Paleobiology

Euomphalina

According to Wagner's (1999) species-level phylogenetic hypothesis of Late Cambrian to Silurian gastropods, based mainly on characters of the teleoconch, two great clades diverged from a stem group during the Early Ordovician. One of these, the “euomphalinae,” corresponds to the taxon Euomphalina. The Ordovician (late Arenig to Richmondian) genus Scalites, assigned by Wagner (1999) to the Raphistomatidae, is the only member of this clade said to have an inhalant siphonal indentation (Linsley 1978a).

Murchisoniina

All other siphonate gastropods belong to Wagner's (1999) other great clade, the “murchisoniinae.” Four Early Paleozoic groups represent early siphonate offshoots from this great clade. Most or all of them diverged before the origin within “murchisoniinae” of the Orthogastropoda, the clade that contains all living gastropods except the “euomphalinae”-derived Patellogastropoda (limpets). These four early offshoots are (1) the Ordovician (late Tremadoc) genus Plethospira in the family Hormotomidae, belonging to Wagner's plesiomorphically nonsiphonate “plethospiroid” clade; (2) the Ordovician (late Tremadoc to Richmondian) family Subulitidae with four genera, belonging to the “subulitid” subclade of the “hormotomoid” clade; (3) the Early Ordovician (Early Arenig) genus Donaldiella in the Lophospiridae, belonging to Wagner's “goniostrophine” subclade of the “cyrtostrophid” clade within the larger “hormotomoid” clade; and (4) the Late Silurian to Early Devonian (Early Ludlow to Emsian) genus Diplozone in the Loxonematidae, belonging to Wagner's “omospirine” subclade of the “cyrtostrophid” clade. According to Wagner's (1999) phylogeny, these groups represent four independent acquisitions of the siphonal indentation among closely related clades, each of which is primitively nonsiphonate. Although the Subulitidae resemble Caenogastropoda (a large derived clade within Orthogastropoda), they did not give rise either to them or to other later Paleozoic subulitid-like gastropods (Wagner 1999; Nützel et al. 2000; Wagner and Erwin 2006; Bandel 2002a,b). Instead, the ancestors of Orthogastropoda are interpreted to be nonsiphonate members of the Loxonematoidea, which belong to Wagner's (1999) “cyrtostrophid” subclade of the “hormotomoid” clade.

Vetigastropoda

In the Vetigastropoda, one of the major orthogastropod clades, a single small Devonian subclade has evolved a siphonal indentation. This small clade comprises the genera Tylozone, Buechelia, and Arizonella, collectively ranging from the Emsian to the Frasnian (see Linsley 1968, 1978a,b). Although Linsley (1968, 1978a,b) placed this group in Raphistomatidae, P. J. Wagner (personal communication December 2006) believes that they are possible vetigastropods that should be placed in a new family.

Other Vetigastropoda have a nonsiphonate apertural margin whose outer lip is a logarithmic spiral expanding adapically. This configuration also applies to early Apogastropoda, the clade containing all Orthogastropoda other than Vetigastropoda. The Apogastropoda is divided, perhaps in a trichotomy, into the Neritaemorpha (= Neritopsina), Caenogastropoda, and Heterobranchia.

Early Caenogastropods

Four primarily late Paleozoic gastropod groups have a siphonate indentation. They appear to belong to early primarily nonsiphonate clades that diverged within Caenogastropoda (Frýda 1999, 2001) separately from and before the major crown-group caenogastropod clades discussed below. One of these early groups is the Early Devonian to Triassic family Soleniscidae, high-spired to fusiform gastropods that together with the Meekospiridae form a clade (Nützel et al. 2000). The earliest siphonate soleniscid appears to be the Middle Devonian (Givetian) genus Bensbergia (Heidelberger and Koch 2005). Bandel (2002b) includes the genus Ceraunocochlis in the Soleniscidae, whereas Nützel and colleagues (2000) consider it to be the only siphonate member of the Meekospiridae. A second siphonate group includes Sphaerodoma and Strobeus, which Nützel et al. (2000) assign to Soleniscidae, but which Bandel (2002b) segregates in the unrelated peruneloiean family Sphaerodomatidae. These two genera collectively range from the early Carboniferous (Visean) to the Early Triassic (see also Nützel 2005b). Some early Carboniferous sphaerodomatids, including Globozyga and Globulimorpha, are nonsiphonate, as are Devonian members of the Peruneloidea (Frýda and Bandel 1997; Frýda 1999, 2001; Bandel 2002b). A third siphonate group is the early Carboniferous (Tournaisian) high-spired genus Chlorozyga, which belongs to the otherwise nonsiphonate early Carboniferous family Imoglobidae (Nützel and Cook 2002; Bandel 2002b). Imoglobids are characterized by a peculiarly sculptured larval shell and appear to be a peruneloidean clade separate from the Sphaerodomatidae. Finally, the latest Permian genus Luoguospira, with a low turbiniform to discoidal shell, is a taxonomically and phylogenetically enigmatic gastropod from China with a diffuse, short siphonal canal (Pan and Erwin 2002).

Campaniloidea

The Campaniloidea is an Early Cretaceous (Aptian/Albian) to Recent clade of siphonate herbivorous caenogastropods, represented in the living fauna by the Campanilidae and Plesiotrochidae (Healy 1993; Houbrick 1981a, 1989, 1990b; Kiel et al. 2000; Lydeard et al. 2002). Although they were thought to be part of the clade Cerithioidea (Houbrick 1988; Simone 2001), where they occupy a basal position in most phylogenetic analyses (see also Lydeard et al. 2002), molecular sequences imply derivation from the Architaenioglossa, the most basal crown-group clade (or paraclade) in Caenogastropoda (Harasewych et al. 1998; see also Simone 2004). All known Architaenioglossa other than Campaniloidea lack a siphonal indentation. This could imply that the siphonate condition in Campaniloidea is derived within Architaenioglossa. This interpretation is supported by the nonsiphonate condition of the Ampullospiridae (Early Jurassic to Recent), the only confirmed marine family of Architaenioglossa outside the Campaniloidea (Kase and Ishikawa 2003). Unfortunately, Simone (2004) did not include any marine groups in his anatomy-based phylogeny of Recent architaenioglossans. Alternatively, it is possible that the stem group of Campaniloidea is to be found among early Carboniferous to Permian members of such high-spired families as the Goniasmatidae, Orthonematidae, and Palaeostylidae (Kiel et al. 2000; Bandel 2002b). These gastropods, some of which have an incipient siphonal indentation (e.g. Microlampra, Goniasma, and Palaeostylus), have a protoconch resembling that of Plesiotrochus and Campanile. The Late Triassic Settsassiidae could link these Paleozoic taxa with the Cretaceous and Cenozoic groups. In the present state of uncertainty, I consider the Campaniloidea an independent siphonate clade.

Molecular data (Lydeard et al. 2002) and morphological evidence from anatomy and protoconchs (Bandel and Kowalke 1997a; Simone 2001; Kiel et al. 2000) indicate that the wormlike, sessile, nonsiphonate Vermetidae occupy a highly derived phylogenetic position relative to Campanilidae. This position implies secondary loss of the siphonal indentation in this family, which ranges from the Late Cretaceous (Campanian) to the Recent (see Bandel and Kowalke 1997a; Kiel et al. 2000).

Cerithioidea

Cerithioidea are the next most basal clade of caenogastropods (Ponder and Lindberg 1997; Harasewych et al. 1998; Simone 2001; Lydeard et al. 2002). Phylogenetic analyses of living species, based on either anatomy (Houbrick 1988; Simone 2001) or molecular data (Lydeard et al. 2002), offer contradictory indications about the evolution of the anterior indentation in this large group of grazers and ciliary feeders. Simone's (2001) analysis, with vetigastropods and nonsiphonate basal caenogastropods as outgroups, places the nonsiphonate trochiform Modulidae (latest Eocene or earliest Oligocene to Recent [Mansfield 1937; Petuch 1998]) at the base of Cerithioidea. He and Houbrick (1980) note, however, that this family of microherbivores has many specializations. In the analyses of Houbrick (1988) and Lydeard and colleagues (2002), the weakly siphonate Litiopidae (Campanian Late Cretaceous to Recent [Bandel and Kowalke 1997a]) are basal within Cerithioidea. Neither the Modulidae nor the Litiopidae is as old as the clade Cerithioidea, judging from characters of the protoconch. Whichever phylogenetic interpretation is correct, basal cerithioideans appear to be nonsiphonate. In fact, Simone's (2004) anatomy-based phylogeny of architaenioglossans indicates that the origin of Cerithioidea lies among nonsiphonate members of the paraphyletic Architaenioglossa.

Most Paleozoic and Triassic high-spired gastropods that have been interpreted on the basis of protoconch characters as stem cerithioideans, stem campaniloideans, or stem caenogastropods lack a siphonal indentation (Bandel 1992, 2002b; Bandel et al. 2002; Nützel and Pan 2005). These include the slit-bearing Goniasmatidae (beginning in the early Carboniferous) and the slitless Palaeostylidae (early Carboniferous to late Permian), Orthonematidae (also with early Carboniferous origins), and the Late Triassic Ladinulidae and Prostyliferidae. As noted above, several members of these families have a siphonal indentation, including Goniasma, Microlampra, Palaeostylus, and the Late Triassic Cheilotomona (Batten 1985; Bandel 2002b; Bandel et al. 2002; Nützel et al. 2002; Pan and Erwin 2002; Nützel and Pan 2005). One or another of these siphonate groups likely gave rise in the Triassic to the cerithioidean family Procerithiidae (Bandel 1992; Bandel et al. 2002), which in turn includes ancestors of the major Mesozoic and Cenozoic cerithioidean radiation (Abbass 1973; Bandel 1992, 1993; Bandel et al. 2002). Cerithioideans with a true siphonal canal, including Rhynchocerithium and Occidentocerithium, are present in the Late Triassic (Nützel et al. 2003; Nützel and Erwin 2004). The latest Permian Propupaspiridae, another siphonate family with possible cerithioidean affinities, may be ancestral to Late Triassic siphonate Popenellidae and Procerithiidae (Nützel et al. 2002). The muddled evidence indicates that the siphonal indentation of Cerithioidea has a late Paleozoic origin. At this point it is best to take the highly conservative position that it evolved only once in this group of gastropods.

A tubular, closed abapical canal has evolved at least five times independently in Cerithioidea. These are the Late Cretaceous (Maastrichtian) Parinana from Peru (Olsson 1944), the middle Eocene to Recent cerithiid genus Gourmya (Houbrick 1981b), the taxonomically and phylogenetically enigmatic minute Recent Sherbornia from the Indo-West Pacific (Iredale 1917; Cernohorsky 1981a), the middle Eocene (Lutetian) potamidid Gravesicerithium mitra (Lamarck) (personal observation in collection of B. Landau), and three species in the potamidid genus Terebralia (Fig. 1). In Terebralia, the Recent type species, T. palustris (Linnaeus) from the Indo-West Pacific, has an open canal; a tubular canal occurs in the early Miocene tropical American T. dentilabris (Gabb), the late Miocene to Recent Indo-West Pacific T. sulcata (Born), and the Recent North Australian T. semistriata (Mörch) (Hoerle 1972; Houbrick 1991b). The oldest species of the genus, T. susana Squires and Kennedy from the late Paleocene of California, has an open canal (Squires and Kennedy 1998). Unfortunately, other old species such as T. subcorvina (Oppenheim) from the middle Eocene of Europe are never preserved with the aperture intact, so the nature of their canals cannot be ascertained (Kowalke 2001) and the time of origin of the tubular canal can be constrained only as during or before the early Miocene.

Besides the Modulidae, many other living cerithioideans lack a siphonal indentation. The small-shelled Dialidae (early Miocene to Recent) and Scaliolidae (early Oligocene to Recent [Lozouet 1998]) are interpreted on the basis of molecular data as the sister group to the Modulidae (Lydeard et al. 2002). All three families are close to the older (Late Cretaceous to Recent) siphonate Potamididae (Lydeard et al. 2002). These inferences are consistent with the interpretation that the nonsiphonate condition of dialids and scaliolids is secondary. Nützel and Pan (2005) point out that the protoconchs of these gastropods conform to a type that is plesiomorphic for caenogastropods as a whole, and that similar protoconchs occur in the Paleozoic Goniasmatidae and Orthonematidae. This observation thus leaves open the possibility that the nonsiphonate shell of dialids, scaliolids, and modulids represents the primitive cerithioidean condition.

The nonsiphonate, suspension-feeding, largely sedentary Turritellidae are known from the Early Cretaceous (Valanginian) onward (Schröder 1995) and occupy a derived position in some phylogenetic trees of the Cerithioidea (Houbrick 1988; Lydeard et al. 2002). They may have lost the siphonal indentation, as these phylogenies imply, but likely fossil ancestors of turritellids have not yet been identified.

On the basis of protoconch morphology, the small-shelled, siphonate Litiopidae (Campanian stage of the Late Cretaceous to Recent [Bandel and Kiel 2000]) may form a clade with the indentation-bearing Provannidae, a deep-water group with unknown feeding habits (Bouchet and Warén 1991; Bandel and Kowalke 1997b), known from the Late Cretaceous (Campanian) onward (Bandel and Kiel 2000). The nonsiphonate, worm like, suspension-feeding Dendropomatidae appear to be a Cenozoic group derived from siphonate members of the Litiopidae-Provannidae clade, and would therefore represent another case of secondary loss of the siphonal indentation (Bandel and Kiel 2000).

The nonsiphonate, sedentary, miniaturized Fossaridae (Miocene to Recent [Bandel and Kowalke 1997b]) are close to and likely derived from weakly siphonate Planaxidae (Warén and Bouchet 1988; Houbrick 1990a; Bandel and Kowalke 1997b). The latter family is known from the Eocene onward (Lozouet and Maestrati 1994).Three or four independent lineages of marine cerithioideans invaded brackish and fresh water beginning no later than the Late Jurassic (Bandel 1991a), and secondarily lost the siphonal indentation in each case. These groups are centered on Thiaridae, Pachychilidae, Paludomidae, and Melanopsidae-Pleuroceridae (Houbrick 1988; Bandel 1991b, 1992; Glaubrecht 1999; Simone 2001; Lydeard et al. 2002; Köhler et al. 2004). A siphonal notch is retained in the brackish-water Faunus (Houbrick 1991a) and Melanopsis and in such freshwater taxa as Aylacostoma, Doryssa, Hemisinus, Io, Melanopsis, Paramelania (young only), and Tiphobia. It is very likely that the siphonal indentation was lost more than once in each of the cerithioidean clades that entered fresh water, but the absence of detailed phylogenetic hypotheses prevents a precise estimate from being made. I have therefore accepted three losses of the siphonal indentation in these groups as a highly conservative estimate.

Finally, miniaturization may explain the absence by secondary loss of a siphonal canal in the small bittiine cerithiids Cerithidium (early Miocene to Recent [Lozouet 1998]) and Cassiella (Recent) (see Houbrick 1993).

Ptenoglossa

The remaining Caenogastropoda belong to the clade Hypsogastropoda (Ponder and Lindberg 1997), whose basal subclades again appear to lack a siphonal indentation. An indentation appears to have evolved at least five times in Ptenoglossa, three times in Rissooidea plus Stromboidea, and once in a huge clade comprising Metamesogastropoda and Latrogastropoda (Riedel 2000; Bandel 1993).

As currently understood, the Ptenoglossa are a clade of predominantly parasitic gastropods with origins in the early Carboniferous (Nützel 1998). The Janthinoidea, comprising the planktonic Janthinidae and the benthic Epitoniidae and Nystiellidae, mainly live on cnidarians and lack a siphonal indentation. The Cerithiopsoidea plus Triphoroidea forms a clade of sponge-dwellers with a siphonal indentation, which is usually elaborated into a siphonal canal. The Eulimoidea (Late Cretaceous or Paleocene to Recent), an echinoderm-associated group without a siphonal indentation (Warén 1984) are doubtful members of the Ptenoglossa (Ponder and Lindberg 1997; Nützel 1998). According to Nützel's (1998) interpretation, the Janthinoidea are derived from Triassic members of the mainly nonsiphonate Zygopleuridae (Early Triassic to Late Cretaceous [Squires and Saul 2003a; Nützel 2005b]); whereas the Cerithiopsoidea stem from the Protorculidae (latest Permian or earliest Triassic to Late Triassic [Nützel 1998, 2005b]). The Triphoroidea, a mainly or exclusively Cenozoic group, is derived from Cerithiopsoidea (Nützel 1998). The nonsiphonate Paleozoic Pseudozygopleuridae is either the sister group (Nützel 1998) or stem group (Nützel 2005b) of all later Ptenoglossa. The siphonal indentation in Protorculidae arose either in the early Carnian stage of the Late Triassic (Nützel and Erwin 2004) or, if the siphonate genus Prodiozoptyxis is a protorculid (Nützel 1998), in the latest Permian. Batten's (1985) interpretation of Prodiozoptyxis as an early nerineid seems implausible in view of the long stratigraphic gap between its late Permian occurrence and the next oldest nerineid in the Early Jurassic.

As many as five additional small clades with a siphonal indentation have been tentatively assigned to the Ptenoglossa, and may represent independent acquisitions of the indentation. These include (1) the pseudomelaniid genera Liocium (Campanian Late Cretaceous of California) and Bayania (Eocene of Europe) (Squires and Saul 2003b); (2) the Cretaceous (Albian to early Campanian) genus Alamirifica of uncertain taxonomic position (Saul and Squires 2003); (3) the Late Triassic zygopleurid subfamily Andangulariinae (Nützel and Erwin 2004); (4) the taxonomically enigmatic Late Triassic genera Idahospira and Spirocyclus (Nützel and Erwin 2004); and (5) the Abyssochrysidae, represented in the living fauna by two deep-sea species (Houbrick 1979), perhaps descended from the Zygopleuridae (Nützel 1998). The Jurassic genus Diatrypesis may also be a zygopleurid (Nützel and Erwin 2004) and belong to one of the five clades above, although Kollmann (2005) assigns this genus to the metacerithiid subfamily Nerineopsinae, which is either a cerithioidean group (Kollmann 2005) or in Campaniloidea (Kiel et al. 2000). The position of the Pseudomelaniidae, to which the siphonate Liocium and Bayania have been assigned, is also uncertain. Bandel (1993) and Nützel (2005b) very tentatively linked this mostly nonsiphonate group to the mostly siphonate Soleniscidae of the late Paleozoic and Early Triassic. This would mean secondary loss of the indentation in basal Jurassic pseudomelaniids and a regain in the two siphonate members. More plausibly in my view, Pseudomelaniidae are Ptenoglossa derived from a zygopleurid or pseudozygopleurid stock (Squires and Saul 2003b; Kaim 2004). Finally, Kowalke (2001) assigns the siphonate estuarine genus Bayania to the cerithioidean family Thiaridae, whose freshwater representatives have lost the canal. For now, I accept four cases listed and discussed above as independent acquisitions of the siphonal indentation in Ptenoglossa: the Protorculidae-Cerithiopsoidea-Triphoroidea clade, Zygopleuridae + Abyssochrysidae, Pseudomelaniidae, and Spirocyclus-Idahospira. The genus Alamirifica and the subfamily Andangulariinae could be additional instances, but they could also have arisen from one or more of the above siphonate clades.

A closed tubelike anterior canal has evolved at least once in the Ptenoglossa. It characterizes most members of the subfamilies Triphorinae (sinistral) and Triforinae (dextral) of the Triphoridae, including the two earliest known species of middle Paleocene age belonging to the genus Epetrium (see Kosuge 1966; Marshall 1977, 1983; Nützel 1998). Species with an open canal belong to the Metaxiinae (dextral) and are scattered in the Triphorinae among such genera as Paramendax (Triforinae), Inella, Latitriphora, Talophora, and Brucettriphora (Marshall 1977, 1983; Beu 2004).

Rissooidea and Stromboidea

The apparently plesiomorphic condition in the very large middle Permian to Recent caenogastropod clade Rissooidea is an abapically rounded aperture without a siphonal indentation (Ponder 1985, 1988). A siphonal notch has arisen at least three times independently in this clade. Two instantiations occur in the Early Jurassic: (1) the iravadiid genus Kalchreuthia of late Pliensbachian age (Gründel and Nützel 1998); and (2) The Palaeorissoinidae, which appear in the Toarcian Stage (Harzhauser and Kowalke 2001; Gründel and Kowalke 2002). Members of the Rissoinidae (or Rissoininae), including such genera as Isseliella, Microstelma, and Stosicia, also have a siphonal notch (Ponder 1985), but Kaim (2004) regarded the rissoinids as derived in the Middle Jurassic from palaeorissoinids. A notch is also present in the early Miocene (Aquitanian) brackish-water Greveniellinae, interpreted by Harzhauser and Kowalke (2001) as a subfamily of Palaeorissoinidae. The inference that a siphonal notch arose only once in the Palaeorissoinidae-Rissoinidae-Greveniellinae is therefore highly conservative. A third separate instantiation occurs in the freshwater South American cochliocopine hydrobiid genus Sioliella, known from the middle Miocene to the Recent (Wesselingh 2000).

Kaim (2004) has made the plausible suggestion that the mostly siphonate Stromboidea (Early Jurassic to Recent) are derived from or sister to the Rissoinidae. Early members of both Rissoinidae and Stromboidea have an abapical as well as an adapical indentation. Kaim (2004) further suggested that rissoinids have ancestors among Pseudozygopleuridae, and that such genera as Plocezyga could be intermediates. Bandel (1991a) advocates a somewhat similar scenario, deriving the earliest (aporrhaid) stromboideans from Late Triassic nonsiphonate, high-spired Cassianozyga. As Kiel and Bandel (2002) noted, however, these Triassic genera lack the elaborations of the outer lip that characterize stromboideans. Bandel's scenario, like Kaim's, remains plausible but tentative. Bandel (1993) thought that the Stromboidea are phylogenetically connected to the Purpurinidae, which in turn might be related to the Rissoinidae. The earliest purpurinid, the Early Triassic (Olenekian) genus Werfenella, has a siphonal indentation (Nützel 2005a). Among these three scenarios, I favor a Late Triassic to Early Jurassic derivation of Stromboidea from siphonate Rissoinidae or Palaeorissoinidae, because the latter two groups often have a thickened lip indicating determinate growth, which characterizes nearly all Stromboidea including the earliest aporrhaids. As discussed below, purpurinids (which lack determinate growth) are more likely ancestral to siphonate latrogastropods. The siphonate condition of Stromboidea therefore is part of the same radiation that gave rise to living siphonate rissoinids, and arose only once in this clade near the Triassic/Jurassic boundary.

I suggest that secondary loss of the siphonal indentation has occurred at least once in Stromboidea. The trochoidal carrier shells (Xenophoridae), known from the Late Cretaceous (Campanian) to the Recent (Dockery 1993; Perrilliat and Vega 2001), are deposit feeders (Ponder 1983) in which the rounded aperture shows no trace of a siphonal indentation. In Simone's (2005) anatomy-based phylogeny, the Xenophoridae are derived relative to both Aporrhaidae (Early Jurassic to Recent [Kaim 2004]) and Struthiolariidae (early Oligocene to Recent [Darragh 1991]). I am unaware of plausible candidates for the ancestry of xenophorids in the fossil record, but fossil evidence would seem to preclude struthiolariids. Some struthiolariids show substantial reduction in the siphonal indentation, perhaps related to their ciliary feeding habits; but there is always some trace of the indentation. The most likely origin of xenophorids therefore must be sought among the paraphyletic Aporrhaidae, which have been siphonate since their first appearance in the Early Jurassic.

Latrogastropoda

Most siphonate caenogastropods belong to Riedel's (2000) clade Latrogastropoda. These include such well known groups as animal-grazing cowries and their relatives (Cypraeoidea and Lamellarioidea) and predatory Tonnoidea, Ficoidea, and Neogastropoda. Although latrogastropods are overwhelmingly a mid-Cretaceous to Recent group, likely siphonate ancestors date back to the Triassic. The Maturifusidae, represented by the single genus Astandes (= Maturifusus; Norian Late Triassic to Late Cretaceous [Nützel and Erwin 2004; Kaim 2004]), have been interpreted as the stem group of either the Neogastropoda (Kaim 2004) or, more plausibly in my view, of the clade comprising Laubierinoidea, Tonnoidea, Ficoidea, and Neogastropoda (Riedel 2000). Taylor and colleagues (1980) suggested that neogastropods (and therefore perhaps by extension latrogastropods) have ancestors among the Mesozoic family Purpurinidae, whose earliest member (Werfenella, Olenekian Early Triassic [Nützel 2005a]) is siphonate. As noted above, purpurinids may have been derived from a Permian pseudozygopleurid ancestor. If so, the siphonate condition of Werfenella, Astandes, and undoubted latrogastropods arose in the Early Triassic, independently of the origin of the indentation in Rissoinidae-Stromboidea.

At least nine separate cases of formation of a closed siphonal tube in the adult shell are known in neogastropods. Seven of these occur in Muricidae. Among Trophoninae, there are two examples: the late Miocene Xymenopsis peninsularis Brunet from Argentina (Brunet 1997) and the Californian Recent Ocenotrophon painei (Dall) (McLean 1995). The Muricinae contains at least two instantiations. These are Purpurellus (late Oligocene to Recent) from tropical America and West Africa (Emerson and D'Attilio 1969; Vokes 1970a) and a group of species including some members of the genus Pterynotus including the early Miocene P. raulini (Cossmann and Peyrot) and the Recent Caribbean P. tripteroides (Lamarck) (Cossmann 1897; Cossmann and Peyrot 1924) and the derived middle Eocene to Recent Tripterotyphinae. Although most tripterotyphines have a closed canal (Ponder 1972; D'Attilio and Hertz 1988; Vokes 1989), some species of Pterotyphis and Tripterotyphis have a narrowly open one (Ponder 1972; Houart 1996). The early Eocene to Recent Typhinae likewise usually have a sealed canal, but again there are some exceptions (Vokes 1964, 1989; Ponder 1972; D'Attilio and Hertz 1988). Their origins are obscure but are likely to be independent of those of the Pterynotus-Tripterotyphinae group. At least some species of the muricopsine genus Homalocantha (late Oligocene to Recent [Merle 2002]) from throughout the Tropics have a closed canal. Many genera in the Ocenebrinae have a ventrally sealed canal, including Ceratostoma (Vermeij and Vokes 1997; Amano and Vermeij 1998a), Crassilabrum (DeVries 2005), Fenolignum (Vermeij and Vokes 1997), Genkaimurex (Matsukuma 1977), Hadriania (Vermeij and Vokes 1997), Inermicosta (Vermeij and Houart 1999), Jaton (Vermeij and Houart 1996), Kestocenebra (Vermeij 1998b), Microrhytis (Vermeij and Vokes 1997), Muregina (Vermeij 1998b; DeVries 2005), Ocenebra (Vermeij 1998b), Ocinebrellus (Amano and Vermeij 1998b), Ocinebrina (Vermeij and Vokes 1997; Vermeij 1998b), Odontopurpura (Vermeij and Vokes 1997), Poropteron (Houart 1991), Pteropurpura (Vermeij and Vokes 1997), and Pterorytis (Vermeij and Vokes 1997). The earliest ocenebrine with a sealed canal is Ocinebrina rarisulcata (Deshayes) from the middle Eocene of France (Merle 2005). It is highly likely that the sealed canal evolved multiple times in Ocenebrinae, and that there were also frequent reversals to a narrowly open canal (Marko and Vermeij 1999). Here I take the conservative position that the closed canal arose only once in this group.

Two other small neogastropod groups evolved a sealed siphonal canal. These are the photine buccinid genera Terajana and Neoteron from tropical America, known from the late Miocene onward, the Recent representatives being restricted to the eastern Pacific (Vokes 1970b; Cernohorsky 1981b); and the vasid Siphovasum latiriforme (Rehder and Abbott), living in the Gulf of Mexico (Vokes 1966).

Two species said to have a sealed canal have not been included in the analysis. These are Murex (Ocenebra) aschersoni, described by Quaas (1902) from the Late Cretaceous of Libya; and Vredenburg's (1928) Murex sindiensis, a late Paleocene species from Pakistan, a possible member of the genus Pterynotus. Preservation of these species is insufficient to be sure if the canal is indeed sealed.

The phylogenetic position of the nonsiphonate, shell-drilling, predatory Naticidae (Campanian Late Cretaceous to Recent [Kase and Ishikawa 2003]) remains unclear. Gründel (2001) suggested ancestry in the purpurinoid nonsiphonate family Tripartellidae of the Jurassic. Data are insufficient to decide whether naticid latrogastropods are primitively or secondarily nonsiphonate.

Secondary loss of the siphonal indentation characterizes many independent latrogastropod clades. Living Capulidae, which include coiled Trichotropinae (Cenomanian to Recent [Bandel 1993]) and limpetlike Capulinae (Paleocene to Recent [Warén and Bouchet 1991; Bandel 1993]) are more or less sedentary parasites (Matsukuma 1978) or facultative cleptoparasites and suspension feeders (Graham 1954; Pernet and Kohn 1998; Iyengar 2002, 2005). In trichotropines, the abapical part of the shell is extended, and the aperture is narrowed into what appears to be a siphonal canal, but there is no siphonal indentation (Fig. 1). Trichotropines appear to be derived from a siphonate stock that also gave rise to Tonnoidea (Warén and Bouchet 1991; Riedel 2000). The sedentary, nonsiphonate Haloceratidae, with unknown feeding habits and without a known fossil record, likewise appear to be derived from a trichotropine-like ancestor (Warén and Bouchet 1991; Riedel 2000). The nonsiphonate, limpetlike Hipponicidae (Campanian Late Cretaceous to Recent, or Bathonian Middle Jurassic to Recent [Bandel and Riedel 1994; Kaim 2004]) and Calyptraeidae (Santonian Late Cretaceous to Recent [Kiel and Bandel 2003]) may also be derived from trichotropine-like gastropods (Bandel and Riedel 1994; Riedel 2000; Collin 2003). This scenario, based on fossil and molecular evidence, conflicts with Simone's (2002) anatomy-based phylogeny, which derives these two families from Vanikoridae, which are coiled, nonsiphonate, sedentary, often miniaturized particle feeders. Bandel and Kowalke (1997b) proposed that the Vanikoridae and the related nonsiphonate Pickworthiidae (both early Paleocene to Recent) primitively lack the indentation and are derived from Late Triassic Prostyliferidae. In their review of vanikorids and pickworthiids, Warén and Bouchet (1988) and Le Renard and Bouchet (2003) remain agnostic about the ancestry of these enigmatic gastropods.

The ascidian-associated Velutinidae, with a nonsiphonate rounded aperture and a thick periostracum, have often been interpreted to be primitive cypraeoideans or lamellarioideans (Simone 2005; Vermeij 2005). Golikov and Gulbin (1990) regarded them as derived from a capulid-like ancestor, a view that is consistent with the presence of a thick periostracum and with the relatively young age (early Miocene [Gladenkov and Sinelnikova 1990]) of the earliest velutinid. Whichever scenario is correct, velutinids most probably represent a case of secondary loss of the siphonal indentation. It remains unclear whether this loss occurred in ancestors that also gave rise to other nonsiphonate groups (see above) or whether it is independent of that in Capuloidea. I tentatively accept the first of these alternatives.

The Late Cretaceous genus Weeksia, with a radial nonsiphonate aperture and a nearly planispiral shell, is interpreted by Bandel and Dockery (2001) on the basis of protoconch characters to be related to and derived from the siphonate Lowenstamia, which Bandel and Dockery (2001) assigned to the latrogastropod stem-group Sarganidae (Cenomanian to Maastrichtian in the Cretaceous). Weeksia, with a likely sedentary mode of life, therefore appears to represent independent loss of the siphonal indentation.

Reversion to the nonsiphonate condition characterizes at least six subclades of Neogastropoda. Two of these are coralliophiline Muricidae: the limpetlike Galeropsis (= Quoyula), an early Miocene to Recent genus associated with pocilloporid corals (Lozouet and Le Renard 1998), and the early Miocene to Recent coral-boring genus Magilus. Some Cancellariidae from middle Eocene time onward lack a siphonal canal and indentation in the adult shell. Nonsiphonate cancellariids have been assigned to two subfamilies, the Trigonostomatinae (including Trigonostoma, Coptostoma, and Junghuhnia) and Admetinae (including Admete and Admetula). Cancellariids are inferred to be suctorial feeders (Harasewych and Petit 1982, 1986; Petit and Harasewych 1986). In the family Pseudolividae, the genera Benthobia, Fusulculus, and Zemira have either an obsolete siphonal indentation or none at all (Bouchet and Vermeij 1998; Vermeij 1998a). Nothing is known about the habits of these gastropods. Finally, the common large Recent sand-dwelling conid Conus (Lithoconus) litteratus Linnaeus from the Indo-West Pacific is unique among conoideans in that the outer and inner lips of the aperture meet abapically in a rounded continuous margin that does not deviate from the apertural plane as an indentation (Fig. 1). This species is a slow-moving worm-eater whose feeding habits differ little from those of its siphonate nearest relatives such as Conus leopardus Hwass (Duda and Kohn 2005). With the exception of Fusulculus (Late Cretaceous to Recent), all secondarily nonsiphonate neogastropods have Cenozoic origins.

Heterobranchia

A siphonal indentation has evolved at least three times in the Heterobranchia, a major clade recognized by the change in coiling direction from the protoconch to the teleoconch. The Heterosubulitidae is based on a single late Carboniferous species, Heterosubulites blatta (Knight) from the Henrietta Shale of Missouri, characterized by a fusiform shell and very shallow anterior notch (Bandel 2002a). In the Early Triassic (Scythian) or Middle Triassic (Ladinian) to Recent, high-spired family Mathildidae, most members of which are nonsiphonate (Bieler 1995), several Triassic genera including Promathilda and Siphonilda have evolved a siphonal canal (Nützel and Erwin 2004). The Late Triassic Trachoecidae, which may be related to these siphonate mathildids, also have such a canal (Bandel 1994; Nützel and Erwin 2004). The much larger-shelled Nerineoidea (Early Jurassic to Late Cretaceous [Kollmann 1976, 2005; Bandel 1994]) also have a distinct anterior indentation or canal. Bandel (1994) suggested that this prominent Mesozoic group arose from a mathildid-like ancestor. I therefore tentatively regard the siphonate condition of these high-spired mathildids, trachoecids, and nerineoideans as having arisen only once, either in the Middle or Late Triassic. The third siphonate heterobranch clade is the opisthobranch family Ringiculidae, known from the Middle Jurassic (Callovian) onward (Kaim 2004) and well represented since Early Cretaceous (Valanginian) time (Chavan 1947; Bandel 1994; Squires and Saul 2001). The earliest heterobranchs, of Devonian age (Frýda and Blodgett 2001), lack a siphonal indentation. The indentation is therefore interpreted as derived in this major gastropod clade.