Center for the Study of Evolution and the Origin of Life (Institute of Geophysics and Planetary Physics) and NASA Astrobiology Institute, University of California, Los Angeles, California 90095-1567, USA
Center for the Study of Evolution and the Origin of Life (Institute of Geophysics and Planetary Physics) and NASA Astrobiology Institute, University of California, Los Angeles, California 90095-1567, USA
Department of Earth and Space Sciences, Molecular Biology Institute, University of California, Los Angeles, California 90095-1567, USA
Abstract. Bryozoans are among a diverse range of invertebrates capable of secreting calcium carbonate skeletons. Relatively little is known about biomineralization in bryozoans, despite the importance of understanding biomineralization processes for nanotechnology and the threats imposed by ocean acidification on organisms having calcareous skeletons. Ten species of cheilostome bryozoans that are reported to have bimineralic skeletons of calcite and aragonite are studied here using Raman spectroscopy. This technique allowed identification of the two mineral phases at submicron spatial resolution, allowing the distributions of calcite and aragonite within bryozoan skeletons to be determined with unprecedented precision. Confirming previous findings based on the use of chemical stains, most of the bimineralic species analyzed exhibited a calcitic skeletal framework, composed of basal, vertical, and inner frontal walls, having aragonite deposited subsequently onto the outer surfaces of the frontal walls. In one species (Odontionella cyclops), aragonite formed the superstructure above the autozooids, and in two others, traces of aragonite were detected on the undersides of the frontal shields. Using Raman spectroscopy, it was possible for the first time to determine the mineralogy of small-scale structures, including orificial rims, condyles and hinge teeth, avicularian pivotal bars and rostra, and ascopore rims and sieve plates. Even when surrounded by aragonitic frontal shields, these structures were found typically to be calcitic, the two exceptions being the aragonitic avicularia of Stylopoma inchoans and O. cyclops. Unexpectedly, the first-formed part of the basal wall at the distalmost growing edge of Pentapora foliacea was found to consist mainly of aragonite. This may point to a precursory phase of biomineralization comparable with the unusual mineralogies identified previously in the earliest-formed skeletons of members of some other invertebrate phyla.
Calcium carbonate (CaCO3) is used as a skeletal material by a wide range of invertebrates, including foraminifera, molluscs, brachiopods, serpulid polychaetes, echinoderms, and crustaceans. An organic matrix is generally present within these skeletons (Towe 1972; Mann 2001), resulting in chemical and physical complexities not present in abiogenic carbonates. Three mineral polymorphs of CaCO3 having different crystal structures can be formed biogenically: calcite, aragonite, and vaterite. Vaterite is a rare biomineral, but calcite and aragonite are common. Some species with calcareous skeletons secrete calcite, others aragonite, and still others have bimineralic skeletons in which calcite and aragonite are secreted in distinct skeletal layers. Considerable interest has focused on the mechanism(s) that control the polymorphic switch from calcite to aragonite in bimineralic taxa (e.g., Belcher et al. 1996; Thompson et al. 2000). Such research has centered on gastropod and bivalve molluscs in which, because of their relatively thick shells, the layers composed of each polymorph can be readily discerned and studied. A better understanding of the processes of biomineralization could be applied usefully in the field of biofabrication (Fritz et al. 2002). Similarly, more information is needed about the formation and nature of the calcareous skeletons of marine invertebrates generally, especially in the light of the threat posed by anthropogenically driven ocean acidification (Raven 2005).
To produce the Raman effect, laser-generated monochromatic light impinges on the material to be analyzed. The great majority of this light is scattered without any change in its wavelength (Rayleigh scattering), whereas a small amount of the scattered light is affected by vibrational changes in the chemical bonds of the analyzed material that result in a shift in its frequency (the Raman shift). Because the amounts of such shifts are characteristic of specific molecular structures, Raman spectra recorded from such wavelength-shifted light provide the information necessary to identify the material analyzed.
Raman analyses can be performed both by measuring the point spectrum of a submicron-sized area of interest (whether on the surface of a sample or in the interior of a translucent specimen) and by mapping a selected area of a sample to produce a two-dimensional Raman image (e.g., Schopf et al. 2005) that shows the distribution of particular molecular structures at submicron spatial resolution. Applications of Raman spectroscopy in the biological and geological sciences have increased appreciably in recent years (e.g., Nasdala et al. 2001; Schopf et al. 2005; Burchell et al. 2006).
For the study of calcareous skeletons, Raman spectroscopy has several advantages over other methods of analysis. In particular, it is relatively rapid, non-destructive, non-invasive, and allows micron-scale, in situ analysis of materials that can subsequently be imaged optically or by scanning electron microscopy (SEM). The relative merits of Raman spectroscopy and X-ray diffraction (XRD) for studying the mineral phases of CaCO3 have been evaluated by Kontoyannis & Vagenas (2000) and by Dickinson & McGrath (2001). According to these resrearchers, while XRD provides a more precise quantification of carbonate minerals in bulk samples, only Raman can map the distribution of these minerals at small scales. Furthermore, grain size variations do not markedly affect identification of carbonate minerals in Raman spectroscopy (Herman et al. 1987). Before this work, Raman spectroscopy has been used in several studies to analyze CaCO3 biominerals (Bischoff et al. 1985; Urmos et al. 1991; Roberts & Murray 1995; Cuif & Dauphin 1998; Weiss et al. 2002; Addadi et al. 2003; Raz et al. 2003; Weiner et al. 2003). Heretofore, however, this technique has not been applied to bryozoans.
Bryozoa are a diverse phylum of mostly marine, colonial invertebrates, abundant today and having a rich fossil record that stretches back to the Ordovician. Nearly all species secrete CaCO3 skeletons. Smith et al. (2006) recently surveyed the carbonate mineralogy of bryozoans and found that two-thirds of the analyzed colonies had calcitic skeletons. Of the remainder, about half were aragonitic and half were bimineralic. All bimineralic species belong to the Cheilostomata, the dominant order of bryozoans in modern seas. Bimineralic forms are taxonomically widespread among the cheilostomes, occurring in many different families. According to Sandberg (1983), in bimineralic cheilostomes the basic structural box (zooecium) of each zooid is calcitic, aragonite being added later in ontogeny as elaborations or reinforcements, commonly on frontal exterior surfaces, although in free-living species aragonite may occur as a basal thickening.
Here, we use Raman spectroscopy to study the spatial distribution of CaCO3 polymorphs in ten species of bimineralic cheilostome bryozoans. These studies show that Raman point analyses and two-dimensional mapping enable the distributions of calcite and aragonite to be determined at submicron scales. Our results generally corroborate previous findings for bimineralic bryozoans, i.e., that calcite forms their skeletal framework, with aragonite being added subsequently to outer frontal surfaces. However, they show also, for the first time, that calcitic structures can occur within zones of aragonite secretion, that avicularia are constructed of aragonite in some and of calcite in other species, and that an aragonite-rich precursory layer is present at the growing edge in one bryozoan species.
Fragments of bryozoan colonies were bleached in diluted commercial sodium hypochlorite to remove organics. Some samples were surface-analyzed without further preparation, but others were first embedded in epoxy and thin sectioned in planes perpendicular to the colony frontal surface to a thickness of ∼30 μm. Such sectioned surfaces were roughly polished and left uncovered for Raman studies.
Spectroscopic data were obtained using a T64000 triple-stage laser-Raman system (JY Horiba, Edison, NJ, USA) having macro-Raman and confocal micro-Raman capabilities. This system allowed acquisition both of individual point spectra and of Raman images that display the two-dimensional spatial distribution of molecular-structural components. Owing to the confocal capability of the systems, the use of a 100 × objective (having an extended working distance of 3.4 mm, a numerical aperture of 0.8, and not requiring immersion oil) provided a horizontal resolution of ≤1 μm and a vertical resolution of 1–3 μm. A Coherent Innova 90 argon ion laser (Santa Clara, CA, USA) equipped with appropriate optics provided several laser wavelengths in the blue-green region of the visible spectrum. We used a single spectral window centered at 1400 cm−1 that, for the laser excitation used, at 488 nm, provided coverage from 140 to 2650 cm−1, a range that contained all major Raman bands both of calcite and of aragonite.
For analysis, specimens were centered in the path of the laser beam projected through an Olympus BX41 microscope (Olympus, Center Valley, PA, USA). The typical laser power was 21–8 mW over a 21 μm spot, an instrumental configuration well below the threshold causing radiation damage to specimens of the kind studied here. To acquire two-dimensional Raman images, a rectangular area enclosing a part of a sample was selected for imaging; the backscattered Raman spectra obtained within such rectangles were collected through the optical system described above along μm-resolution scan lines, and the x–y registrations of the data obtained were then automatically recorded to provide a pixel-assigned array of spectral elements (“spexels”). During such mapping, the alignment of the laser beam was fixed and the specimen was moved incrementally in the x and y directions using a motorized high-precision microscope stage (SCAN 75 × 50, Märzhäuser GmbH & Co. KG, Wetzler-Steindorf, Germany). To obtain an acceptable signal-to-noise ratio, the acquired spectral images were typically composed of 40 × 40 spectral elements, each analyzed for 2–4 s, resulting in a total data collection time for each image of 22 h. The several hundred spexels comprising each such image were then processed (using the PC program, LABSPEC version 4.14) by constructing a map of the intensity in the spectral window corresponding to the major Raman bands either of calcite (∼283 cm−1) or of aragonite (∼205 cm−1).
Although cheilostome bryozoans have complex skeletons, three basic topological components can typically be distinguished in the box-shaped skeletons of their zooids: a basal wall that forms the floor of the zooid box; vertical walls that form the sides of such boxes; and a frontal wall (or frontal shield) that forms the lid of the box-like structure and contains an opening (orifice) through which the lophophore of the zooid is present. The presence of ovicells for brooding larvae, of non-feeding zooidal structures such as avicularia, and of small-scale structures associated with the orifice (such as hinge teeth and condyles), and in some species an additional opening (ascopore) into the compensation sac, adds to the complexity of cheilostome skeletons. To investigate the skeletal intricacies of such specimens, multiple Raman spectra were obtained of each of the taxa studied here.
Specimens of ten species of bimineralic recent cheilostomes were selected for study from the collections of the Natural History Museum, London (NHM): (1) Pentapora fascialis (Pallas 1766), 40–41 m, between San Giovanni and Little San Giovanni Islands, Rovinj, Croatia, Adriatic Sea, F.K. McKinney Collection 1995; (2) Pentapora foliacea (Ellis & Solander 1786), Cornwall, England, part of NHM No. 22.214.171.1241; (3) Schizoporella errata (Waters 1878), buoy south of Tema, Ghana, P.L. Cook Collection, part of NHM No. 19126.96.36.199; (4) Stylopoma inchoansTilbrook 2000, Malta, J. Borg Collection; (5) Microporella ordo Brown 1952, New Zealand, NZOI Collection; (6) Calpensia nobilis (Esper 1796), Naples, Italy, Norman Collection, part of NHM No. 188.8.131.524; (7) Odontionella cyclops (Busk 1854), craypot, Waitangi, Chatham Island, New Zealand, P.D. Taylor Collection 1999; (8) Poricella celleporoides (Busk 1884), Torres Strait, Australia, part of NHM No. 18184.108.40.206; (9) Escharoides contorta (Busk 1854), South Africa, part of NHM No. 220.127.116.11; and (10) Hippopodina tahitiensis Leca & d'Hondt 1993, Nagasawe, Etafe, Vanuatu, part of NHM No. 2004.2.4.2. Each of these samples comprised several fragments presumed to have derived from the same colony.
Lines corresponding to molecular and lattice vibrational modes were well separated in the Raman spectra of bryozoan calcite and aragonite. In the high-frequency region of the Raman spectra of these minerals, the band corresponding to a totally symmetric vibrational mode of carbonate anion CO32− dominated, situated at ∼1085 cm−1 for calcite and at ∼1084 cm−1 for aragonite. Although the small difference in location (∼1 cm−1) between these bands made them difficult to use to distinguish between these two CaCO3 polymorphs, the low-frequency Raman bands of the two minerals were very well separated: ∼283 and ∼205 cm−1 for calcite and aragonite, respectively. These major Raman bands were therefore used to distinguish between calcite and aragonite in the bryozoans analyzed, whether in thin sectioned specimens or in pristine or broken skeletal surfaces of colonies.
Table 1 summarizes our findings from 169 Raman spectral point analyses and 29 two-dimensional maps. On request, the full database of optical images, point spectra, and maps can be obtained from P.D.T.
Table 1. Summary of calcite and aragonite distributions in ten bimineralic cheilostome species as determined by Raman spectroscopy. A, aragonite; C, calcite.
Distalmost growing edge is composed mostly of aragonite.
This ascophoran cheilostome, a member of the Bitectiporidae, has erect, bifoliate colonies with strap-like branches. The back-to-back basal walls of the two layers of its zooids form a median lamina that is exposed at the distal growing edges of its colonies. The autozooidal frontal shields in P. fascialis are heavily calcified and a small minority of its zooids have suboral adventitious avicularia. Ovicells were absent in the studied material.
Raman spectra of the basal and vertical walls of the specimens analyzed showed them to be composed of calcite. As expected on the basis of previous studies of Pentapora (Carson 1978), the inner layer of the frontal shield consisted of calcite (Fig. 1D) but the subsequently deposited outer layer was aragonitic (Fig. 1E). The boundary between the calcitic and aragonitic layers was sharp, with no evidence of a mixed zone. Studies by transmitted light microscopy showed that although the aragonitic layer was typically more fibrous and more brown in color than the overlying calcite and the boundary between the two layers was clearly delineated in some regions by a dark line, in other regions it was indistinct (Fig. 1F). Surface analyses of older zooids having thickened frontal shields showed that aragonite covered the ends of the vertical walls at zooidal boundaries. However, orificial rims and condyles were calcitic, even in older zooids that exhibited aragonitic frontal shields. Suboral avicularia were composed of calcite.
This species, regarded by some researchers as a junior synonym of P. fascialis, is typically foliaceous, having broad bifoliate fronds joined at their edges to form a boxwork colony. In the specimen studied here, avicularia and ovicells were not observed.
As in P. fascialis, above, the older basal walls in P. foliacea were usually calcitic. However, Raman analyses of the youngest part of a pristine basal wall that formed a 20-μm-wide zone at its distalmost growing edge showed it to be comprised predominantly of aragonite (Fig. 2D). Vertical walls were found to be calcitic, in some cases retaining this mineralogy when forming zooidal boundaries on the surface of the colony. Frontal shields in vertical sections had an inner calcitic layer that transitioned sharply into an outer aragonitic layer (Fig. 1A,B). Low mounds in the calcitic layer of the studied specimen were amplified to form distinct pustules in the overlying aragonitic layer (Fig. 1A–C). Surface analyses of the undersides of frontal shields in old zooids revealed the presence of calcite, as expected, but traces of aragonite were also detected. Orificial rims were composed of calcite but contained traces of aragonite in older zooids.
Analyzed material of this ascophoran (Schizoporellidae) comprised delicate tubes, mostly unilamellar, but locally having multilamellar overgrowths. Although adventitious avicularia were analyzed, no ovicells were detected.
Basal and vertical walls were calcitic, whereas the frontal shield had an inner calcitic and an outer aragonitic layer (as in the previously analyzed species). The frontal shield of a newly formed, frontally budded zooid was found to have an outer surface that was mostly aragonitic, but that also contained significant amounts of calcite. Orificial rims, condyles, and the rostra and pivotal bars of adventitious avicularia were all composed of calcite.
Like those of the closely related Schizoporella errata, above, the studied specimens of this schizoporellid species were tubular, but the tubes were considerably more robust, comprising stacks of about five layers of frontally budded zooids. Both adventitious and interzooidal avicularia were present. Ovicells were not observed.
Basal and vertical walls were calcitic. Some basal walls had two optically distinct layers, both composed of calcite. Thickened basal walls containing deep pits were present in some zooids that formed the innermost layer of the tubular colony. Despite their resemblance to the aragonitic frontal shields of some other cheilostome taxa, such thickened basal walls were found to be entirely calcitic. Raman analyses showed the frontal shields in S. inchoans to be composed of a thin inner layer of calcite and a secondarily deposited thicker outer layer of aragonite. Orificial rims, as well as the deeply seated ribbed condyles characteristic of this species (Tilbrook 2000), were shown to be composed of calcite (Fig. 3). In contrast with this predominance of calcite, and unlike the results obtained from analyses of the closely related S. errata (above), all of the many avicularian structures analyzed in S. inchoans were shown to be aragonitic. These included cross-bars and rostra of adventitious avicularia as well as the outer frontal shield and ligulum on the cross-bar of larger interzooidal avicularia.
Unlike most of its congeners (see Taylor & Mawatari 2005), this species of the ascophoran cheilostome Microporella (Microporellidae) has erect colonies formed of bifoliate fronds. Ovicells, which are large and bulbous in M. ordo, and adventitious avicularia were both present in the analyzed material.
Raman spectra showed that the basal and vertical walls in M. ordo are calcitic and that its frontal shield is composed of a calcitic inner layer and an aragonitic outer layer. The aragonitic layer could be discerned optically on the surfaces of young zooids where it occurred as a white layer partly coating the surface of the underlying gray calcitic layer. The ovicells were shown to be composed of calcite, and because the basal walls of such ovicells immediately overlie the fully formed frontal shields of the zooids distal to the maternal zooids, vertical sections through this part of the skeleton revealed it to be a layered calcite–aragonite–calcite “sandwich”. Aragonite overgrowths were observed to encroach the flanks of the ovicells, but were never seen to cover the entire surface. Raman Point analyses demonstrated the presence of calcite in orificial rims, hinge teeth, and the rim and sieve plate covering the ascopore (Fig. 3). Avicularia were also primarily calcitic, as indicated by analyses of pivotal bars and rostra, but Raman analyses of the rostra of older avicularia also showed the presence of aragonite.
The anascan C. nobilis (Microporidae) forms fragile unilamellar or multilamellar colonies. Its autozooids have a well-defined semicircular orifice and an extensive, porous, cryptocystal frontal shield that contains deep opesiules for passage of the parietal muscles. Neither ovicells nor avicularia are present in C. nobilis.
The basal and the vertical walls in C. nobilis were found to be calcitic, whereas Raman studies showed that the frontal shield was composed of an inner layer of calcite and an outer layer of aragonite. White material partly covering the frontal shields of young zooids proved to be aragonite, the underlying gray layer being calcite. The rim of the orifice, as well as the smooth platform laterally and distally continuous with it, were calcitic.
Studied material of this variable anascan cheilostome (see Gordon 1986), a member of the Calloporidae and sometimes placed in the genus Foveolaria, consisted of broad, bifoliate fronds. The frontal surfaces of the autozooids were mostly hidden beneath a robust avicularium-bearing superstructure.
Apart from the superstructure and its avicularia, which were found to be aragonitic, all analyzed structures were shown to be composed of calcite. These structures included basal, vertical and frontal walls, the mural rim, and the tiny teeth on the comb-like structure at the proximal edge of the opesia. The ovicells in O. cyclops, which have roofs level with the superstructure, are composed of calcite overlain by aragonite.
This ascophoran (Arachnopusiidae), formerly placed in Tremogasterina (see Cook 1977), exhibits multilamellar colonies. The autozooids of the material studied have one or more large frontal pores characterized by spinose rims. Adventitious avicularia of variable size and ovicells were present.
Raman analyses showed that basal walls exposed at the growing colony edge, as well as fractured vertical walls, are calcitic. The spines around the rim of the frontal pores and the orificial condyles were also composed of calcite. Similarly, only calcite could be detected in the frontal shields of young zooids, including the ribs between areolae around the circumference of the zooids. Although an analysis of the underside of a frontal shield of an older zooid revealed only calcite, the upper surfaces of the frontal shields of such zooids were found to contain both calcite and some aragonite. The ovicells of P. celleporoides consisted of calcite, overgrown by aragonite; and the entooecia of broken ovicells were calcitic. Avicularia were also calcitic, as shown by analyses of their pivotal bars and palatal shelves.
The analyzed bifoliate colonies of this ascophoran (Exochellidae) had thickly calcified autozooids and abundant adventitious avicularia but no ovicells.
Both the basal and the vertical walls of E. contorta were found to be calcitic, whereas frontal shields had an inner calcitic layer overlain by aragonite. Calcite was detected in point analyses of the undersides of zooid frontals, both in the region of the umbonuloid frontal shield and proximal to the ring scar. Oral spines and the mucro on the proximal edge of the secondary orifice were calcitic. From analyses of their pivotal bars and rostra, both large and small adventitious avicularia were found to be calcitic.
Studied material of this fragile encrusting ascophoran (Hippopodinidae) contained ovicells and adventitious avicularia.
Whereas the basal wall in H. tahitiensis was found to be entirely calcitic, Raman analyses of vertical walls showed the presence of both calcite and aragonite. Its frontal shields also contained both calcite and aragonite, the former being detected around the pseudopores and the latter on both the outer and inner surfaces of such shields. The ovicells were found to be calcitic, despite their similarity in appearance to adjacent aragonitic frontal shields (Fig. 4). Orificial condyles were calcitic, whereas Raman analyses of the outer surfaces of orificial rims yielded data consistent with the presence either of calcite or aragonite. The pivotal bars and rostra of avicularia were shown to be composed of calcite.
As shown here, Raman spectroscopy is a powerful technique by which to establish, at submicron-scale resolution, the spatial distribution of the two CaCO3 polymorphs, calcite and aragonite, within the skeletons of bimineralic bryozoans. The Raman spectra of these two minerals were clearly and unequivocally distinguishable in virtually all of the analyses carried out on such specimens, for the first time allowing mineralogical analysis of minute bryozoan structures such as the hinge teeth and condyles of autozooids and the pivotal bars of avicularia. Sharp mineralogical boundaries between calcitic and overlying aragonitic layers were shown to be typical of such specimens, even in those where an optical distinction between the two layers was scarcely apparent.
The studies reported here show that bimineralic cheilostomes have a calcitic framework skeleton overlain by aragonite. Basal and vertical walls are always calcitic, except for a thin initial layer of mixed mineralogy detected in Pentapora foliacea (discussed below) and the presence of some aragonite in the vertical walls of Hippopodina tahitiensis. Frontal shields of the anascan Calpensia nobilis and the ascophorans P. foliacea, Pentapora fascialis, Schizoporella errata, Stylopoma inchoans, Microporella ordo, Poricella celleporoides, Escharoides contorta, and H. tahitiensis were all found to be composed of an inner layer of calcite overlain by an outer layer of aragonite. In some specimens, the aragonitic layer occurred as white patches spreading over the gray calcitic frontal shields of young autozooids. Aragonite was also detected on the undersides of the frontal shields in P. foliacea and H. tahitiensis. A similar pattern of distribution was recorded by Sandberg (1976, 1983), who noted the existence of a zooecial lining of aragonite in an unnamed bimineralic cheilostome. In the anascan Odontionella cyclops, aragonite was found to occur only in the skeletal superstructure that overlies the frontal surface of the autozooids.
In addition to providing mineralogical data on the main topological components of bryozoan skeletons (viz., the basal, vertical, and frontal walls), the Raman-based studies reported here have yielded the first direct in situ measurements of the mineralogical composition of small-scale skeletal structures. Orificial rims and associated structures such as teeth and condyles were found to be entirely or predominantly calcitic, even when surrounded by aragonitic frontal shields. The rim of the ascopore and the sieve plate that partly occludes the ascopore opening in M. ordo were also shown to be composed of calcite. In the several species in which they were studied, ovicells proved to be basically calcitic, although those of O. cyclops and P. celleporoides can be overgrown by aragonite. The avicularia in five species (M. ordo, S. errata, P. celleporoides, E. contorta, and H. tahitiensis) were established to be calcitic, but in two other taxa (S. inchoans and O. cyclops) they were shown to be aragonitic.
Because aragonite is a metastable form of CaCO3, it is relatively rarely preserved in fossil bryozoans. Dissolution of the aragonitic layers of bimineralic bryozoans can cause a dramatic change in the appearance of their skeletons (e.g., Greeley 1969). This can potentially lead to taxonomic confusion. For example, Pliocene specimens of the thickly calcified ascophoran Schizostomella dubia Busk 1859, in which the aragonitic frontal shields were lost by leaching during diagenesis, were originally placed in the weakly calcified anascan genus Flustra (see Buge 1957). More subtle changes can occur in taxa in which the frontal shields are bimineralic and only the outer aragonitic layer is lost. This may be the case for many fossil occurrences of the common genus Microporella, in species of which the calcitic avicularia and ascopores can be expected to have become more prominent after diagenetic loss of the outer aragonitic layer of the frontal shields of the autozooids. Two of the ten bimineralic species studied here have been demonstrated to have aragonitic avicularia. These mineralogically metastable structures can be expected to be lost entirely after diagenetic leaching, with such a loss having potentially important consequences in that the presence or the absence of avicularia is used routinely in cheilostome taxonomy.
Perhaps the most surprising result of this study was the discovery of a narrow zone composed predominantly of aragonite at a pristine distal growing edge of the basal wall of P. foliacea. Two-dimensional Raman mapping showed an abrupt transition from an aragonite-rich to an entirely calcitic composition as the wall aged and thickened. Neither contamination nor post-mortem alteration provides satisfactory explanations for the presence of the aragonite zone in view, respectively, of the excellent preservation of microstructure under SEM and the improbability of alteration from calcite to the less stable polymorph aragonite. Aragonite has not been reported previously in the basal walls of biomineralic cheilostomes, although it is of course present in such walls of cheilostomes that have wholly aragonitic skeletons. The anomalous zone of aragonite mineralogy in the basal wall of P. foliacea deserves further study as its location, the site of initial skeletal formation in this taxon, suggests that it may have relevance to understanding processes of biomineralization in cheilostomes. Studies of the skeletal structure of members of other invertebrate phyla have similarly revealed the presence of unusual mineralogies in their early developmental stages. For example, Williams et al. (1998) found the larval shells of apatitic brachiopods to contain siliceous tablets, and Constantz & Meike (1990) reported the presence of calcite centers of biomineralization in a scleractinian coral having an otherwise aragonitic skeleton (but cf. Cuif & Dauphin 1998). In addition, transient amorphous CaCO3 has been reported to play an important role in early biomineralization both of the calcitic spicules of larval sea urchins and the aragonitic larval shells of molluscs (Weiss et al. 2002; Addadi et al. 2003; Raz et al. 2003).
Under normal marine conditions, aragonite is appreciably more soluble than calcite and, thus, will be affected first as acidification of the oceans continues and the secretion of aragonite becomes physiologically increasingly disadvantageous. For bimineralic cheilostomes, the inferred consequences of such an acidification are a thinning of their frontal shields, shown here to have a high content of aragonite, and the dissolution of avicularia from species in which such structures are composed of aragonite. Both of these effects can be predicted to result in a weakening of bryozoan skeletons against predators, while the thinning of their frontal shields may also increase their vulnerability to breakage in strong flow regimes.
Acknowledgments. P.D.T. is grateful for the support provided him as a Visiting Scholar by the IGPP-Center for the Study of Evolution and the Origin of Life (CSEOL) at UCLA where the Raman spectroscopy for this article was carried out, funded by CSEOL and NASA Exobiology Grant NAG5-12357 (to J.W.S.). We thank Alan Cheetham for discussion, and two anonymous reviewers for helpful remarks.