/. MolL Stud. (1997), 63,65-77 © The Malacological Society of London 1997

FORMATION AND HOMOLOGY OF RADULAR TEETH; A CASE STUDY USING COLUMBELLID GASTROPODS (: )

ROBERT GURALNICK and MARTA J. de MAINTENON Department of Integrative Biology and Museum of Paleontology, University of California, Berkeley, CA Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 94720-3140 USA (Received 10 May 1996; accepted 8 August 1996)

ABSTRACT mature tooth row. The evolutionary trans- formation from one radular type to another Radular morphology has contributed prominently to occurs by the gain, loss or specialization of one molluscan systematic classification, especially that of or more teeth of a particular type (Haszprunar, the . Inferences of homology have been based on the shape, number and position of radular 1988; Ponder & Lindberg, 19%). In this study, teeth. In this study, we demonstrate the significance we evaluate the homology of radular struc- of the mode of radular formation in assessing tures, and in particular the middle tooth, of hypotheses of homology. The columbellid radula, the columbellids using not only positional considered representative of the stenoglossate type, and morphological information but also is used as a case study. formational data. The columbellid middle tooth is acuspate, Our intent is to show that positionally spanning part of a wide membrane between two equivalent radular structures may be formed in relatively large, complex lateral teeth. We examined more than one way, using the secretion of radulae from several species with acuspate or multi- cuspate middle teeth, using histological and whole rachidian structures in columbellid gastropods mount techniques, to compare the modes of forma- as an example. The constructional process tion. Our results show that acuspate rachidian struc- determines morphology and position of the tures in the species we examined are not secreted by radular structures, and so provides a more odontoblastic fields as has been shown for radular meaningful criterion for evaluating hypotheses teeth in other taxa, but are instead produced of homology. We also discuss alternative independently by other secretory cell types. The mechanisms for the formation of radular structure can not, therefore, be considered homolo- structures using evidence from examination of gous to rachidian teeth, illustrating how different whole and sectioned columbellid radulae. formational mechanisms can produce positionally equivalent structures. We suggest not using the term Since we are focusing both on the columbellid 'rachidian' to refer to the central plate in the radula and on mechanisms by which the radula columbellid radulae. is formed, we feel it is important to provide some background information on each of these topics.

INTRODUCTION The columbellid radula The radula has been considered one of the Columbellidae Swainson, 1840 is included as a most useful structures for molluscan system- family of the buccinoid neogastropods. The atics and classification. In gastropods, different columbellid radula is stenoglossate, with a radular types, congruent with anatomical char- middle plate and a single pair of lateral teeth acteristics such as the number of columellar (radular formula given as 1-1-1). Within this muscles and the number and type of ctenidia, formula the columbellid radula is distinct, and have been used to diagnose non-euthyneuran conserved in the group. Although three taxa at the ordinal level and higher (Troschel, radular structures are present, the rachidian is 1856-1893). Tooth types and formulae carry generally acuspate, and the lateral teeth are connotations of implied homology. Features of large, curved, pincer-like, and capable of these radular types include the number of rotation on bases that are narrower than those teeth and their shape and position in each of lateral teeth in other taxa. The radular 66 R. GURALNICK & MJ. de MAINTENON ribbon is long, with 80 to over 300 tooth rows membranoblastic secretion does not appear (Marcus & Marcus, 1962), and the posterior to have a profound effect on the final tooth end may protrude from the pleurembolic morphology; it primarily stiffens structures proboscis into the visceral cavity. already present (Mischor & Markel, 1984). Because many columbellid shell charac- The shape of radular components is deter- teristics are shared with other buccinoid taxa, mined by the shape, position and secretory radular morphology provides a basis for differ- rate of the apocrine cell fields that form them entiating columbellids from other buccinoids. (Mackenstadt & Markel, 1987). The radula is used extensively for classification within the family. Although the acuspate middle plate varies little and is not considered MATERIALS AND METHODS Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 to be of great systematic value, Radwin (1977) used the morphology of the lateral teeth to Representatives of four nominal columbellid taxa distinguish membership in two subfamilies, were examined to evaluate the distribution of and variation within those types has been used different modes of formation of the rachidian structure. All specimens used were adults, or nearly for generic level classification. For example, full grown. The species examined, and the sources Pseudanachis basedowi (Hedley, 1918), one of of material are: the taxa used in this study, has been included in Columbellidae based on features of lateral (1) Columbella major Sowerby, 1832: CAS 085593, tooth morphology, but it differs in other Mazatlan, Sinaloa, Mexico (two specimens) and aspects of its anatomy (de Maintenon, unpub- Bahia de las Animas, Baja California, Mexico (two lished) and its phylogenetic affinities are specimens). (2) Columbella paytensis Lesson, 1830: CAS uncertain. 096169, Paita, Peru (three specimens). (3) Euplica bidentata (Menke, 1843): specimens collected at Nancy Cove, Rottnest Island, Western Radula formation Australia (three specimens). Radular formation has been studied primarily (4) Pseudanachis basedowi (Hedley, 1918): ANSP in pulmonate gastropods (Runham, 1963; A3390, South Vietnam (three specimens); and ANSP A8874, Malaysia (two specimens). Mischor & Markel, 1984; Mackenstadt & Two methods were used in this study to provide a Markel, 1987). Mischor & Markel (1984) preliminary description of radular formation. First, demonstrated that radular formation in the generative and mature regions of whole radulae Pomacea bridgesi, a species of ampullariid, were viewed in all four species, to compare their differs significantly from radular formation in morphology during formation and maturation. pulmonates. The formational process they Second, histological sections were examined to described for Pomacea is also applicable to evaluate the cell fields associated with different Patella coendea (Peters, 1979), and is assumed stages of formation. to be true for all non-euthyneuran gastropod taxa (Mischor & Markel, 1984). /. Whole radulae In Pomacea bridgesi, the radular teeth and membrane are secreted by two different types Radulae dissected from preserved specimens were macerated for twelve hours in 25% potassium of apocrine cells, odontoblasts and mem- hydroxide (KOH). The radulae were dehydrated branoblasts (Mischor & Markel, 1984). The through an ethanol series, air dried, and mounted on membranoblasts secrete the radular membrane stubs using double-sided tape. They were examined continuously while the odontoblasts secrete using an Electroscan model E3 environmental teeth periodically. Both odontoblasts and scanning electron microscope, uncoated, at 10 kV membranoblasts are arrayed in the back of the with a chamber pressure of 3-5 Torr water vapor. radular sac, a blind pouch formed as an out- pocketing of the stomodaeum (Raven, 1958). //. Histological radular sections The radula is continually renewed post- eriorly, pushing previously formed structures We dehydrated and cleared radular sacs, probosces anteriorly toward the functional region, where or whole of each species, and embedded the teeth are used and worn away. After the them in paraffin. The blocks were sectioned at 7 to 10 jim using a rotary microtome. The sections were teeth and membrane are formed, muco- stained primarily in eosin/alum haematoxylin. Speci- polysaccharides and minerals are secreted onto mens were oriented to obtain either cross sections or them by supraradular and subradular epithelia sagittal sections. Sections were viewed with a light (Runham, 1963; Mischor & Mfirkel, 1984). In microscope and photographed using an attached taxa previously studied, post-odontoblastic and camera mount and Canon AE-1 camera. RADULAR HOMOLOGY 67 RESULTS caused by different timing of tooth secretion, with lateral teeth on one side forming before Whole radulae those on the other. The odontoblastic fields that secrete these structures, however, appear Micrographs of fully developed, unworn to be bilaterally symmetrical in the radular sac, radular teeth typical to each species are shown with minor variation due to cell movement in figures 1A, 2A, 3A and 4A. For direct during secretion. comparison, the posterior, generative portion Membranes in all taxa are extensive, and from the same specimens are shown in figures entirely enclose the radular ribbon such that

IB, 2B, 3B and 4B. The mature regions of the the lateral membranes meet dorsally over top Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 radula do not deform during air-drying. In of the lateral teeth. The thickness of the contrast, the newly secreted regions are not membrane varies; the membrane forming hardened, and the membranes wrinkle when the lateral tooth sockets is thick and the rest of dried. The amount of radular deformation the membrane is much thinner. differs between the four species we examined; Pseudanachis basedowi shows little deforma- tion, while the generative regions of the two Histological radular sections species of Columbella and Euplica bidentata In Columbella major, Columbella paytensis are soft and flexible over a relatively great and Euplica bidentata, the posterior portion of distance. the radular sac is bisected by a column of The four species also differ in the timing of weakly staining cells (Figs 7A, 8A), which we appearance of the middle teeth; the multi- call the dorsal-ventral ridge. Anteriorly, the cuspate rachidian of Pseudanachis basedowi ridge of non-secretory cells splits, creating a forms very early, with the lateral teeth (Fig. horseshoe shaped extracellular space (Fig. 7B). 4B). In Columbella paytensis, in contrast, the Two odontoblast fields, placed laterally to the rachidian structures do not definitively appear dorsal-ventral ridge (Figs 6A, 7B, 8A), secrete until far forward on the radular membrane, the lateral teeth. These fields and associated and the posterior portion of the radula is secretory products appear more posteriorly poorly organized (Fig. 2B). The other two than any other radular structures in all three species are intermediate. In Columbella major, species. No central odontoblastic field is a weak rachidian plate may be visible near the present. posterior edge of the membrane, but another In Pseudanachis basedowi, the posterior is not definitively visible until the fourth lateral portion of the sac is not bisected. Instead, tooth row (Fig. IB). Euplica bidentata appears three odontoblast fields are present (Fig. 5A): to generate its radula over a longer distance a center field, placed ventrally, and two lateral than C. major, but a shorter distance than C. fields. The first tooth structure that appears in paytensis (Fig. 3B). The middle structures in histological sections is the rachidian, though Columbella are longitudinally striated when this may vary with the stage of secretion. they first appear, however they have a smooth Membranous lateral tooth sockets and the surface when mature. ventral membrane appear after the rachidian The two Columbella species (Figs 1A, 2A) tooth, anterior and ventral to the center have similar mature radular morphologies. odontoblastic field (Fig. 5B). The lateral odon- The lateral teeth of Euplica bidentata (Fig. 3A) toblastic fields in P. basedowi extend three or are proportionately taller and lack the external four tooth rows anteriorly, rather than being basal cusps of the lateral teeth characteristic of confined to the posterior-most portion of the Columbella and related taxa (Radwin, 1977). sac (Fig. 5C). The radular morphology of Pseudanachis base- After the lateral teeth and membranous dowi resembles those of the previous species sockets are secreted, the ventral opening of the primarily in the narrow width of the lateral sac forms a centrally located continuous tooth bases. The mature teeth are otherwise ventral cell field and a dorsal central fold different in form and orientation. The rachid- between the lateral teeth (Figs 6A, 7B, 8B). ian tooth of Pseudanachis is multicuspate, with The ventral cell field that appears after the a larger center cusp (Fig. 4A). dorsal-ventral ridge splits in Columbella and As the shown in Figs. 1A, 2A, 3A, 4A, the Euplica is membranoblastic, based on its individual tooth rows of all four species are position. Lateral membranoblastic fields also slightly asymmetrical, with one side of the row appear following odontoblastic secretion, more posterior than the other. Asymmetry is generating the membrane into which the teeth 68 R. GURALNICK & MJ. de MAINTENON socket. The central membrane is thin while the After initial deposition of the ventral tooth socket membrane is continuously thick. membrane, the ventral cell fields change to The posteriormost membranoblasts of the another cell type, having large cells with nuclei acuspate species are tall, thin cells with visible placed uniformly in the center of the cell. The secretory vesicles and heavily stained nuclei change can be seen on the left side of the located at varying heights in the lower three ventral membrane in figure 7C. The new cell quarters of the cell (Figs 7B, 7C) similar to field may be the subradular epithelium, how- membranoblasts in other species (Peters, 1979; ever it is responsible not only for thickening Mischor & Markel, 1984). These cells are already formed membrane but also possibly bordered laterally on both sides by cells having for secreting new membrane, so the identity of variously basal nuclei and more heavily stained the field is uncertain. With the appearance of Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 secretory products. Both types appear to this cell field, renewed secretion occurs across deposit a very thin sheet representing only the the ventral membrane, combined with mineral- center region of the membrane, where the ization. Secretion from either the central cell rachidian structure will appear. Two mem- field or more laterally placed membranoblasts branoblast fields may be involved in these results in addition of material to either side of phases of secretion, because the very thin part the center ventral membrane, widening the of the membrane may be maintained over a membrane appreciably on both sides (compare distance, and then thickened suddenly, but no Figs 8B, C, D), and in secretion of muco- corresponding change in cell types was polysaccharides onto the ventral lateral detected. Where the membranoblasts are corners of the membrane (Figs 6C, D), recog- active, the dorsal fold lies very close to the nizable by its staining with haematoxylin. forming membrane. In P. basedowi, mem- Though secretion appears to occur across the branoblastic cells appear after the odonto- whole width of the membrane, the result is a blasts. The membranoblasts are long and thin, uniformly wider membrane rather than a with large nuclei in the basal two-thirds of the thicker one. cells (Fig. 5B). Dorsally, the odontoblasts are replaced by The lateral membranes are formed by the the supraradular epithelium, with large cell secretion of continuous sheets of membrane nuclei located in the apical third of the cells. In to the outer sides of the lateral tooth Pseudanachis basedowi, the subradular and sockets (Figs 6C, 7B, 8B). The cells that supraradular epithelia continue to actively secrete these membranes are low with secrete material, but the general shape and uniformly basal nuclei, unlike those of the position of the teeth and membrane do not ventral membranoblast fields. This portion of change. In the other three species, the shape the membrane may be homologous with the and position of central radular structures will alary processes described by Lutfy and continue to change with epithelial secretions Demian (1964) for Marisa cornuarietis, how- and mechanical deformation of the unmineral- ever these processes are composed of two ized membrane. separate units, one of which does not appear Supraradular and subradular epithelia are attached to the ventral radular ribbon itself. responsible for both secretion of new material The alary process of M. cornuarietis is also and mineralization of material already present, broad and semi-circular, unlike the condition and the two processes are often difficult to in columbellids. The lateral membranes of disentangle. We will restrict the term mineral- Columbella paytensis are secreted concurrently ization to deposition of material that stains with the central membrane. In Columbella differentially with eosin, and probably consists major (Fig. 6C), Euplica bidentata, and of inorganic compounds. In C. paytensis, Pseudanachis basedowi, the lateral membranes secretion of material that stains dark blue with appear later. haematoxylin appears several tooth rows from

Figures 1-4. Scanning electron micrographs (ESEM) of uncoated radulae of species examined. Micrographs were taken of the anterior (mature) region of the radula and of the posterior (forming) portion. After capture, digital images were processed using Adobe Photoshop 2.5.1. 1A. Anterior (mature) portion of the radula of Columbella major, scale bar = 100 (im. IB. Posterior (forming) region of the radula of Columbella major, scale bar = 100 u.m. 2A. Anterior portion of the radula of Columbella paytensis, scale bar = 100 u.m. 2B. Posterior portion of the radula of Columbella paytensis, scale bar =100 UJTI. 3A. Anterior portion of the radula of Euplica bidentata, scale bar = 100 (un. 3B. Posterior portion of Euplica bidentata, scale bar = 100 Jim. 4A. Anterior portion of the radula of pseudanachis basedowi, scale bar = 50 ujn. 4B. Posterior region of Pseudanachis basedowi, scale bar = 50 |xm. RADULAR HOMOLOGY 69 Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 70 R. GURALNICK & MJ. de MAINTENON Figures 5-6. Light micrographs of approximate cross sections of Pseudanachis basedowi and Columbella major. The Columbella major specimen is cut slightly off cross section. Photographs were taken using a micro- scope camera mount and converted to digital form using a Hewlett Packard ScanJet 3c/T. The digital images were processed with Adobe Photoshop 2.5.1. Sections show progression from posterior to anterior. 5A. Pseu- danachis basedowi 30 u,m from the start of the radular sac showing three odontoblast groups, scale bar = 50 u,m. 5B. Pseudanachis basedowi showing two odontoblast groups and membranoblasts, scale bar 50 jim. 5C Pseudanachis basedowi 120 ji.m from the beginning of the sac, scale bar = 50 p.m. 6A. Columbella major 210 jim from the back of the sac, scale bar = 100 u.m. 6B. Columbella major 280 y.m from the back of the sac, scale bar = 100 nm. 6C Columbella major 550 Jim from the back of the sac, scale bar = 100 urn. Note secretory products lying on top of membrane. 6D. Columbella major mature radula, scale bar = 100 \im. Abbreviations: lm, lateral membrane; It, lateral teeth; mb, membranoblasts; od, odontoblasts; rt, 'rachidian' tooth; se, sub- radular epithelium; spre, supraradular epithelium; tsm, tooth socket membrane; vm, ventral membrane. Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 the beginning of the sac, in the lateral teeth, mineralization (Fig. 9). Ventral addition of central region, and portions of the ventral mucopolysaccharides onto the lateral corners membrane surface. This dark blue material is of the membrane limits flexibility to the middle likely a mucopolysaccharide or glycoprotein portion where mineralization of the middle that may strengthen or harden membrane. The tooth will take place. Once the rachidian secreted material lies on top of the ventral structure is present and mineralized, the whole membrane in antero-posterior strips and the radula becomes much less flexible. The maxi- supraradular cells appear to be separated into mum extent of buckling in the median radular discrete fields above that point (Fig. 8C). The region is greater than the thickness of the strips visible in histological sections may be the membrane in C. paytensis (Fig. 9). striations seen in whole radulae. In later The ventral radular membrane also changes sections, however, traces of heavy secretion in width, measured between the lateral teeth. do not consistently or cyclically show up. Initially, the membrane is fairly narrow (Figs In some specimens, a thin layer of material 7C, 8B, 8C). It widens considerably with can be differentiated on top of the ventral secretion of the central lateral portions. membrane at different points (Figs 6B, C). Further to the anterior, near the point of This material does appear cyclically and in mineralization, the membrane gradually numerous sections and may possibly form the compresses laterally and appears to shrink. rachidian plate. In all species, the material The membrane expands by about three- lying on top of the membrane is thin and quarters of its original width during elongation flexible when it appears, and difficult to and contracts only a small amount during discriminate from the membrane except by shrinking. During forward movement of the slight differential staining. Several tooth rows radula, the membrane and lateral teeth also anterior, teeth and the rachidian region stain change in dorso-ventral position in relation to pale pink with eosin, marking the beginning of the sac. Though it originates low in the sac, the mineralization. Further to the anterior, the ventral membrane folds dorsally in the middle teeth stain bright red with eosin (Fig. 6D). The near the point of mineralization and-bulges up mature radular teeth do not stain or stain very between the lateral teeth. Its final position, little with eosin/haematoxylin, and the mature after mineralization, is approximately across membrane stains uniformly purple (Fig. 8E). the middle of the sac. This dorsal shift is best Mineralization of the rachidian structure in seen in C. paytensis (Fig. 8D). Columbella and Euplica appears both surficially and also internally in the membrane, resulting in an embedded structure of some DISCUSSION thickness. Some portions of the rachidian structure in these taxa do not appear to break through the membrane surface (Fig. 8D). In Homology of the rachidian structure Pseudanachis, the rachidian tooth mineralizes Homology is similarity due to common deep red on top of the radular membrane, ancestry. Rieppel (1988) suggested what he which remains light pink. Mineralization never called three 'tests' to determine whether appears internally. features of different organisms are homolo- In the species with acuspate plates, the gous (i.e., whether they are states of the same whole radular membrane is flexible and shows character): similarity, connectivity and con- definitive buckling until after the point of gruence. Similarity is morphological likeness RADULAR HOMOLOGY 71. Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021

of a feature in two organisms, and connectivity ogy. The recognition of a systematic character, implies likeness of number and position. generally by one or more criteria, constitutes a Cracraft (1981) and De Pinna (1991) argue hypothesis of homology. Any such initial that similarity is less a test and more a factor hypothesis of homology must still pass the test that compels recognition of potential homol- of congruence, using a principle like parsi- 72 R. GURALNICK & MJ. de MAINTENON Figures 7-9. Light micrographs of approximate cross sections of Euplica bidentala and Columbella payiensis. The Euplica specimen is cut slightly off cross section. 7A. Euplica bidentata 100 |xm from the back of the sac. Note the weakly staining dorsal-ventral ridge of cells that bisect the sac into two halves, scale bar = 100 (j.m. 7B. Euplica bidentata 140 jim from the back of the sac, scale bar = 100 \xm. 7C. Euplica bidentata 190 jtm from the back of the sac. Note change of cell field progressing from left to right ventrally, scale bar = 100 ji.m. 8A. Columbella paytensis 120 mm from the back of the sac. Note the weakly staining dorsal-ventral ridge of cells that bisect the sac into two halves. Scale bar = 100 (j.m. 8B. Columbella paytensis 220 urn from the back of the sac, scale bar = 100 jun. 8C. Columbella paytensis 330 p,m from the back of the sac. Note secretory products lying on top of the ventral membrane, scale bar = 100 u,m. 8D. Columbella paytensis 570 JJLITI from the back of the sac, scale bar = 100 \im. 8E. Columbella paytensis mature radula, anterior end, scale bar = 100 jim. 9. Columbella paytensis sagittal section. Note the buckling of the radular membrane and the middle plate at the high points of the buckles, scale bar = 200 \un. Abbreviations: dvr, dorsal/ventral ridge; lm, lateral membrane; It, lateral teeth; mb, membranoblasts; od, odontoblasts; rt, 'rachidian' tooth; se, subradular Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 epithelium; spre, supraradular epithelium; tsm, tooth socket membrane; vm, ventral membrane. mony, to differentiate true homologies from The strongest evidence for a different mech- homoplasies (Patterson, 1982). anism in secretion of the middle structure in The radula is a complex structure with Columbella and Euplica species comes from discrete parts that may or may not be homolo- comparison with Pseudanachis. In Pseu- gous between taxa. The criteria generally danachis, a central odontoblast field is located relied upon to infer similarity of radular teeth in the posterior of the sac concurrent with are simple, intuitive measurements such as the two odontoblastic cushions secreting the tooth number, position and shape in each row. lateral teeth. The position of odontoblastic An indication of the ubiquity of such informa- cushions in Pseudanachis is similar to that of tion as criteria in determining similarity is the non-euthyneuran taxa (Lutfy & Demian, 1964; existence of radular formulae, in which three Peters, 1979; Mischor & Markel, 1984). In the different kinds of teeth are recognized; lateral species of Columbella and Euplica we exam- teeth, marginal teeth and the rachidian. The ined, the sac is bifurcated, forming two lateral recognition of these tooth types in different odontoblastic fields. No middle odontoblastic taxa and their use as character states in phylo- cell field is present. Only anteriorly does the genetic analyses implies their homology, even dorsal-ventral ridge split, forming a horseshoe though the same tooth type in different taxa are shaped extracellular space. The posteriormost generally not morphologically similar. Addi- ventral secretory cells are located in the same tionally, the formulae themselves are assumed position as membranoblasts in other groups to indicate a hypothesis of homology, in terms studied (Peters, 1979; Mischor & Markel, of the number of each tooth type per row. 1984) and not arrayed along the back edge of We suggest that the mechanism of radular the sac like odontoblasts. This evidence formation is a stronger criterion in recognizing suggests that odontoblastic secretion is not similarity than positional information. We do responsible for formation of the middle tooth. not focus on morphological criteria in this Evidence based on the morphology of the study, because in columbellids, the middle membrane and teeth also supports a different plate is generally acuspate, and has been mechanism for middle structure formation. In described as reduced or suppressed (Ponder, Pseudanachis, the multicuspate rachidian 1973). The middle structure is called a rachid- appears concurrently with the lateral teeth. ian tooth on the basis of its position relative to Once formed, it lies on top of the radular the other teeth. The shape, size, and position membrane. Until the point of mineralization of of radular teeth, however, are determined in the rachidian, the membrane and rachidian most gastropod taxa by the position, size, and move ventrally to lie low in the sac but other- secretion rate of the odontoblastic cell fields wise do not change shape or position, except which form them (Mischor & Markel, 1984). If for thickening from the supraradular and cell fields are comparable between taxa, then subradular epithelium. the teeth should be comparable as well. Con- The rachidian structure of Columbella and structional information, then, is fundamental Euplica species may appear initially as a thin to understanding tooth morphology, especially layer of material on top of the already in situations where different formational thickened membrane. This layer is difficult to modes may result in positionally and morpho- distinguish from membrane in histological logically similar structures. cross sections, and in some cases is not defini- RADULAR HOMOLOGY Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 74 R. GURALNICK & MJ. de MAINTENON tively present. It differs from the mature ian' not be used when discussing the middle middle structure in being entirely surficial. The plate in columbellids we have examined, or mineralized rachidian plates are partially that it made clear that the 'rachidian' plate of embedded in the radular membrane, and are columbellids is not homologous with the still difficult to differentiate from membrane rachidian tooth in other gastropod taxa. material, except by differential staining. How The mechanism by which the middle such a structure might attain a thickness and structure forms in acuspate taxa is not known. depth in the membrane without increasing We present two different models that might membrane thickness itself is unknown. The explain the appearance of a regularly spaced forming rachidian plates of Columbella also middle structure on the radular membrane

appear to be longitudinally striated; this could (Figs 10A, 10B). Each model assumes that the Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 be caused by supraradular epithelium deposit- 'rachidian' plate is partly formed by membrane ing in discrete anterior-posterior strips or by material and supraradular epithelium, but are folding of material. otherwise quite different. One model operates The radular membranes in Columbella and under the assumption that mechanical defor- Euplica undergo a large amount of change mation and supraradular secretion is necessary after deposition, in contrast to Pseudanachis. and sufficient to explain the formation of The biphasic secretion of the ventral mem- middle structure (Fig. 10A), while the other brane with the initial membrane forming of the central region and then later forming of lateral regions seen in Columbella and Euplica has not been reported in other taxa. The narrow- ing of the membrane is also novel. After this occurs, the membrane becomes somewhat more brittle, and colours slightly yellow in untreated specimens; this change may be caused by protein tanning or sclerotization. Such a process would tend to enhance the mechanical strength of the radula, as well as its resistance to chemical or enzymatic break- down (Hopkins & Kramer, 1992). Runham (1961) showed that the radula is composed primarily of proteins and chitin, supporting the notion that sclerotization or tanning might occur. If the rachidian plate in the acuspate species is secreted by odontoblasts, we would expect to see odontoblastic fields in the back of the radular sac. We would also expect to see the rachidian plates definitively present on the posteriormost portion of the radular membrane in all species. The data we have collected does not support odontoblastic secre- tion of the acuspate rachidian plate. The data Figure 10. Schematic sagittal section showing the does support formation of the middle plate by two different models of middle plate formation. A. secretion from the supraradular epithelium Model showing mineralization of the middle plate onto the membrane (see discussion of models formed only by buckling of radular membrane due of tooth formation below). In a phylogenetic to tensile forces exerted by the lateral teeth and analysis, we would code presence/absence of change of position of the membrane from ventral to the middle odontoblast field and presence/ dorsal. Arrow 1 shows the point where mineraliza- absence of middle plates secreted by supra- tion occurs. B. Model showing cyclical secretion of a radular epithelium as separate, independent middle plate by a modified membranoblastic or characters. Therefore, the plate in Columbella supraradular cell field, at arrow 1. This portion of the membrane differentially signals mineralization, at and Euplica should not be considered homolo- arrow 3. Arrow 2 shows where a middle precursor tgous with the rachidian tooth of Pseudanachis, exists, but is not visible. Abbreviations: A, or probably with those of other molluscan supraradular epithelium; B, membranoblasts; C, sub- taxa. We suggest that either the term 'rachid- radular epithelium. RADULAR HOMOLOGY 75 model is based on the assumption that miner- widening of the membrane occurs after this alization from supraradular epithelium is initial sheet is deposited. Second, the existence signaled by prior cyclical secretion of one of of several different kinds of membranoblastic the secretory cell types and does not require a cell fields and several phases of mineralization mechanical explanation (Fig. 10B). These suggests the potential for different depos- models are not mutually exclusive and forma- itional products. The initial sheet of membrane tion may incorporate aspects of both models. is somewhat larger than the 'rachidian' plate, Lutfy & Demian (1964) and Mischor & but appears to be generated by more than one Markel (1984) demonstrate that the radular ventral cell type. Part of it may be cyclically membrane alternates between thick and thin, formed and act as a tooth primordium, or may with the thin parts lying underneath the be chemically structured in such a way that Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 radular teeth, and the thick parts lying allows it to mineralize, while the rest of the between rows of radular teeth. In Pomacea membrane is unable to do so. Any such bridgesi, the radular membrane is tallest differentiation of this region may not posterior to the tooth bases, and is about forty necessarily be visible until mineralization microns higher than the thinnest parts, which begins. are twenty microns thick. The height differ- The 'rachidian' plates are generally of a ence is explained by the anchoring of teeth, species-specific form, with definite edges; which is resistant to tensile stresses (Mischor & suggesting that the mechanism by which they MSrkel, 1984). Lutfy & Demian (1964) showed are formed is relatively consistent. We could that the upper surfaces of the highest parts of test both models for 'rachidian' plate forma- the radular membrane in Marisa corunarietis, tion by selectively destroying cell fields in live another ampullariid, are in contact with the animals using either mechanical or chemical overlying parts of the supraradular epithelium, means (Isarankura & Runham, 1968) and from which they apparently receive additional determining if the plate remains. If the secretion. mechanical model is correct, the only cell field In the acuspate columbellids, the distance responsible for secretion is the supraradular between the lateral teeth is relatively wide. epithelium and knocking out the other fields Sagittal sections of the material do not show would not alter the presence of the rachidian. different thicknesses of the central membrane If the cyclical secretion model is correct, then similar to the ampullariid taxa mentioned destroying cell fields other than the supra- above, but do show regular membrane buck- radular epithelium should affect middle plate ling. Between each row of teeth, it is possible formation. that the membrane in the acuspate groups is The functional significance of the acuspate thrust up higher than the surrounding mem- median plate of columbellids is unknown. Dis- brane where the lateral teeth are socketed. cussions of columbellid radulae (Bandel, 1974; This raised part is not mineralized until the Nielsen & Lethbridge, 1989) have not included point that the whole radula moves dorsally a detailed description of radular mechanics, such that it comes in contact with the except to note the unusual flexibility of the supraradular epithelia (Fig. 10A, arrow 1), lateral teeth. Bandel (1974) suggests the which moves at the same rate as the radula columbellid radula to be 'constructed for (Mischor & Markel, 1984). The middle struc- cutting, tearing and hooking' by rotation of the ture lies right behind, not in line with, the lateral teeth inward. The animals are known to lateral teeth, supporting such a mechanism. ingest a number of different food items, includ- The other formational mechanism is based ing algae in some taxa (Marcus & Marcus, on the possibility of cyclical secretion by a 1962; Bandel, 1974; Nielsen & Lethbridge, modified membranoblastic or supraradular cell 1989). Marcus & Marcus (1962) suggested that field (Fig. 10B, arrow 1), which may then cue radular morphology may be consistent with differential mineralization (Fig 10B, arrow 3). dietary variation between taxa; carnivorous Two main lines of evidence support this taxa tend to have square median plates and hypothesis. First, and most important, initial thin lateral teeth, and herbivorous taxa tend to central membrane formation in Columbella possess broad, thin median plates and rela- and Euplica produces only a thin sheet of tively larger lateral teeth. The median plate material in the center of the ventral membrane may help to stabilize the lateral teeth during region (Fig. 10B, arrow 1). This appears to usage; wear and breakage of the lateral teeth t>e formed through both dorsal and ventral suggests that they forcefully come into contact secretion. Thickening and, more importantly, with the substrate. The median plate does not 76 R. GURALNICK & MJ. de MAINTENON show wear, however, and is probably not assessment of similarity based solely on shape active in food gathering. Although there may and position of structures in the tooth row is be one or more potential adaptive uses for the inaccurate. median plate, it is equally possible that it Assessing similarity of radular parts may represents a phylogenetic artifact remaining also depend on changes in the number of cell after the loss of the odontoblastic field. fields, not just their presence or absence. Hick- man (1980) illustrates aberrant individuals of Solariella nuda Dall, 1896, in which the rachid- ian is fused with a lateral tooth on one side. If Systematic implications fusion of odontoblast cell fields occurs in the On the basis of anatomical evidence, colum- evolution of a lineage, the resulting tooth will Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 bellids are included in the muricoidian neo- not be distinguishable from a single tooth field gastropods, and are differentiated by several that evolved from the loss of cell fields. In characters of the gonoducts and their unique one case the resulting structure is homologous radular morphology (Ponder, 1973; de Main- with two teeth, while in the other case, it is tenon, unpublished). They are probably homologous with a single tooth. The timing of derived from a group in which a central tooth different cell field secretions can also vary is present. Loss of the odontoblastic field independently. For example, the timing of appears to be the major evolutionary change appearance of the lateral membranes in the which has occurred to produce the columbellid taxa we studied varies with respect to the radula, as the capability to mineralize a central appearance of the ventral membrane, suggest- structure, however it is secreted, still exists. ing that the secretion timing of the lateral Acuspate middle plates are present in a few membranes may be evolutionarily decoupled other taxa such as the buccinids Volutopsius from the ventral membrane. Our results show nonvegicus (illustrated in Bouchet & War6n, that radular formation occurs through a num- 1985) and Liomesus dalei (drawn in Ponder, ber of processes acting in concert, that may act 1973); whether they are formed similarly to very differently in different taxa. Our under- those of columbellids is unknown, but remains standing of the pattern of evolution of radular an intriguing question. In contrast, many structures will be greatly enhanced by under- gastropod taxa, such as harpids and volutids, standing the formational processes that lead to are often said to have evolutionarily 'lost' morphologies. entire radular teeth. The implications of our study are most important in these situations where teeth are traditionally assumed to be present, absent or reduced. Comparision of ACKNOWLEDGEMENTS structures and cell fields across taxa may give additional information concerning character We would like to acknowledge the California coding. Academy of Sciences and the Academy Our results concerning radular homology for of Natural Sciences of Philadelphia, and the columbellids are likely applicable to other particularly Gary Rosenberg, for providing gastropod taxa. For example, data collected some of the material for this project. We thank for more basal gastropod taxa show that posi- Carole Hickman, Brian Simison and Chris tion and shape information is not sufficient to Meyer for collecting specimens used in this assess homology in the radula. Moskalev study, and David Lindberg and Carole (1970; see Fig. 1A) has shown that the uncini Hickman for reviewing drafts of the manu- or 'marginal' teeth of Collisella, a patello- script and providing many helpful suggestions. gastropod, are formed on the membrane and Leigh Anne McConnaughey drew one of not on the basal plates like the lateral teeth. the figures and Karen Wetmore provided Basal plates are formed on top of the mem- assistance with the ESEM. We gratefully brane by basal plate cells, a different secretory acknowledge the Museum Informatics Project, cell type. Sollas (1907) also noted that in and Glen Jackson in particular, for making Patella vulgata the contrast in Bethe's stain available a Textronix printer for production between marginal and the lateral teeth is of the illustrations. A predoctoral NSF grant strikingly different. These data suggest that the to R. Guralnick partially supported this uncini are not equivalent to marginal teeth of research. This is contribution number 1649 of other taxa, but are modifications of basal the University of California Museum of plates, another example where an initial Paleontology. RADULAR HOMOLOGY 77 REFERENCES MOSKALEV, L.I. 1970. Gastropod of the Colisella (Prosobranchia, Acmaeidae) of BANDEL, K. 1974. The radulae of Caribbean and the fringing Asian seas of the Pacific Ocean. other Mesogastropoda and Neogastropoda. Transactions of the P. Shirshov Institute of Zoologische Verhandelingen, 214:1-188. Oceanology, 88:174-212. BOUCHET, P. & WAREN, A. 1985. Revision of the NIELSEN, J. & LETHBRIDGE, R. 1989. Feeding and the northeast Atlantic bathyal and abyssal neo- epiphyte food resources of gastropods living on gastropoda excluding Turridae (Mollusca, Gas- leaves of the seagrass Amphibolis gnfpthii in tropoda). Bollettino Malacologico, Supplemento 1: south-western Australia. Journal of the Malaco- 123-296. logical Society of Australia, 10: 47-58. CRACRAFT, J. 1981. The use of functional and adap- PATTERSON, C. 1982. Morphological characters and

tive criteria in phylogenetic systematics. American homology. In: Problems of phylogenetic recon- Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021 Zoologist, 21: 21-36. struction (K.A. Joysey & A.E. Friday, eds), 21-74. DE PINNA, M.C.C. 1991. Concepts and tests of Academic Press, London. homology in the cladistic paradigm. Cladistics, 8: PETERS, W. 1979. Basal bodies in the odontoblasts of 367-394. limpets, Patella coerulea L. (Gastropoda). Cell and HASZPRUNAR, G. 1988. On the origin and evolution Tissue Research, 202: 295-301. of major gastropod groups, with special reference PONDER, W.F. 1973. The origin and evolution of the to the Streptoneura. Journal of Molluscan Studies, Neogastropoda. Malacologia, 12: 295-338. 54:367-441. PONDER, W.F. & LINDBERG, D.R. 1996. Gastropod HICKMAN, C.S. 1980. Gastropod radulae and the phylogeny—challenges for the 90s. In: Origin and assessment of form in evolutionary paleontology. evolutionary radiation of the Mollusca (J. Taylor, Paleobwlogy, 6: 276-294. ed.), 135-154. Oxford University Press. HOPKINS, T.L. & KRAMER, K.J. 1992. Insect cuticle RADWIN, G.E. 1977. The family Columbellidae in sclerotization. Annual Review of Entomology, 37: the Western Atlantic. Veliger, 19: 403-417. 273-302. RAVEN, C.P. 1958. Morphogenesis: the analysis of ISARANKURA, K. & RUNHAM, N.W. 1968. Studies molluscan development. Pergamon Press Inc., on the replacement of the gastropod radula. London. Malacologia,T.l\-9\. RIEPPEL, O. 1988. Fundamentals of comparative LUTFY, R.G. & DEMIAN, E.S. 1964. The histology of biology, BirlchSuser, Basel and Boston. the radula and the radular sac of Marisa cornuari- RUNHAM, N.W. 1961. The histochemistry of the etis {L.).A'in Shams Science Bulletin, 10: 97-118. radula of Patella vulgata. Quarterly Journal of MARCUS, E. & MARCUS, E. 1962. Studies on Colum- Microscopical Science, 102: 371-380. bellidae. Faculdade de Filosofia, Ciencias e LetrasRUNHAM, N.W. 1963. A study of the replacement da Universidade de Sao Paulo. Boletim 261, mechanism of the pulmonate radula. Quarterly Zoologia, 24: 335-384, 8 pis. Journal of Microscopical Science, 104: 271-277. MACKENSTADT, U. & MARKEL, K. 1987. Experimen- SOLLAS, I.BJ. J907. The molluscan radula: its tal morphology of radular renewal in pulmonates chemical composition and some points in its (Mollusca, Gastropoda). Zoomorphology, 107: development. Quarterly Journal of Microscopical 209-239. Science, 51: 115-136. MISCHOR, B. & MARKEL, K. 1984. Histology and TROSCHEL, F.H. 1856-1893. Das Cebiss der regeneration of the radula of Pomacea bridgesi Schnecken zur Bcgrilndung einer natiirlichen (Gastropoda, Prosobranchia). Zoomorphology, Classification. 2. Berlin. 104: 42-66. Downloaded from https://academic.oup.com/mollus/article/63/1/65/978311 by guest on 23 September 2021