicata involved elongation posterior to neurogenic placodes for buccal ganglia, an area that corresponded to the anterior esophagus in the other littorinimorphs. However, foregut elongation occurred anterior to neurogenic placodes for buccal ganglia in two buccinoidean and one muricoidean neogastropod. The elongated foregut within the pleurembolic proboscis of these neogastropods qualifies as anterior esophagus only if the definition of the anterior esophagus is expanded to include the dorsal folds that run down the roof of the buccal cavity. Regardless of how the anterior esophagus is defined, comparative developmental data do not support the hypothesis of homology between the elongated adult foregut regions in C. fornicata and in neogastropods with a pleurembolic proboscis.