Volume 274, Number 3, March 2013 J O U R N A L O F ISSN 0362-2525 Editor: J. Matthias Starck JOURNAL OF MORPHOLOGY 274:243–257 (2013) Skeletal Ontogeny in Basal Scleractinian Micrabaciid Corals Katarzyna Janiszewska,1 Jakub Jaroszewicz,2 and Jarosław Stolarski1* 1 2 Institute of Paleobiology, Polish Academy of Sciences, PL-00-818 Warsaw, Poland Faculty of Materials Science and Engineering, Warsaw University of Technology, PL-02-507 Warsaw, Poland ABSTRACT The skeletal ontogeny of the Micrabaciidae, one of two modern basal scleractinian lineages, is herein reconstructed based on serial micro-computed tomography sections and scanning electron micrographs. Similar to other scleractinians, skeletal growth of micrabaciids starts from the simultaneous formation of six primary septa. New septa of consecutive cycles arise between septa of the preceding cycles from unique wedge-shaped invaginations of the wall. The invagination of wall and formation of septa are accompanied by development of costae alternating in position with septa. During corallite growth, deepening invagination of the wall results in elevation of septa above the level of a horizontal base. The corallite wall is regularly perforated thus invaginated regions consist of pillars inclined downwardly and outwardly from the lower septal margins. Shortly after formation of septa (S2 and higher cycles) their upper margins bend and fuse with the neighboring members of a previous cycle, resulting in a unique septal pattern, formerly misinterpreted as ‘‘septal bifurcation.’’ Septa as in other Scleractinia are hexamerally arranged in cycles. However, starting from the quaternaries, septa within single cycles do not appear simultaneously but are inserted in pairs and successively flank the members of a preceding cycle, invariably starting from those in the outermost parts of the septal system. In each pair, the septum adjacent to older septa arises first (e.g., the quinaries between S2 and S4 before quinaries between S3 and S4). Unique features of micrabaciid skeletal ontogeny are congruent with their basal position in scleractinian phylogeny, which was previously supported by microstructural and molecular data. J. Morphol. 274:243–257, 2013. Ó 2012 Wiley Periodicals, Inc. KEY WORDS: basal Scleractinia; skeletal ontogeny; microtomography Micrabaciidae; INTRODUCTION Recent molecular studies revealed that solitary and exclusively azooxanthellate Micrabaciidae and Gardineriidae form the most deeply diverging clade of scleractinian corals (Kitahara et al., 2010; Stolarski et al., 2011). According to relaxed molecular clock analyses calibrated by scleractinian fossils, the basal clade splits deeply in the Palaeozoic, around 425 Ma. One can expect that skeletal features of corals with such a long and separate evolutionary history may differ from those of other recent Scleractinia. Indeed, Stolarski (1996) proved Ó 2012 WILEY PERIODICALS, INC. that the skeletal architecture of gardineriids (i.e., by the presence of thick, exclusively epithecal wall) is distinctly different from that of other modern corals. In turn, Janiszewska et al. (2011) showed that micrabaciid thickening deposits (structures forming one of the main microstructural regions in scleractinian skeletons; see Stolarski, 2003) are composed of an irregular meshwork of short and thin fibers organized into small, chiplike bundles, what distinguishes them from other Scleractinia. Micrabaciids form cupolate skeletons (up to 5 cm in diameter) with a peculiar lace-like pattern of septa, costae, and porous corallum wall (Moseley, 1876; Squires, 1967; Cairns, 1989). Calices are everted, often with distinct calicular rim (marginal shelf; see Cairns, 1989), and with septa growing upwards and outwards from the corallum centre. Septa are arranged in a distinct flower-like pattern of oval chambers around the columella (Moseley, 1881). Repeatedly branching costae correspond with the number of septa and alternate in position with them (Wells, 1933). The arrangement of septa as observed in adult coralla suggests formation by symmetrical division of septa of the preceding cycle. This lead Cairns (1982, 1989, 1995, 2001; Cairns and Zibrowius, 1997) to propose that the micrabaciid septal system consists of members of the first two cycles and multiple bifurcations of Additional Supporting Information may be found in the online version of this article. Contract grant sponsor: European Regional Development Fund (Innovative Economy Operational Programme); Contract grant number: POIG.02.02.00-00-025/09; Contract grant sponsor: National Science Centre (Poland); Contract grant number: DEC-2011/03/N/ST10/06471 (K.J.); Contract grant sponsor: European Union SYNTHESYS; Contract grant number: NL-TAF-947 (J.S.). *Correspondence to: Jarosław Stolarski; Institute of Paleobiology, Polish Academy of Sciences, PL-00-818 Warsaw, Poland. E-mail: stolacy@twarda.pan.pl Received 11 August 2012; Revised 28 August 2012; Accepted 2 September 2012 Published online 15 October 2012 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/jmor.20085 244 K. JANISZEWSKA ET AL. the Laboratory of Microtomography, Institute of Paleobiology, Polish Academy of Science, Warsaw. Voltage, current, and number of projections were varied depending on the type of sample (density, skeletal size etc.). Radial projections were reconstructed with Skyscan NRecon and XMReconstructor (for XRadia data) software. The 3D images of corallites and serial micro-CT sections were obtained with Skyscan Data Viewer and Avizo software, respectively. Skeletal micromorphology was visualized using a scanning electron microscope (SEM). The specimens were first sputter coated with platinum and then photographed with a Philips XL 20 SEM in the Institute of Paleobiology PAS, Warsaw. Terminology Fig. 1. Diagrammatic representation of one system of septa (solid lines) and costae (dotted lines) in Leptopenus discus. According to Cairns, higher order septa arise by branching of the third cycle septa. Successive bifurcations of tertiaries were desigIII nated as SI3 , SII 3 , and S3 . Figure adapted from Cairns 1982. tertiary septa (Fig. 1). This model was consistently applied for 30 years despite it being based mostly on adult specimens with fully developed septal cycles. Thus, the goal of this article is to clarify the nature of the micrabaciid skeletal ontogeny based on a series of specimens in various stages of growth and their serial sections. MATERIAL AND METHODS Material We have examined all the micrabaciid genera: recent representatives of Leptopenus Moseley (1881), Letepsammia Yabe and Eguchi (1932), and Rhombopsammia Owens (1986b), both recent and fossil representatives of Stephanophyllia Michelin (1841), and known only from the Cretaceous Micrabacia Milne-Edwards and Haime (1849). Despite the differences between micrabaciid taxa (the robust skeleton of Micrabacia vs. fragile Leptopenus, different number of septa in adult specimens, porosity of septa, etc.), they share the same arrangement of septa and costae. Thus, to propose a general model of micrabaciid ontogeny, we selected specimens that most accurately show the successive phases of corallite growth, regardless of their generic affiliation. The taxonomic details are given in figure captions, whereas locality data are provided in Supporting Information, Table S1. Museum abbreviations: NMNH—National Museum of Natural History, Smithsonian Institution, Washington, D.C.; RMNH—Netherlands Centre for Biodiversity Naturalis, formerly Rijksmuseum van Natuurlijke Historie, Leiden; ZPAL— Institute of Paleobiology, Polish Academy of Sciences. Methods The micrabaciid coralla were scanned with a micro-computed tomography (micro-CT) system. Radial projections were then used to reconstruct virtual slices and three-dimensional (3D) models of corallites. Thus, thin sections (traditionally used in studies of coral ontogeny) were replaced by serial micro-CT sections (microtomograms). To interpret the septal insertion pattern, horizontal (perpendicular to the corallite axis) sections were used. Micro-CT data were collected with Skyscan 1172 at the Faculty of Materials Science and Engineering, Warsaw University of Technology system and with XRadia MicroXCT-200 system in Journal of Morphology To date, the terminology used for descriptions of mostly adult coralla of micrabaciids has been largely adopted from those describing skeletal structures of other scleractinians. Our ontogenetic observations of micrabaciid coralla show, however, that the formation and structure of some of the skeletal elements are very different from traditional interpretations. The following glossary provides an overview of the terminology used previously and herein (including new terms) for micrabaciid skeleton descriptions. Costa (plural: costae). Radial structure outside the corallite wall. Usually, costae appear as prolongations of septa (Budd and Stolarski, 2011). However, in a few genera (e.g., micrabaciids, turbinoliids: Idiotrochus and Dunocyathus) costae alternate in position with septa (Cairns, 1989); costae-like structures protruding outside the calice outline, but formed as convexity of the wall are called crests and occur in some flabellids (Stolarski, 1995). Septum (plural: septa). Radially arranged longitudinal partition of corallite, in Scleractinia hexamerally arranged in cycles. New septa are inserted between members of the preceding cycle. Septa of successive cycles are designated as primaries, secondaries, tertiaries, quaternaries etc. (abbreviated S1, S2, S3, S4, respectively). Septum S1 and all septa between it and adjacent septum S1 (comprising one-sixth of a corallite) form a system of septa. According to Cairns (1982, 1989), new septa in Micrabaciidae arise from the division of the preexisting septa of the third cycle. Thus, he introduced a new notation, in which the septal system in Micrabaciidae consists of septa of the first (S1) and second (S2) cycles and multiple bifurcations of the third cycle designated as S3’, S3’’, S3III, S3IV, S3V, and S3VI (Fig. 1). The data presented in this article show that micrabaciid septa are not formed by bifurcation, but are inserted separately from the neighboring septa (Fig. 2B); therefore the ‘‘traditional’’ notation of septa is used. Septa within a single cycle do not appear simultaneously, thus roman numerals are used to show the minor differences between the time of septal formation, for example, for quinaries: S5I, S5II, S5III, S5IV, whereby S5I appear as first. Synapticula (plural: synapticulae). Rod or bar-like structure connecting opposing faces of adjacent septa and perforating mesenteries between them; synapticulae are formed by joining of two granules growing toward each other from adjacent septa (Budd and Stolarski, 2011; Jell, 1974; Nothdurft and Webb, 2007) Synapticulotheca. Corallite wall formed by one or more rings of synapticulae (Wells, 1956) Pillar (new term). Bar-like structure, which is a part of invaginated micrabaciid wall; converse to synapticulae, pillars growth simultaneously with the formation of septa and appear in pairs (thus plural form—pillars—is herein most frequently used (see also Discussion). Wall (or theca). Skeletal deposit enclosing the column of the polyp and uniting the outer edges of septa (Wells, 1956). RESULTS Formation of Septa New septa are formed on the wedge-shaped invaginations of the wall between septa of the SKELETAL ONTOGENY OF MICRABACIIDS 245 Fig. 2. Early stage of septal growth in recent Leptopenus (ZPAL H.25/7-576): (A) enlarged in (B) lateral and (C) distal view of a corallite fragment. New septum is formed on invagination of wall between septa of preceding cycles; invaginated parts of the wall composed of two sets of pillars inclined downwardly and outwardly from the lower septal margin. Upper margin of septum curves toward neighboring member of a lower cycle; (B) desmocyte attachment scars (arrows) cover an inner part of the wall. (D) proximal view of a corallite (fragment of the same specimen): costae composed of two rows of granules outlining the invaginations of wall and alternating in position with septa. (SEM micrographs). preceding cycle (Fig. 2). In its invaginated parts, the wall is composed of two sets of pillars inclined downwardly and outwardly from the lower septal margins (Fig. 4C). In the course of corallite growth (from the centre of the base) and deepening invagination of the wall, the lower margins of septa are gradually lifted above the level of the horizontal base. In the early stages of growth, septa are not connected to neighboring septa (Figs. 2B,C, and 5C). However, their upper margins strongly bend toward the adjacent septa (Fig. 2B) and, in the next stage of growth, join the nearest members of a former cycle. This connection occurs shortly after the formation of septa, thus, usually independent septal blades are not observed in adult skeletons, unless last-cycle septa are still in the process of completion. The pseudobifurcate septal pattern observed in most specimens is the outcome of the ontogenetic development of septa, which in the earliest growth phases occur as separated blades. Because the neighboring septa are fused in the juvenile stages of growth, the septa of consecutive cycles in adult coralla are morphologically indistinguishable from each other (Fig. 3E). In addition to pillars which are herein recognized as parts of the invaginated wall, the bar-like elements connect faces of opposite septa in the upper parts of a corallite (Figs. 4A and 5C,F). Those structures, representing typical synapticu- lae, arise after the formation of septa and grow perpendicular to the septal planes (Fig. 5F). Although various micrabaciid taxa may differ in porosity of septa (highly porous septa of Letepsammia vs. imperforate in Rhombopsammia; Cairns, 1989), their common features are a series of oval foramina between the lower edge of each septum and regularly arranged pillars and costae (e.g., Fig. 2A). The inner part of the wall between septa is covered with rows of oval depressions. These shallow pits, few tens of micrometers in diameter, are arranged parallel to the septal planes (Figs. 2B and 5H). Costae The outer part of the micrabaciid wall is covered with granules or ridges forming radially arranged costae. Granules occur along the edges of wall invaginations which, as described above, are the regions where the new septa are formed (Figs. 2D and 5E,I). Because the new septa appear at the top of each invagination, costae alternate in position with septa. Formation of septa and simultaneous invagination of the wall occurs in the interseptal spaces (Fig. 5H); thus costae associated with previous cycles of septa divide symmetrically and two rows of granules outline the new invaginations. In this way, the bifurcate pattern of costae Journal of Morphology 246 K. JANISZEWSKA ET AL. Fig. 3. Cretaceous Micrabacia in different stage of skeletal development. (A–C) specimen (ZPAL H.II/9-36) with 36 fused septa in (A) distal and (B) proximal view and (C1–C5) horizontal micro-CT sections. (D–F) Adult specimen (ZPAL H.II/10-38) having 96 septa: (D1–D5) micro-CT sections of a corallite, (E) distal and (F) proximal view. (G–I) specimen (ZPAL H.II/11-449; enlarged in Fig. 4) with 24 septa: (G1–G5) micro-CT sections, (H) distal and (I) proximal view. Solid central part of the base in all specimens (B,F,I) is disrupted by 12 invaginations of wall reflecting arrangement of first two cycles of septa on the other side of the base. In horizontal micro-CT sections (C1,D1,G1) first cycle septa, formed on wedge-shaped invaginations of wall, are visible as six V-shaped structures. (A,B,E,F,H,I. SEM micrographs). Journal of Morphology SKELETAL ONTOGENY OF MICRABACIIDS 247 Fig. 4. Early juvenile Cretaceous Micrabacia sp. (specimen ZPAL H.II/11-449): (A) distal view of a corallite; septa of three cycles arranged in typical micrabaciid fashion—septa of higher cycle fused with adjacent members of a lower cycle by their inner margins. Synapticulae connect first and third cycle septa (arrows), columella unites inner ends of primaries (S1). (B) proximal view; 12 porous invaginations of wall reflect arrangement of first two septal cycles. (C) lateral view; septa ‘‘supported’’ on oblique pillars of wall, costae alternating with septa. (SEM micrographs). emerges (Figs. 5E,I and 6E, Supporting Information, Fig. S1B). The number of costae is always equal to the number of septa. Division of costae (Fig. 5E) coincides precisely with the formation of incipient septa (Fig. 5C) on the other side of the base. Double ridges forming the marginal parts of costae in juvenile Stephanophyllia (Fig. 5I) correspond to two sets of pillars, each associated with a different invagination and formation of another septal cycle. Costal ridges that appear on the sides of younger (higher cycle) septa are a little less developed than their counterparts on the sides of older (lower cycle) septa, which results in asymmetrical ornamentation of costae (Fig. 5G). At the calicular rim of specimens with incomplete septal cycles that are often bifurcated or have split ends of costae can be discerned (Figs. 6F and 7D). These structures represent incipient skeletal modifications, which during further growth would be followed by invagination of the wall and formation of septa (Fig. 6D). Micrabaciid taxa differ in thickness and ornamentation of costae. In taxa with thick costae (e.g., Leptopenus: Fig. 2D, Stephanophyllia: Fig. 7C,D), two separate rows of granules (each corresponding to one set of neighboring pillars) outline the wide portions of nonporous horizontal wall. In the genera with a more porous base (Letepsammia: Fig. 6B,E; Supporting Information, Fig. S1B; Rhombopsammia: Supporting Information, Fig. S4B), the wall between invaginations is reduced to linear elements, covered with one row of granules located directly at the junction of two neighboring sets of pillars (Fig. 6F, Supporting Information, Fig. S1C). In juvenile Micrabacia (Figs. 3B,F, and 4B), there are no well-defined ridges of costae, but the horizontal parts of the wall between invaginations have slightly elevated margins. A nonporous wall with a cornflower outline forms the central part of the base. In the next stages of growth, outward from the corallite axis, the base becomes more porous and only fragile linear structures of costae are discerned (Fig. 3B,F). Micrabaciid costae, together with concentric rows of pillars form the characteristic porous base of a corallite (e.g., Figs. 5D, 6B, 7CD, and 8C). In all examined specimens, the small (1–2 mm in diameter) central part of the base has a nonporous surface with no trace of attachment (Figs. 6C and 7C, Supporting Information, Fig. S1B). With increasing corallum diameter, the solid, compact structure of the base is disrupted by wall invaginations associated with pores. The consecutive invaginations of wall, accompanied with bifurcation of costae, reflect the order of formation of septa and allow recognition of septa of successive cycles even if they do not differ in size or ornamentation (Fig. 4B, Supporting Information, Figs. 2, 3, and 4). The youngest of two neighboring septa is the one in the youngest (farthest from the centre) fork of costae (Figs. 6B and 7D, Supporting Information, Fig. S1). Septal Insertion Pattern The septal insertion pattern in Micrabaciidae is herein interpreted based on a series of specimens (SEM observations) and virtual sections (microCT). As described above, the septal formation occurs on the invaginations of wall between septa of preceding cycles, and not by bifurcations of third cycle septa. Thus, the traditional notation of sclerJournal of Morphology 248 K. JANISZEWSKA ET AL. Fig. 5. Arrangement of septa in juvenile Miocene Stephanophyllia. (A–E) specimen ZPAL H.27/1-3 (S. elegans) with four cycles of septa: (A1–A4) horizontal micro-CT sections; note solid central part of a base (A1). (B,C) distal and (D,E) proximal view of a corallite. Arrows indicate incipient septa of fifth cycle (C) corresponding to bifurcation of costae on the other side of the base (E). (F–I) specimen ZPAL H.27/2-14 (Stephanophylllia sp.): (F, enlarged in H) desmocyte attachment scars (arrows) cover inner part of the wall between adjacent septa. Note the onset of wall invaginations in the peripheral part of a corallite (arrowheads). In the upper part of a corallite, synapticulae connect faces of neighboring septa (e.g., F-asterisk). (G) lateral and (I) proximal view; different time of invagination (formation of pillars) result in asymmetrical ornamentation of corresponding costae (arrowheads); (B-I. SEM micrographs). actinian septa is used to designate septa of successive cycles in micrabaciids. Scanning electron microscope. The youngest of the specimens studied herein has three cycles of septa (Figs. 3G–I and 4, Supporting Information, Fig. S1). The secondary septa are inserted between Journal of Morphology septa of the first cycle and septa of the third cycle—between first and second cycle septa. Secondaries join adjacent primaries with their upper margins. Two S2 join one S1, the opposite S1 remains free, and each one of the remaining S2’s joins one of the four S1’s (Fig. 4A). The bilateral SKELETAL ONTOGENY OF MICRABACIIDS 249 Fig. 6. Arrangement of septa in recent specimen of Letepsammia formosissima (ZPAL H.25/21-621) with 60 septa: (A) distal and (B) proximal view of a corallum; bifurcations of costae reflects septal insertion pattern (dotted lines). (C) close-up to the solid central part of the base. (D–F) new septa (first two quinaries in each system) appear between bifurcate ends of a costae (arrows) (C–F. SEM micrographs). symmetry of the corallite, now gently accentuated, is emphasized in the next stages of growth by the elongated columella, developing at the junction of the primaries (Figs. 3A,E, 5B, and 8B). The third cycle is comprised of 12 septa. Pairs of tertiaries join each S2. In the subsequent step (36 septa), the first 12 quaternary septa appear in the spaces next to primaries and join S3’s at their upper/inner margins (Fig. 3A–C). In micrabaciids with a complete fourth cycle of septa (48 septa), the S4’s adjacent to primaries join the S3’s closer to the corallite axis than the S4’s growing in spaces between S3 and S2 (Figs. 5A and 6B). In the next stage of skeletal growth (60 septa, see Fig. 6), the first quinaries arise in the spaces between S1 and S4. The next 12 quinaries appear on either side of these S4’s (based on Cairns’ 1982 drawing of Leptopenus but modified using traditional notation; Fig. 1). Again, younger septa (those between S3 and S4) join quaternaries further from the axis than septa of the same cycle formed beside S1. In the case of the specimens with 84 septa, quinaries develop also between S2 and S4 (Fig. 7). In accordance with the previous pattern, these 12 quinary septa join quaternaries farthest from the corallite axis. In a juvenile specimen with an incomplete fifth cycle of septa (lack of septa between S3 and S4 adjacent to S2), quinaries differ in height and thickness (septa adjacent to primaries are higher and thicker than septa between S4 and S3 and between S2 and S4). Also neighboring septa differ in thickness and development of ornamentation (Fig. 7F). Despite the incompleteness of the fifth cycle of septa in juvenile Letepsammia (the lack of septa between S3 and S4), 12 septa in the spaces assigned to S6 are discerned (septa appear in the interseptal spaces closest to the primaries; Supporting Information, Fig. S2). If four S6’s are present in the septal system (specimens with 120 septa—Supporting Information, Fig. S3), S6 septa adjacent to primaries join S5’s closer to the center of a corallite than S6’s inserted between S5 and S4. In Micrabaciidae with 144 septa (Supporting Information, Fig. S4), pairs of S6’s flank S5 septa lying in the next (in the direction from primaries) spaces. Septa between S3 and S5 join quinaries closer to the corallite center than S6’s inserted between S5 and S4. Depending on the species, coralla of adult micrabaciids consist of 96 to 144 septa (with equal numbers of costae), arranged in up to 6 cycles (the last incomplete). Cairns and Zibrowius (1997) mentioned one specimen of Letepsammia with 228 septa (that suggests the occurrence of a seventh, incomplete cycle of septa). Journal of Morphology 250 K. JANISZEWSKA ET AL. Fig. 7. Arrangement of septa in recent specimen of Stephanophyllia complicata (RMNH Coel. 23396) comprising of 84 septa (fifth cycle incomplete).: (A1–A5) horizontal micro-CT sections; note 12 notches in the spaces lacking last septa of fifth cycle (A5arrowheads). (B) distal and (C) proximal view, enlarged in (D)—bifurcation pattern of costae reflects arrangement of septa on the other side of the base (dotted lines). Arrowheads indicate places that lack quinaries between S3 and S4– note splitted ends of costae; (E, enlarged in F—SEM micrograph) lateral view; neighboring septa are different in thickness and size of ornamentation. Note also differences in size of quinaries. Virtual micro-CT sections. The sections of a lower part of corallites show six radially arranged V-shaped elements (Fig. 3C1,D1, and G1). In the upper sections, these elements are replaced by Journal of Morphology simple linear structures. According to our SEM observations, these elements are invaginations of wall which, in the upper part of a corallite, continue as unbranched septal blades. In a series of SKELETAL ONTOGENY OF MICRABACIIDS 251 Fig. 8. Arrangement of septa in recent specimen of Stephanophyllia complicata (ZPAL H.25/22-635) with 96 septa (five complete cycles): (A1–A6) horizontal micro-CT sections of corallite (one system of septa was bold), (B) distal and (C) proximal view of a skeleton; (D) diagrammatic representation of one system of septa based on A4 section. sections, septa of the first cycle appear simultaneously and merge in the centre of a corallite. In the upper sections, septa of the second cycle are visible between primaries. At first free and strongly curved toward adjacent septa of a former cycle, the secondaries finally join the primaries revealing the bilateral symmetry of the corallite (one pair of S2 join one of primaries, an opposite S1 remains free, the other four S2 join each remaining S1; Figs. 3C4, 7A3, and 8A2,3). The pattern of septal appear- ance in the subsequent sections of juvenile corallites corresponds with those from the lowest (oldest) parts of the skeletons in the later stages of growth. Twelve third cycle septa appear between septa of the first and second cycle (Figs. 3C3, D3, and 8A2). The S3 septa are at first straight and free, but strongly curve toward the nearest S2 in the upper part of a corallite (Fig. 3C4, D4). In the next sections, pairs of tertiaries join secondaries. AfterJournal of Morphology 252 K. JANISZEWSKA ET AL. wards, 12 quaternaries appear between S1 and S3 and, in the upper sections, another 12 S4 between S2 and S3. Twenty-four septa of the fourth cycle join neighboring tertiaries (S4 appearing in the lower part of a skeleton join S3 closer to the corallite axis). The fifth cycle comprises of 48 septa. Initially, 12 S5 septa appear adjacent to primaries and join S4 septa closest to the corallite axis. Then, the quinaries between S4 and S3 arise followed by quinaries adjacent to secondaries. The septa located between S3 and S4 (closer to secondaries) are located furthest from the axis, and last in the subsequent sections (Figs. 3D and 8A4,D). In juvenile specimen with incomplete fifth cycle of septa, the 12 spaces that ‘‘lack’’ of the last quinaries (septa between S3 and S4) are clearly visible (Fig. 7A5). The micro-CT sections (Figs. 3C5,D5, 7A5, and 8A5,A6) also exhibit regularly spaced connections between adjacent septa, arranged perpendicular to the septal planes. On the basis of these sections, it is not possible to interpret their origin (whether they were formed by granules growing toward one other from faces of opposite septa), but, based on SEM observations, they are interpreted as true synapticulae. Micrabaciid costae are arranged almost horizontally, thus the data provided by micro-CT scanning are similar to SEM micrographs of the basal part of the corallite described above. Costae are radially arranged in the interseptal spaces. The bifurcation pattern of costae corresponds to insertions of septa in slightly upper sections of a corallite (Fig. 8A2–A3). DISCUSSION Formation of Septa, Costae and the Nature of the Corallite Wall The most important outcome for scleractinian taxonomy is the observation that micrabaciid septa do not form by repeated peripheral branching of septa. Instead, they are formed in the peripheral zone of a corallite, on the invaginations of the wall between septa of preceding cycles (Fig. 9A). During the course of corallite growth, series of oval foramina appear between the lower margins of septa, regularly arranged pillars and costae, giving the wall a porous character (Fig. 2B and 9C,E,F). According to Moseley (1881, p 203), ridges of soft tissue located in the intercostal grooves connect with the bases of mesenteries lying above the costae through these foramina (interpreted as ‘‘perforations’’ in the base). Indeed, the inner part of the wall in the interseptal spaces (of both fossil and recent micrabaciids) bear shallow pits arranged in zones parallel to the septal planes (Figs. 2B and 5H). Similar structures, called desmocyte attachment scars, occur in many other scleractinians and correspond with the position of mesenteries in living polyps (Wise, 1970; Muscatine et al., 1997). Journal of Morphology The above observations challenge previous interpretations of the micrabaciid wall. The corallite wall (often called the base, e.g., Stephenson, 1916) was described as regularly perforated and composed of numerous concentric and radial trabeculae (Moseley, 1881; Duncan, 1884; Fowler, 1888). Concentric trabeculae were next interpreted as synapticular rings, and the wall, consequently, as synapticulothecal (Owens, 1986b; Cairns, 1989; Baron-Szabo, 2008). Our observations show that skeletal ‘‘rings’’ are parts (pillars) of invaginated wall that grow simultaneously with septa, in contrast to synapticulae which develop as a result of fusion of granulation of two opposite septa (Wells, 1956). Therefore, growing in structural continuity with septa, the micrabaciid wall is not synapticulothecate as formerly suggested (Duncan, 1884; Stephenson, 1916; Squires, 1967; Cairns, 1989) but bears resemblance to a marginotheca in flabellids and traditional caryophylliids (Mori and Minoura, 1980; Stolarski, 1995). Moreover, formation of septa in flabellids is preceded by ‘‘inward warping’’ of the wall (Mori and Minoura, 1980: 323), similar to that in Micrabaciidae (‘‘invagination of wall’’). Concentric rows of pillars, representing the successive stages of base growth (Figs. 3B,F, 5B, 7A3,C, and 8C) may be analogous to thecal rings known from corals attached to substrata (Durham, 1949). However, even the oldest, central part of a base in juvenile specimens has a nonporous smooth surface with no trace of attachment (Figs. 6C, Supporting Information, Fig. S1B). The micrabaciid basal plate has not been observed so far and further studies (and findings) are required to decipher the initial phase of corallum growth. The main difficulty is that micrabaciid polyps completely enwrap the corallum (Cairns, 1989), and the initial stages are entirely overgrown by the skeleton secreted during subsequent growth steps. The specimens examined so far do not show any type of different substrate (sand grain or the shell fragment) than is observed in some other corals as initial substrate (Cairns, 1989). The micrabaciid costae are formed along the edges of wall invaginations. As the septa develop at the top of invaginations, micrabaciid costae alternate in position with septa (Figs. 2D, 6C, and 9A,C,E,F), instead of being their prolongations as in the majority of Scleractinia. Wells (1933) considered costae as formed by the fusion of two sets of synapticulae (skeletal elements herein recognized as pillars of invaginated wall). Indeed, when costae are narrow, single rows of granules seem to be formed directly at the junction between two adjacent sets of pillars (Fig. 6C, Supporting Information, Figs. 2B and 3B). However, in species with broader costae, invaginations are clearly separated by wide fragments of nonporous horizontal wall (Figs. 2D, 4B,C, and 7D), thus there is no fusion between neighboring sets of pillars. SKELETAL ONTOGENY OF MICRABACIIDS 253 Fig. 9. Model of septal growth and insertion pattern in Micrabaciidae. (A) Incipient septum arises at the top of invagination of wall; granules arranged along the edges of invagination form costae. The invaginated part of the wall is composed of two sets of pillars that raise the septum above the level of a base. In the next stages of growth, septum bends toward adjacent septum of a former cycle (C, F—oblique view) and joins it by its upper margin (E). (B,D) Diagrammatic representation of one system of micrabaciid septa: according to Cairns (1982) higher cycle septa arise by branching of the third cycle septa; septa of the first cycle remain free (B—adapted from Cairns, 1989; modified). In contrast, according to the model proposed herein (A,C–F), septa of successive cycles are inserted as independent blades; pseudo-bifurcate pattern is an effect of fusion of neighboring septa. Roman numerals indicate the order of formation of septa within single cycle. Journal of Morphology 254 K. JANISZEWSKA ET AL. Septal Arrangement A flower-like pattern of micrabaciid septa, which results from multiple connections of septa of various cycles, was noticed already by 19th century researchers and was used as a criterion for the higher rank classification of the group. For example, Milne-Edwards and Haime (1850) described Stephanophyllia and Leptopenus as eupsamiids (synonym of dendrophylliids) based on septal development. Micrabaciids were later considered a distinct family but their septal arrangement was still described as of ‘‘dendrophylliid type’’ (Vaughan and Wells, 1943). Seeming similarity of septal arrangement in micrabaciid and dendrophyllid corals led Chevalier (1987) to classify both families within suborder Dendrophylliina, a view that was not supported by further morphological (Cairns, 1989) and molecular (Kitahara et al., 2010) studies. Moseley (1881) was the first to point out the unique bifurcation pattern of micrabaciid septa. This regular arrangement of septa was so striking that Moseley (1881, p 207) was ‘‘at a loss to designate the quinary, quaternary and tertiary septa in each system’’ in some specimens, and suggested that these ‘‘terms would seem hardly to apply.’’ Nonetheless, the traditional notation of septal cycles was still in use (Alcock, 1902; Keller, 1977). When Cairns (1982) resurrected Moseley’s idea of bifurcating septa, he proposed a new interpretation, in which the septal system in Micrabaciidae consists of septa of the first two cycles and multiple bifurcations of third cycle septa. This model was later extended to all fossil and extant species (e.g., Owens, 1986b,1994; Cairns, 1989). The bifurcate pattern of septa being very different from dendrophylliid fusion of septa (Pourtalès plan) was suggested as a micrabaciid synapomorphy. Contrary to Cairns’ model, our data show that micrabaciid septa do not arise by symmetrical division of older septa but are formed as independent septal blades between members of preceding cycles as in other Scleractinia (Fig. 9A). Shortly after the formation of septa, their upper margins bend and fuse with adjacent septa of a former cycle (Fig. 9E), yielding a pseudobifurcate pattern observed in adult specimens. Noteworthy, despite some similarity in fusion of septa, the micrabaciid arrangement of septa still differs from that of dendrophylliids. The dendrophylliid Pourtalès plan is a form of septal substitution (Cairns, 2001), which does not take place in the ontogeny of micrabaciids. ‘‘Fusion of septa’’ in the Pourtalès plan refers to members of the same cycle, which grow simultaneously and join by its inner ends (Wells, 1956), whereas newly formed micrabaciid septa join the septa of the preceding cycle. Finally, the lower cycle septa of dendrophylliids are shorter than the members of a higher cycle (Wells, 1956: Fig. 239) contrary to the micrabaciid septal insertion pattern (discussed below). Journal of Morphology The septal ontogeny of micrabaciids bear some resemblance to that of fungiids. In both groups, septa of a higher cycle may join those of a lower cycle. However, in fungiids, first cycle septa remain free, whereas in micrabaciids primaries are united with the second cycle septa. Moreover, pairs of higher cycle septa, which in fungiids consistently merge with septa of lower cycles (starting from fourth cycle up), flank symmetrically the older septa and join them at the same distance from the corallite axis (Boschma, 1923). In contrast, in micrabaciids in each pair, one septum appears earlier and joins older septum closer to the corallite axis than the other. In both groups, septa of a higher cycle may be formed before preceding cycle is completed: first members of a new cycle are inserted in the vicinity of S1’s (Boschma, 1923: Fig. 1 for fungiids, and Supporting Information, Fig. S2 for micrabaciids). However, the order of formation of the other septa within single cycle is different in both groups. In fungiids, septal insertion pattern is taxon-specific: often each sector of the same corallite has different number and/ or arrangement of septa. For example, in fungiid taxa with elongated calices (e.g., Ctenactis echinata, Herpolitha limax, Polyphyllia talpina), consecutive higher cycles septa are formed along axial S1’s, whereas there is a lack of septa of those cycles in the vicinity of other S1’s (Boschma, 1923; Hoeksema, 1989 Fig. 41). In contrast, in all micrabaciids, septa successively appear in all spaces along the septa of preceding cycle, invariably following the same order for each sector of the calice (see ‘‘Septal insertion pattern’’). Septal Insertion Model The serial micro-CT sections (septa appearing in the lower part of a corallite and closer to its axis were considered as older) combined with SEM micrographs of consecutive ontogenetic stages allow reconstruction of the septal insertion model. As in most Scleractinia, skeletal growth begins with simultaneous (judging from their equal size) formation of six primary septa. The members of successive cycles are inserted between previously formed septa in the peripheral part of a corallite. The septa of each cycle join with those of the preceding cycle by their upper margins. Fusion of primaries and secondaries gives a bilateral symmetry to the corallite (four S2’s join four S1’s and two of S2’s join one S1, leaving an opposite S1 free). Only the septa within the first three cycles appear simultaneously. The succession for the fourth cycle is twofold. The septa next to the primaries appear first, then those inserted between S2 and S3. The next cycle begins with pairs of quinaries flanking S4’s nearest to the primaries. The insertion of S5 adjacent to primaries is followed by development of a septum between S4 and S3 septa. Thereafter SKELETAL ONTOGENY OF MICRABACIIDS the remaining quinaries arise: septa in the spaces between S2 and S4 septa and those between S3 and S4 septa afterwards. Species having 120 septa have additional pairs of S6 septa inserted along S5 septa lying in the outermost part of the septal system (S6 between S1 and S5 appear first). In micrabaciids with 144 septa, new S6 septa are inserted on each side of the next S5 septum (in the direction from the primaries). Following the previous pattern: S6 septa appear between S3 and S5 septa before they appear between S4 and S5 septa (Fig. 9D). To sum up: the micrabaciid septa are inserted as follows: 6S1, 6S2, 12S3, 12S4I, 12S4II, 12S5I, 12S5II, 12S5III, 12S5IV, 12S6I, 12S6II. In every cycle, the insertion starts in the outermost part of the septal system where pairs of newly formed septa flank members of a former cycle adjacent to primaries. Then, septa along the next (in the direction from primaries) septa of a previous cycle arise. In each pair of septa, those inserted between older septa are formed first (e.g., quinaries between S4 and S2 before quinaries between S4 and S3). Thus, the final composition of a half-system of micrabaciid septa is: S1, S6I, S5I, S6II, S4I, S6IV, S5II, S6III,S3, S5IV, S4II,S5III, S2 (for specimen with 144 septa; Fig. 9D). In juvenile specimens, neighboring septa differ in height, thickness, and development of ornamentation (Fig. 7F). As the skeleton grows, the differences between adjacent septa become blurred, and fused septa of two cycles may appear as if they formed by branching of the older septum (Figs. 3D–F, 8B, and 9D). However, the above-mentioned dissimilarities between septa in the early stages of development preclude formation of septa by symmetrical division of a lower cycle septa as suggested by Owens (1986b). Phylogenetic and Taxonomic Implications Since the beginning of studies on Scleractinia, skeletal features have been used as the main criteria for taxonomic and phylogenetic interpretations of the group. However, traits originally considered taxonomically significant often diminish in importance and are replaced by new ‘‘unique’’ features. A good example of such an intricate quest for useful diagnostic characters are studies on micrabaciids. Despite the striking similarity in skeletal architecture (e.g., Duncan, 1884; Fowler, 1888), the species that are now grouped in the Micrabaciidae were originally assigned to different coral taxa. Micrabacia was considered as a member of the Fungiidae (a family only consisting of reef corals; see Hoeksema, 1989; Gittenberger et al. 2011), whereas Stephanophyllia and Leptopenus were described as Eupsammiidae (Milne-Edwards and Haime, 1850). When Vaughan (1905) established the new family Micrabaciidae, only Micrabacia, 255 Diafungia, Microsmilia, Podoseris, and Antilloseris were included. In 1933, Wells revised the diagnosis of a family which included Stephanophyllia, Micrabacia, and Leptopenus (Letepsammia and Rhombopsammia were added by Owens in 1986a,b) and pointed out that the feature that distinguishes micrabaciids from other scleractinians is the alternate position of costae and septa. However, this feature is not unique for micrabaciids—costae alternating in position with septa have been described also in some turbinoliids (Dunocyathus and Idiotrochus; Cairns, 1989). From all morphological features that were considered as supporting monophyletic status of Micrabaciidae, only the ‘‘bifurcation pattern of septa and costae’’ (Cairns, 1989) has not yet been challenged. However, as it is shown herein, the dichotomous pattern is not created by true division of septa, but is the result of fusion of newly inserted septa with adjacent members of a former cycle at an early stage of development. Nonetheless, such fusion is unknown in other corals, as well as the micrabaciid-specific insertion pattern of septa within single cycles. Thus, paradoxically, although the in-depth analysis of ontogeny does not support features traditionally considered as unique for micrabaciids, it indicates other traits associated with the same skeletal elements, which seem to be true autapomorphies of the group. This taxonomic aspect of the paper is summarized in the emended diagnosis of Micrabaciidae appended at the end. The presence of so many features that distinguish micrabaciids dovetails with their unique basal position in scleractinian phylogeny, which is suggested by molecular analysis of recent species (Kitahara et al., 2010; Stolarski et al., 2011). It also shows that morphology, when properly recognized, remains a reliable tool for interpreting the phylogeny and taxonomy of corals. The above described details of skeletal ontogeny, particularly the fusion of newly inserted septa at a very early stage of development, show the necessity to scrutinize the skeletal traits of those groups of corals (Palaeozoic kilbuchophyllids) that were considered similar to micrabaciids precisely because of the bifurcation of septa (see discussion in Stolarski et al., 2004, 2011). The results also encourage deeper analysis of the still poorly known anatomy and physiology of deep-water corals that are involved in the formation of such unique macro and microstructural skeletal features. Taxonomic Note: Emended Diagnosis of Micrabaciidae Corallum solitary, cupolate, free, completely enwrapped by the polyp tissue. Septa haxamerally arranged, formed on wedge-shaped invaginations of wall; invaginated parts of the wall porous, composed of pillars inclined downwardly and outJournal of Morphology 256 K. JANISZEWSKA ET AL. wardly from the lower septal margins. Each septum (S2 and higher cycles) joins the nearest member of a former cycle at the upper margin resulting in a pseudobifurcation pattern; starting from the fourth cycle, septa within a single cycle do not appear simultaneously but are inserted in pairs which successively flank the members of the preceding cycle, invariably starting from the outermost part of septal system. In each pair, the septum adjacent to older septa arises first (e.g., the quinaries between S2 and S4 before quinaries between S3 and S4) and joins the neighboring septum closer to the corallite axis. Costae alternate in position with septa by successive bifurcation, and match the number of septa. Microstructure of thickening deposits, unique in Scleractinia: aragonite fibres organized into small, chip-like bundles forming an irregular meshwork within the skeleton. ACKNOWLEDGMENTS The authors wish to thank Tina Molodcova for help in access to coral collection of Institute of Oceanology, Russian Academy of Sciences, Moscow. Ann Budd (The University of Iowa) provided useful comments on the manuscript. The authors also thank two anonymous reviewers who made very helpful comments on the manuscript. LITERATURE CITED Alcock AW. 1902. Report on the deep-sea Madreporaria of the Siboga-Expedition. Leiden. p 52. Baron-Szabo RC. 2008. Corals of the K/T-boundary: Scleractinian corals of the suborders Dendrophylliina, Caryophylliina, Fungiina, Microsolenina, and Stylinina. Zootaxa 1952:1–244. Boschma H. 1923. The Madreporaria of the Siboga Expedition. Part 4. Fungia patella. Siboga-Expeditie 16d:1–20. Boschma H. 1934. On the septal arrangement in Fungid corals. In: Proceedings of 5th Pacific Science Congress, Victoria, Vancouver, pp 4199–4206. Budd AF, Stolarski J. 2011. Corallite wall and septal microstructure in scleractinian reef corals: Comparison of molecular clades within the family Faviidae. J Morphol 272:66–88. Cairns SD. 1982. Antarctic and Subantarctic Scleractinia. Antar Res S 34:1–74. Cairns SD. 1989. A revision of the ahermatypic Scleractinia of the Philippine Islands and adjacent waters, Part 1: Fungiacyathidae, Micrabaciidae, Turbinoliinae, Guyniidae, and Flabellidae. Smithson Contrib Zool 486:1–136. Cairns SD. 1995. The marine fauna of New Zealand: Scleractinia (Cnidaria: Anthozoa). N Z Oceanogr Inst Mem 103:1–210. Cairns SD. 2001. A Generic revision and phylogenetic analysis of the Dendrophylliidae (Cnidaria: Scleractinia). Smithson Contrib Zool 615:1–75. Cairns SD, Zibrowius H. 1997. Cnidaria Anthozoa: Azooxanthellate Scleractinia from the Philippine and Indonesian regions. Mém Mus Natl Hist Nat 172:27–243. Chevalier J.-P, Beauvais L. 1987. Ordre des Scléractiniaires. In: Grasse P.P. (ed.) Traité de Zoologie, Cnidaires, Anthozoaires. Masson: Paris. pp 403–764. Duncan M. 1884. On the internal structures and classificatory position of Micrabacia coronula. Q J Geol Soc 40:561–566. Journal of Morphology Durham JW. 1949. Ontogenetic Stages of Some Simple Corals. Bulletin of the Department of Geological Sciences, California University Publications Berkeley CA, Vol. 28, pp 137–172. Fowler GH. 1888. The anatomy of the Madreporaria: IV. Q J Microsc Sci 28:415–430. Gittenberger A, Reijnen BT, Hoeksema BW. 2011. A molecularly based phylogeny reconstruction of mushroom corals (Scleractinia: Fungiidae) with taxonomic consequences and evolutionary implications for life history traits. Contrib Zool 80:107– 132. Hoeksema BW. 1989. Taxonomy, phylogeny and biogeography of mushroom corals (Scleractinia: Fungiidae). Zool Verh 254:1– 295. Janiszewska K, Stolarski J, Benzerara K, Meibom A, Mazur M, Kitahara MV, Cairns SD. 2011. A unique skeletal microstructure of the deep-sea micrabaciid scleractinian corals. J Morphol 272:191–203. Jell JS. 1974. The microstructure of some scleractinian corals. In: Proceeding of 2nd International Coral Reef Symposium, Brisbane, Australia. Vol. 2. pp 301–320. Keller NB. 1977. New species of the genus Leptopenus and some peculiarities of deep-sea ahermatypic corals. Trudy Inst Okeanol 108:37–43. Kitahara MV, Cairns SD, Stolarski J, Blair D, Miller DJ. 2010. A comprehensive phylogenetic analysis of the Scleractinia (Cnidaria, Anthozoa) based on mitochondrial CO1 sequence data. Plos One 5(7):e11490. Michelin JLH. 1840–1847. Iconographie zoophytologique, description par localités et terrains des Polypiers fossiles de la France et pays environnants. 2 vol. P. Bertrand: Paris. pp 1–348. Milne Edwards H, Haime J. 1849. Observations sur les polypiers de la famille des astréides. C R Acad Sci 29:465–470. Milne-Edwards H, Haime J. 1850. Introduction; corals from the Tertiary and Cretaceous Formations. A monograph of the British fossil corals, First Part. London: Palaeontographical Society. p 171. Mori K, Minoura K. 1980. Ontogeny of ‘epithecal’ and septal structures in scleractinian corals. Lethaia 13:321–326. Moseley HN. 1876. Preliminary report to Professor Wyville Thomson, F.R.S., director of the civilian staff, on the true corals ddredged by H.M.S. Challenger in Deep Water between the dates Dec. 30th, 1870, and August 31st, 1875. Proc R Soc Lond 24:544–569. Moseley HN. 1881. Report on certain hydroid, alcyonarian, and madreporarian corals procured during the voyage of H.M.S. Challenger, in the years 1873–1876. Part 3. On the deep-sea Madreporaria. In: Thomson CW, editor. Report of the Scientific Results of the Voyage of HMS Challenger during the Years 1873–76 under the command of Captain George S Nares and Captain Frank Tourle Thomson. London. p 248. Muscatine L, Tambutte E, Allemand D. 1997. Morphology of coral desmocytes, cells that anchor the calicoblastic epithelium to the skeleton. Coral Reefs 16:205–213. Nothdurft LD, Webb GE. 2007. Microstructure of common reefbuilding coral genera Acropora, Pocillopora, Goniastrea and Porites: constraints on spatial resolution in geochemical sampling. Facies 53:1–26. Owens MJ. 1986a. On the elevation of the Stephanophyllia subgenus Letepsammia to generic rank (Coelenterata: Scleractinia: Micrabaciidae). P Biol Soc Wash 99:486–488. Owens MJ. 1986b. Rhombopsammia, a new genus of the family Micrabaciidae (Coelenterata, Scleractinia). P Biol Soc Wash 99:248–256. Owens MJ. 1994. Letepsammia franki, a new species of deepsea coral (Coelenterata: Scleractinia: Micrabaciidae). P Biol Soc Wash 107:586–590. Squires DF. 1967. The evolution of the deep-sea coral family Micrabaciidae. Stud Trop Oceanogr 5:502–510. Stephenson LW. 1916. North American Upper Cretaceous corals of the genus Micrabacia. Prof Pap US Geol Surv 98–J:115– 131. SKELETAL ONTOGENY OF MICRABACIIDS Stolarski J. 1995. Ontogenetic development of the thecal structures in caryophylliine scleractinian corals. Acta Palaeontol Pol 40:19–44. Stolarski J. 1996. Gardineria—A scleractinian living fossil. Acta Palaeontol Pol 41:339–367. Stolarski J. 2003. Three-dimensional micro- and nanostructural characteristics of the scleractinian coral skeleton: A biocalcification proxy. Acta Palaeontol Pol 48:497–530. Stolarski J, Roniewicz E, Grycuk T. 2004. A model for furcate septal increase in a Triassic scleractiniamorph. Acta Palaeontol Pol 49:529–542. Stolarski J, Kitahara MV, Miller DJ, Mazur M, Meibom A, Cairns SD. 2011. The ancient evolutionary origins of Scleractinia revealed by azooxanthellate corals. BMC Evol Biol 11:316. Vaughan TW. 1905. A critical review of the literature on the simple genera of the Madreporaria Fungida, with a tentative classification. Proc US Natl Mus 28:371–424. 257 Vaughan TW, Wells JW. 1943. Revision of the suborders, families, and genera of the Scleractinia. Geol Soc Spec Paper 44:1–363. Wells JW. 1933. Corals of the Cretaceous of the Atlantic and Gulf coastal plains and western interior of the United States. Bull Am Paleo 18:85–288. Wells JW. 1956. Scleractinia. In: Moore RC, editor. Treatise on Invertebrate Paleontology, Part F: Coelenterata. New York: Geological Society of America and University of Kansas Press. p F328–F444. Wise SW, Jr. 1970. Scleractinian coral exoskeletons: Surface microarchitecture and attachment scar patterns. Science 16:978–980. Yabe H, Eguchi M. 1932. Some Recent and fossil corals of the genus Stephanophyllia H. Michelin from Japan. Sci Rep Tohoku Univ, Ser 2 (Geology) 15:55–63. Journal of Morphology