he ne LE Am ah he nn A alana, Huhn end en al ni eh RCIP CIC ER PET EC PCT EN
HR ER EN,
at ne eis ana EUR OPA RS
CA 1s Be wip tare
CON An Th A Para Det vo
REN SI re u tue an 4
CES)
APCE PL TEE
CET En
CCC
CPP
nr ame
= heme
ae
AAO
EN
eee
Are
Sow EEE LU à ER ba
= ~
An, : FAA PAS à 8 laa DS
Paleontological
Research
ISSN 1342-8144
Formerly Transactions and Proceedings of the Palaeontological Society of Japan
Vol. 5 No.1 April 2001
The Palaeontological Society of Japan + :
Co-Editors Kazushige Tanabe and Tomoki Kase Language Editor Martin Janal (New York, USA)
Associate Editors
Jan Bergström (Swedish Museum of Natural History, Stockholm, Sweden), Alan G. Beu (Institute of Geological and Nuclear Sciences, Lower Hutt, New Zealand), Satoshi Chiba (Tohoku University, Sendai, Japan), Yoichi Ezaki (Osaka City University, Osaka, Japan), James C. Ingle, Jr. (Stanford University, Stanford, USA), Kunio Kaiho (Tohoku University, Sendai, Japan), Susan M. Kidwell (University of Chicago, Chicago, USA), Hiroshi Kitazato (Shizuoka University, Shizuoka, Japan), Naoki Kohno (National Science Museum, Tokyo, Japan), Neil H. Landman (Amemican Museum of Natural History, New York, USA), Haruyoshi Maeda (Kyoto University, Kyoto, Japan), Atsushi Matsuoka (Niigata University, Niigata, Japan), Rihito Morita (Natural History Museum and Institute, Chiba, Japan), Harufumi Nishida (Chuo University, Tokyo, Japan), Kenshiro Ogasawara (University of Tsukuba, Tsukuba, Japan), Tatsuo Oji (University of Tokyo, Tokyo, Japan), Andrew B. Smith (Natural History Museum, London, Great Britain), Roger D.K. Thomas (Franklin and Marshall College, Lancaster, USA), Katsumi Ueno (Fukuoka University, Fukuoka, Japan), Wang Hongzhen (China University of Geosciences, Beijing, China), Yang Seong Young (Kyungpook National University, Taegu, Korea)
Officers for 1999-2000 President: Kei Mori
Councillors: Kiyotaka Chinzei, Takashi Hamada, Yoshikazu Hasegawa, Itaru Hayami, Hiromichi Hirano, Noriyuki Ikeya, Junji Itoigawa, Tomoki Kase, Hiroshi Kitazato, Itaru Koizumi, Haruyoshi Maeda, Ryuichi Majima, Makoto Manabe, Hiroshi Noda, Ikuwo Obata, Kenshiro Ogasawara, Terufumi Ohno, Tatsuo Oji, Tomowo Ozawa, Yukimitsu Tomida, Tsunemasa Saito, Takeshi Setoguchi, Kazushige Tanabe, Akira Yao
Members of Standing Committee: Hiroshi Kitazato (General Affairs), Tatsuo Oji (Liaison Officer), Makoto Manabe (Finance), Kazushige Tanabe (Editor in Chief, PR), Tomoki Kase (Co-Editor, PR), Ryuichi Majima (Planning), Hiromichi Hirano (Membership), Kenshiro Ogasawara (Foreign Affairs), Haruyoshi Maeda (Publicity Officer), Noriyuki Ikeya (Editor, "Fossils"), Yukimitsu Tomida (Editor in Chief, Special Papers), Tamiko Ohana (Representative, Union of Natural History Societies).
Secretaries: Masanori Shimamoto, Takao Ubukata (General Affairs), Hajime Taru (Planning), Tokuji Mitsugi (Membership), Shuko Adachi (Foreign Affairs), Kazuyoshi Endo, Yasunari Shigeta, Takenori Sasaki (Editors of PR), Akira Tsukagoshi (Editor of "Fossils"), Naoki Kohno (Editor of Special Papers), Hidenori Tanaka (Publicity officer)
Auditor: Nobuhiro Kotake
Notice about photocopying: !n order to photocopy any work from this publication, you or your organization must obtain permission from the following organization which has been delegated for copyright for clearance by the copyright owner of this publication.
Except in the USA, Japan Academic Association for Copyright Clearance (JAACC), 41-6 Akasaka 9- chome, Minato-ku, Tokyo 107-0052, Japan. Phone: 81-3-3475-5618, Fax: 81-3-3475-5619, E-mail: kammori@msh.biglobe.ac.jp
In the USA, Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. Phone:
(978)750-8400, Fax: (978)750-4744, www.copyright.com
Cover: Idealized sketch of Nipponites mirabilis Yabe, a Late Cretaceous (Turonian) nostoceratid ammonite. Various reconstructions of the mode of life of this species have been proposed, because of its curiously meandering shell form (after T. Okamoto, 1988).
All communication relating to this journal should be addressed to the PALAEONTOLOGICAL SOCIETY OF JAPAN c/o Business Center for Academic Societies, Honkomagome 5-16-9, Bunkyo-ku, Tokyo 113-8622, Japan Visit our society website at http://ammo.kueps.kyoto-u.ac.jp/palaeont/
Paleontological Research, vol. 5, no. 1, pp. 1, April 30, 2001 © by the Palaeontological Society of Japan
Morphological approaches in paleobiology
This special issue is partly based on the workshop of the Palaeontological Society of Japan on “Fossils and Mor- phology” held at the Misaki Marine Biological Station, University of Tokyo, in Miura City, Japan, from April 21 to 24, 2000. The workshop particularly focused on morphological approaches to paleobiological studies, from both fundamen- tal and practical points of view. Most of the 34 participants of the workshop were young paleontologists interested in morphology.
Research on organic evolution over the Earth’s history necessarily depends on the morphology of hard tissues pre- served as fossils. Many paleontologists have developed and refined various methodologies for handling the raw mor- phological data presented by fossils and related extant or- ganisms. However, there has been little reciprocal interaction or feedback in this process. For instance, the skills of an expert in comparative anatomy and the compli- cated methods handled by a specialist in mathematical mor- phology are still too far apart from each other. It would seem to be an important goal to make a wide variety of mor- phological methods accessible to all paleontologists, particu- larly younger ones.
The workshop setting made it possible to examine the morphological aspects of fossils from a number of view- points and provided an opportunity for young paleontologists to learn various methods in morphology. Most major topics concerning morphology, such as functional morphology, constructional morphology, morphometrics, biometry, theo- retical morphology, developmental constraints, developmen- tal genetics, heterochrony, cladistics, comparative anatomy, histology, biomineralization, etc., were covered. Topics in- cluded current research on various zoophyla, such as molluscs, arthropods, echinoderms and vertebrates, but there were also abstract discussions of methodology or prin- ciple that did not deal with particular groups of organisms. We did not attempt to construct a synthesis of those diverse topics. However, the workshop seems to have succeeded in addressing why so many morphological methods are valu- able for paleobiology.
In consequence, the workshop was also a good occasion on which to plan a special issue of Paleontological Research on the use of morphology in paleobiology. The present issue partly reflects the activities at the workshop but does not aspire to represent the complete proceedings. The six contributions offered herein cover only a part of the topics presented at Miura. Among the senior authors of this issue, Enrico Savazzi, Takenori Sasaki and Takao Ubukata pre- sented their papers in the workshop, and Richard Reyment and Kazushige Tanabe were welcome additions to the con- tributors to this special issue. All contributions in this issue underwent the regular review and editing process of the
(JUL 12 2001 > LIBRARIES _
journal. In addition, several papers:which were prepared for this special issue but did not complete the review process in time will be treated as regular submissions.
This collection is diverse, to the extent that perhaps there is no coherent theme. It covers specialized topics such as the morphometrics of ammonites, morphodynamics of an endolithic gastropod, macrosymbiosis in bivalves, early shell morphology of ammonoids, comparative anatomy of gastro- pods, and the theoretical morphology of bivalve shell struc- ture. Although the papers contained in this issue differ in scope, each touches on the phylogenetic, functional and/or morphogenetic aspects of an organic form. These three as- pects may be conceived of as the parameters of a Seilacher's triangle of constructional morphology, in which organic form is postulated to result from the interplay of the three factors. An integrated approach focusing on several aspects of organic form is becoming more and more impor- tant as paleobiological researches broaden out to include subjects of evolutionary biology. This issue will serve as a benchmark of the present state of this field and indicate lines of inquiry for future research to follow.
It should be noted that most of the approaches in the pa- pers contained in this special issue were based on handy, “low-tech” methods such as observation or mathematical analysis, and required neither cutting-edge high technology nor the supporting framework of a large project. This col- lection indicates that morphology based on simple and eco- nomical techniques remains an exciting and creative field of science. This fact should encourage the young paleontolo- gist who might feel that his or her work is mere handicraft in comparison to what colleagues who participate in large- scale projects in fashionable high-tech fields are doing. | hope that the topics presented here will be of interest to all paleontologists, and that this special issue will awaken the interest of many students in the field of morphology. | be- lieve that new approaches to morphological studies will play a key role in paleontology in the 21st century.
| thank Tomoki Kase and Kazushige Tanabe, co-editors of Paleontological Research, for their help in the course of the editorial process of this issue. Thanks are also due to Rihito Morita for his cooperation in organizing the workshop, to the staff of Misaki Marine Biological Station for their kind hospitality during the workshop, to all participants in the workshop for their active and valuable discussions, and to all of the authors for their thoughtful contributions to this special issue. Although this is only the first time that Paleontological Research has collected papers on a specialized theme, | hope it will not be an isolated instance but the first of many to come.
—TAKAO UBUKATA JANUARY 15, 2001
LT oe éd
. en : en
Loney oo, 2
y.
> = > + cut = Mi ay |e L | u 7 13 F 46
Ù née ve (AY. gal DEL Hat head MR ze aaa ee Pr. aide. NON
tu orbite cf em: 7
. #10 a er a ae) avoit dot Stews Tye sliaeglt vs
N ue! 4 +e) ich! f er) Ni EN PE 4 rar | u > UE. Met Meg j “ei
À “sh - 4 ? we hrs Fab ren a “« Bere "oy a. = ee ell
ak Tie 18 an QO Omi ; - Vy Te OMT SUET RE 5
5 SA Che bel A ES à
Paleontological Research, vol. 5, no. 1, pp. 3-11, April 30, 2001 © by the Palaeontological Society of Japan
Morphodynamics of an endolithic vermetid gastropod
Enrico Savazzi Hagelgränd 8, S-75646 Uppsala, Sweden. (e-mail: enrico.savazzi@usa.net)
Received 29 May 2999; Revised manuscript accepted 15 November 2000
Abstract. In coiled mollusc shells, the apex typically is located near the centre of a whorl spiral and the aperture at its outer perimeter (exceptions do occur in molluscs with determinate or peri- odic growth, but they do not invalidate this general principle). This geometry satisfies simple growth and functional requirements. The Recent vermetid Dendropomasp. defies these require- ments with its inside-out shell geometry, in which the aperture is located along the axis of the spiral and earlier whorls coil around it. In addition, this species is unique among the Vermetidae in being fully endolithic in the adult, and is one of very few endolithic molluscs with the shell cemented to the substrate during growth. While Dendropoma is typically semi-endolithic, several species ap- pear to have secondarily returned to epifaunal coiling as a response to crowding of the substrate. In D. sp., this was prevented by the immediate environment, subjected to a high rate of erosion. This is likely the factor that triggered the onset of endolithic coiling into the substrate. The change from a semi-endolithic to a fully endolithic life habit in this form was probably sudden, since the
preadaptations of Dendropoma prevent a functioning intermediate stage.
Key words: endolithic, evolution, functional morphology, Gastropoda, Mollusca, Vermetidae.
Introduction
In the adult stage, almost all vermetid gastropods are ce- mented by the shell to a solid substrate (Morton, 1955, 1965; Keen and Morton, 1960; Keen, 1961; Savazzi, 1996, 1999a, and references therein). As a consequence of this habit, vermetid shells are irregularly coiled or lose all traces of coil- ing in the adult stage (above references, and below). Most vermetids remove a small amount of material from calcare- ous substrates before cementing the growing shell margin to it (Savazzi, 1996, and therein). This has the apparent adaptive value of removing superficial layers weakened by encrusting organisms and microborers, and of improving the adhesion between shell and substrate. Observation on free-living gastropods with regularly coiled shells shows that a comparable phenomenon, consisting of a secondary re- moval of periostracum and a thin outermost layer of shell material, is common at the junction between adjacent whorls. In the latter case, this phenomenon likely enhances adhesion between successive shell whorls. It is reasonable to suppose that removal of the superficial layer of substrate in vermetids may have evolved from the latter, widespread feature (Savazzi, 1996). It is not known in detail how this process is carried out by vermetids, but, as in most in- stances of secondary shell resorption in molluscs, the proc- ess is likely of a chemical, rather than mechanical nature.
A few vermetids remove a larger amount of substrate, and excavate a trench sunk into the surface of the substrate, into which subsequently they secrete their shells. This trait is
particularly developed in the genus Dendropoma, which can be characterised as semi-endolithic because a substantial portion of the shell lies below the original surface of the sub- strate (Keen and Morton, 1960; Keen, 1961; Savazzi, 1996, 1999a; references therein, and below). In several species of this genus, the shell cavity is covered by a thickened “roof’ of shell material, level with or slightly projecting from the surface of the substrate (Figures 1A, 2, and below).
This paper discusses the adaptations and evolution of a Recent species -or ecomorph -of Dendropoma with a unique combination of morphologic and ontogenetic characters. The taxonomic position of this form, and whether or not its adaptations and morphology warrant the introduction of a new supraspecific taxon, are discussed summarily, but a de- cision on these matters is not taken. This organism is hereby referred to as Dendropoma sp. because its closest affinities are clearly with this genus. This form differs from typical species of Dendropoma in that its adult shell is com- pletely embedded in the substrate, except for the shell open- ing (see below). This habit can be characterised as fully endolithic, as opposed to the semi-endolithic habit typical of this genus (see above, and below). It is unique among vermetids.
In order to appreciate the unique character of the shell ge- ometry in Dendropoma sp., it is useful to review the principal geometric properties of shells that grow by marginal accre- tion. Coiled mollusc shells typically have the apex located on the axis of a planispiral or helicospiral formed by later whorls. The aperture, on the other hand, is located at the
Enrico Savazzi
Figure 1. Exterior view of substrate (B-D) and view of aperture and internal shell cavity seen from below the surface of the substrate (E-G). C,E:x4;D, G: x 6.
outer periphery of the spiral. This geometry satisfies simple functional and ontogenetic requirements: it allows continued growth, since earlier shell portions are not in the way of fur- ther growth, and yields a relatively compact shell geometry (compatibly with accretionary shell growth). Exceptions to this rule are known, but they do not invalidate the general principle. These exceptions are typically associated with a
A. Dendropoma annulatum, Harrington Sound, Bermuda, x 2. B-G. Dendropoma sp., Hilotongan Island, the Philippines.
B: x 2;
“count-down” growth process (Seilacher and Gunji, 1993), in which further shell growth is impossible. This growth proc- ess is characterised by “preparatory” morphological changes that precede the attainment of the adult size and the cessa- tion of further growth. For instance, several land snails bring the adult aperture close to the coiling axis in the adult stage, and a few bend the aperture back toward the apex, or
Morphodynamics of vermetid gastropod
direction of secretion direction of etching
latest growth increment
substrate to be etched next
Figure 2. Schematic section through apertural region of typi- cal semi-endolithic Dendropoma. The surface of the substrate is at the top. The shell cavity has been cut longitudinally through a plane perpendicular to the surface of the substrate and bisecting the shell aperture (the original outline of the aperture is indicated by a dashed line). Note the growth lines on the surface of the shell “roof’ and within its structure. Arrows indicate the direction of migration of the aperture along the surface of the substrate during growth, caused by the twin processes of substrate etching (stippled area) and shell secretion (cross-hatched area).
away from it (e.g., Paul, 1999, and references therein). The same principle applies to several heteromorphic ammonoids with count-down shell geometries. A few gastropods (e.g., Distorsio), avoid this problem by returning periodically to a more conventional geometry of the last shell whorl, thus al- lowing the repetition of the count-down process. Dendropoma sp., instead, seems to defy the laws of shell geometry and growth by keeping the apertural region of the shell located at the centre of the whorl spiral, and earlier whorls coiled around it (see below) throughout its adult stage, while at the same time extending the length of its shell and increasing its depth within the substrate.
Material and methods
The species described in this paper is abundant (Figure 1B) in a band extending from high water mark to about 1 m below it on a rocky shore on the east coast of Hilotongan Island, the Philippines (123° 59° 10”E, 10° 12° 50°°N, as measured on Philippine government maps). Population
density exceeds 30,000 adult individuals per m” in the dens- est patches. The surface of the substrate is inclined to the vertical and relatively even. The substrate is a Quaternary, coarse-grained, poorly sorted and somewhat recrystallised biogenic calcirudite, containing occasional large bioclasts (mostly fossil scleractinians). Other areas of this island, as well as most of the nearby islands, were visited repeatedly by the writer, but this species was not observed elsewhere. Samples of the substrate containing the endolithic vermetid were fixed in dilute buffered formaldehyde and sub- sequently kept in 5% ethanol. The substrate was bathed in dilute chlorine and mechanically scrubbed to remove adher- ent algae prior to drying for observation and photography.
Morphology and development of Dendropoma
In order to appreciate the affinities of Dendropoma sp. with other representatives of this genus, as well as the unique characters of this form, it is necessary to discuss the life habits and morphology of typical Dendropoma species.
Sexes are separate. Female individuals of Dendropoma house a small number of large embryos within the mantle cavity (Keen, 1960; Hughes, 1978). A veliger stage is absent, and the juvenile passes through a crawling stage be- fore cementing to the substrate. Dispersion of juveniles may be facilitated by wave or current action, aided by the se- cretion of a mucus thread that retards sinking. However, the sinking rate of juveniles is too high to regard passively transported juveniles as planktonic (Hughes, 1978). Typically, the unattached juvenile shell is trochospiral, brown in colour, and consists of about two whorls.
As in other vermetids (above references), the juvenile shell initially becomes cemented to the substrate by its outer lip. Subsequent shell growth continues along a regular helicospiral trajectory, and lifts the aperture away from the substrate. At this point, the aperture bends toward the apex and reattaches to the substrate by its dorsal region. Shortly afterwards, the mollusc starts partially to sink the shell within the substrate, by eroding the substrate in the region adjacent to the aperture (Savazzi, 1996, 1999a, and references therein).
In adult Dendropoma (except for the form in question), two shell geometries and modes of growth are represented. In the first, which is exclusive to this genus, the shell follows an irregular path on the substrate, partly sunk within its surface (Figure 1A; Keen and Morton, 1960; Keen, 1961; Hadfield et. al., 1972; Savazzi, 1996). Because of the recessed po- sition of the shell within the substrate, the body whorl is bent outward at an angle of roughly 90° in the region near the ap- erture, so the plane of the aperture is parallel to the surface of the substrate and perpendicular to the growth direction of the shell (Figures 1A, 2). As a consequence, the aperture moves sideways during growth. Shell secretion takes place along half of the perimeter of the aperture, while erosion of the substrate takes place simultaneously along the other half (Figure 2). The exposed portion of the shell constitutes a “roof’ flush with or slightly projecting from the surrounding substrate. Earlier positions of the aperture remain visible as coarse, falcate growth rugae on this roof (Figures 1A, 1C, 3A). The morphology of this relief-pattern is used as a taxo-
6 Enrico Savazzi
Figure 3. Juvenile shell showing initial cementation scar (A) and adult operculum in interior view (B) and schematic cross-section
(C-D) of Dendropoma sp. Scale bars represent .5 mm.
nomic character at the species level. This mode of growth is common among immature individuals of most species of Dendropoma, and in the adult stage of solitary species and morphs.
The second mode of growth in this genus consists of piling successive whorls on top of each other, forming an irregu- larly coiled helicospiral shell. This mode of growth is well represented in other vermetid genera, such as Serpulorbis, Tripsycha and Petaloconchus (e.g., Keen, 1961; Savazzi, 1996, 1999a). In Dendropoma, this morphology is frequent in gregarious species and on crowded substrates.
In the adult stage, vermetids exhibit dextral coiling, or no coiling at all (above references). Since the adult shell por- tion is cemented to the substrate, or to earlier whorls, by its dorsal surface (see above), the whorls coil in the counter- clockwise direction when viewed from above. Sharp bends of the shell in a sinistral (i.e., clockwise) direction are uncom- mon, and vermetids generally do not exhibit the meandering in alternate directions which is frequently seen in cemented serpulid polychaetes. As discussed by Savazzi (1996), sinistral bending in the Vermetidae is probably prevented by the laterally asymmetrical placement of the columellar mus- cle.
Dendropoma possesses a corneous operculum, the mor- phology of which is used as a species-level taxonomic char- acter (Keen and Morton, 1960; Keen, 1961; Morton, 1965; Hadfield et al., 1972). Common to all species of Dendro- poma is a secondarily thickened central portion of the operculum, forming an inward projecting boss or elongated “handle” (above references, and below).
Several species of Dendropoma, as other vermetids, are gregarious and form exclusive or olygotypic associations in the intertidal and shallow subtidal zones, sometimes building small reefs (Keen and Morton, 1960; Keen, 1961; Kempf and Laborel, 1968; Hadfield et al., 1972; Hughes, 1978, and personal observations).
Morphology and growth of Dendropoma sp.
Larval and juvenile development
Females incubate few (from 6 to 12) large eggs within the mantle cavity. Mature embryos contain a well developed protoconch consisting of approximately 2 whorls. This shell is translucent, smooth when observed under a dissection mi-
croscope, and brown-coloured.
Samples of the substrate inhabited by adults of this spe- cies were found to harbour numerous juvenile specimens, usually nested in small crevices. As typical of vermetids (above references), juveniles first cement to the substrate by the outer shell lip. Immediately afterwards, the shell bends in the apical direction, usually lifting free of the substrate, and subsequently reattaches to the substrate by the (homologically) adapical surface of the whorl.
The size of the shell in the smallest attached juveniles is comparable to that of the largest larval shells. Therefore, a veliger stage is apparently lacking. This agrees with obser- vations on other species of Dendropoma (see above refer- ences).
Immature shell
At this stage, the shell is light brown to white with occa- sional brown patches (especially on its interior surface), car- ries coarse falcate ridges on its exposed surface, and is slightly sunk within the substrate. Except for the rather small size of D. sp., young specimens (central portion of Figure 1B) are essentially identical, in general appearance, to the adult stages of several other species of Dendropoma (e.g., Figure 1A).
Some of the individuals of D. sp. afterwards continue to wander in an apparently random fashion during growth, per- haps as a response to a crowded or irregular substrate. Most of the individuals, however, after an irregular early stage build an evenly curved, counterclockwise arch (Sur- rounding the adult shell aperture of several individuals in Figure 1D). The radius of this arch, which gradually ex- tends to one full whorl, is rather constant in length among different individuals, and the free space enclosed within the arch has a diameter comparable to that of the adult shell ap- erture. While building this arch, the apertural region of the shell becomes gradually embedded deeper within the sub- strate, thanks to a gradual increase in etching activity. Earlier portions of the body whorl do not change their posi- tion relative to the substrate.
Adult shell
Upon reaching one whole, broadly umbilicate whorl, the apertural portion of the shell, instead of rising onto and over- lapping the earlier portion of the whorl as in other vermetids
Morphodynamics of vermetid gastropod 7
(see above), dips beneath it and into the substrate. At the same time, the shell aperture bends toward the umbilicus, thus avoiding the erosion of earlier portions of the arch. It is convenient to characterise these events as the beginning of the adult growth stage of the shell, although they may have no connection with reproductive maturity.
From this moment onward, the shell aperture opens within the umbilicus (Figure 1D-G). Further growth continues in the same fashion, with subsequent whorls becoming embed- ded deeper within the substrate, and the aperture located at the bottom of an umbilical well surrounded by earlier whorls (Figures 1C-G, 5). As a result, the shell is coiled in an ap- parently sinistral fashion. However, since the shell initially attaches to the substrate by the adapical side of the whorl, this coiling geometry is not sinistral, but hyperstrophic dex- tral.
Substrate-etching continues to take place within the aper- ture, which revolves around its axis and moves deeper within the umbilical well. This results in the formation of a pattern of spiralling ridges on the walls of the umbilical well (Figure 1D, bottom of Figure 1C, centre of Figure 5A). These ridges are homologous to the falcate ridges on the exposed surface of the shell in semi-endolithic species of Dendropoma. The largest individuals of D. sp. display an internal shell cavity consisting of about two whorls. The earliest, most superficial of the preserved shell whorls are commonly eroded (Figure 1C), with the shell cavity exposed and closed off by internal septa (rightmost portion of Figure 1D). Thus, this portion of the shell is gradually vacated and sealed off by the mollusc. In some specimens, judging from the diameter of the shell cavity, at least one or two additional whorls were originally present, and were destroyed by rapid erosion of the substrate. This is further supported by the frequent occurrence of radular scratch marks on the surface of the substrate (probably caused by chitons and archaeogastropods, which are both frequent in crevices and sheltered areas), and by a general lack of cemented epifauna.
Operculum
The operculum (Figure 3B-D) is corneous, white with red- dish-brown periphery and nucleus. It is variable in general shape, with the outer surface slightly concave to highly domed (Figure 3C-D), carrying irregular concentric ridges but no visible spiral ornamentation. The perimeter consists of a thin lamina surrounding a thickened and bulging ring (Figure 3B-D), which, unlike the rest of the outer surface, is smooth and shiny. The inner surface of the operculum is smoother and carries an elongated axial “handle” (Figure 3C-D) and a semicircular or horseshoe-shaped dark patch near the basis of the handle (Figure 3B).
Soft parts
The general appearance of the body (Figure 4) is consis- tent with that of other species of Dendropoma (above refer- ences). The body is rather short, with a well developed columellar muscle. The mantle is slitted in the female, whole in the male. The pedal and cephalic tentacles ap- pear to be rather short. A detailed anatomical investigation was not carried out.
operculum
cephalic tentacles
pedal Lash
tentacles
foot
mantle
columellar muscle
Figure 4. in lateral view.
Soft parts and operculum of male Dendropoma sp., Scale bar represents 5 mm.
Affinities
The anatomy of the soft parts, morphology of the operculum and general shell characteristics (above, and Figures 1, 3-5) show this species to be a member of the vermetid genus Dendropoma Morch, 1861 (see also Keen, 1961). The gregarious habits and morphology of the operculum are consistent with placement in the subgenus Novastoa Finlay, 1927. However, as noted by Keen (1961) this subgenus may be artificial. Furthermore, D. sp. is con- spicuously different from all other Dendropoma in its shell geometry, mode of growth and endolithic habit. These dif- ferences are at least as important as those on which the ex- isting subgenera of Dendropoma (i.e., Dendropoma s.s., Novastoa and Elliptovermetus Cossmann and Peyrot, 1922) are based. If future observations show that the shell mor- phology and life habits of D. sp. as described in this paper are a constant feature of this species, rather than an ecotypic response to a particular environment, it might be justified to erect a new subgenus to accommodate its unique character. However, a review of Dendropoma from the Philippines would be necessary to settle the problems con- nected with the specific and subgeneric placement of this genus, and such a study lies outside the scope of the pre- sent paper.
Evolution
The evolution of the endolithic habit in D. sp. cannot be explained as a process characterised by a gradual increase
8 Enrico Savazzi
B
Figure 5. Internal mould of shell cavity (A) and cross-section through the deepest portion of the shell cavity (B), cut through a plane parallel to the surface of the substrate, in Dendropoma sp. In A, the early portion of the shell cavity is closed off by septa. A sculpture of coarse oblique rugae is visible on the walls of the apertural well (vertical portion of shell cavity located at the centre of the spiral). The shell aperture proper is located at the bottom of the apertural well.
in depth of the trench excavated by Dendropoma. This sim- ple mechanism would produce an external roof above the whorls, which is absent in D. sp.
As an alternative, one could envision a growth process in which earlier portions of the shell gradually migrate deeper within the substrate during growth, by eroding substrate along one side of the shell cavity and secondarily depositing shell material on the opposite side (i.e., in a manner analo- gous to the sideways ontogenetic migration of the aperture). However, this process would leave telltale growth lines in the shell structure indicating earlier positions of the whorls within the substrate. This situation can be compared with the backwards and/or sideways boring process displayed by endolithic lithophaginid bivalves when they move away from their original position within the substrate. This process leaves meniscus-shaped layers of calcareous material filling the vacated region of the borehole (e.g., see Savazzi, 1999b, fig. 17.2B). Such structures, instead, are absent in D. sp., thus ruling out the above process and showing that the position of earlier shell portions within the substrate does not change.
A more unconventional growth mechanism (Figures 5B, 6) explains the observed features. The apertural portion of the shell, which is located at the bottom of the apertural well, grows in the manner typical of Dendropoma, i.e., by etching substrate along half of its circumference, and depositing shell material along the opposite half. However, in D. sp. the shell aperture, instead of migrating along a path parallel to the surface of the substrate, revolves around the bottom of the apertural well. As a result, the aperture rotates and progressively deepens within the substrate, leaving on the walls of the apertural well a set of spiralling, curved ridges homologous to those on the roof of semi-endolithic species (albeit arranged in a different geometric pattern because of the different trajectory of the aperture). At the same time, the rest of the shell cavity becomes coiled around the
A
B
Figure 6. Schematic section through the bottom portion of the shell cavity of Dendropoma sp. along a plane parallel to the surface of the substrate (which is located above the plane of the figure), showing the shell growth process. Substrate erosion takes place along a por-
tion of shell cavity in proximity with the apertural well (A).
The direction of growth follows a circular trajectory (arrow in B) around the
apertural well, gradually moving deeper within the substrate at the same time. This process is recorded by growth lines in the shell wall that surrounds the apertural well (B). Note that, with the exception of the gradual deepening of the apertural well, neither the apertural well nor earlier coiled portions of the shell cavity change their position relative to the substrate during growth.
Morphodynamics of vermetid gastropod 9
apertural region because it is connected to the revolving ap- erture by a short bent portion (Figures 5A, 6). As a result, successive whorls are located progressively deeper within the substrate, trailing behind the aperture like a spiral stair- case. Instead of coiling outwards from the surface of the substrate, like regular vermetids, D. sp. coils into the sub- strate. The direction of coiling, however, remains dextral as in epifaunal vermetids. Since in all vermetids the adapical shell surface is cemented to the substrate, the different di- rection of translation of the whorls along the coiling axis in D. sp. results in an inverted direction of translation of the helicospiral. Therefore, the coiling of D. sp. is not sinistral as could appear from a superficial observation, but hyperstrophic dextral (see above). Early portions of the shell cavity are vacated and closed off by septa, and thus the soft parts effectively migrate deeper within the substrate, in spite of the shell being cemented to it.
This growth process achieves the combined results of deepening the apertural portion of the shell (which is the only growing shell portion, since earlier shell portions remain immobile with respect to the surrounding substrate) without moving it laterally along the surface of the substrate. In ad- dition, it makes a fully endolithic habit possible without in- creasing substantially the volume of substrate that must be eroded, compared to a semi-endolithic habit. Finally, it achieves these results without requiring substantial changes in the nature of the mechanisms involved in the shell con- struction and growth of Dendropoma.
Morphodynamics
The evolutionary process that led to this peculiar morphol- ogy and ontogeny remains to be explained. Biological morphodynamics (Seilacher, 1991) is a _ conceptual framework stating that the morphology and evolution of an organism can be understood as the interplay of four factors: function (i.e., all aspects of morphology with a direct adap- tive significance), construction (including morphogenesis, structural materials and building principles), phylogenetic tradition (the evolutionary history, preadaptations and Bauplan; see below) and immediate environment (the char- acteristics of the environment in close proximity with the or- ganism). The Bauplan is a set of constructional “building blocks” and morphologic characters that constitute the shared features of a lineage or taxonomic group (although they may not be expressed in all its members).
Morphodynamics is a recent extension of constructional morphology (sensu Seilacher, 1970; an alternative definition of the latter exists, but is not discussed here), which includes only the first three of the above factors. Both morpho- dynamics and constructional morphology were meant as practical frameworks to guide and summarise one’s reason- ing while carrying out an analysis of morphology, not as ab- stract philosophical generalisations. Therefore, the merits of these frameworks should be judged in the context of spe- cific instances of functional analysis, like the present one.
In both conceptual frameworks, lateral migration of the shell aperture in typical Dendropoma is the preadaptation that allowed the evolution of the unique ontogenetic mecha- nism of D. sp. The spiral coiling of this species likewise origi-
nates from the generic gastropod Bauplan. While adult Dendropoma is facultatively not coiled, its larval shell is obli- gatorily coiled, just as the adult shell in noncemented ances- tors of the Vermetidae was regularly coiled. Spiral shell coiling is still visible in the facultative or obligatory (albeit somewhat irregular) adult coiling of several vermetid genera (Keen, 1961; Savazzi, 1996, 1999a, and therein). Dextral shell coiling is also related to the laterally asymmetric place- ment of the columellar muscle in the Vermetidae (including those that are noncoiled in the adult stage). Thus, it is le- gitimate to state that spiral coiling is part of the vermetid Bauplan.
In the present case, morphodynamics may display a prac- tical advantage over constructional morphology. The latter provides an explanation for the morphology of D. sp. in an adaptive context, but does not explain clearly the cause of its evolution. Morphodynamics, instead, offers a better framework for explaining both. The rock substrate inhab- ited by D. sp. is characterised by a high rate of erosion (as evidenced by grazing tracks and by the fact that early por- tions of the shell of this species are often eroded away; Figure 1E-F). Significantly, the only other sessile inverte- brate common in the same substrate is a rock-boring foraminiferan that lives in a shallow pit (dark grey patches in Figure 1F) and continuously deepens it in order to remain protected against erosion (J. Whittaker and E. Savazzi, un- published). It can be noticed also that the substrate is crowded (Figure 1B, E-F). These are the key environ- mental factors to explain the evolution of D. sp.
Fast erosion of the substrate encouraged a switch from a semi-endolithic to an endolithic habit, because the latter offers a better protection against erosion, and thereby an en- hanced probability of survival. Most endolithic gastropods and bivalves fight erosion of the substrate by boring deeper within the substrate during growth (e.g., Savazzi, 1999a, 1999b, and references therein). This requires mainly behavioural modifications, rather than morphologic ones, since these molluscs are not cemented to the substrate, can move within their boreholes, and already bore deeper into the substrate during growth, in order to accommodate their ontogenetic increase in size. Dendropoma, instead, is con- strained by shell cementation to the substrate in its possible evolutionary “choices”.
Excavation of a deeper trench with a thicker roof (cf. Figure 2) would seem to be a straightforward route into the endolithic habit. However, this solution is not feasible be- cause it involves a substantial increase in the volume of sub- strate that must be removed, and of shell material that must be secreted (i.e., the “roof” of shell material covering the trench). Alternatively, faster horizontal growth along the surface of the substrate, resulting in an increased length of the shell, would also help to fight off erosion by continually moving the organism to fresh areas of the substrate, but like- wise involves an increase in the energy spent boring and se- creting shell material, besides being unfeasible in crowded substrates. In D. sp., a secondary return to coiling avoided these problems, and provided a working solution to seem- ingly contrasting necessities: cementing the organism to the substrate, allowing it to move deeper into the substrate dur- ing growth, not increasing the used surface area of sub-
10 Enrico Savazzi
strate, and not increasing the volume of removed substrate substantially, compared with a semi-endolithic habit.
The secondary return to coiling in D. sp. confronted this species with a “choice” between two mutually exclusive life habits: epifaunal coiling above the surface of the substrate, or endolithic coiling beneath its surface. No working inter- mediate choices are possible (short of abandoning coiling and reverting to an uncoiled semi-endolithic habit), because this would cause the mollusc to bore into earlier portions of its own shell, still occupied by the soft parts. Thus, the onset of coiling and of fully endolithic habits must have been a sudden evolutionary event, made possible by the facts that coiling was already available in the vermetid Bauplan, and that the accompanying coadaptations did not require the evolution of substantially new morphological or behavioural traits.
Alternatively, an intermediate paedomorphic stage, adap- tive in lessening the negative effects of erosion by reducing the life span, may have been involved. Such a stage likely had a life habit comparable to that of juvenile vermetids, and therefore displayed no substantial morphologic innovation. The subsequent return to a larger size forced a switch from epifaunal to endolithic habits and hyperstrophic coiling. Thus, also this alternative process involves a sudden evolu- tionary change.
The facts that several species of Dendropoma display fac- ultative or obligatory epifaunal coiling (see above), while all Dendropoma are capable of semi-endolithic boring, suggest an evolutionary scenario in which epifaunal coiling was ini- tially lost in Dendropoma, and subsequently reappeared secondarily multiple times during the evolution of this genus, rather than representing the uninterrupted maintenance of this character from epifaunal ancestors. The factor that “tipped the scales” in favour of endolithic boring in D. sp., in- stead of resulting in yet another instance of parallel evolution of epifaunal coiling, is likely the substrate being subjected to a high rate of wave erosion and/or bioerosion, which makes epifaunally coiled individuals excessively vulnerable. Thus, the immediate environment appears to be the trigger of the evolutionary processes that led to the unique shell morphol- ogy and growth mechanism in D. sp. In turn, the presence of a solid substrate surrounding the organism, which characte- rises the endolithic environment, allowed the evolution of a growth mechanism that would be impossible in an epifaunal organism.
To the knowledge of the writer, the clavagellid bivalve Bryopa is the only other boring mollusc that permanently ce- ments the shell (in particular, the left valve) to the substrate without losing the capability of migrating deeper within the substrate during growth (Savazzi, 2000). Bryopa does so by continuously elongating the shell and shifting the position of the soft parts within the left valve, abandoning its early portions. Thus, its adaptations are partly convergent with those of D. sp.
Conclusions Dendropoma sp. is unique among vermetids in being
fully endolithic in the adult stage. During the juvenile, semi- endolithic stage, it builds a broadly umbilicated whorl partly
embedded in the substrate. Subsequently, instead of build- ing successive whorls upward and elevating its shell above the substrate, as in several epifaunal vermetids, it places new whorls deeper within the substrate and underneath ear- lier ones, like a descending spiral staircase. The aperture moves into the shell umbilicus and does not migrate further along the surface of the substrate during growth. Instead, it gradually deepens within an umbilical apertural well. This process results in the formation of a characteristic spiralling pattern of ridges on the walls of the well.
While the shell morphology of D. sp., at first sight, appears to violate the fundamental laws of gastropod shell geometry, the adaptiveness and evolution of this morphology can be successfully explained within the framework of biological morphodynamics. The demands posed by a substrate sub- jected to rapid erosion appear to be the factor that triggered the evolution of this form. Its growth mechanism and most of its unique adaptations are related to the sudden change in immediate environment caused by an evolutionary switch from a semi-endolithic to a fully endolithic life habit. This switch was a sudden event, because the preadaptations of Dendropoma allow no feasible intermediate stage between a semi-endolithic and a fully endolithic life habit.
Acknowledgments
| wish to thank the Japan Society for the Promotion of Science and Prof. Kazushige Tanabe, my host at the University of Tokyo, for a grant that allowed me to write the present paper and to carry out additional research in Japan in April-May, 2000.
References
Hadfield, M. G., Kay, E. A., Gillette, M. U. and Lloyd, M. C., 1972: The Vermetidae (Mollusca: Gastropoda) of the Hawaiian Islands. Marine Biology, vol. 12, p. 81-98.
Hughes, R. N., 1978: Notes on the reproductive strategies of the South African vermetid gastropods Dendropoma corallinaceum and Serpulorbis natalensis. The Veliger, vol. 21, p. 423-427.
Keen, A. M., 1960: Vermetid gastropods and marine intertidal zonation. The Veliger, vol. 3, p. 1-2.
Keen, M., 1961: A proposed reclassification of the gastropod family Vermetidae. Bulletins of the British Museum (Natural History) Zoology, vol. 7, p. 183-213.
Keen, A. M. and Morton, J. E., 1960: Some new African spe- cies of Dendropoma (Vermetidae: Mesogastropoda). Proceedings of the Malacological Society of London, vol. 34, p. 36-51.
Kempf, M. and Laborel, J., 1968: Formations de vermets et d’algues calcaires sur les côtes du Brésil. Recueil des Travaux de la Station Marine d’Endoume, vol. 43, p. 9-23. Morton, J. E., 1955: The evolution of vermetid gastro- pods. Pacific Science, vol. 9, p. 3-15.
Morton, J. E., 1965: Form and function in the evolution of the Vermetidae. Bulletins of the British Museum (Natural History) Zoology, vol. 11, p. 585-630.
Paul, C. R. C., 1999: Terrestrial gastropods. In, Savazzi, E. ed., Functional Morphology of the Invertebrate Skeleton, p. 149-167. Wiley and Sons, Chichester.
Savazzi, E., 1996: Adaptations of vermetid and siliquariid gas- tropods. Palaeontology, vol. 39, p. 157-177.
Savazzi, E., 1999a: Cemented and embedded gastropods. In, Savazzi, E. ed. Functional Morphology of the Invertebrate Skeleton, p. 183-195. Wiley and Sons, Chichester.
Savazzi, E., 1999b: Boring, nestling and tube-dwelling bi- valves. In, Savazzi, E. ed., Functional Morphology of the Invertebrate Skeleton, p. 205-237. Wiley and Sons, Chichester.
Savazzi, E., 2000: Morphodynamics of Bryopa and the evolu- tion of clavagellids. Geological Society Special
Morphodynamics of vermetid gastropod
Publications, no. 177, p. 313-327.
Seilacher, A., 1970: Arbeitskonzept zur Konstruktions- Morphologie. Lethaia, vol. 3, p. 393-396.
Seilacher, A., 1991: Self-organizing mechanisms in morpho- genesis and evolution. /n, Schmidt-Kittler, N. and Vogel, K. eds, Constructional Morphology and Evolution, p. 251-271. Springer-Verlag, Berlin.
Seilacher, A. and Gunji, Y. P., 1993: Morphogenetic count- downs in heteromorph shells. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen, vol. 190, p. 237-265.
102 0 er; in ii ee A A eee sperren eb ei, À» sul
vii ae CN Drag.
À elle.
u Ey
he NE Par xt { “
fl Ip | em, A ales
© eek ptet) (
tbe fg. . Ko Growling
Ms sta ate hat ro i
’ BEE
ee ET
sur, ever hance: A ieee
0 4) ha sthe sel ee
je ‘i cubed: vw a wo
| jé
hi iris
(eT) Re & N | ; “2 > Cover ml Lu x \ ; ET N LT. idole 0 et 7
in uchs Mure au LE ih inten ie MI
’ Free vive hues WO d à on b W if LU à at rat PEL cé Mie: ba
7
Paleontological Research, vol. 5, no. 1, pp. 13-19, April 30, 2001 © by the Palaeontological Society of Japan
External features of embryonic and early postembryonic shells
of a Carboniferous goniatite Vidrioceras from Kansas
KAZUSHIGE TANABE’, CYPRIAN KULICKI’, NEIL H. LANDMAN’ and ROYAL H. MAPES*
‘Department of Earth and Planetary Science, Graduate School of Science, University of Tokyo, Hongo 7-3-1, Tokyo 113-0033, Japan (e-mail:tanabe @eps.s.u-tokyo.ac.jp)
®Institute of Paleobiology, Polish Academy of Sciences, ul. Twarda 51/55, 00-818 Warsawa, Poland (e-mail: kulicki @ twarda.pan.pl)
“Division of Paleontology (Invertebrates), American Museum of Natural History, Central Park West at 79” Street, New York, New York 10024-5192, U.S.A. (e-mail: landman @amnh.org)
‘Department of Geological Sciences, Ohio University, Athens, Ohio 45701, U.S.A.
(e-mail: mapes @oak.cats.ohiou.edu)
Received 7 August 2000; Revised manuscript accepted 4 December 2000
Abstract. The ornamentation and dorsal wall structure of Vidrioceras (Cycloloboidea, Goniatitina)
in the early ontogenetic stage are described on the basis of specimens from the Upper Pennsylvanian in Kansas, USA. The exposed surface of the embryonic shell is smooth, without any trace of ornamentation or growth lines. Regularly spaced lirae abruptly appear on the early postembryonic shell just adoral of the primary constriction. The inner surface of the dorsal wall in the embryonic and early postembryonic stages exhibits a distinct ornament consisting of evenly spaced, longitudinal ridges, which are replaced adorally by the typical wrinkled ornament in the subsequent stage. Our observations are in accord with those of goniatites from the Upper Carboniferous Buckhorn Asphalt of Oklahoma, suggesting that in the Goniatitina, the outer surface
of the embryonic shell is smooth.
Comparison with the embryonic shell formation of extant
Nautilus suggests that in the Goniatitina, the embryonic shell was uniformly secreted by the shell
gland on the posterior side of the embryo.
Key words: Ammonoidea, Carboniferous, development, embryonic shell, Goniatitina, Kansas,
Vidrioceras
Introduction
The embryonic shell of ammonoids (termed the ammonitella by Druschits and Khiami, 1970) consists of a spherical initial chamber and approximately one subsequent whorl with a thick nacreous swelling (primary varix) at the aperture (for references see Landman et al., 1996). The boundary between embryonic and postembryonic stages is marked by the primary constriction. Most previous studies on the external morphology and microstructure of embryonic shells of the Ammonoidea have been based on Mesozoic material with aragonitic preservation (Druschits and Doguzhaeva, 1981; Bandel, 1982, 1986; Bandel et al., 1982; Landman, 1982, 1985, 1987; Landman and Bandel, 1985; Tanabe, 1989; Kulicki and Doguzhaeva, 1994; Landman et al., in press, among others). In Paleozoic ammonoids, the microstructure of the embryonic shell is rarely preserved in most fossils due to diagenesis, except for aragonitic
goniatites studied by Kulicki et al. (in press) from the Carboniferous Buckhorn Asphalt in Oklahoma, USA. Previous authors have indicated that the embryonic shells of Paleozoic ammonoids display ornamentation and features of several internal shell characters different from those of Mesozoic ammonoids (Beecher, 1890; Miller, 1938; House, 1965; Mapes, 1979; Tanabe et al, 1993, 1994; Doguzhaeva, 1996; Landman et al., 1996; Landman et al., 1999; Klofak et al., 1999; Kulicki et al., in press). These studies have revealed a variety in the embryonic shell fea- tures in the Ammonoidea that can be used for higher-level phylogenetic analysis.
In this paper, we describe the ornamentation and the dor- sal wall structure of goniatites at the embryonic and early postembryonic stages based on specimens from the Upper Pennsylvanian of Kansas, USA. Furthermore, the result of our observations is compared with data on other ammonoids and extant Nautilus and discussed for its implications for
14 Kazushige Tanabe et al.
systematics and embryonic shell formation. Some of the specimens utilized have been studied by Tanabe et al. (1993), but those observations are partly reevaluated in this
paper. Material and methods
Tanabe et al. (1993) discovered an embryonic ammonoid assemblage in a carbonate concretion recovered from the Virgilian (Upper Pennsylvanian) offshore shale in Pomona, Kansas, USA. These authors classified the embryonic ammonoids into two morphotypes by the difference in the size and shape of their initial chambers, namely into a large and globular and a small and ellipsoidal one. They further assigned these large and small morphotypes to Vidrioceras (Vidrioceratidae, Cycloloboidea) and Aristoceras (Thalassoceratidae, Thalassoceratoidea) respectively, on the basis of comparison with their initial chambers with those of larger specimens of these two genera from the same con- cretion.
About one hundred well preserved specimens of the two genera at embryonic and early postembryonic stages were removed without etching from the weathered portion of the concretion by the wet-sieving method. They were coated with platinum and observed by scanning electron micros- copy. Although the embryonic shells of Aristoceras occur more abundantly than those of Vidrioceras, we did not ob- serve early shell features of the former genus because of poor preservation of the dorsal wall sculpture in the available specimens. As already pointed out by Tanabe et al. (1993), the goniatite specimens from Pomona preserve calcified shell material, and the recrystalized condition of their shell wall prevents study of the shell ultrastructure. Our observa- tions are, therefore, mainly restricted to the external features of these specimens. All of the specimens utilized are housed in the University Museum, the University of Tokyo (UMUT).
Observations
Embryonic shells
General morphology and ornamentation. The embry- onic shells of Vidrioceras examined are all globular in overall shape and consist of a spindle-shaped initial chamber and approximately one subsequent planispiral whorl (Figure 1). In median section, a thick nacreous swelling (primary varix) appears on the inner side of the prismatic layer in the apertural region (see Tanabe et al., 1995, figure 2A). The first whorl is much broader than high, covering the greater portion ot the initial chamber (Figures 1, 2). The spiral length of the embryonic shell (=ammonitella angle of Landman et al., 1996) is relatively long, measuring about 360° in median section. The embryonic shell diameter in the specimens examined ranges from approximately 720 um (UMUT PM19872-1; Figure 1.1) to 780 um (UMUT PM19872-3; Figure 1.3a). The exposed surface of the shell at the embryonic stage is smooth without any trace of ornamentation or growth lines (Figure 1).
Dorsal wall of the first whorl.— In several embryonic and early postembryonic specimens, part of the first whorl has
been lost due to mechanical destruction during the taphonomic process or due to the wet-sieving procedure. In those specimens, evenly spaced longitudinal ridges are visi- ble on the dorsal side of the missing whorl portion (Figures 2.1-2.3, 3.1, 3.2). They never occur on the lateral flanks of the initial chamber that are free from the first whorl. This fact indicates that the longitudinal ridges represent the sculpture on the inside surface of the dorsal wall of the first whorl. A weaker ridge is occasionally intercalated between the longitudinal ridges (Figure 2.3b).
Remarks.— Tanabe et al. (1993, fig. 3A) described evenly spaced longitudinal ridges on the ventrolateral side of sev- eral embryonic shells of Vidrioceras from the same locality and interpreted them as the surface ornamentation. However, our reexamination of these and additional speci- mens reveals that the longitudinal ridges are not the surface ornamentation but the microornamentation on the inner side of the dorsal wall, and that the exposed surface of the em- bryonic shell of Vidrioceras is in fact smooth.
Early postembryonic shells
The embryonic shell margin is easily visible from outside by the presence of the slightly depressed primary constric- tion followed by the sharp apertural edge (see arrows, Figure 1.1, 1.3a). Fine transverse lirae abruptly appear on the adoral side of the primary constriction. They are initially rectiradiate in the early postembryonic stage (Figure 1.1, 1.3a, b), but become prorsiradiate and gently convex at the venter in a later stage. Each lira is asymmetric in cross sec- tion with a steep edge on the adoral side and is gently in- clined adapically (Figure 1.3b).
The change of the inner surface sculpture of the dorsal wall is visible in several early postembryonic shells whose body whorl is partly lost secondarily (Figure 2.1-2.3). In one of these specimens, shown in Figure 2.2, the longitudi- nally ridged ornament disappears on the dorsal side of the embryonic shell aperture and a wrinkled ornament similar to a human fingerprint pattern begins to appear on the adoral side of the primary constriction. In the embryonic or early embryonic specimen shown in Figure 2.4, the wrinkled orna- ment already exists on the dorsal side near the shell margin. In another two specimens, shown in Figure 2.1 and 2.3, the longitudinally ridged ornament extends for a half whorl be- yond the primary constriction. These observations clearly indicate that Vidrioceras exhibits some variation in the ontogenetic change of the dorsal wall ornament.
Comparison with other ammonoids
On the basis of observations on excellently preserved ma- terial from the Pennsylvanian (Desmoinesean) Buckhorn Asphalt in Oklahoma, Kulicki ef a/. (in press) reported that the outer surface of the embryonic shells of goniatites is smooth without any trace of ornamentation or growth lines, as is the case of Vidrioceras described herein. These ob- servations strongly suggest that in the Goniatitina the em- bryonic shell is smooth.
Ornamentation on embryonic shells of the Goniatitina dif- fers from those of other ammonoid suborders. In the Devonian Agoniatitina, Anarcestina and Tornoceratina, the
Goniatite early shell features 15
Figure 1. Vidriocerassp. 1. Oblique view of early postembryonic specimen showing slightly depressed primary constriction and ad- jacent embryonic shell edge (pointed by an arrow). The embryonic shell is smooth, while the postembryonic shell is sculptured by fine trans- verse lirae. UMUT PM19872-1. 2. Lateral view of embryonic shell with smooth surface. UMUT PM19872-2. 3a, b. UMUT PM19872-3. Lateral view of early postemebryonic shell showing abrupt appearance of fine transverse lirae after primary constriction (pointed out by an arrow) (3a) and close-up of lirae (3b). Scale bars: 100 pm (1, 2, 3a) and 10 um (3b)
embryonic shells are characterized by fine transverse lirae parallel to the aperture (Beecher, 1890; Miller, 1938; House, 1965; Landman et al., 1996; Klofak et a/., 1999). In the un- coiled Late Silurian and Devonian Bactritina and partly coiled Devonian ammonoids, the straight whorl after the ini- tial chamber is also covered with transverse lirae (Erben, 1964). Mapes (1979) reported both smooth and longitudi- nal ornament on the initial subspherical chamber and early shaft of “bactritids” from the Carboniferous, and Mapes (1979) and Doguzhaeva (1996) reported a reticulate orna- mentation on the earlier embryonic shaft portion of “bactritids” from the Carboniferous and Permian, respec- tively. However, Doguzhaeva et al. (1999) suggested that some of these specimens may eventually be reassigned to
the Coleoidea. In the Ceratitina, Phylloceratina, Lyto- ceratina, Ammonitina, and Ancyloceratina, the embryonic shell lacks lirae and instead is covered with minute tubercles (Kulicki, 1974, 1979; Bandel, 1982, 1986; Bandel et al, 1982; Landman and Waage, 1982; Landman, 1985, 1987; Tanabe, 1989; Kulicki and Doguzhaeva, 1994; Landman et al., in press). To sum up these previous descriptions, at least three kinds of embryonic shell ornamentation have been recognized in the Ammonoidea excluding Upper Paleozoic “bactritids’. Each ammonoid suborder, excluding the doubtful taxon Bactritina, appears to have its own chara- cteristic pattern of ornamentation.
ow — © ® © © = F ® D = [2] 2 N © ST
Goniatite early shell features 17
Figure 3. gitudinal ridges.
walls is discernible in this specimen. UMUT MM19872-8.
Discussion
Observations of the early embryonic shells of extant Nautilus provide a reference point for discussions about the embryonic development of ammonoids. According to Arnold and Landman (1993) and Tanabe and Uchiyama (1997), the early embryonic shell development of Nautilus can be divided into two stages with different shell microstruc- ture and ornamentation. In the first stage (=early organogenetic stage), a low cap-shaped shell with a distinct median depression (called cicatrix) is secreted in the se- quence of outer conchiolin and inner prismatic layers. The cicatrix lacks growth lines, indicating uniform shell secretion by the mantle primordium (shell gland) on the posterior side of the embryonic body (see Tanabe and Uchiyama, 1997, fig. 1A). In the second stage (=middle organogenetic stage), a new shell consisting of outermost conchiolin, outer prismatic, middle nacreous, and inner prismatic layers ap- pears at the outer margin of the cicatrix, leaving a disconti- nuity in the shell structure at the boundary. It is sculptured by transverse growth lines and radial undulations. At this stage of development, the anterior mantle margin is well dif- ferentiated and possesses three folds, where shell secretion
Vidrioceras sp. 3a. Oblique view of embryonic shell with preserving interior dorsal wall sculpture of the first whorl with lon- 3b. Close-up of longitudinal ridges on interior side of dorsal wall.
Note that the boundary between ventral and dorsal shell
occurs (Tanabe et al., 1991).
The absence of transverse lirae on the embryonic shells of the Goniatitina strongly suggests that the walls of the initial chamber and the first whorl were secreted synchronously by the undifferentiated shell gland. Bandel (1982) and Kulicki and Doguzhaeva (1994) hypothesized this kind of embryonic shell development in Mesozoic ammonoids, relying upon ob- servations about the biomineralization of embryonic shells of modern “archaeogastropods”. The longitudinal dorsal layer was probably secreted at a late stage of embryogenesis and occasionally at an early postembryonic stage. The appear- ance of transverse lirae on the postembryonic shell indicates an accretionary mode of growth. This event does not occur synchronously with the development of the dorsal wall with wrinkled ornamentation (Figure 2.1-2.4; Kulicki et al., in press, pl. 2, fig. 3). Such a wrinkled dorsal wall has been extensively recognized in Paleozoic ammonoids in the postembryonic stage (House, 1971; Walliser, 1970). Kulicki (1979, 1996) and Kulicki et al. (2001) have pointed out that the dorsal wall of ammonoids consists of two components, namely, the outer component consisting of organo-prismatic material, sometimes with a wrinkled texture on the outside, and the inner prismatic component that covers the outer
@ Figure 2.
second whorls exhibits longitudinally ridged ornament on the inner side.
Vidrioceras sp. 1. Lateral view of incomplete postembryonic shell, part of whose body whorl is lost.
Dorsal wall of the first- UMUT MM19872-4. 2. Ventral view of early postembryonic shell
showing change of interior dorsal wall sculpture from longitudinally ridged pattern to wrinkled pattern at embryonic shell/postembryonic shell boundary. UMUT MM19872-5. 3a,b. UMUT MM19872-6. 3a. Oblique view of incomplete postemebryonic shell, showing change of in- terior dorsal wall sculpture from longitudinally ridged pattern in embryonic stage to wrinkled pattern in postembryonic stage. 3b. Close-up of longitudinally ridged dorsal wall structure in early postembryonic stage of same specimen. 4a, b. UMUT MM19872-7. 4a. Ventral view of embryonic (or early postembryonic) shell showing interior feature of dorsal wall with wrinkled ornamentation. 4b. Close-up of wrinkled or- namentation in same specimen. Scale bars: 100 um (1-4a) and 40 um (4b). Arrows in 1, 2, 3a, and 4a point to the approximate position of primary constriction.
18 Kazushige Tanabe et al.
component on the adapical side of the body chamber. Our observations indicate that the inner dorsal wall component is absent in the embryonic shells of Vidrioceras. It presuma- bly begins to appear in the postembryonic stage.
In view of the absence of transverse lirae, the mode of embryonic shell formation in the Ceratitina, Phylloceratina, Lytoceratina, Ammonitina, and Ancyloceratina may also be explained by the “archaeogastropod model” of Bandel (1982) and Kulicki and Doguzhaeva (1994). The presence of fine transverse lirae on the relatively large embryonic shells of the Devonian suborders Agoniatitina, Anarcestina and Tornoceratina and on their postembryonic shells (see Landman et al., 1996, appendix |!) is, however, problematic. One possibility is that in the embryonic stage the mantle al- ready was differentiated in the embryonic stage to secrete a shell with growth lines at its anterior margin. This type of embryonic shell development would be described as accretionary growth. A second possibility is that the embry- onic shell was rapidly mineralised and an accretionary mode of growth characterized only the postembryonic shell (Klofak et al., 1999). Future research utilizing well preserved mate- rial will resolve this problem.
Acknowledgements
We thank John Arnold, Klaus Bandel and Larisa Doguzhaeva, Sigard von Boletzky and Yasunari Shigeta for critical discussion, and Curtis Faulkner and Susan Klofak for help in collecting and preparing the specimens utilized. This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for Promotion of Science (no. 12440141 for 2000).
References
Arnold, J. M. and Landman, N. H., 1993: Embryology of Nautilus: Evidence for two modes of shell ontogeny. Journal of Cephalopod Biology, vol. 2, p. abst. 1.
Bandel, K., 1982: Morphologie und Bildung der frühontogene tischen Gehäuse bei conchiferen Mollusken. Facies, vol. 7, p. 1-198, pls. 1-22.
Bandel, K., 1986: The ammonitella: a model of formation with the aid of the embryonic shell of archaeogastropods. Lethaia, vol. 19, no. 2, p. 171-180.
Bandel, K., Landman, N. H. and Waage, K. M., 1982: Microornamentation on early whorls of Mesozoic ammonoids: Implications for early ontogeny. Journal of Paleontology, vol. 56, no. 2, p. 386-391.
Beecher, C. E., 1890: On the development of the shell in the genus Tornoceras Hyatt. American Journal of Science, vol. 40, p. 71-75.
Doguzhaeva, L. A., 1996: Shell ultrastructure of the Early Permian Bactritella and Ammonitella, and its phylogenetic implication. Jost Wiedmann Symposium on Cretaceous Stratigraphy, Paleobiology and Paleobiogeography, Abstract Volume, Report of Geologisch- Paläontologisches Institut der Universitat Kiel, no. 76, p. 19-24.
Doguzhaeva, R. A., Mapes, R. H. and Mutvei, H., 1999: A Late Carboniferous spirulid coleoid from the southern Mid- Continent (USA). In, Olöriz, F. and Rodriguez-Tover, F. J. eds., Advancing Research on Living and Fossil
Cephalopods, p. 47-57. Kluwer Academic/Plenum, New York.
Druschits, V. V. and Doguzhaeva, L. A., 1981: Ammonity pod Elektronym Mikroskopom [Ammonites under the Electron Microscope]. 238 p., 43 pls. Moscow University Press, Moscow. (in Russian)
Druschits, V. V. and Khiami, N., 1970: Stroyeniye sept, stenki protokonkha i nachal’nykh oborotov rakoviny nekotorykh rannemelovykh ammonitov [Structure of the septa, protoconch walls and initial whorls in Early Cretaceous ammonites]. Paleontologicheskii Zhurnal, 1970, no. 1, p. 35. (in Russian)
Erben, H. K., 1964: Die Evolution der altesten Ammonoidea. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen, vol. 120, no. 2, p. 107-212.
House, M. R., 1965: A study in the Tornoceratidae: The suc- cession of Tornoceras and related genera in the North American Devonian. Philosophical Transactions of the Royal Society of London, Series B, vol. 250, no. 763, p. 79-130, pls. 1-11.
House, M. R., 1971: The goniatite wrinkle-layer. Smithsonian Contribution to Paleobiology, vol. 3, p. 23-32, pls. 1-3.
Klofak, S. M., Landman, N. H. and Mapes, R. H., 1999: Embryonic development of primitive ammonoids and the monophyly of the Ammonoidea. In, Oloriz, F. and Rodriguez-Tovar, F. J. eds., Advancing Research on Living and Fossil Cephalopods, p. 23-45. Kluwer Academic/Plenum, New York.
Kulicki, C., 1974: Remarks on the embryogeny and postembryonal development of ammonites. Acta Palaeontologia Polonica, vol. 19, p. 201-224.
Kulicki, C., 1979: The ammonite shell: its structure, develop- ment and biological significance. Palaeontologia Polonica, no. 39, p. 97-142, pls. 24-48.
Kulicki, C., 1996: Ammonoid shell microstucture. /n, Landman, N. H., Tanabe, K. and Davis, R. A. eds., Ammonoid Paleobiology, p. 65-101. Plenum Press, New York.
Kulicki, C. and Doguzhaeva, L. A. 1994: Development and cal- cification of the ammonitella shell. Acta Palaeontologia Polonica, vol. 39, no. 1, p. 17-44.
Kulicki, C., Tanabe, K., Landman, N. H. and Mapes, R. H., 2001: Dorsal shell wall in ammonoids. Acta Palaeon- tologia Polonica, vol. 46, no. 1, p. 23-42.
Kulicki, C., Landman, N. H., Mapes, R., Tanabe, K. and Haney, M., in press: Buckhorn Asphalt goniatites ammonitellae, the oldest, preserved with primary mineral composition. Abhandlungen der Geologischen Bundesanstalt, Special Volume (Proceedings of the 5th International Symposium, Cephalopods-Present and Past, Vienna, 1999).
Landman, N. H., 1982: Embryonic shells of Baculites. Journal of Paleontology, vol. 56, no. 5, p. 1235-1241. Landman, N. H., 1985: Preserved ammonitellas of Scaphites (Ammonoidea, Ancylocerataceae). American Museum
Novitates, no. 2815, p. 1-10.
Landman, N. H., 1987: Ontogeny of Upper Cretaceous (Turonian-Santonian) scaphitid ammonites from the Western Interior of North America: Systematics, develop- mental patterns, and life history. Bulletin ofthe American Museum of Natural History, vol. 185, no. 2, p. 118-241.
Landman, N. H. and Bandel, K., 1985: Internal structures in the early whorls of Mesozoic ammonites. American Museum Novitates, no. 2823, p. 1-21.
Goniatite early shell features
Landman, N. H. and Waage, K. M., 1982: Terminology of structures in embryonic shells of Mesozoic ammonites. Journal of Paleontology, vol. 56, no. 5, p. 1293-1295.
Landman, N. H., Tanabe, K. and Shigeta, Y., 1996: Ammonoid embryonic development. /n, Landman, N. H., Tanabe, K. and Davis, R. A. eds., Ammonoid Paleobiology, p. 343- 405. Plenum Press, New York.
Landman, N. H., Mapes, R. H. and Tanabe, K., 1999: Internal features of the embryonic shells of Late Carboniferous Goniatitina. /n, Oloriz, F. and Rodriguez-Tovar, F. J. eds., Advancing Research on Living and Fossil Cephalopods, p. 243-254. Kluwer Academic/Plenum Publishers, New York.
Landman, N. H., Bizzarini, F., Tanabe, K. and Mapes, R. H., in press: Microornamentation on the embryonic and postembryonic shells of Triassic ceratites (Ammonoidea). American Malacological Bulletin.
Mapes, R. H., 1979: Carboniferous and Permian Bactritoidea (Cephalopoda) in North America. University of Kansas, Paleontological Contribution, Article 64, p. 1-75, pls. 1- 41.
Miller, A. K., 1938: Devonian ammonoids of America. Geological Society of America, Special Papers, no. 14, p. 1-262.
Tanabe, K., 1989: Endocochliate embryo model in the
Mesozoic Ammonitida. 183-196.
Tanabe, K. and Uchiyama, K., 1997: Development of the em- bryonic shell structure in Nautilus. The Veliger, vol. 40, no. 3, p. 203-215.
Tanabe, K., Tsukahara, J., Fukuda, Y. and Taya, Y., 1991: Histology of a living Nautilus embryo: Preliminary obser- vations. Journal of Cephalopod Biology, vol. 2, no. 1, p. 13-22.
Tanabe, K., Landman, N. H., Mapes, R. H. and Faulkner, C. J., 1993: Analysis of a Carboniferous embryonic ammonoid assemblage from Kansas, U.S.A.-Implications for ammonoid embryology. Lethaia, vol. 26, p. 215-224.
Tanabe, K., Landman, N. H. and Mapes, R. H., 1994: Early shell features of some Late Paleozoic ammonoids and their systematic implications. Transactions and Proceedings of the Palaeontological Society of Japan, New Series, no. 173, p. 384-400.
Tanabe, K., Shigeta, Y. and Mapes, R. H., 1995: Early life his- tory of Carboniferous ammonoids inferred from analysis of shell hydrostatics and fossil assemblages. Palaios, vol. 10, p. 80-86.
Walliser, ©. H. 1970. Über die Runzelschicht bei Ammonoidea. Goettinger Arbeiten zur Geologie und Paläontologie, vol. 5, p. 115-126, pls. 1-4.
Historical Biology, vol. 2, p.
+ eg, An EM e ® » à d 2 FRE 4 x > +1 un le nigh ay ol! aA N mpm} leu $y wun
“a tom
= 4 Ve tam’ ars
’ é à i
mi 25
f = en PAR A A E | wi AD - am? pans er, b LL 6 arr ü a aie k oe m a N — Ss - u À mi CU | st sh Liha, CE - | h wean = i CTA PER hee
a ene
Br N o Bi
’ e z; ip PAL ù | ai
\
Paleontological Research, vol. 5, no. 1, pp. 21-31, April 30, 2001
© by the Palaeontological Society of Japan
Macro- and microstructure of shell and operculum in two Recent gastropod species, Nerita (Theliostyla) albicillaand Cinnalepeta pulchella
(Neritopsina: Neritoidea)
TAKENORI SASAKI
The University Museum, The University of Tokyo,
7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan (sasaki@um.u-tokyo.ac.jp)
Received 3 October 2000; Revised manuscript accepted 26 December 2000
Abstract.
The shell and opercular structures of Nerita (Theliostyla) albicilla and Cinnalepeta
pulchella were described and compared with those of other extant members of Neritopsina. The
shell of N. (T.) albicilla is composed of four layers: the outermost prismatic layer, followed by the simple crossed-lamellar layer, the myostracum, and the inner complex crossed-lamellar layers. The operculum consists of three prismatic layers deposited on both sides of an organic layer. C. pulchella also has a four-layered shell, but lacks an operculum. The outer layer is a homologous structure. The shells of Recent neritopsine families can be categorized into a four-layered group (Neritiliidae, Neritidae, and Phenacolepadidae) and a three-layered group (other families). In con- trast, opercular structure is markedly variable in the Neritopsina, and little correlation can be estab-
lished in the light of phylogenetic evolution or adaptation.
Key words: Neritidae, Neritopsina, operculum, Phenacolepadidae, shell structure
Introduction
Neritopsina is a phylogenetically distinct gastropod clade which originated in the Ordovician (Bandel and Fryda, 1999). The Recent members share characteristic apomorphies of odontophoral cartilages and muscles, ante- rior digestive tract, reproductive organs, and nervous system (Haszprunar, 1988; Ponder and Lindberg, 1997; Sasaki, 1998) and exhibit successful adaptive radiation in a wide range of habitats in deep-sea hydrothermal vent or hydro- carbon seep, submarine cave, intertidal rocky shore, and nonmarine aquatic and terrestrial environments (Ponder, 1998; Sasaki, 1988). It is also particularly interesting that neritopsines have rich fossil records since the early Paleozoic (Bandel, 1992; Bandel and Fryda, 1999), and that ancient taxa like Neritopsis have survived as relics in cryptic habitats (Kase and Hayami, 1992).
Among hard-part characters, shell structure, along with larval shell morphology, is of primarily taxonomic importance for studies uniting fossil and Recent taxa of Gastropoda (Bandel, 1982; 1988; 1991; Bandel and Geldmacher, 1996). In Gastropoda, Patellogastropoda are known to exhibit re- markable diversification of the shell microstructure (MacClintock, 1967; Lindberg, 1988, 1998). Grouping by
shell structure corresponds well to the anatomical division of supraspecific taxa, and therefore enables paleontologists to allocate fossilized taxa within an anatomy-based systematic scheme (Kase, 1994; Kase and Shigeta, 1996; Lindberg and Hedegaard, 1996; Hedegaard et al., 1997). As for other gastropod higher taxa, Vetigastropoda are typically charac- terized by the apomorphic occurrence of columnar nacreous structure, although some taxa have supposedly lost it secon- darily (Hedegaard, 1997). In the Apogastropoda, which is a huge clade including Caenogastropoda and Hetero- branchia, the shells are composed primarily of several layers of crossed-lamellar structure and differences between dis- tantly related subclades within it are in general minor (Bog- gild, 1930; Bandel, 1979; Togo and Suzuki, 1988). At rela- tively lower rank, however, some striking differences can be revealed by detailed comparison, as was shown in littorinid genera by Taylor and Reid (1990).
Despite its unique phylogenetic status among gastropods, little discussion has been devoted to exoskeletal evolution of Neritopsina as compared to other major taxa of molluscs. The current Knowledge of neritopsine hard-part structures is derived from only a limited amount of literature, and in addi- tion, most data have been documented in simple format with a few or no illustrations. To increase data quality and
22 Takenori Sasaki
quantity on neritopsine hard parts, this study aims to de- scribe the shell and opercular structures of two Recent spe- cies in detail from the macro- to microscopic level. The results of the observations were compared with the pub- lished data of other neritopsines in the literature, and their similarities and dissimilarities were discussed from phylogenetic and adaptational viewpoints.
Material and methods
The shells of Nerita (Theliostyla) albicilla were collected alive from an intertidal zone at Banda, Tateyama, Chiba Prefecture, central Japan, and living specimens of Cinnalepeta pulchella from Tosashimizu, Kochi Prefecture, southwest Japan were provided by Dr. Shigeo Hori. In the laboratory, the macroscopic morphology and the distribution of shell layers based on texture were first observed under a binocular microscope. Then, the shell was crushed with a hammer, and the original position of fragments was labeled before cleansing in bleach for 12 hours and later in running water for 30 minutes. The fresh fracture of shell fragments was observed with a scanning electron microscope (SEM).
The description of microstructure was made on layer dis- tribution, boundary between layers, form of crystal aggrega- tion, and orientation and morphology of first- to third-order units (major to minor structural arrangement). The descrip- tive terminology of microstructure follows Carter and Clark (1985) and Kano and Kase (2000b). The two terms, “cros-
Outer Layer
1018]S0d
Inductura
sed lamellar” and “complex crossed lamellar,” are abbrevi- ated as “CL” and “CCL.”
Terms for orientation were based on the following criteria: (1) “anterior-posterior” direction was determined by body axis of the animal, and (2) “adaxial-abaxial” distinction rela- tive to coiling axis of whorls was used to indicate relative po- sition along inner-outer lips of the aperture.
The samples used in this study are preserved in the University Museum, the University of Tokyo (UMUT).
Description
Order Neritopsina Cox and Knight, 1960 Superfamily Neritoidea Rafinesque, 1815 Family Neritidae Rafinesque, 1815 Nerita (Theliostyla) albicilla Linnaeus, 1758
Figures 1-5
Shell.—The shell is elongate along the anterior-posterior axis of the animal. The spire is completely depressed as a part of rounded whorls, and the external part of the shell is mostly occupied by a large body whorl and an extended ap- erture.
The outer lip of the aperture is thickened and indented with elongate denticles, being arranged parallel to the apertural margin. The inner lip spreads widely over the body whorl to form a robust inductura.
The shell wall near the outer lip consists of three layers;
Inner Layer
DS AS uch —. De ù rn © 2) y A . N % ie AU / N ae I, RA iy) Denticle
5mm
Figure 1. Apertural view of the shell of Nerita (Theliostyla) albicilla, The sculpture and texture of the surface are depicted slightly sche- matically. The dark outer layer is distinguished by fine-grained smooth surface, the middle layer exhibits dense linear pattern of CL structure,
and most of the inner layer is visible as irregularly oriented lines.
i.e. outer and middle layers and a distinct denticular zone
(Figure 1). The outer layer contains black pigments, encir- cles the apertural margin, and makes a clear contrast with the pale inner layer. Microstructurally this layer is formed of a thick aggregate of short prisms (Figure 3A), which can be identified as blocky prismatic structure. The prisms attain less than 4 um in length and 1 um in width. The middle layer is composed of comarginal simple CL structure (Figure 3B-E). The linear lines of the first-order units are clearly visible on the inside of the aperture even at low magnifica- tion (Figures 1, 2). The denticular zone is built up of a thin layer of irregularly crossed fine crystals (CCL structure). The denticles do not continue spirally toward the inside of the aperture but remain in the identical position due to resorption.
The inner lip margin is shallowly curved with two to four small nodules (Figure 1). The surface of the inductura is roughened with nodules of various size, and their number and distribution are considerably variable intraspecifically. The inductura, especially near the inner lip, shows an irregu- lar texture reflecting the first-order arrangement (Figure 1). There is no sharp boundary between the CL structure of the middle layer and the CCL structure of the inductura. The linear patterns of the middle layer gradually merge into the irregular patterns of the inductura (Figure 1).
Shell muscle scars are separated into a disjunct pair cor- responding to right and left shell muscles (see Sasaki, 1998:
Septum
Prismatic Sublayer
Inner Lip
Middle Layer
Figure 2. Schematic representation of inner part of whorls of Nerita (Theliostyla) albicilla, seen from the outer lip of the aperture.
Left Muscle Scar
Shell structure of neritopsine gastropods 23
fig. 73a). The left scar is located on the basal side near the inner lip (Figure 2), while the right one lies on the opposite, apical side. The scars are deeply impressed on the interior shell surface and reflect the form of muscles which are di- vided into bundles (Figure 2). The myostracum from the left scar is formed as a vertical stack of irregular prisms (Figure 3F), and its thickness exceeds 80 ym in an adult shell. The myostracum from the right scar is immediately resorbed and not traceable in most sections.
The inside of the visceral part of the shell is extensively resorbed and reorganized as a hollow space without a true columella. A platy septum connecting the inner lip and api- cal wall of the shell is secreted and inserted into the narrow space between head-foot and uncoiled visceral mass of the animal (Figure 2; see also Sasaki, 1998: fig. 73a). The sep- tum, inner lip, inductura, and the interior of whorls are all constructed as a continuity of the inner layer (Figure 2). The layer has CCL structure: lathy third-order units are set radially to form fan-shaped second-order units, which in turn are vertically stacked to form wedgelike first-order units which are irregularly oriented and interdigitate with one an- other (Figure 3G, H). The structure appears as spinous crossing prisms at the initial stage of formation near its growth front. Several very thin prismatic sublayers are in- serted in a thick CCL layer of the inductura (Figure 2).
Operculum.—The operculum is heavily calcified with a distinct apophysis (Figure 4A, B). The exterior surface is
Inductura
Outer Layer
S RE WLM OR ere >
A ee BG
VAGUE = 29 z » Middle Layer 57 52 KA N Inner Layer
ESS
Myostracum
3 mm
Outer Layer
Shell
layers on cut planes are illustrated based on the results of observations with SEM.
RZ & 2] is)
ao
= o c ® = © bE
Se N
Shell structure of neritopsine gastropods
Apophysis
Nucleus
Organic Layer
Outer Layer
Outer Layer
Organic Layer
25
Right Opercular Muscle Scar
Apophysis
Left Opercular Middle Muscle Scar Layer
3mm
Figure 4. Operculum of Nerita (Theliostyla) albicilla. A. Exterior view. B. Interior view. C. Vertical section showing the growth direc-
tion of prisms in three calcified layers.
covered with small nodules which tend to be arranged spi- rally (Figure 4A). The nucleus lies on the adaxial basal side at the origin of the spiral growth line.
The calcified part of the operculum can be divided into three layers (Figure 4C). The outer and inner layers are both composed of slightly inclined, nearly vertically ordered prisms (Figure 5B, D). In the middle layer, prismatic crys- tals are arranged in a spherulitic form (Figure 5C). A very thin organic layer, the homologue of the noncalcified operculum of other gastropods, is mostly concealed be- tween the outer and middle layers and appears only along the abaxial margin (Figure 4A, B).
The adaxial side of the operculum is partially embedded in the pedal musculature of the animal and marks clear depres- sions of muscle scars (Figure 4B). The left scar is small and very irregular with nearly concentric lines (Figure 5A). The right scar is elongated along the abaxial margin and smoothened.
Cut position is shown in Figure B with solid lines.
Family Phenacolepadidae Rafinesque, 1815 Cinnalepeta pulchella (Lischke, 1871)
Figures 6, 7
Shell.—The shell is completely limpet-shaped and elon- gated along the anterior-posterior axis (Figure 6). The apex is situated at the posterior end of the shell. The larval shell in the original position of the apex is involved between whorls and inner lip during growth and no longer visible in a fully matured adult shell. The inside of the apex is slightly remoulded by resorption, but a septumlike structure is not constructed in this species.
The shell consists of four layers, including the myostracum. The outer layer is composed of fine homoge- neous crystals (homogeneous structure) (Figure 7A, B). The middle layer is of commarginal simple CL structure. It is somewhat transparent, and clearly demarcated from the
ae a a ee a eh oh ey nn
æ Figure 3. SEM micrographs of shell microstructure of Nerita (Theliostyla) albicilla (UMUT RM27950). A. Blocky prismatic structure of outer layer. Scale =5 m. B. Outcrop pattern of simple CL structure near outer lip of aperture. Scale = 20 um. C. Oblique view of the frac- ture of simple CL structure in the middle layer. Scale = 20 um. D. Horizontal view of the fracture of the same layer. Scale=20 um. E. Enlarged view of the same layer, showing the arrangement of the third-order units of CL structure. Scale=5 um. F. Vertical fracture of the myostracum inserted between inner (above) and middle (below) layers. M = myostracum. Scale = 40 um. G. Outcrop pattern of CCL struc- ture on interior surface of the whorls. Scale = 20 um. H. Vertical fracture of the same structure in inner layer. Scale = 20 um.
26 Takenori Sasaki
Figure 5. SEM micrographs of opercular microstructure of Nerita (Theliostyla) albicilla(UMUT RM27951). A. Surface of left opercular
muscle scar with irregularly concentric lines.
Scale = 250 um. B. Vertical fracture of outer prismatic layer.
Scale = 10 um. C. Vertical
fracture of the middle layer, showing spherulitic arrangement of prisms. Scale = 10 um. D. Oblique view of a section of inner prismatic layer.
Scale = 5 um.
brown outer layer. Linear patterns of the first-order units of the middle layer are parallel near the shell margin and in- crease in irregularity toward the center (Figure 6). In the typical regular simple CL structure near the apertural mar- gin, the crossing angle of the second-order lamellae is ap- proximately 125 degrees. The inner layer consists of a CCL structure with fan-shaped second-order and lath-type third- order units (Figure 7D, E).
The inner lip projects inside along the posterior apertural margin and is formed as an extension of the CL structure of the middle layer (Figure 6). It lies between the ventral pos- terior of the visceral mass and the dorsal posterior of the foot of the animal.
The muscles scars are distributed in an elliptical form, keeping an almost constant distance from shell margin (Figure 6). They are inserted by two kinds of muscles of the animal: the thicker horseshoe-shaped part is the attachment of pedal retractor muscles (including head retractors in part), and the thinner anterior part is that of pallial muscle hanging the mantle onto the interior of the shell. The myostracum from these muscle scars is a thin layer of vertically oriented columnar prisms (Figure 7C). The surface of the pedal muscle scar is deeply impressed and exhibits a ridgelike
rough sculpture (Figure 7F).
Operculum.—The operculum is completely absent and was not found in any section of pedal musculature as al- ready described by Sasaki (1998: 120).
Discussion
Recent forms of neritopsine gastropods comprise nearly 120 genera and subgenera (Vought, 1989). Although differ- ent opinions exist regarding suprageneric systematics, the Recent forms can be grouped into at least seven families, namely Neritopsidae, Hydrocenidae, Helicinidae [this family may be divided into Ceresidae, Proserpinidae, and Helicinidae (Thompson, 1980)], Titiscaniidae, Neritiliidae, Neritidae, and Phenacolepadidae [“Shinkailepadidae” is probably included here] (Ponder, 1998; Sasaki, 1998; Bandel and Fryda, 1999; Kano and Kase, 2000a, b; see also Sasaki, 1998 for their anatomical basis). Information on their shell and opercular structure can be summarized as fol- lows.
Shell structure of Recent Neritopsina Shells of only ten genera belonging to six families have
Anterior
Outer Layer Pallial Muscle Scar
Middle Layer
Inner Layer
Apex Posterior
2 mm
Figure 6. Apertural view of Cinnalepeta pulchella. The sculpture texture of the surface is illustrated slightly schematically. The outer layer is separated from other parts by deep brown color, the middle layer is represented by concentric lines of CL structure, and the inner layer is visible as fine irregular lines of CCL structure.
been investigated at the microstructural level (Table 1). Major differences among suprageneric taxa are found mainly in the number of shell layers, the microstructure of each layer, and the crystal forms of carbonate calcium (aragonite-calcite).
(1) Neritopsidae: This family is characterized by intact inner upper whorls in contrast to other families with resorbed, hollow whorls inside (Bandel and Fryda, 1999). The shell wall of Neritopsis radula was described as “having two crossed-lamellar layers” by Batten (1979), but according to Suzuki et al. (1991), it is composed of an aragonitic outer layer of CL structure and an aragonitic inner layer of “proto- crossed lamellar, irregular prismatic, homogeneous, and complex crossed lamellar structures.” The mixture of four microstructures in the inner layer of N. radula is, therefore, a unique feature among Neritopsina.
(2) Hydrocenidae: The shell of Georissa japonica has three layers including the myostracum: the thicker aragonitic outer layer is of CL structure, and the thinner inner aragonitic
Shell structure of neritopsine gastropods
layer is primarily of irregular prismatic structure and subsidiarily of “protocrossed-lamellar’ structure (Suzuki et al., 1991).
(3) Helicinidae: The shell of Waldemaria japonica (Helicininae) has almost the same structural design as that of Georissa japonica, but the inner layer is mainly occupied by “protocrossed-lamellar” structure (Suzuki et a/., 1991). Microstructural data have not been provided for any other member of these families.
(4) Titiscanidae: This family totally lacks the shell at least at the adult stage (Bergh, 1890; Taki, 1955).
(5) Neritilidae: This group had been extremely poorly known taxonomically but was redefined by Kano and Kase (2000a, b) as small neritiform gastropods with (i) spiral ridges on the protoconch, (ii) the inclination of the protoconch against the teleoconch, and (iii) perpendicular, not inclined, prisms in the outer shell layer. The shell of Pisulina species consists of four layers: an outer layer of simple irregular prisms, middle layer of simple CL structure, myostracum, and inner layer of CCL structure with prismatic sublayers (Kano and Kase, 2000b; figs. 6, 7).
(6) Neritidae: All neritid taxa so far investigated share a four-layered shell consisting of an outer layer of calcitic pris- matic structure, middle layer of aragonitic CL structure, myostracum, and inner layer of aragonitic CCL structure (Table 1). Their shells can be further classified into two types based on relative thickness of shell layers: marine species have a thicker outer layer with a_ thinner periostracum, while nonmarine species have a reduced outer layer with a well developed periostracum (Suzuki et al. 1991). This difference is, however, considered to be in- duced by environmental factors, because thin shells pro- tected by a thick periostracum occur in various distantly related brackish and freshwater mollusks.
(7) Phenacolepadidae: The only description for this family was given for Cinnalepeta pulchella in this study, and the dif- ference from other families lies in the homogeneous struc- ture of the outer layer.
Operculum of Recent Neritopsina
Neritopsine operculum exhibits a great diversification in the number of calcified layers, the coverage of the organic layer, the position of calcareous layer(s) on one or both sides of the organic layer, the presence or absence of apophysis, and the morphology of muscle scars.
(1) Neritopsidae: The exterior surface of the operculum of Neritopsis radula is covered with a thick callus without a nu- cleus and spiral lines; the interior is divided into a smooth semilunar zone at the abaxial side and a large projection with a radial striation at the adaxial side (Thiele, 1929: fig. 55; Wenz, 1938: fig. 1001; Knight et al., 1960: fig. 182; Ponder, 1998: fig. 15.71C, D). This projection may be the hypertrophied homologue of the apophysis of other neritopsines, with its origin shifted toward the center. The absence of a spiral line on both surfaces and a large projec- tion from the interior center is quite unique among Gastropoda. However, no microstructural data for this fam- ily have been published to date.
(2) Hydrocenidae: The operculum of Georissa japonica is calcified with an apophysis and three-layered: the organic
28
Takenori Sasaki
Figure 7. SEM micrographs of shell microstructure of Cinnalepeta pulchella (UMUT RM27952). A. Vertical fracture of the outer layer.
Scale = 20 um. B. Enlarged view of homogeneous structure.
Scale = 5 um. C. Vertical fracture of pedal muscle scar showing the
myostracum overlying the middle layer. M= myostracum. Scale= 10 um. D. Vertical fracture of CCL structure near the center of the shell. Scale = 10 um. E. Oblique fracture (below) and outcrop surface (above) of CCL structure. Scale=50 um. F. Oblique view of pedal muscle
scar near inner lip, showing irregular rough surface.
layer on the outermost surface is underlain by two aragonitic layers of irregular prismatic structure (Suzuki, et al., 1991: fig. 5).
(3) Helicinidae: This family is highly specialized for Neritopsina in that Ceresinae and Proserpininae completely lack an operculum, while Helinicinae have a calcified operculum without an apophysis (Thompson, 1980). The
Scale = 200 pm.
operculum of Waldemaria japonica has a single layer of calcitic blocky structure on an organic layer (Suzuki, et al, 1991). Helicinids are strikingly different form other neritopsines in that calcification occurs only on the exterior surface, not interior of the organic layer.
(4) Titiscanidae: This shell-less taxon also lacks an operculum and is completely sluglike (Bergh, 1890; Taki,
Table 1. Published data on shell microstructure of the Recent Neritopsina. crossed lamellar, HO = homogeneous, IPR = irregular prismatic, PCL = protocrossed lamellar, PR = prismatic.
Shell structure of neritopsine gastropods
29
BL = blocky, CCL = complex crossed lamellar, CL = simple
Shell microstructu
re
Family Genus Outer layer(s) Inner layer Reference
Neritopsidae Neritopsis CL CL Batten (1979)
Neritopsis CL PCL, IPR, HO, CCL Suzuki et al. (1991) Hydrocenidae Georissa CL PCL, IPR Suzuki et al. (1991) Helicinidae Waldemaria CL PCL, IPR Suzuki et al. (1991) Neritiliidae Pisulina PR CL CCL Kano and Kase (2000b) Neritidae Nerita PR CL CCL Boggild (1930); Gainey and Wise (1980);This study
Nerita BL CL PCL Suzuki et al. (1991)
Neritina PR CL CCL Boggild (1930)
Neritina BL CL PCL Suzuki et al. (1991)
Clithon BL CL PCL Suzuki et al. (1991)
Neripteron BL CL PCL Suzuki et al. (1991)
Septaria BL CL PCL Suzuki et al. (1991) Phenacolepadidae Cinnalepeta HO CL CCL This study
1955).
(5) Neritiliidae: In the species of Pisulina, the operculum is secreted with exterior corneous and interior calcareous lay- ers (Kano and Kase, 2000b: figs. 8, 9, 11). The small initial part containing the nucleus on the corneous layer is demar- cated from the remaining part. The muscle scars are di- vided into three areas: two elongate zones along the adaxial and basal margins, and a central one extending between the nuclear zone and apophysis. The inner calcareous layers are formed of spherulitic prisms (Kano and Kase, 2000b: fig. 6F). (6) Neritidae: The gross morphology of the neritid operculum is classified into two distinct types (see e.g. Starmühlner, 1993; Komatsu, 1986): (i) The operculum of Septaria (subfamily Septariinae) is embedded in the dorsal part of the foot. The anterior left corner has a sharp projec- tion which is supposedly homologous to the apophysis of other neritopsines on the left side. (ii) The rest of the neritids (subfamilies Neritinae and Smaragdiinae) have a semilunar external operculum with a prominent apophysis.
The opercular microstructure of neritids is known to be variable, depending on habitats. The marine neritids have a common opercular plan with a single exterior aragonitic prismatic layer, an organic layer, and two interior aragonitic prismatic layers (Suzuki et a/., 1991: fig. 5; this study: Figs. 4,5). The opercula of nonmarine species (Neritina and Clithon) have only two aragonitic prismatic layers covered by a well developed organic layer (Suzuki et al., 1991).
The operculum of Bathynerita naticoidea (tentatively treated as a neritid here) dwelling exclusively in deep-sea hydrocarbon seeps is “partly calcified with a thicker calcare- ous layer where it is attached to the foot’ (Warén and Bouchet, 1993), and the apophysis is absent. Its operculum with only partial calcification is greatly different from those of shallow-water neritids.
(7) Phenacolepadidae: The opercular morphology of phenacolepadids can be divided into three distinct states. (i) The presence of a vestigial internal operculum with inte-
rior calcification and rudimentary apophysis was docu- mented by Fretter (1984) in Phenacolepas omanensis and observed with SEM for the first time by Kimura and Kimura (1999: fig. 7C, D) in Phenacolepas sp. (ii) The opercula of so-called “Shinkailepedidae” has double structure of calci- fied anterior and noncalcified posterior parts as described in Shinkailepas by Okutani et al. (1989: fig. 12) and Beck (1992: pl. 1, fig. 4) and in Olgasolaris by Beck (1992: pl. 5, fig. 4). Double-layered nail-shaped operculum is strikingly convergent with that of the neritid genus Septaria. (iii) The operculum is absent in Cinnalepeta, as described above.
Implication of neritopsine hard-part microstructures
It has been generally accepted that microstructural char- acters of the shells are useful for the understanding of molluscan higher taxonomy. It is, however, necessary to check the correlation between taxonomic distribution of structural morphotype and phylogenetic relationships in the Recent taxa before comparing extant and extinct forms directly.
(1) Shell structure: As reviewed above, the patterns of shell structure of the Recent Neritopsina can be categorized into two major types: (i) Genera of three families, Neritidae, Phenacolepadidae, and Neritiliidae, all have four-layered shells in which inner CCL and middle CL layers are overlaid by a prismatic (in Neritidae and Neritiliidae) or homogeneous (in Phenacolepadidae) shell layer. (ii) By contrast, other families such as Neritopsidae, Helicinidae, and Hydro- cenidae secrete three-layered shells which consist mostly of CL/CCL structures and lack an additional outer layer. This apparent difference between the two groups may be viewed as expressing the distinctness of their relationship, but it is still premature to present phylogenetic implication because of the lack of a reliable phylogenetic hypothesis. The phylogenetic analysis has been conducted only in a part of Neritopsina by Holthuis (1995) and Sasaki (1998), and the phylogenetic status of Neritopsidae, Helicinidae, and Hydrocenidae is totally unknown. The scarceness of struc-
30 Takenori Sasaki
tural data relative to the number of existing genera is also problematic so far as testing the stability of character states within each family.
Another unresolved problem is whether all of the taxa with a four-layered shell share a calcitic outer layer or not. Suzuki et al. (1991) revealed that at least the shell of neritids is constructed from a calcitic outer layer and otherwise aragonitic layers in contract to the entirely aragonitic shell of other families. They argued that this bimineralic composi- tion is attributable to adaptation to a shallow aquatic environ- ment, probably as a means of reinforcing the mechanical strength of the shell. Thus, it should be tested as a next step whether non-neritid four-layered shells are also made of two crystal forms of calcium carbonate. Concerning the outermost shell layer, Taylor and Reid (1990) revealed the parallel homoplastic addition of a calcitic outer layer in some genera within littorinid gastropods. This means that the convergence in conchological characters should necessarily be considered also at the microstructural level in other groups of molluscs.
(2) Opercular structure: The opercular structure can be di- vided into several types as a result of the above comparison: (i) three (single exterior and two interior) aragonitic layers in marine neritids, (ii) two interior aragonitic layers in Hydrocenidae and nonmarine neritids, (iii) single calcitic ex- terior layer without interior calcification in Helicininae, (iv) interiorly calcified operculum with unknown layer distribution in Neritiliidae and a part of Phenacolepadidae, and (v) ab- sence of an operculum in Titiscaniidae and in part in the Phenacolepadidae and Helicinidae. Thus, a single similar state often occurs across several different families, and also several different states can coexist within the same family. At the family level the similarity and dissimilarity in opercular structure are very difficult to explain in the phylogenetic con- text.
In connection with nonphylogenetic factors, the less calci- fied opercula in nonmarine neritids as compare with marine confamilial members may be explained as a consequence of adaptation to low-salinity environments (Suzuki et al., 1991). However, in other taxa, there is no Clear correlation between opercular structure and habitat selection. The marked dif- ferences in neritopsine opercular structure is difficult to un- derstand also in terms of adaptation. The possession of apophysis is presumably under phylogenetic control within Neritopsina, and at the same time, it is a convergent state also found in caenogastropod rissoideans. The peglike structure has possibly arisen to increase the area of muscu- lar attachment in this case.
Because of the insufficient resolution of neritopsine phylogeny and the lack of mineralogical data in part, the evolutionary scenario of neritopsine hard parts remains largely speculative at present. For further studies, exoskeletal structure including mineralogical characters should be investigated more comprehensively in whole ex- tant and extinct neritopsines together with the comparative anatomy and molecular phylogeny of the Recent species.
Acknowledgments
| deeply thank Kazushige Tanabe (Department of Earth
and Planetary Science, University of Tokyo), Yasunori Kano (Department of Biological Sciences, University of Tokyo) and two anonymous reviewers for their constructive com- ments and suggestions. The samples of Cinnalepeta pulchella were kindly provided by Shigeo Hori (Kuroda Chiromorphology Project, Japan Science and Technology Corporation). This work was partly supported by the Grant- in-Aid from the Japan Society for Promotion of Science (No. 12440141).
Reference
Bandel, K., 1979: Übergänge von einfacheren Strukturtypen zur Kreuslamellenstruktur bei Gastropodenschalen. Bio- mineralization, vol. 10, p. 9--38.
Bandel, K., 1982: Morphologie und Bildung der frühonto- genetischen Gehäuse bei conchiferen Mollusken. Facies, vol. 7, p. 1-198.
Bandel, K., 1988: Early ontogenetic shell and shell structure as aids to unravel gastropod phylogeny and evolution. Malacological Review, Supplement, vol. 4, p. 267-272.
Bandel, K., 1991: Ontogenetic changes reflected in the mor- phology of the molluscan shell. /n, Schmidt-Kittler, N. and Vogel. K., eds., Constructional Morphology and Evolution, p. 111-230. Springer, Berlin.
Bandel, K., 1992: Platyceratidae from the Triassic St. Cassian Formation and the evolutionary history of the Neritomorpha (Gastropoda) Paläontologische Zeitschrift, vol. 66, p. 231-240.
Bandel, K. and Fryda, J., 1999: Notes on the evolution and higher classification of the subclass Neritimorpha (Gastropoda) with the description of some new taxa. Geologica et Palaeontologica, vol. 33, p. 219-235.
Bandel, K. and Geldmacher, W. 1996: The structure of the shell of Patella crenata connected with suggestions to the classification and evolution of the Archaeogastropoda. Freiberger Forschungshefte, vol. C464, p. 1-71.
Batten, R, L., 1979: Gastropods from Perak, Malaysia Part 2. The trochids, patellids, and neritids. American Museum Novitates, no. 2685, p. 1-26.
Beck, L. A., 1992: Two new neritacean limpets (Gastropoda: Prosobranchia: Neritacea: Phenacolepadidae) from hydrothermal vents at hydrothermal Field 1 “Wienerwald” in the Manus Back-Arc Basin (Bismarck Sea, Papua New Guinea). Annalen des Naturhistorischen Museums in Wien B, vol. 93, p. 259-275.
Bergh, R., 1890: Die Titiscanien, eine Familie der rhipido- glossen Gastropoden. Morphologische Jahrbücher, vol. 16: p. 1-26.
Boggild, ©. B., 1930: The shell structure of the mollusks. Det kongelige Danske Videnskabernes Selskabs Skrifter. Naturvidenskabelig og Mathematisk Afdeling, ser. 9, vol. 2, p. 231-326.
Carter, J. G. and Clark, G. R. Il, 1985: Classification and phylogenetic significance of molluscan shell microstruc- ture. In, Broadhead, T. W. ed, Molluscs, Notes for a Short Course. Studies in Geology no. 13, p. 50-71. University of Tennessee.
Fretter, V., 1984: The functional anatomy of the neritacean limpet Phenacolepas omanensis Biggs and some com- parison with Septaria. Journal of Molluscan Studies, vol. 50, p. 8-18.
Gainey, L. F., Jr. and Wise, S. W., Jr., 1980: Convergent shell
morphology in intertidal gastropods. Journal of
Molluscan Studies, vol. 46, p. 192-207.
Haszprunar, G., 1988: On the origin and evolution of major gastropod groups, with special reference to the Streptoneura. Journal of Molluscan Studies, vol. 43, p. 367-441.
Hedegaard, C., 1997: Shell structures of the Recent Vetigastropoda. Journal of Molluscan Studies, vol. 63, p. 369-377.
Hedegaard, C., Lindberg, D. R. and Bandel, K., 1997: Shell microstructure of a Triassic patellogastropod limpet. Lethaia, vol. 30, p. 137-148.
Holthuis, B. V., 1995: Evolution between Marine and Freshwater Habitats: a Case Study of the Gastropod Neritopsina, 286 p. Doctoral dissertation, University of Washington. [Abstract in Bulletin of the Malacological Society of London, no. 29 (1997), p. 9]
Kano, Y. and Kase, T., 2000a: Pisulinella miocenica, a new genus and species of Miocene Neritiliidae (Gastropoda: Neritopsina) from Eniwetok Atoll, Marshall Islands. Paleontological Research, vol. 4, p. 69-74.
Kano, Y. and Kase, T., 2000b: Taxonomic revision of Pisulina (Gastropoda: Neritopsina) from submarine caves in the tropical Indo-Pacific. Paleontological Research, vol. 4, p. 107-129.
Kase, T., 1994: New species of Patella (Gastropoda: Mollusca) from the Miocene and Pliocene of Japan: A clade extinct from the Pacific. Bulletin of the National Science Museum, Tokyo, Series C, vol. 20, p. 53-65.
Kase, T. and Shigeta, Y., 1996: New species of Patello- gastropoda (Mollusca) from the Cretaceous of Hokkaido, Japan and Sakhalin, Russia. Journal of Paleontology, vol. 70, p. 762-771.
Kase, T. and Hayami, |., 1992: Unique submarine cave mollusc fauna: composition, origin and adaptation. Journal of Molluscan Studies, vol. 58, p. 446-449.
Knight, J. B., Cox, L. R., Keen, A. M., Batten, R. L., Yochelson, E. L. and Robertson, R. 1960: Systematic description. /n, Moore, R. C. ed, Treatise on Invertebrate Paleontology. Part |. Mollusca 1, p. 1169-1310. Geological Society of America and Kansas University Press, Colorado and Kansas.
Kimura, S. and Kimura, T., 1999: The gastropod fauna of the marshes of the reed (Phragmites australis (Cav.)) in the estuaries in Mikawa Bay and Ise Bay, Japan. Japanese Journal of Benthology, vol. 54, p. 44-56. (in Japanese with English abstract)
Komatsu, S., 1986: Taxonomic revision of the neritid gastro- pods. Special Publication of the Mukaishima Marine Biological Station 1986, p. 1-69.
Lindberg, D. R., 1988: The Patellogastropoda. Malacological Review, Supplement, vol. 4, p. 35-63.
Lindberg, D. R., 1998: Order Patellogastropoda. /n, Beesley, P. L., Ross, G. J. B. and Wells, A. eds, Mollusca: The Southern Synthesis. Fauna of Australia, vol. 5, p. 639- 652. CSIRO Publishing, Melbourne.
Lindberg, D. R. and Hedegaard, C., 1996: A deep water patellogastropod from Oligocene water-logged wood of Washington State, USA (Acmaeoidea: Pectinodonta). Journal of Molluscan Studies, vol. 62, p. 299-314.
Linnaeus, C., 1758: Systema Naturae, Editio decina, iv+823 p. L. Salvius, Stockholm.
Lischke, C. E., 1871: Japanische Meeres-Conchylien. Ein
Shell structure of neritopsine gastropods
Beitrag zur Kenntnis der Mollusken Japans, mit besonderer Rücksicht auf die geographische Verbreitung derselben, vol. 2. 184 p., pls. 1-14. Theodor Fischer, Cassel.
MacClintock, C., 1967: Shell structure of patelloid and bellerophontoid gastropods (Mollusca). Peabody Museum of Natural History, Yale University, Bulletin, vol. 22, p. 1-140.
Okutani, T., Saito, H. and Hashimoto, J., 1989: A new neritacean limpet from a hydrothermal vent site near Ogasawara Islands, Japan. Venus, vol. 48, p. 223-230.
Ponder, F. W., 1998: Superorder Neritopsina. /n, Beesley, P. L., Ross, G. J. B. and Wells, A. eds., Mollusca: The Southern Synthesis. Fauna of Australia, vol. 5, p. 693- 703. CSIRO Publishing, Melbourne.
Ponder, F. and Lindberg, D. R., 1997: Towards a phylogeny of gastropod molluscs: an analysis using morphological characters. Zoological Journal of the Linnean Society, vol. 119, p. 83-265.
Rafinesque, C. S. 1815: Analyse de la Nature, ou Tableau de l'Univers et des Corps Organisés, 244 p. Ralermo, Italy.
Sasaki, T., 1998: Comparative anatomy and phylogeny of the recent Archaeogastropoda (Mollusca: Gastropoda). The University Museum, The University of Tokyo, Bulletin, no. 38, p. 1-223.
Starmühlner, F., von, 1993: Ergebnisse der österreichischen Tonga-Samoa Expedition 1985 des Instituts für Zoologie der Universität Wien: Beiträge zur Kenntnis der Süß- und Brackwasser-Gastropoden der Tonga- und Samoa-Inseln (SW-Pazifik). Annalen des Naturhistorischen Museums in Wien B, vol. 94/95, p. 217-306.
Suzuki, S., Togo, Y. and Y. Uozumi, S., 1991: Exoskeletal ad- aptations for life environment in neritacean gastropods. In, Suga, S. and Nakahara, H. eds, Mechanisms and Phylogeny of Mineralization in Biological Systems, p. 397-401. Springer-Verlag, Tokyo.
Taki, I., 1955: New species of Titiscania, a rare archaeogastropd mollusk from Hachijo-jima. Bulletin of the National Science Museum, Tokyo, vol. 2, p. 50-53.
Taylor, J. D. and Reid, D. G., 1990: Shell microstrucutre and mineralogy of the Littorinidae: ecological and evolutionary significance. Hydrobiologia, vol. 193, p. 199-215.
Togo, Y. and Suzuki, S., 1988: Shell structure and phylogeny of the Gastropoda. /n, Omori, M, Suga, S., and Goto, M. eds, Biomineralization and Phylogeny of Marine Organisms, p. 113-134. Tokai University Press, Tokyo. [in Japanese]
Thiele, J., 1929: Handbuch der systematischen Weichtier- kunde. Teil 1 (Loricata; Gastropoda: Prosobranchia), 134 p. Gustav Fischer Verlag, Jena.
Thompson, F. G., 1980: Proserpinoid land snails and their re- lationships within the Archaeogastropoda. Malacologia, vol. 20, p. 1-33.
Vaught, K. C. 1989: A Classification of the Living Mollusca, 189 p. American Malacologists, Florida.
Waren, A. and Bouchet, P., 1993: New records, species, gen- era, and a new family of gastropods from hydrothermal vents and hydrocarbon seeps. Zoologica Scripta, vol. 22, p. 1-90.
Wenz, W., 1938: Gastropoda. Teil 2: Prosobranchia. Handbuch der Paläozoologie. Band 6, p. 241-477. G ebrüder Borntraeger, Berlin.
tank
nf Un Ni Ana .
jan «ay
ioe wur A I Ira toe
hr | , u à PCT 2 "dite N Fr ar : À = re Bert MR? : x
u eur q En ey TEN 6. a EN ur * Var tri inet htt i Pen te eit
| PCA TT Pr) 4) ‘nie? Ds uy) ot:
DOME vey VOR Be eit
MU Du gi Ri
uly et We jit dnt Nas AN x.
à a aes fh ety bal ‘4 jae NIEREN rs ahi
w ke ty TE te ER : i pa ‘ Wi LE OT Or Meas REN , et ee ds pes Seay wet Soy a : yer wart Wick LS à : CVS { oy fi Ke PS aM L
gt PAT nes Er wee ORE NA NER re wer‘ c PA és à“ Sov te 4. she fd CRT LS ah! ane f i où qu: ls hier” CH yO M = MAS DE k A SE a one Ceara) A Lge 9 i ees | oe Se A:
ef PL CIE <i >
Fe} SL A. dt Tann, cp « . sv ‘Lew a L
è n
Paleontological Research, vol. 5, no. 1, pp. 33-44, April 30, 2001 © by the Palaeontological Society of Japan
Geometric pattern and growth rate of prismatic shell structures in Bivalvia
TAKAO UBUKATA
Institute of Geosciences, Shizuoka University, Oya 836, Shizuoka, 422-8529, Japan (sbtubuk @ipc.shizuoka.ac.jp)
Received 11 October 2000; Revised manuscript accepted 27 November 2000
Abstract. The distribution patterns, sizes and nucleation sites of aggregated prisms on the outer shell surface were examined in 16 species of Bivalvia and modeled theoretically. Biometric analy- sis shows a negative correlation between the median size and variation of sizes of calcitic simple prisms. In species with aragonitic vertical composite prisms, instead, the density of prisms tends to decrease when their nucleation sites are randomly distributed. Comparison of the results of computer simulations with those of biometric analyses reveals the following: 1) a positive correla- tion between growth rate of prisms and the probability of nucleation for simple prisms, and 2) a limit of the number of nucleations per unit time in vertical composite prisms. Prism size correlates with the growth rate of the entire shell or prisms, and increases as the shell grows faster or prisms grow slower.
Key words: biomineralization, bivalves, prismatic structure, shell growth rate, theoretical morpho-
logy
Introduction
The microscopic features of a molluscan shell record cer- tain physiological conditions of the organism at the time the shell is formed. Carbonate minerals within a molluscan shell crystallize and grow under physicochemical conditions controlled by the physiology of the organism. Wada (1972, 1985) reported a seasonal change of shape and size of aragonite crystals in the nacreous layer of such bivalves as Pinctada fucata, Pinna attenuata and Hyriopsis schlegeli. He suggested that the topography of growing crystals may reflect the rate of crystal growth or the degree of supersaturation of the extrapallial fluid. Quantitative analy- sis of the relationship between size and/or shape of crystals and the rate of crystal growth within a shell provides a reli- able basis to understand the ‘paleophysiology’ of fossil or- ganisms.
The goal of this study is to clarify the relationship between the geometry of bivalve shell microstructure and the relative growth rate of crystals or of the entire shell. For this pur- pose, the present study focuses on the geometry of the outer shell surface of a simple or vertical composite pris- matic shell layer. Although they differ in the ultrastructure of prisms, those two prismatic structures both consist of many parallel-arrayed columnar units (Carter and Clark, 1985; Carter et al, 1990). Each prism is surrounded and bounded by an organic matrix showing a honeycomb-like appearance on the outer shell surface.
For understanding the rule or algorithm forming the geo- metric pattern of shell microstructure, theoretical morphol- ogy is particularly useful (Ubukata, 1997a, b, 2000). Inthe present study, a biometric analysis of size and nucleation sites of prisms in actual shells was carried out in 16 species. Furthermore, a theoretical morphological modeling of growth kinematics of aggregated prisms was attempted, and the computer simulations of that model were compared with the results of the biometric analyses.
Biometric analyses
Material and methods
The outer shell surface of a simple prismatic or vertical composite prismatic outer shell layer was examined in 16 extant species of Bivalvia (Table 1). Each species was rep- resented by a single specimen, except Anodonta woodiana. Most of them were collected at various localities around the Japanese Islands and the Philippines. All the specimens examined are stored at Shizuoka University (SUM).
In order to remove the periostracum from the shell com- pletely, the shells examined were bleached for one day. Pieces of them were washed, dried in air, coated with gold using a JEOL JFC-1500 ion coater, and examined with a JEOL JSM-5800LV scanning electron microscope operated at 15kV and interfaced to a computer (Dell Optiplex Gxa EM).
In order to analyze size-frequency distribution of prisms,
34
Takao Ubukata
Table 1. List of material examined. All specimens have the prefix SUM. 2 Family species locality specimens
Pteriidae Pteria penguin (Roding) San Luice, Bathangas, Philippines HM-B-0014
Pinctada maculata (Gould) Iriomote Is., Okinawa, southwest Japan HM-B-0015 Isognomonidae /sognomon perna (Linnaeus) Iriomote Is., Okinawa, southwest Japan HM-B-0016
I. ephippium (Linnaeus) San Luice, Bathangas, Philippines HM-B-0017 Malleidae Malleus regula (Forskäl) Iriomote Is., Okinawa, southwest Japan HM-B-0018 Pinnidae Atrina pectinata (Linnaeus) Ariake, Saga, western Japan HM-B-0019
A. vexillum (Born) Honda Bay, Palawan, Philippines HM-B-0020 Ostreidae Crassostrea gigas (Thunberg) Misaki, Kanagawa, Central Japan HM-B-0021 Margaritiferidae Margaritifera laevis (Haas) Nakagawa, Hokkaido, northern Japan HM-B-0022 Unionidae Inversidens reiniana (Kobelt) Lake Biwa, Shiga, Central Japan HM-B-0023
Unio biwae Kobelt Lake Biwa, Shiga, Central Japan HM-B-0024
Lanceolaria oxyrhyncha (Martens) Lake Biwa, Shiga, Central Japan HM-B-0025
Anodonta woodiana (Lea) Lake Biwa, Shiga, Central Japan HM-B-0026, -0027
A. calypygos Kobelt Lake Biwa, Shiga, Central Japan HM-B-0028
Cristaria plicata (Leach) Lake Biwa, Shiga, Central Japan HM-B-0029 Trigoniidae Neotrigonia margaritacea (Lamarck) French Is., Australia HM-B-0030
EF IE OY. = GE ar
TED
areas of prisms on the outer shell surface were measured at 6-12 positions along a growth increment on the shell sur- face. An SEM image of the measured portion was saved as a computer bitmap file (Figure 1A). Next, the boundaries between prisms were traced on a NEC PC-9821 V166 per- sonal computer using Microsoft PowerPoint 7.0, and then each prism was colored differently using Justsystem Hanako PhotoRetouch (Figure 1B). Subsequently, the area of each prism was measured by counting pixels. For this counting, a program written in VISUAL BASIC was used on a personal comouter.
Since the size-frequency distribution of the areas of prisms is generally right-skewed, the mean and standard de- viation are not suitable for representing the distribution of the areas of prisms. Therefore, the “average” area of prisms on a shell is represented by the median of the areas (5), and the variation of the areas is expressed as a standardized hinge spread (Q.), which is defined as follows:
_~G=Q Q, = a (1) where Q, and Q, are the first and third quartiles of the areas of prisms, respectively (Hoel, 1976). S and Q, were both estimated in all shell portions examined.
On the outer surface of a vertical composite prismatic shell layer, microgrowth increments are clearly visible within a bleached prism (Figure 2A). In a simple prismatic shell layer, growth increments within a prism are faintly observed on the bleached shell surface (Figure 2B). In either case,
Figure 1. A. SEM photograph of the outer shell surface of the vertical composite prismatic layer in Anodonta woodiana (SUM- HM-B0027), scale: 50 um. B. Trace of the outlines of prisms on the SEM image of A.
Geometry of bivalve prismatic shell 35
GR TOURS | LIE fet tae aS Dim 22
Wala ir | Y L
Figure 2. A. SEM photograph of the outer shell surface of the vertical composite prismatic shell layer in A. woodiana (SUM-HM-B0027) showing clear growth increments within individual prisms, scale: 20 um. B. The simple prismatic layer in Pinctada maculata (SUM-HM- B0015). Arrows indicate nucleation sites, scale: 20 um. C. The SEM image of A. woodiana subdivided into squares, scale: 50 um. D.
Distribution of nucleation sites of prisms in C.
the center of the circular growth increments is regarded as the nucleation site of the prism.
The distribution of nucleation sites of prisms was analyzed in all shell portions in which the areas of prisms were meas- ured. An SEM image was divided into a squared grid of ap- propriate size for each square to include an average of two nucleation sites (Figure 2C). Next, the number of nuclea- tion sites was counted in each square (Figure 2D). The numbers of squares and nucleation sites in each square both define the /; index of Morisita (1959), which represents the nonuniformity of a distribution independent of the size of the quadrates. The /; index is defined as:
FT (2)
where n is the number of squares, x, is the number of sites in the square i. The value of /; is zero when the distribution is perfectly uniform (Figure 3A). /; increases as the distribu- tion becomes nonuniform (Figure 3B), and it is expected to be one when sites are distributed randomly (Figure 3C). When the distribution of sites is biased considerably, /; has a large value (Figure 3D). The value of /; was estimated in every shell position examined, and the nonuniformity of the distribution of nucleation sites was represented by Js.
Results
The biometric analyses revealed a negative correlation between $ and Q, in a single specimen of /sognomon perna, and a positive correlation in a single specimen of Malleus regula (Figure 4B). In another species, no correla- tion was observed, though the number of measured portions
36 Takao Ubukata
Figure 3. Effect of the spatial distribution pattern on the value of Js. A. Perfectly uniform distribution.
including a number of or no dots. Considerably concentrated distribution.
within a single valve was not large enough to be significant. On the other hand, when the data from all species are com- bined, Q. is negatively correlated with S at the 0.01 level of significance, both for simple prismatic and vertical composite prismatic shells (Figure 4A-B). This shows that variation in the size of prisms tends to decrease as prism size in- creases. The negative correlation between Q, and S is clear especially in species with a simple prisms (Figure 4B), while the correlation is more or less obscure in species with vertical composite prisms (Figure 4A).
A positive correlation between Js and ©, is found only in each specimen of Unio biwae, Anodonta woodiana and Malleus regula (Figure 4C, D). However, when all the data are combined, a positive correlation clearly emerges be- tween /; and ©, in species with vertical composite prisms (Figure 4C). In the case of species with simple prisms, Q. is positively correlated with 7; at the 0.05 level of signifi- cance, though the correlation is graphically unclear (Figure 4D). The positive relationship between /; and ©, indicates that variation of the size of prisms tends to increase with in- creasing randomness of the distribution of nucleation sites.
Theoretical morphology
Growing circles model
To better understand the relationships between geometric features of prisms and the growth rate of each prism and/or of the net growth rate of the entire shell, the growth of aggre- gated prisms was modeled theoretically.
C(,=1.07)
In this figure, the area is divided into 16 contiguous squares each B. Nearly uniform distribution. C. Random distribution. D.
For modeling the process of microscopic growth, observ- ing the initial growth stage of prisms helps understand the nature of crystal growth. Formerly, | reported that many small hemispherical incipient prisms occur on the inner sur- face of the periostracum at the growing margin in species possessing simple and vertical composite prisms (Ubukata, 1994, pl. 2, figs. 1-3). Consequently, a growing-circles model, which represents the growth of aggregated prisms, is introduced here.
Growth of a prismatic shell layer consists of three ele- ments, namely, nucleation of prisms, growth of prisms, and accretionary growth of the entire shell. During a single short growth step, the mantle secretes calcium carbonate and nucleation of prisms occurs within the nucleation zone on the inner surface of a periostracum (Figure 5A). The periostracum subsequently secreted by the mantle edge pushes the earlier produced periostracum and its incipient prisms into a more proximal part of the shell (Saleuddin and Petit, 1983). After the prisms pass through the nucleation zone, prisms gradually grow and elongate, forming a colum- nar structure.
Let us consider hypothetical shell growth (Figure 5B). The accretionary growth of the entire shell during a short pe- riod of time is reflected in a shift of the nucleation zone. Growth of a prism is represented by the kinematics of an en- larging circle. Potential nucleation sites are distributed uni- formly within a nucleation zone of width h. The distance between the potential nucleation sites can be expressed by d, which represents the size of a unit cell. Growth compo-
Geometry of bivalve prismatic shell 37
+ M. laevis
x I. reiniana
o U. biwae
x L. grayana + À. woodiana a A. calypygos AC. plicata
= N. margaritacea
0 500 1000
+ M. laevis
x L reiniana
o U. biwae
x L. grayana + À. woodiana
m À. calypygos AC.
plicata
0.5 0.7 0.9
I;
Figure 4. A-B. Relationship between $ and Q. in actual shells. species with simple prisms.
nents are expressed as functions of growth stage s, rather than as functions of time, since the time scale of the growth process is difficult to ascertain in many cases. Over a pe- riod of one growth step, the mantle secretes a periostracum at the shell margin, giving rise to the stippled area in Figure 5B. If the growing margin of the shell shifts downward by d
A. Data from species with vertical composite prisms. C-D. Relationship between /; and Q. for species with vertical composite prisms (C) and simple prisms (D).
= P. penguin + P. maculata Al. perna
+ I. ephippium
o M. regula
m À. pectinata x A. vexillum x C. gigas
0 500
= P. penguin + P. maculata AI. perna
+ I. ephippium o M. regula
m À. pectinata x A. vexillum
x C. gigas
0.7 0.9
B. Data from
during the step, the growth step is regarded to be a unit in- terval of the growth. Then, the growth step As is generally defined as:
AS (3)
38
REES RIRE TETE AR RETIRE PE EE REA
B
Takao Ubukata
Figure 5. A. Schematic diagram of the radial section of the shell and mantle margin of a bivalve showing the position where nucleation
of prisms occurs. B. The growing circles model.
Black bold points indicate initiation sites of produced prisms, and gray ones potential nu-
cleation sites of unborn prisms. During a given growth step As, the shell margin shifts by A/ (stippled area) and the radius of prisms increases
by AR (shaded area).
where the growing margin of the shell shifts by Al during the growth step.
Meanwhile, nucleation of a prism occurs at a potential nu- cleation site within the nucleation zone with a probability of q. Each prism is approximated by a circle which enlarges at a steady rate. During a growth step As, calcium carbonate precipitates along the circumference of each prism giving rise to a new additional rim shown by the shaded portion in Figure 5B, and the radius of each circle increases by AR. As the prisms grow, neighboring prisms come closer and fi- nally in contact with one another, as a result forming a boundary between two prisms. Nucleation of prisms occurs randomly during each growth step, as a result of irregularity of the settling time among prisms. Consequently, a growing circle often occupies the space of nucleation and/or growth of a neighboring ‘unborn’ prism. A newborn prism some- times loses in competition for space between neighboring prisms, and is geometrically terminated (Grigor’ev, 1965).
In the growing-circles model, growth of a shell and prisms is generally expressed by the following three parameters: C
The dotted line indicates the dorsal limit of the nucleation zone.
: the standardized growth rate of prisms, defined as the in- crease of the radius of a prism per growth step, normalized by size of a unit cell (d), P: probability of nucleation per growth step in each potential site, and L: the extent of the nucleation zone standardized by d. Probability of ‘failure’ of nucleation at each site per growth step is expressed as 1- P, and the probability of failure of nucleation during a growth step As (=1-g) is obtained by raising 1-P to As" power. Then, three growth parameters C, P and L are given by:
cer Al Now, we can generally define a growth increment during
an arbitrary growth step if three parameters C, P and Z are given.
(er), L= (4)
Computer simulation
In order to evaluate the effects of parameters C, P and L on the geometric pattern of prismatic structure, computer simulations were performed for growth of prisms based on
Geometry of bivalve prismatic shell 39 |
Figure 6. Color display of the growing circles model. Each prism is identified by its color. As C increases, size variation of the prisms also increases. As P or L increases, the median size of prisms tends to decrease.
Takao Ubukata
40
©
er
tT TELL ts
3
S
a
S
1
=
a
Co
>
Figure 7. Three-dimensional block diagrams showing the relationships among growth parameters (C, P, L) and statistics $ (A-C), Q.
(D-F) or I; (G-1) on a C-P diagram at L values of 1.2, 3.6 and 7.2.
The coordinates of a
nucleation site were recorded on each hypothetical prism for
The dimension of d is ex- calculating the value of J; in a model. Computer simulations
where À is the median of A. pressed as pixels on the computer.
In the theoretical model, the
area of a hypothetical prism was measured as the number of pixels (4) on the display surface, and both the standardized hinge spread (Q,) and the median of a standardized area of
the growing-circles model.
hypothetical prisms (S,,) defined beiow were estimated in
each model:
n VISUAL BASIC by
means of a 64-bit workstation computer (Visual Technology owing circles model.
iyama A702h).
Figure 6 shows a spectrum of geometric patterns of hypo-
=" x
were carried out with a program written VT-Alpha 600) interfaced with a CRT (I thetical prisms that were made by the g
)
5
(
Geometry of bivalve prismatic shell 41
SOUdIOJUI
biometry
OUSJOJUI
Figure 8. Summary of the biometric analyses and computer simulation.
definition of parameters
shell growth rate
biometry simulation
From coupling their results with the consequences from defi-
nition of the parameters, a relationship between growth parameters is inferred either in simple prisms (A) or composite prisms (B).
Each prism is identified by its color in a theoretical prismatic structure. If Cis large, each prism grows considerably while it passes through the nucleation zone, as a result of the great irregularity of the birth time among prisms. Then, as C increases, the variation of prism size also increases. As P or L increases, the number of prisms increases while their size decreases.
Three-dimensional scatter diagrams illustrating the values of $, in relation to C and P, at L=1.2, 3.6 and 7.2 are given in Figure 7A-C, respectively. S.., which means an ‘average’ prism size, seems to be inversely proportional to P, which represents the probability of nucleation. In addition, at the same combination of C and P, 5, tends to increase as L de- creases. This fact indicates that the size of hypothetical prisms increases as the nucleation zone becomes narrow.
Figure 7D-F is a series of diagrams showing the relation- ship among four parameters C, P, Q. and L. When Lis large enough, ©. is positively correlated with C (Figure 7E, F). This fact suggests that variation of prism size tends to in- crease with increasing growth rate of prisms (AR in Eq.1) or with decreasing the accretion rate of the entire shell (A/ in Eq.1). When L is small, no clear relationship exists be- tween pairs of C, P and 9. (Figure 7D). In a high C condi- tion, Q. at a given P also tends to increase with increasing L (Figure 7D-F).
Figure 7G-I shows the variation of J; in relation to C, P and L. When L is large, /; gradually decreases with increasing P. Since /; represents the degree of nonuniformity, this re- sult indicates that the nucleation sites tend to be distributed uniformly as the probability of nucleation increases. In the high P region, /; at a given C decreases as L increases. This fact indicates that nucleation sites tend to be distributed randomly as the nucleation zone becomes narrower, when the probability of nucleation is high enough.
Geometric pattern and growth rate of prismatic shell
As mentioned above, the biometric analyses indicated a negative correlation between $ and Q. especially in the species with simple prisms (Figure 4B), and the computer simulation predicted an inverse relationship between P and S,, (Figure 7A-C) and also a positive correlation between C and ©, (Figure 7E, F). To sum up these results, it can be predicted that P increases as C increases for those species with simple prisms (Figure 8A). That is also inferred from the definition of parameters. Since both P and C are in- versely proportional to the growth rate of the entire shell, a positive correlation between C and P is quite reasonable (Figure 8A) if the size-frequency distribution of prisms is con- trolled mainly by the growth rate of the entire shell. Furthermore, even if both the growth rate of prisms and the probability of nucleation reflect the activity of mantle secre- tion, and if the size of prisms is controlled mainly by the ac- tivity of mantle secretion, the positive correlation between C and P is also expected.
Coupling of the biometrics and the simulation also sug- gests to us a relationship between the parameters P and L. The biometric analyses demonstrated a positive correlation between /; and Q., especially for species with vertical com- posite prisms as shown in Figure 4C. In addition, the com- puter simulation predicted a positive relationship between L and Q,, and a negative correlation between P and /; as shown in Figure 7G-I. These two results suggest that P possibly decreases as L increases in species with vertical composite prisms (Figure 8B). Actually, an inverse relation- ship between L and P is expected if we assume that the fre- quency of nucleation per unit time interval is fixed.
To ascertain this prediction, relationships among simu- lated values of $,, /; and ©, were analyzed for each combina-
42 Takao Ubukata
Ss
0.8 0.9 1 1.1 1.2
Figure 9. Relationships between $, and Q, and between 7; and Q, when three parameters satisfy equations as follows: C=aP, P=bIL, where a and b are coefficients. N represents the total number of samples examined, r exhibits the correlation coefficient, and p is the signifi-
cance level of the correlation.
tion of C, P and L, when a positive correlation between C and P and an inverse relationship between L and P are assumed, as in the following relations:
C=@P, jP==, (6)
where a and b are coefficients. Computer simulations were performed under the following conditions: 1) a is fixed to 3, and b varies between 0.05 and 0.175, 2) a varies between 1 and 3, and bis fixed to 0.05. Figure 9 is a diagram showing the relationships between S,, and Q, and between Land O.. The results of the simulations under a fixed value of a (=3) and various values of b show a negative correlation between
A-B. a=3, 0.05 <b<0.175. C-D. 1 < a < 3, b=0.05.
$, and Q, (p<0.01, Figure 9A), but do not indicate any linear relationship between J; and Q, (Figure 9B). Under even this condition, a negative correlation between S, and Q, and a positive correlation between /; and Q. were approached by a multiple regression analysis, which shows a significant trend for Q, to increase as a function of - $. and Js:
0,'=-0.5735,* + 0.128 1° (r=0.565, F=11.94, p<0.01),
where Q,*, S.° and Js* are standardized variables of Q,, S;, and Js, respectively. In this case, a negative relationship between S, and Q, is especially prominent. This relation- ship, concordant with that of the biometric analyses as shown in Figure 4B, D, strongly supports conspicuous posi-
Geometry of bivalve prismatic shell 43
tive correlation between C and P assumed as in the Eq. 6, particularly for simple prisms. As stated above, a positive correlation between C and P suggests that both prism size and its variation are mainly controlled either by the growth rate of the entire shell, or by the activity of mantle secretion. If the former is the case, simple prisms must tend to be uni- formly large as the entire shell grows faster. On the other hand, if the latter is the case, density of nucleation and growth rate of prisms both must decrease as the secretive activity of the mantle decreases, and a negative relationship is expected between size and the growth rate of simple prisms.
On the other hand, the results of computer simulations with various values of the coefficient a and the fixed coeffi- cient b (=0.05) show a positive correlation between /; and 0. (p<0.01, Figure 9D), but no significant relationship be- tween S, and ©, (Figure 9C). A multiple regression analysis for Q. on S. and /; provides a significant trend to increasing Q. as a function of $. and Js:
Q.'=-0.3935."+0.686 Is’ (r=0.582, F=9.99, p<0.01).
In this equation, a positive relationship between /s and Q, is more striking than a negative relationship between S., and Q.. This is concordant with the results of the biometric analyses in species with vertical composite prisms shown in Figure 4A, C. This fact supports the assumption that P has an inverse relationship with L as defined in Eq. 6, particularly in vertical composite prisms. This assumption implies that the number of nucleations per unit time interval is fixed to a constant value. In the species having a large nucleation zone, in which the value of L is large, maintenance of the probability of nucleation inevitably causes a large number of nuclei. If we assume an upper limit of the total number of nuclei per unit time interval, the probability of nucleation will vary inversely with the width of the nucleation zone. If the total number of nuclei per unit time interval is fixed, the den- sity of nucleation is expected to decrease as the total shell grows faster. In this case, the median size of prisms is con- trolled mainly by the growth rate of the entire shell. This fact suggests that, in the case of vertical composite prisms, prism size tends to increase as the entire shell grows faster.
Discussion and conclusion
Wada (1961) studied the size of crystals in a nacreous shell layer of Pinctada martensii in relation to the rate of cal- cium carbonate deposition. He demonstrated that when the rate of shell deposition is at a maximum, a large number of small crystals occurs on all the nacreous surfaces, while larger crystals occur as the rate of deposition decreases. In addition, Wada (1972, 1985) also reported an inverse relationship between them for such bivalves as Pinctada fucata, Pinna attenuata and Hyriopsis schlegeli. Wada as- sumed that the primary factor determining size of crystals was the degree of calcium carbonate concentration in the extrapallial fluid. He regarded the rate of calcium carbonate deposition as the rate of crystal growth, and thought that larger crystals tend to grow slowly at a low degree of supersaturation of the fluid, at which the frequency of nu- cleation diminishes. The results of the present study may
partly support those of Wada, since the negative correlation between size and growth rate of crystals is also expected in this study if the size of prisms is assumed to be controlled by the activity of mantle secretion.
Unlike prismatic structure, nacreous structure does not form the outermost shell layer in bivalves. In the nacreous layer, deposition corresponds to thickening of the shell or growth of crystals, rather than growth of the entire shell. Therefore, the size of crystals in the nacreous layer corre- lates with the rate of crystal growth rather than with the growth rate of the entire shell. On the contrary, the net growth rate of the entire shell, which reflects the growth rate of the soft parts, seems to be significant for the size of crys- tals which constitute the outermost shell layer.
Ubukata (1994) claimed that relatively rapid growth of prisms produces prisms prominently inclined to the outer shell surface, because of retardation of the initiation of their forward growth relative to the radial direction. Although such inclined prisms are commonly found in species belong- ing to Unionidae and Ostreidae (Ubukata, 1994), their prisms also characteristically fell in low (small size) and high ©. (irregular in size) regions in Figure 4A, B (refer to Table 1 for higher taxonomy). The present study suggests that such a pattern is produced under the condition of either rapid growth of prisms or slow growth of the entire shell. Ubukata (1994) demonstrated that the relative growth rate of prisms to that of the entire shell determines the orientation of elongation of the prisms.
This study suggests that the sizes of prisms and their size variability correlate with the growth rate of the shell in bi- valves. It is well known that internal microgrowth incre- ments often provide a high-resolution record of growth rate and physiological condition of a bivalve (Lutz and Rhoads, 1980), particularly in species with crossed lamellar and hori- zontal composite prismatic shells. In simple prismatic and vertical composite prismatic shells, however, it is hard to es- timate the growth rate of the shell from an analysis of inter- nal microgrowth increments, because periodic growth incre- ments tend to be obscure. The size-frequency distribution of prisms may be a promising index for estimating the growth rate of a fossil bivalve shell.
This study indirectly estimates the relationship between the size of prisms and the growth rate of the shell or prisms from computer simulations, but the direct observation of the growth rate of the shell or prisms remains to be performed. An experimental study on growth of a shell and/or crystals may be required for clarifying the relationship between the crystal size and the growth rate more clearly, and for obtain- ing a regression equation to predict the growth rate from prism size. The growth rate of the shell and/or crystals is clearly related to the physiological condition of a fossil or- ganism, and has the potential to in spire a research field of ‘paleophysiology’.
Acknowledgments
| thank Enrico Savazzi (Uppsala University), Kazushige Tanabe (University of Tokyo), Rihito Morita (Natural History Museum and Institute, Chiba) and Joseph G. Carter (University of North Carolina) for critical review of the manu-
script and for valuable comments. This study has been supported by Grant-in-Aid for Scientific Research (A) from the Japan Society for the Promotion of Science (No. 10304040).
References
Carter, J. G. and Clark, G. R. Il, 1985: Classification and phylogenetic significance of molluscan shell microstruc- ture. In, Broadhead, T. W. ed., Mollusks: Notes for a Short Course, p. 50-71. Department of Geological Sciences, Studies in Geology, 13. University of Tennessee, Knoxville.
Carter, J. G., Bandel, K., de Buffrénil, V., Carlson, S. J., Castanet, J., Crenshaw, M. A., Dalingwater, J. E., Francillon-Vieillot, H., Géraudie, J., Meunier, F. J., Mutvei, H., de Ricqlés, A., Sire, J. Y., Smith, A. B., Wendt, J., Williams, A. and Zylberberg, L., 1990: Glossary of skeletal biomineralization. In, Carter, J. G. ed., Skeletal Biomineralization. Patterns, Processes and Evolutionary Trends, vol. 1, p. 609-671. Van Nostrand, New York.
Grigor'ev, D. P., 1965: Ontogeny of Minerals, 250p. Israel Program for Scientific Translations Ltd., Jerusalem.
Hoel, P. G., 1976: Elementary Statistics, 4th ed., 361p. John Wiley and Sons, Inc., New York.
Lutz, R. A. and Rhoads, D. C., 1980: Growth patterns within the molluscan shell. An overview. In, Rhoads, D. C. and Lutz, R. A. eds., Skeletal Growth of Aquatic Organisms, p. 203-254. Prenum Press, New York.
Takao Ubukata
Morisita, M., 1959: Measuring of dispersion of individuals and analysis of the distributional patterns. Memoirs of the Faculty of Science, Kyushu University, Series E, vol. 2, p. 215-235.
Saleuddin, A. S. M. and Petit, H. P., 1983: The mode of forma- tion and the structure of the periostracum. In, Saleuddin, A. S. M. and Wilbur, K. M. eds., The Mollusca, vol. 4, Physiology, pt. 1, p. 199-234. Academic Press, New York.
Ubukata, T., 1994: Architectural constraints on the morphogenesis of prismatic structure in Bivalvia. Palaeontology, vol. 37, p. 241-261.
Ubukata, T., 1997a: Microscopic growth of bivalve shells and its computer simulation. The Veliger, vol. 40, p. 165-177.
Ubukata, T., 1997b: Mantle kinematics and formation of commarginal shell sculpture in Bivalvia. Paleontological Research, vol. 1, p. 132-143.
Ubukata, T., 2000: Theoretical morphology of composite pris- matic, fibrous prismatic and foliated shell microstructures in bivalve. Venus, vol. 59, p. 297-305.
Wada, K., 1961: Crystal growth of molluscan shell. Bulletin of the National Pearl Research Laboratory, vol. 7, p. 703- 828.
Wada, K., 1972: Nucleation and growth of aragonite crystals in the nacre of bivalve molluscs. Biomineralization, vol. 6, p. 141-159.
Wada, K., 1985: Growth of CaCO; crystals in bivalve shell mineralization. Journal of the Japanese Association for Crystal Growth, vol. 12, p. 57-70. (in Japanese with English abstract)
Paleontological Research, vol. 5, no. 1, pp. 45-54, April 30, 2001 © by the Palaeontological Society of Japan
Evolution in morphometric traits in North American Collignoniceratinae (Ammonoidea, Cephalopoda)
RICHARD A. REYMENT' and W. JAMES KENNEDY’
‘Naturhistoriska Riksmuseets Paleozoologiska Avdelning, Box 50007, 10405, Stockholm, Sweden *Geological Collections, Oxford University Museum of Natural History, Oxford OX1 3PW, UK
Received 29 August 2000; Revised manuscript accepted 14 November 2000
Abstract. Eight species and one subspecies of two genera of Collignoniceratinae, Collignoniceras and Prionocyclus, are analysed with respect to standard morphological distance measures and rib- frequencies, methods of multivariate statistical analysis, including canonical variate ordination, principal components and generalized distances. It was found that the biostratigraphicaily inferred evolutionary sequence, as currently perceived, is upheld in detail for the Collignoniceras data. The *nearest -neighbour’ relationships between the Prionocyclus part of the sequence is less complete, although links in main branches are supported. The other aspect given consideration, that of “gracile” and “robust” shells (based on visual inspection of the conch), yielded the result that the subjective assignation of shells to gross morphological type is largely, though not unequivocally, upheld by the statistical analysis. Illustrations of typical representatives of the species analysed are provided.
Key words: Ammonites, Collignoniceratinae, compositional analysis, Cretaceous of U.S.A., evolu-
tion, morphometrics
Introduction
Ammonite taxonomy is perforce largely based on external morphology— mode of coiling, tuberculation, ribbing and carination. The diagnostic significance of the suture line seems to have been recognized as being of secondary im- portance for many groups of Jurassic and Cretaceous am- monites (cf. Pérez Claros, 1999). Reyment and Kennedy (1998) and Reyment and Minaka (2000) recorded and de- scribed polymorphism in ornamental properties of Cretaceous ammonites of the genus Neogastroplites. However, it is by no means a trivial matter to distinguish be- tween true polymorphism, in the classical genetic sense of the property (Falconer, 1981, p. 42; Manly, 1985, p. 402; Roughgarden, 1979, p. 259), and the merging of ornamental types in response to ecophenotypic variation of the kind that seems to occur in the present material with respect to the categories referred to as being ‘gracile’ and ‘robust’. The functional significance of such differentiation, if any, remains obscure. A palaeobiological treatment of the problems in- volved, and methods for their analysis, are given in Reyment (1991), chapters 5, 6 and 8.
The primary aim of the present study is directed towards ascertaining to what degree the stratigraphically supported
phylogeny within a subfamily of ammonites arrived at by the traditional methods can be recognized by the quantitative analysis of external morphological characteristics (excluding sutures); that is, mainly distance measures on the shape of the shell, but also ribbing density. With such information available, it should become possible to extrapolate to other groups and to work towards stabilizing phylogenetic relation- ships on less subjective grounds than are yielded by purely descriptive procedures. As far as is known to us, there have been no studies devoted to ascertaining to what extent, if any, wholly distance-related variables are correlated with evolutionary status in ammonites (and, by extension, whether such differentiation can represent a relationship be- tween form and function). The work accounted for in this note shows, with reasonable clarity, that such is indeed the case, at least in the evolution of the Collignoniceratinae. There is, moreover, a more far-reaching consequence to our project. Gross morphology, such as is expressed in coiling and whorl shape, has not been attributed importance of the first order in many evolutionary and taxonomical studies known to us. As we demonstrate in this note, there is infor- mation of evolutionary significance in characters of this kind, but it can only be effectively extracted by means of appropri- ate quantitative procedures in a _ multidimensional
46
Richard A. Reyment and W. James Kennedy
perspective. The detailed classical taxonomy of the mate- rial considered here is scheduled to appear in a separate monograph (Kennedy et al. in press). In order to avoid eventual misunderstanding, we are not concerned with the cladistics/phenetics confrontation; our analysis is solely mul- tivariate morphometric in nature as defined by Blackith and Reyment (1971).
The material analysed
Members of the subfamily Collignoniceratinae that inhab- ited the Cretaceous Western Interior seaway of the United States first appeared in the Lower Turonian and ranged into the Upper Turonian. On a mondial scale, the subfamily arose in the uppermost Cenomanian and died out in the Coniacian. The analysis reported here is based on data being presented elsewhere in Kennedy ef al. (in press), to which monograph reference is made for details of taxonomy, provenance and stratigraphy. Owing to the scarcity of avail- able specimens of some of the species concerned, the phylogenetic study is somewhat less complete than we should have liked it to be. Hence, the conclusions put for- ward here are necessarily of a preliminary nature.
The earliest Collignoniceratinae of the preserved se- quence considered here is the monospecific genus Cibolaites Cobban and Hook 1983, of which neither its whence nor its whither seem to be known with any certainty. The genus is distinguished throughout most of its ontogeny by the presence of umbilical, ventrolateral and siphonal tu- bercles. Its mature body chamber is flat-sided, with pro- gressively weakening tubercles and ribs (Cobban and Hook, 1983, p. 16-18, pl. 2, figs. 1-9; pl. 3, figs 3-8; pl. 8, figs 6-8; pl. 13, figs 1-5; pl. 14, fig. 14).
Collignoniceras woollgari (Mantell, 1822) is believed to have descended from Cibolaites. It is thought to have given rise to several contemporaneous species in western Europe (Kennedy et a/., 1980). In the region under consideration, four successive species seem to have derived from it (pre- sumably via its subspecies C. w. regulare Haas, 1946, al- though the ensuing multivariate study leaves this undecided with respect to the actual route that may have been fol- lowed), of which one is considered in the present analysis, to wit, C. praecox Haas, 1946; the other three are not available in sufficient numbers for study. C. praecox differs from C. woollgari woollgari and C. woollgari regulare by the persis- tence of long and short ribs with ventrolateral tubercles out-
@ Figure 1.
Collignoniceratine ammonoid morphometrics
numbering the umbilical, and a near-continuous siphonal keel. As shown by the results of the morphoevolutionary analysis, the inferred biostratigraphical relationships be- tween successive species may be a simplification of the ac- tual evolutionary sequence of events.
The later Turonian history of the Collignoniceratinae in the U. S. Western Interior is considered to be marked by the evolution from Collignoniceras praecox of species of Prionocyclus. Juveniles of species of the two genera can be distinguished in that the keel of Collignoniceras has siphonal clavi equal in number to the ventrolateral tubercles, whereas in Prionocyclus Meek, 1872, the serrations outnum- ber the ribs. It is significant that Prionocyclus hyatti Stanton, 1894, an early representative of the genus, is morphometrically closely allied to Collignoniceras woollgari regulare and somewhat less so with its putative ancestor, C. praecox. In general terms, there is semiquantitatively mani- fested intraspecific variation in the strength of the ornament of most of the species of the Western Interior Collignoniceratinae, a ‘gracile’ category and a ‘robust’ one. These are not discrete morphological categories such as are recorded by Reyment (1971) for the genus Benueites. Testing the soundness of this interpretation of morphological variability forms an integral part of the analysis presented in the following. Figures 1-3 provide illustrations of typical representatives of the species considered, including exam- ples of shells determined as being robust or gracile.
Methods
Photographs of the specimens passing muster for statisti- cal study were scanned and the coordinates of seven sites (Figure 4) considered diagnostic were recorded, using the digitization program TpsDig written by F. James Rohlf (SUNY at Stony Brook, USA). Only complete specimens were selected for analysis; the numbers of each of the sam- ples of the species are recorded in Table 1. The required distances were computed from these coordinates by simple geometry. This is freely admitted to be an arbitrary manoeuvre and we have desisted from confusing the issue by not calling the coordinate pairs thus obtained “landmarks”. In any event, they cannot be equated to the landmarks used by Johnston et al. (1991) in their analysis of spiral growth in gastropods. The approach utilized in that study is clearly one that should prove eminently useful in fu- ture studies of growth and shape-variation in ammonites, not
a, b. Cibolaites molenaari Cobban and Hook, 1983. USNM 498205, a robust form from USGS Mesozoic locality D8429, sec.
1 and NE”. sec. 12, T. 4N., R. 19W., Cibola County, New Mexico. Mancos Shale, from limestone concretions 24-30 m below the top of the Rio Salado Tongue. c, d. Collignoniceras woollgari woollgari (Mantell, 1822). USNM 356903, a gracile form from USGS Mesozoic locality D 10243, E'/, NE’/. sec. 9, T. 5S., R. 2E., Socorro County, New Mexico, Rio Salado Tongue of the Mancos Shale. e-j. Collignoniceras woollgari regulare (Haas, 1946). e-g, USNM 498237, a gracile form from USGS Mesozoic locality 21792, west of Newcastle, in the NE’/, sec. 31, T. 45N, R. 61W., Weston County, Wyoming, limestone concretions in the Carlile Shale, 18.3 m below the base of the Turner Sandy Member. h-j, USNM 498244, a robust form from USGS Mesozoic locality D9896, NE’/. sec. 35, T. 46M, R. 63W., Weston County Wyoming, limestone concretions in the Carlile Shale, 18.3 m below the base of the Turner Sandy Member. k-n. Collignoniceras praecox (Haas, 1946). k, I, USNM 498272, a gracile form from USGS Mesozoic locality D 13832, sec. 35, T. 8S., R. 1E., Fall River County, South Dakota, from lime- stone concretions in the lower part of the Carlile Shale. m,n, USNM 498266, a robust variant from USGS Mesozoic locality D 10697, S E'/. SE’ sec. 18, T. 9S., R. 2E., Fall River County, South Dakota, limestone concretions in the Carlile Shale 6 m below the base of the Turner Sandy Member. USGS: United States Geological Survey; USNM: U.S. National Museum of Natural History, Washington D.C. All figures are x 0.9.
48
Richard A. Reyment and W. James Kennedy
Collignoniceratine ammonoid morphometrics
Figure 3. Prionocyclus wyomingensis Meek, 1876. Paralectotype, USNM 7729, a gracile form from the Wall Creek Member of the Frontier Formation near Medicine Bow, Carbon County, Wyoming. USNM: National Museum of Natural History, Washington D.C. Natural size.
least because of the finding that mean-forms with similar shapes at the same arbitrary growth increment may have achieved that shape in different ways. The resulting data, suitably corrected for slight differences in magnification from specimen to specimen, were studied by standard multivari- ate analyses made on the distance measures between adja- cent sites and the maximum breadth of the last whorl. The
factor of ornamental complexity was introduced into the analysis by appending rib counts to the data matrix. This lat- ter addition accounts adequately for the morphological cate- gories gracile/robust.
The multivariate statistical methods most useful for our study were found to be (1) principal component analysis of the distances (six in all) and the breadth, maximum directly
@ Figure 2. a-d. Prionocyclus hyatti (Stanton, 1894). a, b, USNM 498308, a robust form from USGS Mesozoic locality D3884, Arroyo Lopez, 1 km. north of Holy Ghost Spring, Sandoval County, New Mexico, Mancos Shale, from lower part of Semilla Sandstone Member. c, d, USNM 498323, from USGS Mesozoic locality D 11208, NE'/. sec. 36, T. 6N., R. 19W., Cibola County, New Mexico, from the D Cross Member of the Mancos Shale. e-h, Prionocyclus macombi Meek, 1876. e, f, USNM 498341, a robust form from USGS Mesozoic locality D4395, north of Rio Gallina in SE'/ sec. 15, T. 20N., R. 1E., Rio Arriba County, New Mexico, Mancos Shale, from the base of the Juana Lopez Member. g, h, USNM 498348, a gracile form from the same locality and horizons as c, f. i-m. Prionocyclus novimexicanus (Marcou, 1858). i-k, USNM 498434, a gracile form, from USGS Mesozoic locality D9833, NW'/, NW’. Sec. 33, T. 40N., R. 82W., Natrona County, Wyoming, from the second ledge forming sandstone below the top of the Frontier Formation. 1, m, USNM 498446, a robust form from USGS Mesozoic locality D6928, NE’ SE'/. sec. 31, T. 22N., R. 75W., Albany County, Wyoming, from the Wall Creek Sandstone Member of the Frontier Formation. n-q, Prionocyclus germari (Reuss, 1845). n, 0, USNM 498458, a gracile form from USGS Mesozoic locality D9118, NW’. NE’ sec. 4, T. 33N., R. 81W., Natrona County, Wyoming, from the uppermost sandstone of the Wall Creek Member of the Frontier Formation. p, q, USNM 498483, from the same horizon and locality as n, 0. USGS: United States Geological Survey; USNM: National Museum of Natural History, Washington D.C. All figures are x 1.
50 Richard A. Reyment and W. James Kennedy
Table 1. The biostratigraphical sequence for the species in- cluded in the analysis and sample sizes.The asterisks denote spe- cies currently used as zonal indices in the Turonian sequence of the Western Interior USA.
Substage Species N
Prionocyclus germari” 24
UPPER Prionocyclus novimexicanus 10
TURONIAN Prionocyclus wyomingensis 6 Prionocyclus macombi* 16 Prionocyclus hyatti’ 19
MIDDLE } : . Collignoniceras praecox 12
TURONIAN : Ë i Collignoniceras woollgari 26
LATE
LOWER Cibolaites molenaari” 7
TURONIAN
Figure 4. Locations of the points observed on the shells superimposed on a gracile specimen of Prionocyclus macombi Meek, 1876.
observed diameter and ribbing density, (2) principal coordi- nate analysis of the coordinate data, and (3) canonical variate analysis of the distance measures in “ordination mode”. Unless otherwise stated, the measures were loga- rithmically transformed. The use of this transformation tends to stabilize the multivariate distribution of the data and to minimize the effects of size differences. Accounts of these procedures applied in like situations are to be found, for example, in Reyment (1991) and Reyment and Savazzi (1999), including examples of the computational procedures applied here and exemplified by ammonite data. Additionally, comparisons between species were made by standard procedures of generalized statistical distance analysis (for the larger samples), backed up by one-way univariate analysis of variance.
The links indicated by the superimposed minimum span-
ning tree on the plots are certainly not infallible statements of fact. For this reason, relationships between species have been further illuminated by means of pair-wise contrasts be- tween latent vectors (principal components) of the logarith- mic covariance matrices of the individual samples. Almost all multivariate “variability” in samples is located to the first latent root. For the most part, the smallest latent roots are almost zero and hence the associated latent vector may be considered as representing an almost invariant linear rela- tionship between variables, as was suggested by Gower (1967). A somewhat similar approach to the study of varia- tion in the shape of ammonite shells is outlined in Neige and Dommergues (1995).
Findings
Ordination of canonical variate means
A useful morphometric tool is often provided by the mini- mum spanning tree superimposed on the plot of the canoni- cal variate means (Reyment and Savazzi, 1999). This gives the standard ordination of multivariate means a nearest- neighbour categorization. A partial analogy with stars in the firmament serves to illustrate this—stars that to the unaided eye seem to be close to each other can actually lie at ex- tremely great distances from each other in the third dimen- sion. The analyses accounted for in this section are designed to see how well morphometric and morphological characters can reproduce the inferred phylogeny of the lineage.
The first set of means was obtained from the canonical variate extraction in covariance space of the logarithms of the distances between seven sites and nine samples, aug- mented by directly measured maximum breadth of the shell and ribbing frequency counts in relation to directly measured diameter. How well then is the inferred evolutionary suc- cession based on traditional procedure mirrored in the coil- ing and ornamental properties of the species of the lineage? In Figure 5, the canonical variate means are seen to divide into two branches. The sequence shown in Figure 5 indi- cates implied relationships with respect to form within the collignoniceratinid lineage. The minimum spanning tree for the means is superimposed on both plots. It is seen that C. molenaari is markedly distant from all other species, as is re- flected in its being a monospecific genus established by traditional criteria. This observation accords with the sup- position that the morphological passage from Cibolaites to Collignoniceras was abrupt (‘punctuated’ as it were). The ordinated linkage order goes from C. w. woollgari to C. w. regulare to P. hyatti, on one branch, and from C. w. regulare to C. praecox to P. macombi Meek, 1876, to P. wyomingensis Meek, 1876, on the second branch. However, we shall see in the following that in some relation- ships, P. hyatti tends to behave as though it were a Collignoniceras, a peculiarity it does not share with the other representatives of the genus included in our study, and which may reflect its middle-of-the-road evolutionary and morphological status between Collignoniceras and Prionocyclus.
A minimum spanning tree cannot be expected to provide a mirror image of an inferred phylogenetic sequence. What
hyatti
u: novimexicanus germari
wyomingensis
regulare
praecox
SECOND CANONICAL MEANS AXIS
-1.9 -1.2 -0.5 0.2 0.9
FIRST CANONICAL MEANS AXIS
Figure 5. Minimum spanning tree superimposed on the plot of the first two canonical variate means for 7 distance measures (6 coordinate-based and maximum breadth of the conch), one fre- quency (ribbing) and 9 groups (the species and subspecies). Analysis made on the covariances of the logarithmically transformed data.
it is designed to seek is morphometric nearness relation- ships free of any obligatory time constraint. It is therefore interesting to be able to record that the multivariate analytically obtained results reflect rather well the phylogenetic indications based in part on biostratigraphical context. Many ofthe samples are small and it is encourag- ing that the results yielded by this analysis are so clearly manifested. In general, it seems clear that the observed stratigraphical ordering of the species is upheld, grosso modo, by the morphometric ordination. We note that, ex- cluding the small samples (to wit, woollgari woollgari, praecox, wyomingensis), all generalized statistical distances between samples are highly significantly different.
In order to test the reliability of results obtained for such small samples, an alternative canonical variate model was used (Reyment and Savazzi, 1999). Here, all values were reduced to standard size by division of them by the maxi- mum diameter of the shell. This standardization has a sub- sidiary effect, notably, that it transforms the data into compositions, and, consequently, the variables to parts. (N. B. it is fairly common practice in ammonite “biometry” to make scatter plots of ratios and then to look for meaningful constellations of points—statistically, this is not sound proce- dure.) This change in statistical properties necessitates a methodology appropriate to (constrained or “closed”) sim- plex space (Aitchison, 1986). The multivariate analysis was therefore made on the log-ratio covariances (which involves the loss of one dimension due to the division by one of them). The ordination illustrated in Figure 6 differs in sev- eral respects from the foregoing (full-space) version in that
Collignoniceratine ammonoid morphometrics
P. germari (Reuss, 1845) is quite out of place in relation to the evolutionary succession. The morphoevolutionary pas- sage from C. molenaari through the two subspecies of Co. woollgari is, however, maintained. We note that P. hyatti is now relegated to a branch of its own. Sidestepping the anomalously located P. germari, there is a branch that runs from C. woollgari s. |. through C. praecox, P. macombi, P. wyomingensis and P. novimexicanus (Marcou, 1858), the latter being registered as a branchoff from P. macombi. Apart from the anomalous location of P. germari, the result illustrated in Figure 6 is in many respects more closely com- patible with the conventionally established succession of species than Figure 5. The main point to be heeded is, however, that the linkages are essentially the same in both representations, notably the Cibolaites-Collignoniceras se- quence and the C. praecox-P. macombi-P. wyomingensis passage.
Results obtained by the principal component analysis of the distances
For the purposes of this part of the analysis, two latent vectors turned out to be of special interest. The latent root attached to the first latent vector is overwhelmingly large in relation to the other roots. The smallest latent root is virtu- ally zero and hence may be considered to be an expression of an almost invariant linear relationship between the vari- ables (Gower, 1967). Such an invariant relationship is of special taxonomic and hence phylogenetic interest because it represents a linear combination that is intrinsically bound to the form under consideration. The species studied in this connexion are: C. woollgari woollgari and subspecies C. w. regulare, C. praecox, P. hyatti, P. novimexicanus, P. macombi and P. germari. We wish to make it clear that the method of comparison of angles (Blackith and Reyment, 1971) is not a statistical technique, being rather a procedure appropriate to ad hoc data-analysis. Nonetheless, it has proven itself to be useful in many taxonometric studies.
The first principal component
This section is concerned with examining how morphometrically divergent succeeding species in the evolu- tionary succession are from each other. The angles be- tween pairs of first latent vectors and sixth latent vectors, respectively, are listed in Table 2. The angles for the first latent vector for the comparisons of C. molenaari and C. w. woollgari, respectively C. w. regulare are almost identical (21.33°, respectively, 21.88°). This interesting result indi- cates not only that the ancestral species is separated by a strongly expressed morphological jump from its descen- dants, but also that the angle between ancestor and descen- dant subspecies is identical and that the two subspecies of C. woollgari are morphologically close to each other. This observation is supported by the small angle between these two subspecies, namely, 2.21°. The species next in the stratigraphical hierarchy is C. praecox, the angle between its first latent vector and both subspecies of C. woollgari being the same, to wit, 9.56°. The angle between C. praecox and P. hyattiis quite small, being 4.82°, whereas that between P. hyatti and P. macombi is 6.24°. The next passage from P. macombi to P. wyomingensis is connected with a small an-
52 Richard A. Reyment and W. James Kennedy
Table 2. Angles between first and sixth latent vectors (covari- ance matrix of logarithmically transformed variables) for critical pairings.The asterisk denotes comparisons made on very small sample sizes.Co=Collignoniceras, Ci=Cibolaites, P.=Prionocyclus.
a Angle for Angle for Comparison between
vector | vector VI Ci. molenaari/Co. woollgari woollgari” 21.33 87.80 Ci. molenaari/Co. woollgari regulare" 22.88 38.75 Co. woollgari woollgari/Co. praecox* 9.56 81.53 Co. woollgari woollgari/Co.regulare 2.21 67.31 Co. woollgari regulare/Co. praecox 9.56 59.54 Co. woollgari regulare/P. hyatti 6.71 20.53 Co. w. regulare/P. germari 3.29 48.23 P. hyatti/P. germari 3.58 33.92 Co. praecox/P. hyatti 4.82 90.00 Co. praecox/P. macombi 5.45 54.06 P. macombi/P. wyomingensis* 3.18 88.10 P. novimexicanus/P. germari 773 46.06
gular difference, notably, 3.18°. There is a notable leap in the angle between the first latent vector of P. wyomingensis and its descendant P. novimexicanus, namely 11.24°. The final link, from P. novimexicanus to P. germari is accompa- nied by an angular divergence of 7.73°. It is significant that the angular divergences between latent vectors of the Prionocyclus sequence are mostly of the same order of magnitude.
Comparisons made for the ‘smallest’ latent vectors are all noticeably greater. This could conceivably indicate that in- trinsic morphometric specific differences are to be found in the (almost) invariant principal component, granted that the first latent vectors, which are all connected to more than 98% of the total variability, are so similar. The more inter- esting of these relationships are:
1. Cibolaites molenaari-Collignoniceras woollgari subspe- cies. The angles for vector | are here almost identical, as reported above. The values for vector VI differ strongly, with the angle between the vectors for molenaari-C. woollgari woollgari being almost a right angle, whereas that for the comparison with regulare is a relatively low 38.75°.
2. Collignoniceras praecox-C. woollgari subspecies. The angular differences for vector VI are large, being essentially greater for the comparison with C. woollgari woollgari. The angular differences for the comparisons between first latent vectors are small, and smallest for the latter subspecies.
3. C. praecox-P. hyatti. The result yielded here is note- worthy. The first latent vectors are almost collinear, which hardly accords with the linkage in Figure 6 yielded by the minimum spanning tree. However, the value for vector VI places these vectors exactly at right angles to each other. We are inclined to accept this result as indicating support for the view that the invariant latent vector of distances harbours taxonomically relevant information.
4. The plot in Figure 6 shows C. w. regulare and P. germari to be located near to each other, but the linkage se- quence does not indicate these two taxa to be close. Comparison of the smallest latent vectors yields a relatively
woollgari
hyatti a ” 3 0.4 TR regulare germari S molinaari © = @ © 5 :0.4 praecox Phe = novimexicanus iS) T [= 9 ® oO -1.2 macombi
wyomingensis
-2.5 -1.7 -0.9 -0.1 0.7
First canonical means axis (constrained)
Figure 6. Minimum spanning tree superimposed on the plot of the first two canonical variate means for the same set of data as in Figure 5, but standardized with respect to size (hence in compo- sitional mode). Analysis made on the log-ratio covariances (cf. Aitchison, 1986).
large angle, namely 48.23°.
5. C. w. regulare-P. hyatti. The biostratigraphical scheme leads one to expect that C. praecox would be more like P. hyattithan C. w. regulare, considering that it has been put forward as the logical ancestor of the former, and this is indeed implied by the angle between the first latent vectors, being among the smallest of the entire set of values reported in Table 2 and, in effect, denoting collinearity. However, the angle between vector VI for these two taxa is by far the smallest of all and it seems that it is this relationship that is reflected in the pattern obtained for the minimum spanning tree (Figure 5).
6. Anomalously large angular differences occur for the pairings macombi-wyomingensis and wyomingensis- novi- mexicanus, probably due to the small sample sizes involved. We note that the sample size for P. wyomingensis is be- neath the level for a resolvable covariance matrix (i.e. more variables than specimens).
Robust vs gracile shells
The routine taxonomic appraisal of the material (Kennedy, 1988) suggested that there are two classes of ornamental categories in the species of the genera considered here, to wit, robustly ornamented shells and finely ornamented shells (the gracile shell-type). These do not appear to be morphs of the kind described by Reyment and Kennedy (1998). Only three species occur in sufficient numbers to permit sta- tistical appraisal of these categories- P. hyatti, P. macombi and P. germari. The same suite of methods as were used for the full set of observations were applied to the reduced set, noting that there are now 6 groups for 6 distance measures. In Figure 7, the canonical variate means for the two ornamental categories for these three species are plot-
hyatti-R
1.5
1 macombi - G
macombi -R
SECOND CANONICAL AXIS FOR MEANS
-0.5 0 0.5 1 FIRST CANONICAL AXIS FOR MEANS
Figure 7. Minimum spanning tree superimposed on the ca- nonical variate means for robust and gracile categories of three spe- cies of Prionocyclus, based on 6 distance measures and 6 groups (three “robust” classifications and 3 “fragile” classifications).
Table 3. Angles between first and sixth latent vectors for the robust/gracile data for critical pairings of three species of Prionocyclus
2 Angle for Angle for
Comparison between vector | vector VI
germari G/germari R 1.99 41.07 macombi G/germari R 1.87 85.40 germari G/hyatti G 2.21 55.62 germari G/macombi G 3.35 7.91 hyatti G/macombi G 5.22 56.69
ted and linked by the superimposed minimum spanning tree. The most notable feature of the plot is that robust and gracile shell-types for each species, respectively, are linked, and that their means do not coincide, or almost coincide in loca- tion. Hence, differences in multivariate means of the sub- jectively identified categories are tangible, but not so great as to confuse taxonomic issues. In Figure 7, we show the plot of the individual points in the plane of the first two ca- nonical variate axes (i.e. the canonical variate scores on the first two axes). The points for the six categories group natu- rally with gracile and robust forms for each species being nearest neighbours. Links between species are the union of robust germari to robust macombi and robust germari to gracile hyatti.
Comparisons of the first and sixth principal components lead to interesting consequences (Table 3). All angles be- tween various comparisons of first latent vectors of the loga- rithmic covariance matrix are low. Note particularly the values for germari and macombi, 1.99° and 1.87° respec-
Collignoniceratine ammonoid morphometrics
tively. The angles for the sixth latent vectors are large and generally of the same order of magnitude as found for the complete data-set, but with one striking exception, notably, the comparison between gracile individuals of germari and macombi, 7.91°. This would seem to point to a close morphometrical relationship between the two species, which is possibly of taxonomical significance.
Doubtlessly, the gracile/robust subsamples yield a more structured multivariate statistical analysis than do the full samples. Granted that the taxonomic integrity of the subsamples was not infringed upon by the partitioning, it may be suggested that the qualitatively arrived at subdivi- sion into two morphological types is largely justifiable.
The question now arises as to how morphometrically simi- lar are the two categories robust and gracile? The material is not sufficiently comprehensive to allow categoric asser- tions. However, the data for P. germari may be taken to serve as a guideline. Robust germari links to robust macombi, thus implying a certain degree of ornamental ho- mogeneity, whereas robust germari links to gracile hyatti. Figure 7 intimates also that germari and macombi are morphometrically more alike than either is with hyatti This observation adds further evidence with respect to the transi- tional evolutionary status of Prionocyclus hyatti, which in many features displays morphometrical properties of Collignoniceras type.
Conclusions
The multivariate-morphometrical analysis of a sequence of Collignoniceratinae from the Turonian of the U. S. Western Interior has brought to light several features of gen- eral interest for the study of evolution in ammonite morphometry with respect to the property of coiling. This result is somewhat unexpected, granted that coiling in shell- bearing cephalopods is under the rigid constraint imposed by the biomechanism determining logarithmic growth. These are:
1. The species of Collignoniceras align in accordance with biostratigraphical observations, thus preserving generic in- tegrity and evolutionary status. The passage from Cibolaites to Collignoniceras is abrupt, a saltation as it were.
2. The passage to Prionocyclus is slightly less well mani- fested in that there is a dichotomy in the relationship implied by the minimum spanning tree with P.hyatti occupying, jointly with respect to linkage (but naturally not time), an evo- lutionary position on a par with C. praecox. In this respect, the former species behaves in some cases as though it were a Collignoniceras. The dichotomous impression is further strengthened by the passage of praecox to macombi which in turn, links to wyomingensis (in agreement with the biostratigraphical inference—cf. Table 1).
3. The second branch in Figure 5 lets hyatti link directly to germari and novimexicanus. In Figure 6, hyatti is located on a separate branch. This result accords with the transi- tional status of this species, as indicated by the results of the present study.
4. The qualitative observations leading to the recognition of robust and gracile shell forms in all species can be upheld in part, as far as can be judged from those samples compris-
54 Richard A. Reyment and W. James Kennedy
ing sufficient material for statistical calculations. Also here, the ambiguous evolutionary status of P. hyatti is manifested. The possible functional significance of the two shell types would seem to be worth detailed study, possibly in relation to foraging aspects (Reyment, 1988).
Acknowledgments
We thank W. A. Cobban (U. S. Geological Survey, Denver) and N. L. Landman (American Museum of Natural History, New York, N. Y.) for advice and criticism. The pho- tographs were made by R. G. Burkholder, formerly of the U. S. Geological Survey, Denver. Kennedy acknowledges the technical assistance of the Department of Earth Sciences, Oxford and the Geological Collections, Oxford University Museum of Natural History. Reyment expresses his thanks to the Royal Swedish Academy of Sciences for support.
References
Aitchison, J. 1986: The Statistical Analysis of Compositional Data, xv+416 p. Chapman and Hall, London.
Blackith, R. E. and Reyment, R. A., 1971: Multivariate Morphometrics, xiv+412 p. Academic Press, London. Cobban, W. A. and Hook, S. C., 1983: Mid-Cretaceous (Turonian) ammonite fauna from Fence Lake area, west- central New Mexico. Memoirs of the Institute of Mining
and Technology, New Mexico, vol. 37, p. 1-50.
Falconer, D. S. 1981: Introduction to Quantitative Genetics, Second Edition, ix+340 p. Longman Inc., New York.
Gower, J. C., 1967: Multivariate analysis and multidimensional geometry. The Statistician, vol. 17, p. 13-28.
Haas, O., 1946: Intraspecific variation in, and ontogeny of, Prionotropis woollgari and Prionotropis wyomingensis. Bulletin of the American Museum of Natural History, vol. 86, p. 141-224.
Johnston, M. R., Tabachnik, R. E. and Bookstein, F. L. 1991: Landmark-based morphometrics of spiral accretionary growth. Paleobiology, vol. 17, p. 19-36.
Kennedy, W. J., 1988: Late Cenomanian and Turonian am- monite faunas from north-east and central Texas. Special Papers in Palaeontology, vol. 39, 131 pp.
Kennedy, J. W., Wright, C. W. and Hancock, J. M. 1980: Collignoniceratid ammonites from the mid-Turonian of northern France. Palaeontology, vol. 23, p. 62-77.
Kennedy, W. J., Cobban, W. A. and Landman, N. H., in press: A revision of the Turonian members of the ammonite sub- family Collignoniceratinae from the United States Western Interior and Gulf coast. Bulletin of the American Museum of Natural History.
Manly, B. F. J. 1985: The Statistics of Natural Selection, xvi+ 484 p. Chapman and Hall, London and New York.
Mantell, G. A., 1822: The Fossils of the South Downs; or illus- trations of the Geology of Sussex, 327 p. Lupton Relfe, London.
Marcou, J., 1858: Geology of North America; with Two Reports on the Prairies of Arkansas and Texas, the Rocky Mountains of New Mexico and the Sierra Nevada of California, 144 p. Zurcher and Furrer, Zurich.
Meek, F. B., 1876: A report on the Cretaceous and Tertiary in- vertebrate fossils of the upper Missouri country. In, Hayden, F. V., Report of the United States Geological Survey on the Territories, vol. 9, Ixiv+629 p.
Neige, P. amd Dommergues, J-L. 1995: Morphometrics and phenetic versus cladistic analysis of the early Harpoceratinae (Pliensbachian ammonites), Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen, vol. 197, no. 3, p. 411-438.
Pérez Claros, J. A. 1999: Analisis morfométrico de la complejidad sutural en ammonites del Juräsico superior en relacion a las caracteristicas del fragmocono y el paleoambiente, 508 p. Tésis Doctoral, Universidad de Malaga, Malaga.
Reuss, A. E., 1845: Die Versteinerungen der böhmischen Kreideformation, 58 p. Schweizerbart‘sche Verlagsbuch- handlung, Stuttgart.
Reyment, R. A. 1971: Vermuteter Dimorphismus bei der Ammonitengattung Benueites. Bulletin of the Geological Institutions of the University of Uppsala, N. S. vol. 3, p. 1-18.
Reyment, R. A. 1988: A foraging model for shelled cephalo- pods. /n, Wiedmann, J. and Kullmann, J. eds., Cephalo- poda-Present and Past, Schweizerbart’sche Verlagsbuchhandlung, Stuttgart, p. 687-703.
Reyment, R. A. 1991: Multidimensional Palaeobiology, ix+414 p. Pergamon Press, Oxford.
Reyment, R. A. and Kennedy, W. J., 1998: Taxonomical rec- ognition of species of Neogastroplites (Ammonoidea, Cenomanian) by geometric morphometric methods. Cretaceous Research, vol. 19, p. 24-42.
Reyment, R. A. and Minaka, N. 2000: A note on reiterated phenotypes in a species of Neogastroplites (Ammo- noidea, Cenomanian, Cretaceous). Cretaceous Re- search, vol. 21, p. 173-175.
Reyment, R. A. and Savazzi, E., 1999: Aspects of Multivariate Statistical Analysis in Geology, 285 p+Compact Disk. Elsevier, Amsterdam.
Roughgarden, J. 1979: Theory of Population Genetics and Evolutionary Ecology: an Introduction, x+634 p. Macmillan, New York.
Stanton, T. W. 1894: The Colorado formation and its inverte- brate fauna. Bulletin of the U. S. Geological Survey, vol. 106, p. 1-288.
Paleontological Research, vol. 5, no. 1, pp. 55-73, April 30, 2001 © by the Palaeontological Society of Japan
A review of symbiosis in the Bivalvia, with special attention to macrosymbiosis
ENRICO SAVAZZI
Hagelgränd 8, S-75646 Uppsala, Sweden. (e-mail: enrico.savazzi@usa.net)
Received 6 November 2000; Revised manuscript accepted 22 January 2001
Abstract. The symbiosis (defined as a strict interspecific association) between bivalves and other organisms is examined. Microsymbiosis (i.e., symbiosis with microorganisms) is frequent among Recent bivalves, and has been proposed to explain the unusual characters of several fossil bi- valves. However, a critical review of the morphological criteria used to infer microsymbiosis in fossil bivalves shows that their application is likely to result in a large number of false positive and false negative results. Symbiosis with macroscopic organisms (i.e., macrosymbiosis), on the other hand, has a better chance of being recognised correctly in fossils, although direct preserva- tion of the associated organisms remains the only completeiy safe criterion. Recent and fossil in- stances of macrosymbiosis are reviewed, and new evidence is presented to clarify the adaptive significance of some of these associations.
Key words: Bivalvia, chemosymbiosis, commensalisms, functional morphology, Mollusca, parasitism, photosymbiosis, Porifera, Scleractinia, symbiosis
Lingulidae,
Introduction
Symbiosis was originally defined as a strict, usually obligatory association of individuals belonging to two or more species (Bary, 1879). This definition does not imply that the association is mutually advantageous to the organ- isms involved. Only subsequently was the term used with the latter meaning. The present paper follows the original definition of symbiosis, since no other concise term is avail- able to characterise an interspecific association without con- notations of usefulness to the involved organisms. Use of the terms commensalism and parasitism, for instance, re- quires that the life habits and the advantages and/or disad- vantages to the species involved are known.
For the purposes of this paper, two categories of symbio- sis can be recognised among bivalves. The first involves microscopic endosymbionts living in a bivalve host. This category can be characterised as microsymbiosis. The second category involves macroscopic organisms associ- ated with bivalves, and can be called macrosymbiosis. This paper concentrates on the latter category, but a discussion of microsymbiosis is useful as an introduction, because this subject has received considerable attention by palaeobiolo- gists (see references below). In addition, this theme illus- trates several of the problems that characterise the recognition of macrosymbiosis in the fossil record.
Repositories. — Unless indicated otherwise in the figure captions, illustrated specimens are in the possession of the
author. Microsymbiosis
Photosymbiosis
Most of the literature on symbiosis in bivalves deals with microscopic endosymbionts hosted by bivalves. In particu- lar, photosymbiotic zooxanthellae and the associated adap- tations of the hosts have been studied in the Recent Tridacnidae (Yonge, 1936; Purchon, 1955; Stasek, 1961, and references therein) and Cardiidae (Kawaguti, 1950, 1968, 1983; Hartman and Pratt, 1976; Jacobs and Jones, 1989; Jones and Jacobs, 1992). The host bivalves show a range of cytological adaptations to the symbionts, as well as microstructural and macroscopic adaptations in shell mor- phology (e.g., see above references, and Seilacher, 1972, 1973, 1990).
Among these bivalves, the Tridacnidae build extremely large and thick shells thanks to photosymbiosis, and their ventral commissure (uppermost in the life position) is modi- fied to maximise exposure of the mantle tissues to sunlight. In some species, sculpture on the external shell surface fur- ther increases the mantle area exposed to light (above refer- ences).
The cardiid Corculum has an antero-posteriorly flattened, semitransparent shell to optimise the exposure of the mantle to sunlfght passing through the shell. The posterior shell slope possesses numerous semitransparent windows, which
56 Enrico Savazzi
result from a peculiar type of shell pigmentation rather than from a specialised microstructure (Watson and Signor, 1986).
Other Recent bivalves are known to possess photosyn- thetic endosymbionts. Among them are the freshwater unionid Anodonta (Goetsch and Scheuring, 1926) and the trapeziid Fluviolanatus subtorta (Morton, 1982). The signifi- cance of these associations has not been studied in detail. Other bivalves are frequently infected by microscopic algae (e.g., the Recent pectinid Placopecten magellanicus; Naidu and South, 1970; Naidu, 1971). Although these can be characterised as instances of parasitism by the alga, they are interesting in that they constitute a possible evolutionary stepping-stone toward a mutually advantageous situation.
The adaptive significance of photosymbiosis is not uniform among bivalves. In the Tridacnidae, photosymbionts are an important food-source for the host, and they allow the con- struction of very large and heavy shells (above references). In all other studied instances, however, the bivalves are rela- tively small, and their shells thin or only moderately thick. In at least part of these cases, the spectrum of sunlight ap- pears to be selectively filtered by the mantle tissues of the host in order to fine-tune the metabolic products of the photosymbionts (above references).
A few palaeontologists (e.g., Kriz, 1979; Yancey, 1982; Yancey and Boyd, 1983; Seilacher, 1990) have proposed photosymbiosis in a broad range of fossil bivalves, using morphological convergence with Tridacna and/or Corculum to support their theses. These fossil bivalves range in age from the Palaeozoic to the Caenozoic, belong to several superfamilies, and possess extremely large and/or thick shells, or antero-posteriorly flattened and presumably trans- lucent Corculum-like shells.
The reliability of these morphologic criteria to infer photosymbiosis in fossil bivalves, however, is questionable. With the exception of Corculum and the Tridacnidae, Recent photosymbiotic bivalves show little or no morphological specialisation of the shell to photosymbiosis. In addition, while photosymbiosis in Tridacna is directly related to in- creased shell secretion and affects its stable-isotope compo- sition (above references), such phenomena are absent in other Recent photosymbiotic bivalves (Jones and Jacobs, 1992). Nonetheless, stable-isotope analysis can be used in fossils when one desires to test whether shell secretion was aided by photosymbiosis, in a manner convergent to the Tridacnidae. Such an analysis led Jones, Williams and Spero (1988) to exclude photosymbiosis (or at least its in- volvement in shell secretion) in the Pliocene Mercenaria “tridacnoides’. This bivalve is a morph of M. campechien- sis, and differs from the latter in a shell with a wavy or zigzag ventral commissure. Seilacher (1990) had earlier proposed photosymbiosis in this form, based on shell morphology.
Finally, one may test the reliability of the above morphol- ogic criteria by applying them to Recent bivalves. There are several living bivalves which have at least one of the mor- phologic characters mentioned above. Among these are large and thick-shelled Ostreidae, Spondylidae, Pectinidae and Arcidae. However, none of these bivalves are known to host photosymbionts.
A flattened and translucent shell in an epifaunal bivalve is
not a reliable indication of photosymbiosis, either. A very good example is the Recent anomiid (or placunid) Placuna placenta. This species possesses a laterally flattened, thin and very translucent shell (Yonge, 1977, regarded this spe- cies as the most flattened bivalve), and is a recliner on the surface of soft sediments in shallow water. Incidentally, the shell of this species was commonly used by the human population of the Philippines for the construction of house windows, before glass became broadly available. In a fossil species, such a shell might be regarded as an indication of photosymbiosis. However, P. placenta apparently hosts no photosymbionts. Although | am not aware of any explicit statement to this effect in the literature, the soft tissues were studied extensively by Yonge (1977), who most likely would not have failed to observe photosymbionts if they had been present.
Other Recent species of Placuna, like P. ephippium, are even larger than P. placenta. In P. ephippium, however, a reddish or brown pigmentation of the internal shell layer re- duces shell translucency, and could be adaptive in shelter- ing the soft tissues from sunlight. Presumably, this pigmentation would be lost in fossilised material, leading an observer to conclude incorrectly that the shell was translu- cent like the one of P. placenta.
A further example is the Recent endolithic pectinid Pedum, which exposes a broad surface of ventral mantle tissues to the ambient light, in a fashion not unlike the Tridacnidae. The mantle tissues of Pedum are heavily pigmented and re- semble the brightly-coloured ones of the Tridacnidae. However, this species hosts no photosymbionts (Savazzi, 1998). The broad expanse of exposed mantle tissues in Pedum does not appear to be adaptive by itself, and is rather a consequence of the mode of growth of this bivalve (see also below).
The presence of a prismatic or fibrous shell microstructure in some large fossil bivalves has been advocated in support of the photosymbiosis hypothesis, on the grounds that such a structure could conduct light well, working like a bundle of optical fibres (e.g., Seilacher, 1990). However, this type of light transmission has not been verified in Recent shell ma- terial. In particular, it is not known whether the sides of the prisms (which in living bivalves are interfaces between cal- cite and organic matrix) act as reflectors, or whether the re- fraction index at the periphery of the fibres is different from the one at its centre. Either condition is indispensable for the fibre to function as a light guide. The fibrous layers of some fossil bivalves do seem to act as optical fibres, but this could be misleading, since the organic shell matrix in this material was likely lost or altered during diagenesis, so that the prism sides are now calcite/air interfaces with optical properties different from the original ones. In addition, the optical properties of fibrous layers could be irrelevant, be- cause sunlight may have been absorbed by additional shell layers with different microstructures (which, in addition, may have disappeared through selective diagenetic solution), or by a pigmented periostracum or organic shell matrix (see also above).
In conclusion, when performing this type of functional re- construction, it should be remembered that morphologic cri- teria alone are unreliable, since flattened and/or thickened
Symbiosis in Bivalvia 5
shells may have several adaptive explanations. Therefore, many instances of photosymbiosis inferred from large and thick shells, translucent shells and/or flattened shell geometries in fossils are likely false positives, while most of the true instances of photosymbiosis in fossils are likely to pass undetected.
A few fossil bivalves with highly specialised morphologic features may have been photosymbiotic. This is the case, for instance, of some rudists (Vogel, 1975; Seilacher, 1998, and references therein). Shell morphology indicates that, in several representatives of this group, a well-developed sys- tem of mantle diverticula occupied cavities within the shell, and in some cases was also exposed to the outer environ- ment. However, it must be stressed that, based on palaeoenvironmental reconstructions, as well as on the lack of the above morphological features in most rudists, photosymbiosis must have been restricted to few represen- tatives (see Jablonski, 1996).
Another candidate for symbiosis is represented by the Triassic wallowaconchids (Yancey and Stanley, 1999), in which the wing-like lateral carinae of the large Corculum- shaped shells were subdivided into partitions by radially- growing septa and, presumably, occupied at least in part by finger-like extensions of mantle tissues. This morphology is compatible with photosymbiosis (assuming the shell was translucent) as well as chemosymbiosis (assuming that the shell cavities housed chemosymbionts). However, septa- tion of the carinae may be functional as a lightweight me- chanical reinforcement, and a critical analysis of the symbiosis hypothesis shows that no septa, or at most a sin- gle septum separating the space within the carina from the rest of the shell cavity, are required for both photosymbiosis and chemosymbiosis to take place.
In these and comparable instances, photosymbiosis should be regarded as a reasonable hypothesis only if (1) the observed morphology satisfies all requirements for photosymbiosis, and (2) alternative functions for the ob- served morphology can be discarded. In the rudists with exposed mantle tissues, for instance, one should try first to eliminate the possibility that the mantle functioned as a ciliated carpet for the collection of food particles. In the case of rudists with shell diverticula, one should exclude al- ternative functions like brood pouches, cavities for the “far- ming” of chemosymbionts, structures for discouraging at- tacks by shell borers, energy-absorbing “bumpers” that would stop impact cracks from propagating to the inner shell layers, and lightweight shell structures like those observed in Recent soft-bottom oysters (e.g., cf. Chinzei, 1995). If these precautions are taken, unusual morphological adapta- tions like those of the rudists and wallowaconchids may be more reliable in inferring instances of photosymbiosis than general criteria based on large massive shells or antero- posterior shell flattening.
Chemosymbiosis
Chemosymbiosis appears to be more common than photosymbiosis among Recent bivalves. In addition to deep-water forms, like Calyptogena, which are associated with hydrothermal vents (e.g., Hashimoto et al., 1989; Horikoshi, 1989) or hydrocarbon seeps (Childress et al.,
1986), several deep-infaunal bivalves from shallower water rely on bacterial chemosymbionts. These forms rely on bacteria that oxidise sulphide or methane (above refer- ences). In several cases (e.g., the Lucinidae, Solemya), the chemicals necessary to feed the symbionts are drawn to the mantle cavity by pumping pore water from deeper, anoxic layers of sediments (Felbeck et a/., 1983; Dando et al, 1985, 1986; Reid and Brand, 1986). At least in some of these bivalves, elemental sulphur can be stored within the organism, possibly as a means to store energy (Vetter, 1985).
Solemya and Lucinidae.—Solemya builds Y-shaped bur- rows, and collects water from underlying, oxygen-poor and nutrient-rich sediment through the lowermost branch of its burrow (Stanley, 1970, pl. 3; Seilacher, 1990). Lucinids use the highly extensible foot to build a system of narrow canals with a comparable function (e.g., see Stanley , 1970, pls. 15-18).
Fossil burrows of Solemya (or bivalves with similar habits) have been described as ichnotaxa (Seilacher, 1990), and the burrows of lucinids are potentially preservable. These ichnostructures can be used to detect indirectly chemosym- biosis in these and similar bivalves. However, one must keep in mind that several burrowing bivalves build a single siphonal gallery directed downwards (e.g., see Stanley, 1970). This gallery leads the exhalant current deep into the sediment, and therefore cannot be related to chemosymbiosis.
Teredinidae. — Most Teredinidae are wood-borers and host symbiotic cellulose-digesting microorganisms in an en- larged gut (Turner, 1966, and references therein). Thus, these bivalves utilise the substrate as a food source. Chemosymbiosis is not directly reflected in morphological adaptations of the skeletal parts. However, the boreholes of the Teredinidae are uniquely long (up to 2 m) and slender (typically 5-20 mm in diameter), and can “snake” around ob- stacles and other boreholes, thus allowing these bivalves to utilise the substrate with a higher efficiency than any other wood-borers.
The Caenozoic to Recent teredinid Kuphus reaches very large sizes (over 1 m in length, with a diameter of up to 60 mm) and builds a thick calcareous tube. However, this form is not a wood borer but a secondary infaunal soft-bottom dweller (Savazzi, 1982a, 1999b). It is not known whether it utilises the substrate as a food source, and therefore its body volume, which is substantially larger than that of any wood-boring teredinid, remains unexplained.
Fresh-water bivalves.— The unionid Pleiodon adami from the African Pleistocene possesses a tube-like structure pro- jecting from the antero-dorsal shell margin and parallel to the elongated hinge line. Seilacher (1990) suggested that this tube functioned like a pipette, in order to funnel pore water from underlying sediment layers into the mantle cavity. This would have avoided direct contact of the soft parts with the anoxic sediment. However, Savazzi and Yao (1992) found that other Recent and Pleistocene fresh-water bivalves of similar overall shell morphology (albeit possessing smaller or no anterior projections) burrow with the commissure plane conspicuously inclined, rather than subvertical as inferred by Seilacher for P. adami. In the latter species, a subvertical
58 Enrico Savazzi
orientation would seem to be optimal for siphoning pore water from deep within sediment, while a substantially in- clined orientation like the one observed in other fresh-water bivalves would place the antero-dorsal pipe in an unfavourable position. Thus, chemosymbiosis would have required Pleiodon to assume a shell orientation unusual for these bivalves. Specimens recorded in the life position could help to shed light on this species.
Conclusions
In spite of numerous attempts, the feasibility of detecting reliably photo- and chemosymbiosis in fossil bivalves ap- pears questionable. All the criteria discussed above for in- ferring photosymbiosis in fossil bivalves are likely to produce a large number of false positive and/or false negative re- sults. Probably, the fossilised burrows of a few chemosymbiotic bivalves are so far the only reliable evi- dence of such life habits. However, it cannot be excluded that careful analyses and new evidence may reveal probable instances of photo- and chemosymbiosis among fossil bi- valves.
Macrosymbiosis
Macrosymbiosis in bivalves has received a lesser atten- tion than microsymbiosis, and it is legitimate to ask whether it has a potential for being recognised in fossil material. Macrosymbiosis can be subdivided into two broad catego- ries. The bivalve may be embedded in a larger organism or attached to its outer surface, or the reverse situation may occur.
Embedded macrosymbiotic bivalves
Lithophaginae.— Several Recent species of the mytilid Lithophaga occur constantly within living scleractinian corals (e.g., Kleemann, 1980). This type of association dates at least from the Palaeogene (Krumm and Jones, 1993). Although each of these Lithophaga species is recorded from several species of host coral, they are never found in dead corals (which are inhabited by other species of Lithophaga, exclusively living in this habitat). The siphonal opening of the borehole is exposed to the external environment, and there is no indication that the bivalve exploits the host as a source of food. Most likely, the living substrate provides better protection (living corals grow, while dead ones are subjected to erosion and/or fouling by encrusters) and possi- bly a lesser degree of competition by other borers (living cor- als may be less subjected to bioerosion than dead ones, and the endolithic fauna of living corals is, at any rate, less di- verse than that of dead substrates).
Boring in living scleractinians requires the veliger to settle on the epithelium of the host and to pierce it to reach the skeleton. Alternatively, the veliger could be ingested by a coral polyp and subsequently pierce its coelenteron lining. In either case, this appears to require a behavioural and/or biochemical specialisation. It is not known whether Lithophaga boring in live coral constitutes a monophyletic or polyphyletic group. These species, at any rate, cannot be distinguished reliably from dead-substrate species on the basis of shell morphology. Several (possibly a majority) of
live-coral Lithophaga possess a secondary calcareous coat- ing on the outer shell surface. In several cases, this coating forms into structures (aristae, or sets of denticles) that pro- tect the posterior shell commissure against predators (e.g., Savazzi, 1999b, and references therein). Most species of Lithophaga that bore in dead substrates are devoid of shell encrustations, but some dead-substrate species possess coatings fully comparable to those of live-coral borers (pers. obs.).
The boreholes of live-coral Lithophaga show distinctive morphological characters. The coral surface immediately surrounding the borehole opening is often depressed into a shallow funnel. This feature is absent in dead-substrate Lithophaga. \n addition, growth of the coral forces the bi- valve to move backwards through the substrate, in order to remain close to the external environment. This, in turn, is required by the relatively inefficient filibranch gills of these bi- valves (Carter, 1978). The backwards-boring process causes the bivalve to vacate the anterior region of the borehole. This results in a long anterior extension of the borehole, partly filled with meniscus-shaped calcareous septa and/or loose calcareous deposits (Figure 5H). In Quaternary deposits along the coast of Hilotongan Island, the Philippines, the writer observed weathered sections of large coral boulders containing Lithophaga backward-boring tracks reaching approximately 1 m in length. Assuming a rate of backward-boring equal to or higher than that of for- ward-boring (because of the presence of the siphonal open- ing, backward-boring necessitates the removal of a smaller volume of substrate per unit of length than forward-boring), the observed length of boring tracks is consistent with obser- vations on the Recent rock-boring species L. lithophaga by Kleemann (1973, and references therein), who reported a boring rate in limestone of up to 12.9 mm per year and a life span of up to about 80 years.
Lithophaga lessepsiana is a small Recent species that bores either in living reef corals or in solitary free-living scleractinians of the genus Heteropsammia (Arnaud and Thomassin, 1976; Kleemann, 1980). Heteropsammia, in turn, is symbiotic with a sipunculid housed in a spirally coiled cavity in the basis of the coral. L. lessepsiana bores within the basis of Heteropsammia, and grows to a shell length comparable to the coral diameter, probably causing the eventual death of the coral (Arnaud and Thomassin, 1976). When boring in Heteropsammia, L. lessepsiana lies with the ventral commissure uppermost (Arnaud and Thomassin, 1976). This species is entirely or almost entirely devoid of secondary calcareous deposits of the external shell surface.
The Recent mytilid Fungiacava went one step further and evolved into an endoparasite of fungiid corals (Goreau et al., 1976, and references therein). The siphonal opening of this small form communicates with the coelenteron of the host, from which the bivalve draws its food. Fungiacava fol- lows the growth of the host by migrating within its borehole to remain near the coelenteron. Like L. lessepsiana, it lies with the ventral commissure uppermost. This habit is recognisable in fossils, because of the placement of the siphonal opening of the borehole in a region of the coral skeleton covered by a considerable thickness of soft tissues. In fact, boreholes of Fungiacava were described in
Symbiosis in Bivalvia 59
inhalant current
Figure 1. sches Institut, Tübingen, Germany, GPIT 1571/2-4). in place within the borehole (C). Recent, Cebu Island, the Philippines. Cretaceous, Somalia (modified from Tavani, 1941).
A-C. Botula hortensis (Lamarck) symbiotic in Pattalophyllia sp., Upper Eocene, Possagno, Italy, (Geologisch-Paläontologi- Host coral with siphonal opening of borehole (A), anterior view of shell (B) and shell Shell length is 19 mm. D. Schematic drawing of Vulsella vulsella (Linnaeus) embedded in host sponge, The surface of the sponge is at the bottom. Shell length is 4 mm. E. Stefaniniella colosii Tavani, Height of the preserved shell portion is 88 mm. The extent of the original shell outline
(indicated with a question mark) is reconstructed by analogy with Recent Vulsella and other malleids.
Pleistocene fungiids (Goreau et al., 1976).
Botula.—The Eocene mytilid Botula hortensis (Figure 1A- C) was a borer obligatorily associated with living solitary cor- als (Savazzi, 1982b). Unlike other boring bivalves (cf. Savazzi, 1999b, and therein), B. hortensis did not secrete a lining onto the walls of the borehole. Instead, the coral host reacted to the presence of the borer by sealing the spaces between adjacent septa with a secondary calcareous secre- tion in the region surrounding the borehole. This reaction by the host enables one to detect that the coral was alive at the time it was bored. The borehole of B. hortensis opens on the side of the coral theca (Figure 1A).
The corals inhabited by B. hortensis were soft-bottom forms attached to a substrate only in their juvenile phase and probably capable of active righting. Their theca is horn- shaped, rather than flattened like fungiids. B. hortensis had a life orientation with the ventral commissure uppermost (like Lithophaga in soft-bottom corals; see above).
Although morphologically similar to other species of Botula, B. hortensis differs in behaviour and autecology. Recent Botula are mechanical borers in soft rocks and packed mud, while B. hortensis appears to be a chemical borer, because its shell is thin, has no specialised sculpture, and yet displays no surface wear (Savazzi, 1982b, 1999b).
It is difficult to imagine an evolutionary pathway leading from typical Botula to B. hortensis. Therefore, it is legiti- mate to suspect that the morphological similarity of this spe- cies with Botula is due to convergence rather than
phylogenetic affinity (i.e., that B. hortensis is not a true Botula), or alternatively, that the life habits of Botula in the past were substantially more varied than those of Recent species.
Gastrochaenidae.— This family contains rock and dead- coral borers, as well as several taxa that evolved secondarily into tube dwellers in soft sediments (Carter, 1978, Savazzi, 1982a, 1999b; Morton, 1983). The boring representatives secrete a calcareous lining onto the inner walls of the borehole, and can extend this lining into a chimney-like pro- jection when threatened by the overgrowth of encrusting organisms. The lining is also functional in protecting the bi- valve when it becomes partly exposed. This lining became a stepping-stone to the evolution of the tube-dwelling habit, in which the bivalves are encased in a calcareous envelope, or crypt (Savazzi, 1982a).
Freneix and Roman (1979) illustrated Tertiary echinoids containing the calcareous linings of gastrochaenids boreholes, and interpreted this as an instance of parasitism by the bivalves. However, an analysis of this instance shows that the echinoids most likely were dead at the time of boring, and that the bivalves utilised the test as a sub- strate. In fact, none of these echinoids visibly reacted to boring by the bivalves, which often passed through the test and built calcareous canopies on both its internal and exter- nal surfaces. This lack of a reaction is highly unlikely in a living echinoid, since its test is a porous dermaskeleton con- taining abundant living tissues. Savazzi (1982a, fig. 3E) il-
60 Enrico Savazzi
Figure 2. A-B. Malleus anatinus (Gmelin), Recent, Cebu Island, the Philippines. of dorsal portion of right valve (B, x 0.7). C,E. Malleus malleus (Linnaeus), Recent, Cebu Island, the Philippines, x 1. Interior of right valve of subadult specimen, not yet showing the secondary shrinkage of mantle tissues (see the text for details). D. Unidentified malleid, Middle Eocene, Verona, Italy, x 4.
lustrated a comparable instance of a gastrochaenid boring in the test of a Recent sand dollar (also dead at the time of bor- ing), and building calcareous canopies on both sides of the thin echinoid test (in the illustrated specimen, the test was subsequently destroyed by erosion, leaving an isolated crypt).
Malleidae.—Typical malleids are epibyssate forms. They are characterised by a monovincular resilium, a byssal notch located close to the umbones (Figure 2A), and in several genera by projections of the shell that function as a stabilising surface in connection with an epibyssate or reclin- ing habit (e.g., Figure 2E; Seilacher, 1984). Yonge (1986) described Malleus as a semi-infaunal orthothetic form, but the writers field observations on living Malleus in the Philippines confirm Seilacher’s interpretation of this genus as an epifaunal recliner. In malleids, the organism often displays a determinate growth pattern, in which the mantle first grows to the full extent of the shell perimeter in order to build the projections along the edge of the shell, and subse- quently shrinks to a substantially smaller area of the inner shell surfaces, leaving behind growth lines on the aban- doned inner shell surfaces that allow one to detect the stages of this growth process (leftmost region of Figure 2B).
The genus Vulsella (Figures 1D,