Poriferan and its implications for palaeobiology

E. A. SPERLING1, D. PISANI2 & K. J. PETERSON3 1Department of Geology and Geophysics, Yale University, P.O. Box 208109, New Haven, CT 06520, USA 2Laboratory of Evolutionary , The National University of Ireland, Maynooth, County Kildare, Ireland 3Department of Biological Sciences, Dartmouth College, Hanover, NH 03755, USA (e-mail: [email protected])

Abstract: Well-supported molecular phylogenies, combined with knowledge of modern biology, can lead to new inferences about the sequence of character acquisition in early , the taxonomic affinity of enigmatic Precambrian and , and the Proterozoic Earth system in general. In this paper we demonstrate, in accordance with previous molecular studies, that are paraphyletic, and that calcisponges are more closely related to eumetazoans than they are to . In addition, our Bayesian analysis finds the Homoscleromorpha, pre- viously grouped with the demosponges, to be even more closely related to eumetazoans than are the calcisponges. Hence there may be at least three separate extant ‘poriferan’ lineages, each with their own unique . Because spiculation is convergent within ‘Porifera’, differences between skeletonization processes in enigmatic Cambrian taxa such as Chancelloria and modern sponges does not mean that these Problematica are not organized around a poriferan body plan, namely a benthic, sessile microsuspension feeding . The shift from the anoxic and sulphidic deep that characterized the mid-Proterozoic to the well-ventilated ocean occurs before the evolution of macrozooplanton and nekton, and thus cannot have been caused by the advent of faecal pellets. However, the evolution and ecological dominance of sponges during this time interval provides both a mechanism for the long-term generation of isotopically-light CO2 that would be recorded in carbon isotopic excusions such as the ‘Shuram’ event, and an alternative mechanism for the drawdown and sequestration of dissolved organic carbon within the sediment.

The ‘explosion’ of and groups near metazoan tree are still largely unknown (Eernisse & the base of the Cambrian remains one of the most Peterson 2004; Halanych 2004). Studies using complex and important questions in historical different markers place sponges, placozoans, cnidar- biology. The heart of the debate is focused on ians and ctenophores in almost every conceivable timing: Is the record a faithful chronicle of relationship except one: —all studies events, with the origin of closely predating (except for some early analyses with limited their appearance, or does the event simply record sampling) unequivocally agree that ctenophores and the appearance of burrowing and biomineralizing cnidarians are more closely related to triploblasts whose stocks diverged deep in the Pre- than they are to sponges. Recent studies using - cambrian (Runnegar 1982)? The ancestors of the coding from mitochondrial genomes (e.g. Cambrian metazoan fauna undoubtedly existed in Wang and Lavrov 2007, and references therein) the Precambrian, and the search for Precambrian recover ‘lower’ Metazoa as monophyletic, although ancestors has focused primarily on the soft-bodied the authors attribute this to a clear artifact related to faunas of Newfoundland, South Australia, rate changes between triploblast and the ‘lower’ Russia and Namibia (Gehling 1991; Narbonne 1998, metazoans. Sponges, which are traditionally regarded 2005). Although these faunas have traditionally been as the most extant metazoans, are always interpreted through direct morphological compari- monophyletic in phylogenetic studies based on mor- sons with the modern biota, interpretations need to phology alone (Zrzavy et al. 1998; Peterson & be more tightly constrained by insights gained from Eernisse 2001). However, analyses of ribosomal both phylogenetic studies of the Metazoa and eco- data (Cavalier-Smith et al. 1996; Collins 1998; logical considerations of modern taxa. Adams et al. 1999; Borchiellini et al. 2001; Medina Although a clearer view of triploblast et al. 2001; Cavalier-Smith & Chao 2003; Manuel is emerging, the relationships at the base of the et al. 2003; Wallberg et al. 2004) as well as protein

From:VICKERS-RICH, P. & KOMAROWER, P. (eds) The Rise and Fall of the . Geological Society, London, Special Publications, 286, 355–368. DOI: 10.1144/SP286.25 0305-8719/07/$15.00 # The Geological Society of London 2007. 356 E. A. SPERLING ET AL.

Fig. 1. The importance of paraphyly: if the eumetazoans and poriferans both represent monophyletic groups, each with a unique trophic mode, it is not possible to polarize feeding strategy, given that the outgroups are all non-metazoan (a). However, if calcisponges are more closely related to eumetazoans than demosponges, polarity can be established (b), suggesting the water-canal system is primitive and the gut is derived. sequence data (Kruse et al. 1998; Peterson & Material and methods Butterfield 2005), suggested that sponges are para- phyletic, with calcisponges more closely related to Although the Peterson & Butterfield (2005) tree eumetazoans than to demosponges. The demon- contained several exemplars of both demosponges stration of poriferan paraphyly represents one of and calcisponges, the taxonomic coverage within the most important insights that molecular systema- these groups was not widespread—the two included tics has given palaeontology because paraphyly calcisponges group closely within the means that former synapomorphies (shared (Manuel et al. 2003), and the three demosponges derived characters, e.g. water canal system [WCS] group in the G4 of Borchiellini et al. (2004: are metazoan symplesiomorphies (shared ancestral this study found four main clades, characters) (Peterson & Butterfield 2005; Peterson which they labelled G1–G4). Although we were et al. 2005). Poriferan paraphyly then gives unable to obtain any calcinean calcisponges, we insight into the biology and ecology of the last were able to analyse two G1 sponges, Darwinella common ancestor of all living metazoans, because mulleri and camera, the G2 Chondria sp., we can now state with confidence that the earliest as well as the putative G3 sponge crown-group metazoans were benthic, sessile, sp. (putative because, although not analysed by microsuspension feeders that extracted dissolved Borchiellini et al. 2004, it groups with the G3 Hali- organic matter and picoplankton out of sea water clona in Nichols 2005) and the G4 sponge Hali- using the WCS (Fig. 1). chondria sp. All were purchased from Gulf Nonetheless, the importance of poriferan para- Specimens Marine Laboratory (Panacea, Florida); phyly extends beyond these palaeoecological except for , which was purchased insights, as it sheds light on interpreting the phylo- from the Marine Biological Laboratory (Woods genetic affinities of Ediacaran organisms, as well as Hole, MA). Total RNA from these taxa was pre- Cambrian ‘Problematica’ including Chancelloria. pared from live animals by using a one-step In addition, sponges provide an alternative mechan- TRIzol method (GIBCO-BRL). Total RNA from ism for the oxidation of the Proterozoic ocean. the homoscleromorph was Here, sponge paraphyly is explored by analysing kindly provided by Dr Scott Nichols (University with Bayesian a concatenated data of California, Berkeley). cDNA synthesis was per- set consisting of seven protein sequences from 30 formed with RETROSCRIPT (Ambion, Austin, eumetazoan and 12 sponge taxa including one TX) using 1–2 mg of total RNA. homoscleromorph. Paraphyly is again found, with Partial sequences of seven nuclear-encoded genes the interesting result that the homoscleromorph were PCR amplified, cloned, and sequenced using Oscarella is more closely related to the eumetazo- standard techniques: aldolase (ALD), ATP synthase ans (cnidarians and triploblasts) than it is to any beta chain (ATPB), catalase (CAT), elongation other sponge . We then explore the palaeo- factor 1-alpha (EF1a), methionine adenosyltransfer- biological implications of this result and argue ase (MAT), phosphofructokinase (PFK), and triose- that poriferan paraphyly gives insight into several phosphate isomerase (TPI). Primer sequences are disparate areas of Precambrian and Cambrian as follows (50-30): ALDf: GGGAARGGNATH palaeobiology. YTNGCNGC; ALDr: GGGGTNACCATRTTNG PORIFERAN PARAPHYLY 357

GYTT; ATPBf: GTNGAYGT NCARTTYGAYGA; were set up, sampling the chains every 1000 ATPBr: NCCNACCATRTARAANGC; CATf: GA cycles. Convergence was monitored by controlling YGARATGDSNCAYTTYGAYMG; CATr: CCN the standard deviation of the split sequences (Ron- ARNCKRTGNMDRTGNGTRTC; EF1Af: AAYA quist et al. 2005). The results of the two Bayesian TYGTNGTNATYGGNCAYGT; EF1Ar: ACNGC runs were summarized by calculating a majority NACNGTYTGNCKCATRTC; MATf: GGNGARG rule consensus of all trees sampled after stationarity GNCAYCCNGAYAA; MATr: CCNGGNCKIA was reached. RRTCRAARTT; PFKf: GAYWSNCARGGNA TGAAYGC; PFKr: CCRCARTGNCKNCCCATN ACYTC; TPIf: GGNGGNAAYTGGAARATGA Results and discussion AYGG; TPIr: GCNCCNCCNACIARRAANCC. -specific primers (50 pmol) and 2 mL of Poriferan systematics cDNA, plus the Amplitaq system (using the 10 buffer with 15 mM MgCl2, Applied Biosys- A Bayesian analysis was completed for the 12 tems)Â were mixed and used in a touchdown style sponge taxa plus the 30 eumetazoan taxa and the PCR. The first touchdown (TD 1) procedure started two outgroups (Fig. 2). All expected higher- the annealing temperature at 528C and then after metazoan relationships (e.g. Metazoa, , two cycles dropped one degree for another two Triploblastica, Protostomia, , , cycles, all the way to 408C, followed by a final ten Deuterostomia, ) (indicated with cycles at 528C. A second touchdown procedure labelled nodes and the light grey boxes) are recov- using 35 cycles at 528C using 1 mL of the TD1 as ered, as are the expected internal relationships template followed if the genes of interest were not within phyla (e.g. , Echinodermata, amplified during TD1. PCR fragments of the , Insecta). The recovery of known external predicted sizes were excised, purified (Qiagen, nodes is important in sponge phylogenetics, as Valencia, CA), ligated at 168C overnight into the hypotheses regarding the are undergoing pGEM-T-Easy vector according to manufacturer’s such flux that there are few internal nodes that can instructions (Promega, Madison, WI), and electropo- confidently be used as an accuracy check for rated into DH10B cells. Clones containing the trees. The posterior probabilities for all nodes are correct insert size were sequenced with an ABI373 indicated on Figure 2. model sequencer according to manufacturer’s The addition of the new sponge taxa does not instructions (Applied Biosystems, Foster City, CA). change the paraphyly of Porifera: calcisponges are We were unable to obtain MAT from Dysidea and more closely related to eumetazoans than they are Halichondria, and PFK from Xestospongia and to demosponges (Fig. 2, dark grey box). And . similar to previous molecular studies (Borchiellini Sequences for the demosponge et al. 2004; Nichols 2005), the G1 clade of keratose queenslandica (formerly Reniera, see Hooper & sponges (which here includes Darwinella and Van Soest 2006) another putative G3 taxon (again Dysidea) is the sister taxon of the G3 G4 not analysed by Borchiellini et al. 2004, but sponges (Fig. 2). Unlike these previous analyses,þ traditionally classified as a chalinid haplo- though, we find strong support for a G1 G2 sclerid closely related to ), as well as (which here includes the taxon Chondrilla)þ clade the anthozoan cnidarian Nematostella vectensis, (Fig. 2). In addition, our analysis does not support the Branchiostoma floridae and Ciona the monophyly of the G3 group as Amphimedon intestinalis, and the Capitella does not group with Xestospongia but instead sp. were downloaded from genomic traces from groups with the two G4 freshwater sponges Ephy- the NCBI Genbank database. Sequences were datia and . Thus, the topology edited, translated, and aligned by using MACVEC- found with ribosomal DNA analyses (Borchiellini TOR, v. 7.2.3 ( Computer Group). Twelve et al. 2004; Nichols 2005) is partially supported different sponge lineages plus 30 eumetazoan with an independent data set. lineages taken from Peterson et al. (2004) and Peter- Previous molecular studies based on ribosomal son & Butterfield (2005) and two outgroups were data (Borchiellini et al. 2004; Nichols 2005) analysed using Bayesian phylogenetics. Bayesian placed the Homoscleromorpha in an unresolved analyses were performed using MrBayes 3.1.2 polytomy that included Demospongiae, Calcarea, (Ronquist & Huelsenbeck 2003) under a mixed and Cnidaria. Wang & Lavrov (2007), model, with the parameters of seven unlinked using sequences from the complete mitochondrial evolutionary models (one for each gene) indepen- genome of Oscarella carmela, recovered this dently estimated during tree search (including the taxon as the of the other sequenced best fitting amino acid substitution matrix). Two demosponges. As mentioned above, current mito- independent runs of four linked MCMC chains chondrial trees recover a monophyletic ‘Lower’ 358 E. A. SPERLING ET AL.

Fig. 2. Bayesian analysis of 30 eumetazoan taxa and 12 ‘sponges’ rooted on the Arabidopsis and the yeast Saccaromyces. Note the paraphyly of ‘Porifera’ (dark grey box) with the homoscleromorph Oscarella more closely related to eumetazoans than to other sponges, forming the clade Epitheliozoa (node 3), and the calcisponges more closely related to the epitheliozoans than to demosponges (node 2). In addition, note the division of demosponges into two clades: G1 (Dysidea, Darwinella) G2 (Chondrilla); and G3 (Xestospongilla, Amphimedon) G4 (Microciona, Halichondria, , and Trochospongillaþ ). Major eumetazoan groups are indicated with lightþ grey boxes. M. edulis and californianus are members of the Mytilus.

Metazoa, a result that morphological, ribosomal, erroneous topology. The Bayesian tree recovered nuclear protein coding and microRNA (Sempere here places the homoscleromorph Oscarella as et al. 2006) data sets strongly refute, and it is diffi- the sister taxon to Eumetazoa (Cnidaria cult to evaluate the placement of taxa within an Triploblastica). Phyla have both a constructionalþ PORIFERAN PARAPHYLY 359 and a historical context, and although the - protistan taxa, also have striated rootlets (Nielsen scleromorph body plan is not exceptionally differ- 2001; Boury-Esnault et al. 2003) implying that ent from other sponges, this result establishes, in either the rootlets of Calcarea Epitheliozoa are effect, a new phylum-level lineage if further ana- not homologous with those of choanoflagellatesþ or lyses with more homoscleromorph taxa continue this trait is plesiomorphic for Metazoa and has to find Homoscleromorpha Eumetazoa to the been secondarily lost in Hexactinellida and exclusion of the other two spongeþ lineages. This Demospongiae. result is not completely unexpected, given that The clade Homoscleromorpha Eumetazoa is homoscleromorphs are unique among sponges in herein recognized as Epitheliozoa (Fig.þ 2, node 3). possessing two eumetazoans characters—true Ax (1996) defined the clade Epitheliozoa for the epithelia (Boury-Esnault et al. 1984; 2003; Boute clade of epithelial animals, and it is usually con- et al. 1996), and a distinct acrosome (Harrison & sidered to include Ctenophora, Cnidaria and triplo- De Vos 1991; Boury-Esnault & Jamieson 1999). blasts (e.g. Wallberg et al. 2004). The position of Although this result must be tested with more the homoscleromorphs as the sister taxa to Eumeta- taxa, congruence of independent morphological zoa, as well as the presence of basal laminae (Boute and molecular data sets provides strong support et al. 1996; Boury-Esnault et al. 2003), suggests for phylogenetic inferences. that the Epitheliozoa should include the Homoscler- omorpha. A second potential apomorphy of the Epitheliozoa is the presence of an acrosome (Harri- Poriferan paraphyly and character son & De Vos 1991; Boury-Esnault & Jamieson acquisition 1999). Thus, of the four primary eumetazoan char- acters—tissues, an acrosome, a nervous system and Donoghue (2005), in his discussion of plant phylo- a gut—the acquisition of tissues and the acrosome geny, demonstrated the usefulness of paraphyly in antedated the last common ancestor of homosclero- unravelling the sequential evolution of what had morphs and eumetazoans. The expression of fea- once appeared to be a number of phylogenetically tures such as epithelia in the adult, along with the coincident character changes. Key changes in acquisition of these new characters (nervous plant evolution occurred not at a single node, but system and gut), and the loss of the WCS, could were spread along many steps. The demonstration be due to a coordinated character change (Jenner that Porifera is paraphyletic and therefore rep- 2004) accompanying the neotenous evolution of a resents an has important impli- non-feeding sponge larva to a predatory eume- cations for the polarization of character states and tazoan. Based on molecular clocks, this coordinated understanding the sequence of character acquisition character loss and acquisition accompanying the at the base of Metazoa. dramatic change in trophic strategies from sponge Metazoa (Fig. 2, node 1) is characterized, in to eumetazoan is likely to have occurred sometime part, by the acquisition of multicellularity and during, or soon after, the melting of the Marinoan the presence of the extracellular matrix, a glaciers c. 635 Ma (Peterson & Butterfield 2005), complex of collagen, proteoglycan, adhesive glyco- and may be tied to the oligotrophic ocean conditions protein and integrin, which mediates motility associated with the Marinoan glaciation. and transitions between epithelial and motile cell types (Morris 1993). Because of sponge paraphyly, the WCS with (itself a likely Poriferan paraphyly and Precambrian plesiomorphic cell type), had also evolved palaeobiology by this point as well. The unnamed clade Calcarea Epitheliozoa (epitheliozoans are the Precambrian fossils are often confusing and conten- homoscleromorphþ eumetazoans, see below) tious and among some of the more enigmatic forms (Fig. 2, node 2) isþ potentially characterized by the are the rangeomorph fauna, preserved primarily in presence of cross-striated rootlets. Most metazoan Ediacaran age strata in the Avalon Zone of New- ciliated cells have a system of cross-striated rootlets foundland. These fossils provide an example of that originates in the ciliary basal body and extends the usefulness of modern systematic studies in con- into the cytoplasm. Calcisponge larvae, as well as straining the possible phylogenetic hypotheses for those of homoscleromorphs, have long, cross- Precambrian fossils. striated cell rootlets (Woollacott & Pinto 1995; Ediacaran fossils were first discovered in the Amano & Hori 2001; Boury-Esnault et al. 2003) middle of the 19th century in Newfoundland, but that were perhaps incorporated into adult eumetazo- these structures were not recognized as biogenic ans through neotenous evolution (Maldonaldo until very recently (Gehling et al. 2000). These 2004). However, the choanoflagellate Monosiga, fossils were deposited between 575 Ma and 560 Ma the placozoan , as well as several other (Benus 1988; Bowring et al. 2003; Narbonne & 360 E. A. SPERLING ET AL.

Gehling 2003) in a deep-water setting well beneath demonstrate that they are heterotrophs that parti- the photic zone (Wood et al. 2003). The rangeo- tioned the water column for suspension feeding, morphs, or ‘fractal vendobionts’ (Seilacher et al. but does not necessarily imply that they were, in 2003), which make up more than 85% of specimens fact, metazoans. Community tiering to extract nutri- in the classic Mistaken Point Formation, are com- ents more efficiently from the water column is a posed of fractally-repeating architectural elements simple ecological strategy, likely to be adopted (Narbonne 2004). The organisms grew by pure regardless of phylogenetic affinity, as shown by inflation, with the first- branches eventually the convergent strategy used by Early Cambrian turning into second-order. Fractal or infaltionary sponges (Yuan et al. 2002). Considering: (1) the systems are a method for increasing the surface deep-water setting which precludes a plant or area-to-volume ratio, as might be expected for an algal status; (2) the lack of any eumetazoan apomor- organism feeding directly from the water-column phies despite exceptional preservation; (3) their without orifices. In the presence of a ‘soup’ of dis- fractal organization; (4) the heterotrophic tiering solved organic carbon (DOC) that likely existed of the community; and (5) the low probability that during the Proterozoic (Rothman et al. 2003), ran- the WCS would have been lost in favour of a geomorphs likely fed primarily on this DOC pool fractal design to feed on the same material, the using direct absorption. Noting their environmental Newfoundland rangeomorph fauna probably rep- setting beneath the photic zone, indeterminate resent members of the (the group growth and lack of movement or taphonomic defined by the last common ancestor of fungi and shrinkage, Peterson et al. (2003) suggested that animals; Cavalier-Smith 1998), but were not many members of the Newfoundland fauna, such crown-group metazoans. It is worth emphasizing as Aspidella, Charnia and Charniodiscus, that they still could be stem-group metazoans, but resembled Fungi rather than Metazoa. Nonetheless, our point is that the most likely explanation of the recent studies have considered the degree of tiering trophic data is that the last common ancestor of similarity between the Newfoundland Ediacaran metazoans and rangeomorphs was unicellular, and ecosystems and both Palaeozoic and Modern eco- the fractal and water canal architectures are two systems sufficient to place the rangeomorphs with different solutions to achieve the same goal, the Metazoa (Clapham & Narbonne 2002; namely feeding upon the abundant DOC available Clapham et al. 2003). The lack of any eumetazoan during the Ediacaran. features such as a gut or mouth, depsite the excep- tional preservation, prompted Narbonne (2005) to consider the rangeomorphs as an extinct architec- Poriferan paraphyly and Cambrian ture that may lie between poriferans and cnidarians ‘problematica’ on the metazoan tree, a position first proposed by Buss and Seilacher (1994). Modern poriferan systematics and biology can also However, in light of the results of sponge para- be used to infer the phylogenetic placement of phyly as discussed above, this phylogenetic place- several groups of enigmatic Cambrian fossils, and ment of the rangeomorphs, while not impossible, suggest guidelines for the of sponge does not represent the most likely evolutionary fossils in general. For example, chancelloriids are scenario. Given that the last common ancestor of a group of early Cambrian spongiform organisms metazoans was something akin to a modern that were originally considered to be sponges but sponge, this implies that a WCS is primitive for have since been the subject of a long history of taxo- Metazoa, and thus must have been lost early in nomic speculation. They range from the earliest the phylogenetic history of rangeomorphs if the Cambrian to the early Late Cambrian, flourishing group is apical to sponges. Although eumetazoans in shallow marine environments, often as com- must have also lost the WCS, the key difference ponents of archaeocyathan mounds (Janussen here is that acquisition of the gut allowed eumetazo- et al. 2002). Although they are usually found as dis- ans to change trophic modes and feed macropha- associated sclerites, complete scleritomes can be gously, whereas rangeomorphs would still feed on found in low-energy depositional environments. the abundant DOC in the Proterozoic ocean in a The scleritomes show sessile, attached, sac-like manner analogous to sponges. While evolution is fossils that are covered by spiny sclerites. The non- not always a predictable pathway, the loss of the anchored end contains a thick tuft of sclerites that WCS by a sponge-like organism in favour of a likely surrounded an apical orifice (Bengtson fractal construction to feed on the same food 2005). The sclerites have thin, originally aragonitic source does not represent a parisimonious interpret- walls surrounding a cavity with a restricted basal ation of the data. The ecological studies of Clapham opening (Bengtson & Missarzhesky 1981). The and colleagues showing similarity to modern and outer surface was covered by a soft epithelial ancient animal ecosystems is merely sufficient to integument, perhaps indicating the presence of PORIFERAN PARAPHYLY 361 desmosomal cell junctions (Bengtson & Hou 2001; and once at either at the base or within Homoscler- Janussen et al. 2002). Chancelloriids were orig- omorpha. Given the clear homoplasy of massive inally described as poriferans due to their sponge- within demo- and calcisponges like body form (Walcott 1920), and some workers (Chombard et al. 1997), convergence of spicule still adhere to this position (Butterfield & Nicholas structure as well should not be too surprising. 1996). Nonetheless, in recent years others have The inescapable conclusion is that an organism suggested possible affinities with ascidians (Mehl cannot be removed from the poriferan grade based 1996), cnidarians (Randall et al. 2005), and with simply on spicule characteristics. An organism is the extinct and most likely polyphyletic (Conway a ‘sponge’ (taxon in quotes to represent a paraphy- Morris & Chapman 1997) Coeloscleritophora letic grade) if it feeds using a WCS, pumping water (Bengtson & Missarzhevsky 1981). Most argu- through a chamber via the power of choanocytes ments against a sponge affinity have focused on and extracting dissolved organic matter and pico- the sclerites (Bengtson 2005; Randall et al. 2005), from the current. Fluid physics causes and because there are clear differences between water flow velocities to slow as cross-sectional the spicules of chancelloriids and sponges, most area increases. Sponges generally have a combined authors agree that the two are not homologous. diameter of chambers greater than the We agree. Most workers argue further that if the incurrent pores, and the combined diameters of spicules are not homologous, then chancelloriids the excurrent oscula are less than both the choano- are not sponges. We disagree. cyte chambers and incurrent pores (Brusca & First, despite the superficial morphological simi- Brusca 2002). This causes water to enter at velocity larity of spicules, clear structural and developmen- x, slow dramatically over the choanoderm for tal differences exist between silicisponges and maximum absorption of nutrients, and then exit calcisponges (Harrison & De Vos 1991; Reitner & the sponge at a velocity far greater than x, sending Mehl 1996; Brusca & Brusca 2002). Silicisponges the water clear of the sponge and avoiding recycling deposit siliceous spicules intracellularly, first problems. Fossil sponges can be recognized, there- secreting an organic carbon axial filament within fore, as organisms with combined incurrent pores an elongated vacuole in a sclerocyte. As the axial greater in diameter than the combined excurrent filament elongates, hydrated silica is secreted into openings. the vacuole and around the filament. Calcareous Chancelloriids were sessile, benthic, radially sponges, on the other hand, deposit their spicules symmetrical organisms constructed with presumed extracellularly and without an organic axis; each excurrent openings having a lesser diameter than spicule is essentially a single crystal of calcium car- the presumed combined incurrent pores. No bonate. Second, the presence of silicified spicules specimen shows evidence of a mouth, gut or any does not characterize all demosponges. The G1 other eumetazoan apomorphy, despite their sponges form their skeleton entirely of spongin co-occurrence with soft-bodied forms in localities fibres and do not secrete a siliceous skeleton. The such as Chengjiang (China), the same is true for most, but not all of the G2s—Chon- (Canada) and the (USA) that drilla lacks megascleres but possesses aster micro- commonly preserve such anatomical features. scleres that are homoplasic with respect to other Finally, the recognition of tissues as a eumetazoan demosponges (Borchiellini et al. 2004). In fact, if plesiomorphy removes an additional difficulty hexactinellids are nested near or within the G3/ with a poriferan affinity (Janussen et al. 2002) G4 sponges, as suggested by some rDNA studies given that, similar to homoscleromorphs, chancel- (Cavalier-Smith & Chao 2003; Nichols 2005), loriids were likely to possess an integument then this would strongly suggest that siliceous (Bengtson & Hou 2001; Janussen et al. 2002). megascleres arose only once; differences between Archaeocyaths, another Early Cambrian fossil the spicules of hexactinellids and demosponges group, provide a historical example of how a could be due to the radically different cellular focus on skeletal characteristics may incorrectly organizations of the two (Leys 2003). And third, preclude a group from the poriferan grade. As the presence of silicified spicules does not charac- sessile, conical marine organisms, archaeocyaths terize all of Homoscleromorpha. Two genera were first formally described as possible sponges, (Oscarella and Pseudocorticium) are aspiculate but later classified as coelenterates, , foramini- (Muricy & Diaz 2002), but whether this is primitive fera or an extinct phylum or , with a non- or not is unknown, as the internal phylogeny of poriferan taxonomic affinity often preferred Homoscleromorpha remains unexplored by because their unique double-walled massive calcar- modern molecular means. Therefore, it is clear eous skeleton was unknown in modern sponges (see from the emerging sponge phylogeny that spicules review in Rowland 2001). However, the discovery arose at least three times within ‘Porifera’: at least of massive, calcareous sponges (sclerosponges and once within Silicispongia, once within Calcispongia Vaceletia) demonstrated that the archaeocyathan 362 E. A. SPERLING ET AL. skeleton was not incompatible with a ‘sponge’ affi- data from carbon and sulphur isotopes in the Huqf nity. Studies of archaeocyath ultrastructure (Kruse Supergroup in Oman suggests that this reorganiz- & Debrenne 1989) and functional ation and the oxidation of the Proterozoic ocean (Savarese 1992; Wood et al. 1992) further cemen- took place during the ‘Shuram’ excursion. Chemo- ted their status as sponges, a point now agreed stratigraphic correlations with other sections world- upon by nearly all archaeocyath and sponge wide suggest that the Shuram event likely occurred workers (Rowland 2001). The prior to 551 Ma (Condon et al. 2005), while detrital of both spicules and massive skeletons in different zircons from Oman date it younger than 610 Ma (Le ‘sponge’ lineages suggests the skeletonization Guerroue et al. 2006a). The faecal pellets of meso- process in archaeocyaths and chancelloriids is zooplankton do not sink, and only the faecal pellets largely irrelevant to their broad- taxonomic of macrozooplankton or nekton are capable of position as compared to recognizing the biological rapidly sinking (Peterson et al. 2005). Given that constraints afforded by a sessile, radially con- the fossil record of planktic (Butterfield structed organism with a probable excurrent 1997), and the actual fossil record of the possible osculum and no visible gut or mouth. predators themselves (Butterfield 1994), indicates that the base of the zooplankton food chain did not exist until after the Tommotian, the evolution Poriferan paraphyly and Precambrian of macrozooplankton could not have occurred oxygenation until significantly after the major oxidation of the dissolved organic carbon reservoir and transition Geochemical evidence suggests that the partial to the Phanerozoic ocean. The origin and diversifi- pressure of atmospheric oxygen rose substantially cation of sponges and associated organisms might, circa 2.3 billion years ago (Holland & Beukes however, represent an alternative mechanism for 1990; Farquhar et al. 2000). While the increase in the drawdown of the dissolved organic carbon. atmospheric oxygen helped to oxygenate the surfi- The Shuram negative carbon isotope excursion, cial layers of the ocean, the mid-Proterozoic deep preserved in mid-Ediacaran strata in Oman, shows a 13 0/00 ocean was likely anoxic and sulphidic (Canfield d Ccarb shift on the order of .15 to values of 1998; Anbar & Knoll 2002). Between the mid- 2120/00 (Le Guerroue et al. 2006b; Fike et al. Proterozoic and the Phanerozoic there was a transi- 2006). It is unique in Earth history in recording tional period from ancient to modern ocean, the long-lasting marine carbonate carbon values with latter characterized by a well-oxygenated mixed presumed primary signal well below the canonical or surface layer, a mildly dysoxic to anoxic oxygen- mantle value of 260/00. Negative excursions of minimum zone from c. 200–1000 metres where the extremely large magnitude, such as those during mineralization of organic matter descending from Snowball Earth episodes (Halverson et al. 2005), surface productivity draws down dissolved oxygen in the Early Cambrian (Maloof et al. 2005) or levels, and an oxygenated deep ocean. Evidence Early (Payne et al. 2004), which do not from carbon isotope studies indicates that the Pro- show carbonate carbon values below 260/00, can terozoic ocean was a ‘soup’ of dissolved organic be explained through normal agents such as matter (Rothman et al. 2003). Logan et al. (1995), changes in the fraction of carbon buried as in their seminal study of sedimentary organic organic carbon and/or changes in isotopic fraction- carbon, suggested that hydrocarbons in Proterozoic ation during photosynthesis. The Shuram excursion, sediments were derived primarily from and however, demands a fundamentally different expla- heterotrophic organisms rather than from photosyn- nation for the long-term production of isotopically- thetic organisms as in the modern ocean. Due to the light carbonates. Fike et al. (2006) suggested that lack of faecal pellets, which increase transport the Shuram excursion was caused by the oxidation speed to the ocean bottom, the degradation of of a large dissolved organic carbon (DOC) pool algal products was unusually complete. Logan hypothesized to have existed throughout much of et al. (1995), therefore, suggested that the slowly the Proterozoic (Rothman et al. 2003), but they sinking algal products were re-mineralized before did not identify any mechanism to account for this they could be sequestered in the sediment, and it event. Here, we posit that the advent of novel was the advent of planktonic organisms with modes of heterotrophy, especially the orgin and faecal pellets that allowed for the drawdown of radiation of sponges and rangeomorphs, could the organic carbon pool and transition to the Pha- account for the Shuram excursion. nerozoic ocean. The timing of this shift as described According to our hypothesis (Fig. 3), in the early in Logan et al. (1995) is not clear due to the lack of , the isotopic composition of marine available radiometric dates at that time, occurring at carbonates is controlled by the normal Phanerozoic 13 some point between the Late Neoproterozoic and paradigm: d Ccarb dw 2 (Ddx f org) where dwis the Early Cambrian (Peterson et al. 2005). New equal to the isotopic¼ composition of mantle-derived PORIFERAN PARAPHYLY 363

0/00 CO2 (generally interpreted to be c. 26 throughout dominance of rangeomorphs in the older (c. 575– Earth history), Dd is the fractionation introduced 565 Ma) Newfoundland Ediacaran fauna, and during photosynthesis, and f org is the fraction their almost complete absence in the younger (and converted to organic carbon during photosynthesis. shallower) Ediacara/White Sea (555 Ma) and Unlike the Phanerozoic, where f org sedimentary Namibian (c. 549–543 Ma) faunas (Narbonne organic matter, a significant percentageffi of primary 2005) may be related to their reliance on passive production is not buried in the sediment (Logan absorption on the disappearing pool of dissolved et al. 1995) but is partly degraded in the water organic carbon in the deep ocean, whereas column and accumulates as dissolved organic carbon sponges, which can essentially bring the ocean (DOC). By the early- to mid- Ediacaran (‘Shuram’ through their body with the water-canal system, time), feeding by sponges and rangeomorphs on would have been less affected by a declining DOC led to volumetrically-significant heterotrophic reservoir. respiration on this DOC pool for the first time in Canfield et al. (2006) suggested the deep ocean Earth history. Unlike in photosynthesis there is no may have become oxygenated around the Gaskiers isotopic fractionation during respiration (oxidation glaciation, c. 580 Ma. Oxidation of the Neoprot- of the DOC), thereby producing large quantities of erozoic DOC pool as discussed above must necess- 0/00 isotopically-light (c. 227 )CO2 that equilibrated arily have resulted in a de-oxygenation of the with the water and caused a long-lived isotopic excur- ocean-atmosphere system, unless a different sion below the canonical mantle value. Eventually the oxidant is invoked, and so oxygenation and oxi- DOC reservoir was depleted, leading to a long isotopic dation are not necessarily linked. Nonetheless, the recovery and explaining why the Shuram is a singular advent of the sponge and rangeomorph body plans event in Earth history. The processing of DOC by allowed DOC to be incorporated into benthic sponges after the Shuram excursion would still biomass that could be buried and removed as an have been important, although the amount of DOC oxygen sink. Given that algal products could not available for feeding would be directly related to be buried as faecal pellets until the Cambrian, the production. The evolution of the modern zooplankton reorganization of biogeochemical cycles, as evi- food chain with faecal pellets near the base of the denced by the Shuram excursion, could reflect the Cambrian finally allowed for the direct origin and subsequent radiation of sponges and consumption of algae and the export of carbon to the rangeomorphs. sediment (Logan et al. 1995) without an intermediate DOC step. This hypothesis is supported by the temporal Conclusions correlation of the Shuram isotope excursion with the origination and ecological dominance of two Our molecular results provide further evidence for unrelated grades of organism, sponges and rangeo- the paraphyly of Porifera and suggest that there morphs. Molecular clocks (Peterson & Butterfield are three ‘sponge’ phyla: Silicispongia, Calcispon- 2005) and the fossil record (McCaffrey et al. gia and Homoscleromorpha. This strengthens the 1994; Love et al. 2006) show that the origin of conclusions of Peterson & Butterfield (2005), and the sponge-grade is not much older than the Edia- previous reports based on ribosomal evidence. caran, and the first rangeomorphs are dated to Further work is required to test this topology, to c. 575 Ma (Narbonne & Gehling 2003). Both provide more concrete evidence for the placement were extremely abundant and physiologically of placozoans and hexactinellids, and to elucidate well-suited to feed directly on DOC, unlike vir- the internal topology of demosponges and homo- tually all other clades of including scleromorphs. Nevertheless, basal metazoan phylo- eumetazoans. The WCS of sponges is extremely geny is becoming clearer, and the paraphyly of efficient at moving water through the body via the sponges allows not only for the polarization of char- power of the choanocytes. For example, a large acter states and an enhanced understanding of the sponge can filter its own volume of water every sequence of character acquisition, but sheds much 10 to 20 seconds (Brusca & Brusca 2002). In situ light on the phylogenetic affinities of long-extinct feeding studies on modern sponges demonstrate taxa, and provides new insights into global that some derive the majority of their food from oceanic oxygenation. Finally, the fact that one DOC, and can remove an average of 10% of the modern clade of demosponges, the cladorhizids, DOC in the water in a single pass through the lost the WCS and feeds macrophagously upon water canal system (Yahel et al. 2003). Rangeo- mesozooplankton using a derived mode of extra- morphs, which must also have fed on dissolved cellular digestion (Vacelet & Boury-Esnault 1995) organic carbon using a convergent solution based may shed light on the analogous loss of the WCS on passive absorption (see above), would also in early eumetazoans (Vacelet & Duport 2004). have contributed to the drawdown. The absolute The concordance between the observations that 364 E. A. SPERLING ET AL. PORIFERAN PARAPHYLY 365 cladorhizids live in oligotrophic environments, and Trace Fossils, Small Shelly Fossils and the Precam- that eumetazoans arose on the heels of the Marinoan brian/Cambrian Boundary. Bulletin of the New ‘snowball Earth’, may be of some significance. York State Museum, Albany, 463, 8–9. BORCHIELLINI, C., MANUEL, M., ALIVON, E., BOURY- We thank S. Nichols for the gift of Oscarella carmela total ESNAULT, N., VACELET, J. & LE PARCO, Y. 2001. RNA, and E. Dunn, C. Fiove, A. Evans and V. Moy for Sponge paraphyly and the origin of Metazoa. Journal technical assistance. We also thank G. Love, of Evolutionary Biology, 14, 171–179. D. Rothman, L. Buss, D.E.G. Briggs and R. Summons BORCHIELLINI, C., CHOMBARD, C., MANUEL, M., for discussion, and A. Collins and G. Narbonne for ALIVON, E., VACELET, J. & BOURY-ESNAULT, N. formal reviews. KJP is supported by the National 2004. Molecular phylogeny of Demospongiae: impli- Science Foundation, NASA-Ames, and Dartmouth cations for classification and scenarios for character College. EAS was supported by a NASA Planetary evolution. and Evolution, Biology Internship grant. 32, 823–837. BOURY-ESNAULT, N. & JAMIESON, B. G. M. 1999. Pori- fera. In:JAMIESON, B. G. M. (ed.) Progress in Male References Gamete Biology. IBH Publishing Co., New Delhi, 1–20. ADAMS, C. L., MCINERNEY, J. O. & KELLY, M. 1999. BOURY-ESNAULT, N., DE VOS, L., DONADEY, C. & Indications of relationships between poriferan classes VACELET, J. 1984. Comparative study of the choano- using full-length 18S rRNA gene sequences. some of Porifera: 1. The Homoscleromorpha. Journal Memoirs of the Queensland Museum, 44, 33–43. of Morphology, 180, 3–17. AMANO, S. & HORI, I. 2001. Metamorphosis of coelo- BOURY-ESNAULT, N., ERESKOVSKY, A., BE´ ZAC, C. & blastula performed by multipotential larval flagellated TOKINA, D. 2003. Larval development in the Homo- cells in the laxa. scleromorpha (Porifera, Demospongiae). Biological Bulletin, 200, 20–32. Biology, 122, 187–202. ANBAR, A. D. & KNOLL, A. H. 2002. Proterozoic ocean BOUTE, N., EXPOSITO, J.-Y., BOURY-ESNAULT, N., chemistry and evolution: a bioinorganic bridge. VACELET, J., NORO, N., MIYAZAKI, K., YOSHI- Science, 297, 1137–1142. ZATO, K. & GARRONE, R. 1996. Type IV collagen AX, P. 1996. Multicellular Animals, Vol. 1. A New in sponges, the missing link in basement membrane Approach to Phylogenetic Order in Nature. Springer- ubiquity. Biological Cell, 88, 37–44. Verlag, Berlin. BOWRING, S., MYROW, P., LANDING, E., RAMEZANI, J. BENGTSON, S. 2005. Mineralized skeletons and early &GROTZINGER, J. 2003. Geochronological con- animal evolution. In:BRIGGS, D. E. G. (ed.) Evolving straints on terminal Neoproterozoic events and the Form and Function: Fossils and Development. Yale rise of metazoans. Geophysical Research Abstracts, Peabody Museum Special Publication. 5, 13219. BENGTSON, S. & HOU, X. 2001. The integument of BRUSCA, R. C. & BRUSCA, G. J. 2002. (2nd Cambrian chancelloriids. Acta Palaeontological edn) Sinauer Associates Inc., Sunderland. Polonica, 46, 1–22. BUSS, L. W. & SEILACHER, A. 1994. The Phylum BENGTSON, S. & MISSARZHEVSKY, V. V. 1981. Coelo- Vendobionta: a sister group of the Eumetazoa? scleritophora—a major group of enigmatic Cambrian Paleobiology, 20, 1–4. metazoans. U.S. Geological Survey Open-file Report, BUTTERFIELD, N. J. 1994. Burgess Shale-type fossils 81–743, 19–21. from a Lower Cambrian shallow-shelf sequence in BENUS, A. P. 1988. Sedimentological context of a deep- northwestern Canada. Nature, 369, 477–479. water Ediacaran fauna (Mistaken Point Formation, BUTTERFIELD, N. J. 1997. Plankton ecology and the Avalon Zone, eastern Newfoundland). In:LANDING, Proterozoic–Phanerozoic transition. Paleobiology, E., NARBONNE, G. M. & MYROW, P. M. (eds) 23, 247–262.

Fig. 3. A biological explanation for the Shuram excursion. Bold arrows or font indicate a relatively more important contribution from that source. SL, sea level; CE, carbonate equilibria (CO2 equilibrates with seawater and 2 combines with ocean Ca þ to form carbonates). (a) In the early Neoproterozoic, the isotopic composition of 13 0/00 marine carbonates (d Ccarb) is equal to that of mantle-derived CO2 (dw; generally interpreted to be c. 26 throughout Earth history) minus the fraction converted to organic carbon during photosynthesis (f org), multiplied by the fractionation introduced during the Rubisco cycle (Dd). Unlike the Phanerozoic, where f org sedimentary organic matter (OM), a significant percentage of primary production is not buried in the sediment butffi is partly degraded in the water column and accumulates as dissolved organic carbon (DOC). (b) By the early- to mid-Ediacaran, feeding by sponges and rangeomorphs on DOC led to volumetrically-significant heterotrophic respiration on this DOC pool for the first time in Earth history. Unlike in photosynthesis there is no isotopic fractionation during respiration 0/00 (oxidation of the DOC), thereby producing large quantities of isotopically-light (e.g. 227 ) CO2 that equilibrated with the water and eventually contributed to a long-lived isotopic excursion below the canonical mantle value. (c) By the Phanerozoic, the carbon isotopic system had achieved its present-day steady state, with a small DOC pool and an isotopic value of marine carbonates controlled primarily by the fraction of carbon buried as organic matter. The evolution of macrozooplankton at the base of the Cambrian finally allowed for the direct processing of algal products and their transport to seafloor as faecal pellets without a DOC intermediate. 366 E. A. SPERLING ET AL.

BUTTERFIELD, N. J. & NICHOLAS, C. J. 1996. Burgess HALANYCH, K. M. 2004. The new view of animal phylo- Shale-type preservation of both non-mineralizing and geny. Annual Review of Ecology and Systematics, 35, ‘shelly’ Cambrian organisms from the Mackenzie 229–256. Mountains, northwestern Canada. Journal of Paleon- HALVERSON, G. P., HOFFMAN, P. F., SCHRAG, D. P., tology, 70, 893–899. MALOOF, A. C. & RICE, A. H. N. 2005. Towards a CANFIELD, D. E. 1998. A new model for Proterozoic Neoproterozoic composite carbon-isotope record. Geo- ocean chemistry. Nature, 396, 450–453. logical Society of America Bulletin, 117, 1181–1207. CANFIELD, D. E., POULTON, S. W. & NARBONNE, G. M. HARRISON, F. W. & DE VOS, L. 1991. Porifera. In: 2006. Late-Neoproterozoic deep-ocean oxygenation HARRISON, F. W. & WESTFALL, J. A. (eds) Micro- and the rise of animal . Science, 315, 92–95. scopic of the Invertebrates. 2: , CAVALIER-SMITH, T. 1998. A revised six-kingdom Porifera, Cnidaria, and Ctenophora. Wiley-Liss, system of life. Biological Reviews of the Cambridge New York, 29–89. Philosophical Society, 73, 203–266. HOLLAND, H. D. & BEUKES, N. J. 1990. A paleoweather- CAVALIER-SMITH, T. & CHAO, E. E.-Y. 2003. Phylo- ing profile from Griqualand West, South Africa: evi- geny of , , and other dence for a dramatic rise in atmospheric oxygen and early eukaryotic megaevolution. Journal of between 2.2 and 1.8 bybp. American Journal of Molecular Evolution, 56, 540–563. Science, 290–A, 1–34. CAVALIER-SMITH, T., ALLSOPP, M. T. E. P., CHAO, HOOPER, J. N. A. & VAN SOEST, R. W. M. 2006. A new E. E.-Y., BOURY-ESNAULT, N. & VACELET, J. of Amphimedon (Porifera, Demospongiae, 1996. Sponge phylogeny, animal monophyly, and the , ) from the Capricorn- origin of the nervous system: 18S rRNA evidence. Bunker Group of Islands, Great Barrier , Austra- Canadian Journal of , 74, 2031–2045. lia: target species for the ‘sponge genome project’, CHOMBARD, C., BOURY-ESNAULT, N., TILLIER, A. & Zootaxa, 1314, 31–39. VACELET, J. 1997. of ‘sclerosponges’ (Por- JANUSSEN, D., STEINER, M. & MAOYAN, Z. 2002. New ifera, Demospongiae) supported by 28S sequences. well-preserved scleritomes of Chancelloridae from the Biological Bulletin, 193, 359–367. Early Cambrian Yuanshan Formation (Chengjiang, CLAPHAM, M. E. & NARBONNE, G. M. 2002. Ediacaran China) and the Middle Cambrian Wheeler Shale epifaunal tiering. Geology, 30, 627–630. (Utah, USA) and palaeobiological implications. CLAPHAM, M. E., NARBONNE, G. M. & GEHLING, J. G. Journal of , 76, 596–606. 2003. Paleoecology of the oldest known animal com- JENNER, R. A. 2004. When molecules and morphology munities: Ediacaran assemblages at Mistaken Point, clash: reconciling conflicting phylogenies of the Newfoundland. Paleobiology, 29, 527–544. Metazoa by considering secondary character loss. COLLINS, A. G. 1998. Evaluating multiple alternative Evolution and Development, 6, 372–378. hypotheses for the origin of : an analysis of KRUSE, M., LEYS, S. P., MU¨ LLER, I. M. & MU¨ LLER, 18S rRNA molecular evidence. Proceedings of the W. E. G. 1998. Phylogenetic position of the Hexacti- National Academy of Sciences, USA, 95, 15458–15463. nellida within the phylum Porifera based on the CONDON, D., ZHU, M., BOWRING, S., WANG, W., amino acid sequence of the protein kinase C from YANG, A. & JIN, Y. 2005. U–Pb ages from the Neo- Rhabdocalyptus dawsoni. Journal of Molecular Evol- proterozoic Doushantuo Formation, China. Science, ution, 46, 721–728. 308, 95–98. KRUSE, P. D. & DEBRENNE, F. 1989. Review of archae- CONWAY MORRIS, S. & CHAPMAN, A. J. 1997. Lower ocyath microstructure. Memoirs of the Association of Cambrian and other Coeloscleritophorans Australasian Palaeontologists, 8, 133–141. from Aksu-Wushi, Xinjiang, China. Journal of LE GUERROUE, E., ALLEN, P. A., COZZI, A., ETIENNE, Paleontology, 71, 6–22. J. L. & FANNING, M. 2006a. 50 Myr recover from the 13 DONOGHUE, M. J. 2005. Key innovations, convergence, largest d C excursion in the Ediacaran ocean. Terra and success: macroevolutionary lessons from plant Nova, 18, 147–153. phylogeny. Paleobiology, 31, Supplement, 77–93. LE GUERROUE, E., ALLEN, P. A. & COZZI, A. 2006b. EERNISSE, D. J. & PETERSON, K. J. 2004. The history of Chemostratigraphic and sedimentological framework animals. In:CRACRAFT, J. & DONOGHUE, M. J. (eds) of the largest negative carbon isotope excursion in Assembling the Tree of Life. Oxford University Press, Earth history: The Neoproterozoic Shuram Formation Oxford, 197–208. (Nafun Group, Oman). Precambrian Research, 146, FARQUHAR, J., BAO, H. M. & THIEMENS, M. 2000. 68–92. Atmospheric influence of Earth’s earliest sulphur LEYS, S. P. 2003. Comparative study of spiculogenesis cycle. Science, 289, 756–758. in demosponge and larvae. Microscopy FIKE, D., GROTZINGER, J., PRATT, L. & SUMMONS, R. Research and Technique, 62, 300–311. 2006. Oxidation of the Ediacaran ocean. Nature, LOGAN, G. A., HAYES, J. M., HIESHIMA, G. B. & 444, 744–747. SUMMONS, R. E. 1995. Terminal Proterozoic GEHLING, J. G. 1991. The case for Ediacaran fossil roots reorganization of biogeochemical cycles. Nature, to the metazoan tree: Geological Society of India 376, 53–56. Memoir, 20, 181–224. LOVE, G. D. & FIKE, D. A. ET AL. 2006. Constraining the GEHLING, J. G., NARBONNE, G. M. & ANDERSON, timing of basal metazoan radiation using molecular M. M. 2000. The first named Ediacaran body fossil, biomarkers and U–Pb isotope dating. Geochimica et Aspidella terranovica. Palaeontology, 43, 427–456. Cosmochimica Acta, 70, A371. PORIFERAN PARAPHYLY 367

MALDONADO, M. 2004. Choanoflagellates, choanocytes, from the end- extinction. Science, 305, and animal multicellularity. Invertebrate Biology, 506–509. 123, 1–22. PETERSON, K. J. & BUTTERFIELD, N. J. 2005. Origin of MALOOF, A. C., SCHARG, D. P., CROWLEY, J. L. & the Eumetazoa: Testing ecological predictions of mol- BROWRING, S. A. 2005. An expanded record of ecular clocks against the Proterozoic fossil record. Early Cambrian carbon cycling from the Anti-Atlas Proceedings of the National Academy of Sciences, Margin, Morocco. Canadian Journal of Earth USA, 102, 9547–9552. Science, 42, 2195–2216. PETERSON, K. J. & EERNISSE, D. J. 2001. Animal phylo- MANUEL, M., BORCHIELLINI, C., ALIVON, E., LE geny and the ancestry of bilaterians: inferences from PARCO, Y., VACELET, J. & BOURY-ESNAULT,N. morphology and 18S rDNA gene sequences. Evolution 2003. Phylogeny and evolution of calcareous and Development, 3, 170–205. sponges: monophyly of Calcinea and Calcaronea, high PETERSON, K. J., WAGGONER, B. & HAGADORN, J. W. level of morphological homoplasy, and the primitive 2003. A fungal analog for Newfoundland Ediacaran nature of axial symmetry. Systematic Biology, 52, fossils. Integrative and Comparative Biology, 43, 311–333. 127–136. MCCAFFREY, M. A., MOLDOWAN, J. M., LIPTON, P. A., PETERSON, K. J., LYONS, J. B., NOWAK, K. S., TAKACS, SUMMONS, R. E., PETERS, K. E., JEGANATHAN, A. & C. M., WARGO, M. J. & MCPEEK, M. A. 2004. Esti- WATT, D. S. 1994. Paleoenvironmental implications metazoan divergence times with a molecular of novel C30 steranes in Precambrian to Cenozoic clock. Proceedings of the National Academy of age petroleum and bitumen. Geochimica et Cosmochi- Sciences, USA, 101, 6536–6541. mica Acta, 58, 529–532. PETERSON, K. J., MCPEEK, M. A. & EVANS, D. A. D. MEDINA, M., COLLINS, A. G., SILBERMAN, J. D. & 2005. Tempo and mode of early animal evolution: SOGIN, M. L. 2001. Evaluating hypotheses of basal inferences from rocks, Hox, and molecular clocks. animal phylogeny using complete sequences of large Paleobiology, 31, Supplement, 36–55. and small subunit rRNA. Proceedings of the National RANDALL, R. D., LIEBERMAN, B. S., HASIOTIS, S. T. & Academy of Sciences, USA, 98, 9707–9712. POPE, M. C. 2005. New chancelloriids from the Early MEHL, D. 1996. Organization and microstructure of Cambrian Sekwi Formation with a comment on chan- the chancellorid skeleton: implications for the celloriid affinities. Journal of Paleontology, 79, of the Chancelloridae. Bulletin 987–996. de l’Institut Oceanographique de Monaco, 14, REITNER, J. & MEHL, D. 1996. Monophyly of the Pori- 377–385. fera. Verhandlungen des naturwissenschaftlichen MORRIS, P. J. 1993. The developmental role of the extra- Vereins in Hamburg (Neue Folgue), 36, 5–32. cellular matrix suggests a monophyletic origin of the RONQUIST, F. & HUELSENBECK, J. P. 2003. MrBayes 3: kingdom Animalia. Evolution, 47, 152–165. Bayesian phylogenetic inference under mixed models. MURICY, G. & DIAZ, D. 2002. Order Homoscleromorpha Bioinformatics, 19, 1572–1574. Dendy, 1905, Plakinidae Schluze, 1880. In: RONQUIST, F., HUELSENBECK, J. P. & VAN DER MARK, HOOPER, J. N. A. & VAN SOEST, R. W. M. (eds) P. 2005. MrBayes. Distributed by the authors. Systema Porifera: A Guide to the Classification of ROTHMAN, D. H., HAYES, J. M. & SUMMONS, R. E. Sponges. Kluwer Academic/Plenum Publishers, 2003. Dynamics of the Neoproterozoic carbon cycle. New York, 71–82. Proceedings of the National Academy of Sciences, NARBONNE, G. M. 1998. The Ediacara-biota: a terminal USA, 100, 8124–8129. Neoproterozoic experiment in the evolution of life. ROWLAND, S. M. 2001. Archaeocyaths—a history of phy- Geological Society of America Today, 8, 1–6. logenetic interpretation. Journal of Paleontology, 75, NARBONNE, G. M. 2004. Modular construction of early 1065–1078. Ediacaran complex life forms. Science, 305, RUNNEGAR, B. 1982. The : animals 1141–1144. or fossils? Journal of the Geological Society of NARBONNE, G. M. 2005. The Ediacara biota: Neoprotero- Australia, 29, 395–411. zoic origin of animals and their ecosystems. Annual SAVARESE, M. 1992. Functional analysis of archaeo- Review of Earth and Planetary Sciences, 33, 421–442. cyathan skeletal morphology and its palaeobiological NARBONNE, G. M. & GEHLING, J. G. 2003. Life after implications. Paleobiology, 18, 464–480. snowball: the oldest complex Ediacaran fossils. SEILACHER, A., GRAZHDANKI, D. & LEGOUTA, A. Geology, 31, 27–30. 2003. Ediacaran biota: The dawn of animal life in NICHOLS, S. A. 2005. An evaluation of support for order- the shadow of giant . Paleontological level monophyly and interrelationships within the Research, 7, 43–54. Demospongiae using partial data from the large SEMPERE, L. F., COLE, C. N., MCPEEK, M. A. & subunit rDNA and cytochrome oxidase subunit I. Mol- PETERSON, K. J. 2006. The phylogenetic distribution ecular Phylogenetics and Evolution, 34, 81–96. of metazoan microRNAs: insights into evolutionary NIELSEN, C. 2001. Animal Evolution: Interrelationships complexity and constraint. Journal of Experimental of the Living Phyla (2nd edn). Oxford University Zoology (Mol Dev Evol), 306B, 576–588. Press, Oxford. VACELET, J. & BOURY-ESNAULT, N. 1995. Carnivorous PAYNE, J. L., LEHRMANN, D. J., WEI, J., ORCHARD, sponges. Nature, 373, 333–335. M. J., SCHRAG, D. P. & KNOLL, A. H. 2004. Large VACELET, J. & DUPORT, E. 2004. Prey capture and diges- perturbations of the carbon cycle during recovery tion in the carnivorous sponge hypogea 368 E. A. SPERLING ET AL.

(Porifera: Demospongiae). Zoomorphology, 123, Newfoundland. Canadian Journal of Earth Sciences, 179–190. 40, 1375–1391. WANG,X.&LAVROV, D. V. 2007. Mitochondrial genome WOOLLACOTT, R. M. & PINTO, R. L. 1995. Flagellar of the homoscleromorph Oscarella carmela (Porifera, basal apparatus and its utility in phylogenetic Demospongiae) reveals unexpected complexity in the analyses of the Porifera. Journal of Morphology, common ancestor of sponges and other animals. Mol- 226, 247–265. ecular Biology and Evolution, 24, 363–373. YAHEL, G., SHARP, J. H., MARIE, D. & GENIN, A. 2003. WALCOTT, C. D. 1920. Middle Cambrian Spongiae. In situ feeding and element removal in the symbiont- Smithsonian Miscellaneous Collections, 67, 261–364. bearing sponge Theonella swinhoei: bulk DOC is the WALLBERG, A., THOLLESSON, M., FARRIS, J. S. & major source for carbon. Limnology and Oceanogra- JONDELIUS, U. 2004. The phylogenetic position of phy, 48, 141–149. the comb jellies (Ctenophora) and the importance of YUAN, X., XIAO, S., PARSLEY, R. L., ZHOU, C., CHEN, taxonomic sampling. , 20, 558–578. Z. & HU, J. 2002. Towering sponges in an Early Cam- WOOD, R. A., ZHURAVLEV, A. Y. & DEBRENNE, F. brian Lagerstatte: disparity between nonbilaterian and 1992. Functional biology and ecology of Archaeo- bilaterian epifaunal tierers at the Neoproterozoic- cyatha. Palaios, 7, 131–156. Cambrian transition. Geology, 30, 363–366. WOOD, D. A., DALRYMPLE, R. W., NARBONNE, G. M., ZRZAVY, J., MIHULKA, S., KEPKA, P., BEZDEK, A. & GEHLING, J. G. & CLAPHAM, M. E. 2003. Paleoenvir- TIETZ, D. 1998. Phylogeny of the Metazoa based on onmental analysis of the late Neoproterozoic Mistaken morphological and 18S ribosomal DNA evidence. Point and Trepassey formations, southeastern Cladistics, 14, 249–285.