Calcareous sponges (Phylum Porifera, Class Calcarea) are known to be taxonomically difficult. Previous molecular studies have revealed many discrepancies between classically recognized taxa and the observed relationships at the order, family and genus levels; these inconsistencies question underlying hypotheses regarding the evolution of certain morphological characters. Therefore, we extended the available taxa and character set by sequencing the complete small subunit (SSU) rDNA and the almost complete large subunit (LSU) rDNA of additional key species and complemented this dataset by substantially increasing the length of available LSU sequences. Phylogenetic analyses provided new hypotheses about the relationships of Calcarea and about the evolution of certain morphological characters. We tested our phylogeny against competing phylogenetic hypotheses presented by previous classification systems. Our data reject the current order-level classification by again finding non-monophyletic Leucosolenida, Clathrinida and Murrayonida. In the subclass Calcinea, we recovered a clade that includes all species with a cortex, which is largely consistent with the previously proposed order Leucettida. Other orders that had been rejected in the current system were not found, but could not be rejected in our tests either. We found several additional families and genera polyphyletic: the families Leucascidae and Leucaltidae and the genus Leucetta in Calcinea, and in Calcaronea the family Amphoriscidae and the genus Ute. Our phylogeny also provided support for the vaguely suspected close relationship of several members of Grantiidae with giantortical diactines to members of Heteropiidae. Similarly, our analyses revealed several unexpected affinities, such as a sister group relationship between Leucettusa (Leucaltidae) and Leucettidae and between Leucascandra (Jenkinidae) and Sycon carteri (Sycettidae). According to our results, the taxonomy of Calcarea is in desperate need of a thorough revision, which cannot be achieved by considering morphology alone or relying on a taxon sampling based on the current classification below the subclass level.
Citation: Voigt O, Wülfing E, Wörheide G (2012) Molecular Phylogenetic Evaluation of Classification and Scenarios of Character Evolution in Calcareous Sponges (Porifera, Class Calcarea). PLoS ONE 7(3): e33417. doi:10.1371/journal.pone.0033417
Editor: Dirk Steinke, Biodiversity Insitute of Ontario - University of Guelph, Canada
Received: December 13, 2011; Accepted: February 8, 2012; Published: March 27, 2012
Copyright: © 2012 Voigt et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by the German Science Foundation (DFG), through grants Wo896/3 and Wo896/6 (SPP 1174 “Deep Metazoan Phylogeny”) to GW. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Calcarea Bowerbank, 1864 is one of the four currently recognized classes of Porifera , . Its relationship to the other main sponge classes, i.e., Demospongiae Sollas, 1885, Hexactinellida Schmidt, 1870 and Homoscleromorpha Bergquist, 1978, has long been unclear, especially because molecular analyses have questioned the monophyly of Porifera. However, the hypothesis of sponge paraphyly was often poorly supported by the data or was hampered by the lack of representatives of what we today refer to as the four sponge classes . More recently, phylogenomic studies found high support for sponge monophyly and a sister group relationship between Calcarea and Homoscleromorpha , .
Calcareous sponges comprise approximately 675 accepted extant species ; therefore, this class is considerably less diverse than for example, the most species-rich class, Demospongiae, which contains approximately 7.000 recognized species . Calcarean species are exclusively marine animals. Most species occur at shallow depths and only few species are known from the deep sea (for an overview see e.g., ). In contrast to other sponges, which have siliceous spicules, all calcareous sponges build calcite spicules, which constitute a synapomorphy of the group . Within Calcarea, the relationships are little understood. However, this small group of sponges has long been of interest to zoologists because of the variety of simple and more complex organization forms found in extant species, and also because of their apparent beauty caused by the occasionally geometrical arrangement of their skeletons. Among the first to be fascinated by the organizational diversity in Calcarea was Haeckel, who for this reason focussed on this group to establish ‘a natural system’ of Calcarea in his monograph ‘Die Kalkschwämme’ – to promote the emerging ideas of Darwinism. Since the days of Haeckel, the most important characters used in the taxonomy of Calcarea are the organization of the aquiferous system and skeletal features.
The aquiferous system of Calcarea
Sponges are filter feeders that create a unidirectional water current through their bodies by the beating central flagella of specialized cells, the choanocytes. The microvilli collar of a choanocyte captures food particles, which are taken up by the cell. Traditionally, four different types of aquiferous systems can be readily distinguished in Calcarea: the asconoid, syconoid, sylleibid and leuconoid grades of organisation. Recently, a fifth type was described, the solenoid system . In asconoid sponges, all internal cavities of the sponge are lined by choanocytes (Fig. 1, A). Such sponges are also called homocoel. All other organization forms of the aquiferous system are heterocoel, i.e., some parts of the internal cavities are lined by pinacocytes, a cell type that also covers the external surfaces of sponges. In syconoid Calcarea, choanocytes are organized in elongated chambers, which are radially arranged around the atrium; water enters the sponge via inhalant canals and the choanocyte chambers via pores. The choanocyte chamber opens to the central cavity (the atrium), which is lined by pinacocytes (Fig. 1, B). Sylleibid sponges have radially arranged choanocyte chambers, which do not open directly into the atrium. Instead, several choanocyte chambers open into a cavity lined by pinacocytes, which itself has an opening to the atrium (Fig. 1, C). The most complex organization of the aquiferous system is the leuconoid grade. Here, water enters the sponge through a system of inhalant canals leading to numerous, approximately spherical choanocyte chambers. These chambers open to exhalant canals, through which the water reaches the atrium (Fig. 1, D). In some leuconoid species, the atrium is strongly reduced. Recently, Cavalcanti and Klautau  introduced the new term solenoid to describe the organisation of the aquiferous system found in the genus Leucascus and to stress the differences with syconoid sponges. The solenoid aquiferous system is characterized by choanocytes, which are restricted to anastomosed tubes, and pinacocytes, which line the atrial cavity. Accordingly, solenoid sponges are also heterocoel. Calcarea is the only sponge class in which all of these different types of aquiferous systems are present.
Figure 1. Different organizations of the aquiferous system in Calcarea.
A: asconoid (Soleneiscus radovani); B: syconoid (Sycon coronatum, from Helgoland, Germany); C: sylleibid (Grantiopsis cylindrica); D: leuconoid (Leucettusa sp. 1). Thin arrows illustrate the direction of water flow in A, B and C. atr = atrium; chc = choanocyte chambers; cx = cortex; eh = exhalant channel; ext: exterior of the sponge; ih = inhalant channel; spt = spicule tract of modified triactines.doi:10.1371/journal.pone.0033417.g001
Skeletal arrangement in Calcarea
Spicule morphology in Calcarea is rather limited compared to the occasionally very elaborate siliceous spicules of other sponges (e.g., Hexactinellida, ). In extant Calcarea, spicules can be categorized in just three types, i.e., diactines, triactines and tetractines, depending on the number of growing rays of the spicule. Pentactines were also reported but are only known from one species, Sycon pentactinale . Modifications of these spicule types can occur. However, in most cases, the arrangement of spicules in the skeleton (in combination with the nature of the aquiferous system) has been considered more phylogenetically informative for higher classifications than the form of the spicules itself.
In the simplest calcareans, the skeleton consists of only one layer of spicules, which supports the pinacoderm on the outer side, and the choanoderm on the inner side of the sponge (e.g., Fig. 1, A). More complex skeletons can be divided into an atrial skeleton, which supports the wall of the atrial cavity, and a choanoskeleton, which supports the choanosome. In sponges with thin walls, the choanosome is only supported by unpaired actines of subatrial spicules and, depending on the sponge, also by actines of subcortical or cortical spicules (Fig. 2, A). Such a skeletal organization is termed inarticulated choanoskeleton. In contrast, articulated choanoskeletons are built from several, roughly parallel rows of similar spicules, usually sagittal triactines with their unpaired actines pointing to the outside of the sponge (Fig. 2, B). With such an arrangement, sponges can build thick walls. The choanosome of thick walled sponges can also be supported by numerous spicules, without apparent order, or by spicular tracts of modified triactines. Reinforced skeletons can be formed by fused (occasionally modified) spicules or by an aspicular calcite mass. A tangential layer of spicules that covers the external surface of the sponge is called the cortex. It can be thin, formed by a single layer of spicules, or thick, and occasionally primarily sustain the sponge wall (Fig. 1, C & D).
Figure 2. Organization of choanoskeleton.
A: inarticulated (Sycettusa aff. hastifera); B: articulated (Grantessa sp. GW974).doi:10.1371/journal.pone.0033417.g002
Classification of Calcarea.
Manuel  offers a short overview of the history of the classification of Calcarea. In the current classification, Calcarea is subdivided into two subclasses, Calcinea and Calcaronea , , . This concept was originally based on a cellular character, the position of the nucleus within the choanocytes, first proposed to separate some asconoid species , . The subclass division has found additional support by several independent characters, such as different larvae types and distinctive development in both subclasses (coeloblastula in Calcinea, amphiblastula with special development in Calcaronea, see ). Furthermore, the two subclasses can be distinguished by different ratios of isotopes incorporated into the spicules during bio-mineralization  and by the analyses of small subunit (SSU) and partial large subunit (LSU) ribosomal RNA genes (rDNA) , , .
Hartman corroborated the Calcinea-Calcaronea concept and provided an order-level taxonomy . In Calcinea, he proposed three orders: (1) Clathrinida for homocoel Calcinea without a cortex, (2) Leucettida for heterocoel Calcinea with cortex or dermal membrane and (3) Pharetronida Zittel 1878 for leuconoid Calcarea with a reinforced skeleton of fused spicules or formed by an aspicular network. However, Vacelet  showed that some members of Pharetronida belong to Calcaronea, whereas others belong to Calcinea, for which he proposed the orders Lithonida and Murrayonida, respectively . In the latest revision of Calcinea , Leucettida was rejected, because the transition from simple homocoel Calcinea to heterocoel Calcinea was interpreted to have occurred independently several times and in different evolutionary pathways , . All Calcinea with free spicules were placed in the order Clathrinida . It has to be noted, that none of the proposed independent evolutionary lines in Calcinea were based on phylogenetic analyses, and as such they are debatable despite being presented in a logical and convincing way.
Within Calcaronea, Hartman  placed homocoel sponges without cortex and dermal membrane in his order Leucosolenida and heterocoel calcaroneans in the order Sycettida Bidder 1898. The current order level taxonomy  differs not only by containing the order Lithonida but also by proposing a new order, Baerida, for Calcaronea with skeletons formed exclusively or in substantial parts by microdiactines . Sycettida was rejected, and its species (with the exception of Baerida) were included in Leucosolenida .
Phylogenies based on morphological and DNA data
The first comprehensive phylogenetic analysis of morphological characters was performed by Manuel, et al. , who showed that little phylogenetic information is present and suggested that morphological characters contain a high level of homoplasy. The analyses of ribosomal rRNA genes found strong support for the monophyly of the two subclasses Calcinea and Calcaronea, but also revealed that many of the classically recognized taxa at the order, family and genus levels were not monophyletic, suggesting that these taxa are artificial groupings , , . Unfortunately, the morphological evolution of the aquiferous system and the skeletal arrangements is difficult to understand considering the phylogenetic hypotheses obtained from molecular phylogenies.
To clarify the evolution of this group of sponges, we included several additional critical taxa in our analyses, especially from members of the families Leucaltidae (Calcinea), Grantiidae and Heteropiidae (both Calcaronea) and analysed a concatenated dataset of the complete SSU rDNA and nearly the complete LSU rDNA. For the latter, previously available nucleotide positions of partial LSU rDNA  were substantially increased. We analyzed our data under various models of RNA specific substitution models and used the resulting phylogenies to evaluate different hypotheses.
Materials and Methods
Sample collection and species identification
Calcareous sponge specimens were collected in the Red Sea (Gulf of Aqaba), with kind permission from the Egyptian Environmental Affairs Agency (EEAA), and in the Great Barrier Reef, with kind permission from the Great Barrier Reef Marine Park Authority (Permit nos G98/142, G98/022, G00/638, G06/16547.1). Additional specimens were obtained from museum collections (Tables 1, 2). To determine the sponges we examined the skeletal arrangements and the nature of the aquiferous system in thin sections, which were prepared as follows.
Table 1. Included specimens of Calcinea, their sample localities and GenBank accession numbers.doi:10.1371/journal.pone.0033417.t001
Table 2. Included specimens of Calcaronea, their sample localities and GenBank accession numbers.doi:10.1371/journal.pone.0033417.t002
Parts of the sponges preserved in 70–96% ethanol (EtOH) were gradually transferred to 30% EtOH in water over a dilution series (70%, 50%, 30% EtOH). Tissues were then stained overnight in a 30% EtOH-fuchsine solution. The stained tissue was dehydrated in a dilution series (50%, 70%, 90%, 99% EtOH-fuchsine-solution). For embedding, the EtOH-fuchsine solution was gradually replaced with LRwhite resin (in dilution steps of 33%, 50%, 66%, 100% LRwhite, all at 4°C to prevent polymerization; the last step had an overnight incubation). For final embedding, LRwhite was exchanged, and after one hour of incubation at 45°C polymerization was induced at 60°C overnight. From the resulting block, we took sections of suitable thickness (10–500 µm; starting with a 200 µm section) from the block with a Leica 1600 saw microtome (Leica, Nußloch, Germany). To stain the cells and nuclei on the surface of the section, we suspended the section for 1:30 min to a 30% EtOH-Touledein blue and 30% basic fuchsine solution; then, we immediately washed off the dye with water. Dried and stained sections were mounted on microscopic slides with Eukitt (Fluka). Spicules were obtained either from the lysis step from the DNA extraction (see below) or by dissolution of tissue with sodium hypochlorite. The obtained spicules were washed five times with water and transferred to a microscopic slide, dried, and mounted with Eukitt. Sections and spicule preparations were observed and documented on a Zeiss Axiolab Microscope equipped with a Canon PowerShot G2 digital camera. The identification of calcarean genera followed available keys . When possible, species were identified by comparing original descriptions to our specimens. Habitus and sections of these newly included specimens are included here or in Figs. S1 and S2.
The re-examination of two specimens included in a previous study leads us to the conclusion that the two specimens were incorrectly determined at the generic level. The specimen QM G313824 was previously considered to be Clathrina cerebrum. We find that the specimen belongs to the genus Ascaltis, because it possesses a (thin) cortex and a large central cavity. Another specimen (QM G316285) was previously identified as Aphroceras sp. Although this specimen does possess larger diactines, they are not longitudinally arranged and do not support the cortex as in Aphroceras. We identified this specimen as Leucandra sp.
The identification of another specimen (SAM-PS0349) was also problematic. It clearly belongs to Grantiidae and possesses large longitudinal diactines that support the cortex and the atrial skeleton. This arrangement is typical of members of the genus Amphiute . But although a syconoid aquiferous system is a diagnostic character for this genus, our specimen shows a leuconoid organization. Therefore, we decided to classify it as Aphroceras sp. Aphroceras is defined by longitudinal diactines in the cortex and a leuconoid aquiferous system , thus its diagnosis does not explicitly exclude the presence of diactines associated with the atrial skeleton.
DNA extraction, PCR, sequencing and alignment
DNA was extracted with the DNeasy tissue kit (QIAGEN) or by standard phenol-chloroform extraction. Template DNA was used in dilutions of 1:1 to 1:500 in PCR reactions, depending on the DNA quantity and quality. Because many of the samples from museum collections yielded only highly degraded DNA, it was necessary to amplify SSU rDNA and LSU rDNA in two and up to five smaller fragments, respectively. PCRs were conducted with the BioTaq (BioLine) as previously described for SSU rRNA  and for LSU rRNA with different combinations of the primers, which are given in Table S1. Purified PCR products were sequenced after cycle sequencing with BigDye Terminator.3.1 (Applied Biosystems) on an ABI 3100 capillary sequencer (Applied Biosystems). Consensus sequences were created in CodonCode Aligner (http://codoncode.com) and submitted to GenBank (SSU: JQ272310–JQ272324; LSU: JQ272256–JQ272309, see Tables 1, 2.). Occasionally, it was not possible to amplify all SSU or LSU fragments for a given sample or the sequences of different fragments did not overlap. In such cases, we combined the sequences by aligning them to the most similar full LSU rRNA sequence, and recoded the missing parts as gaps.
Additional SSU rDNA and LSU rDNA sequences from Calcarea and 41 outgroup taxa were downloaded from GenBank (http://www.ncbi.nlm.nih.gov/; Tables 1, 2 for Calcarea and Table S2 for the outgroup taxa). Outgroup sequences were only included, when both SSU and LSU sequences were available in almost full length (with the exception of hexactinellid 28S sequences due to limited availability). We aligned the sequences in Seaview , taking into account secondary structure information (18S: ; 28S: ). The considered LSU rRNA secondary structure of a typical calcinean sequence is provided in Fig. S3. For our analyses, the C-Domain in LSU was excluded for the outgroup taxa and coded as ‘gaps’ in the alignment, because the homology of sites among all taxa could not be established with certainty. Thus, it was possible to keep calcarean sites of this highly variable region in the analyses. Further sites of uncertain homology were removed from our alignment, and custom-made PERL scripts  were used to generate input files that included secondary structure information suitable for PHASE ,  and RAxML v. 7.2.8 .
Most phylogenetic methods assume that characters in a data matrix evolve independently from each other, but this assumption is clearly violated in the helices of rRNA because paired nucleotides coevolve driven by the selection pressure to maintain the secondary structure, which is pivotal for rRNA function within the ribosome –. By neglecting these coevolutionary processes, phylogenetic inferences can be biased and result in suboptimal tree topologies (e.g., , –). Solutions to this problem are special evolutionary models, which instead of single bases consider the two paired bases of helices, the so-called doublet, as single characters. Such models have been shown to outperform standard 4×4 models of nucleotide evolution in analyses of rDNA data –. Several doublet models that make different assumptions about the evolution of doublets have been proposed (for a comprehensive overview see ).
In contrast to standard 4×4 models of nucleotide substitution, the paired nucleotides in an RNA helix are the single characters in doublet models. Three families of doublet models can be distinguished according to the number of recognized doublets . In 16-state models, all possible pairs are considered. The likelihood is calculated in a 16×16 matrix, resulting in a general reversible model with 119 free rate parameters and 15 free frequency parameters. Such a high number of parameters make general reversible 16-state models impractical to use . Moreover, because mismatch base pairs (MM), i.e. pairs other than Watson-Crick pairs and GU/UG pairs, are rare in real RNA data, these states are pooled into one class (MM) in 7-sate models, or completely ignored in 6-state models. Each model family has a number of different models, which through restrictions and assumptions reduce the number of parameters compared to the most general model. A study with a 5-taxon data set compared the models within each model family and suggested that the most general models are to be preferred over restricted ones . However, comparisons among the doublet model families are not possible because not only the model parameters but also the data matrices are different . We applied 6-, 7- and 16 state models in a likelihood framework using RAxML 7.2.8  and in Bayesian inference using the PHASE software (www.bioinf.manchester.ac.uk/resources/phase/index.html).
For our analyses, we used a concatenated dataset of SSU and LSU rDNA (4,939 positions). Previous studies with data from SSU and a smaller LSU fragment have shown that the combination of both genes lead to a finer phylogenetic resolution, compared to single gene analyses (especially with SSU DNA , ). Furthermore, SSU and LSU rRNA are parts of the ribosomal cistron, which during transcription is transcribed into one pre-rRNA before the splicing of the internal transcribed spacer regions (ITS). We partitioned the combined dataset into two partitions called stem (paired sites) and loop (unpaired sites). For an analysis with a standard 4×4 model under ML, we also applied a different partitioning scheme with one partition for SSU and one for LSU rDNA.
In PHASE, each run had a burn-in phase of 1,000,000 generations, followed by 10,000,000 sampling generations, from which every 200th tree was sampled. We used Tracer v 1.5 (http://tree.bio.ed.ac.uk/software/tracer/) to monitor the parameter sampling of each run. To transform the PHASE output files into a readable Tracer format, we modified the Perl script phase2tracer.pl from Matt Yoder (http://hymenoptera.tamu.edu/rna/download.php) to handle larger PHASE2 output files. The modified script is available on request.
In RAxML, we applied GTR models with gamma distribution to compensate for the rate heterogeneity among sites. For the stem partition, different models of each family were applied (S6A–E, S7A–E, 16A and 16B) in independent analyses using the rapid bootstrapping algorithm with 1,000 bootstrap replicates. The resulting phylogenetic trees were visualized with FigTree v.1.3.1 (http://tree.bio.ed.ac.uk/software/figtree/).
Unfortunately, no a-priori model testing software, such as jModeltest for standard models, is available yet for doublet models or partitioned datasets. Moreover, comparisons among standard models and doublet models, as well as among doublet models of different families are not possible . Following the suggestion by Savill, et al. , we choose the 7A model to discuss most of our results and to test phylogenetic hypotheses. By using this model, we did not ignore the class of mismatches as in 6-state models, nor did we assign an own character class to each of the rare mismatch doublets as in 16-state models. To illustrate model-dependent differences in the topologies, strict consensus trees for results under each family of doublet models were calculated in PAUP* 4.0b10  and are presented in Figs. S4 and S5.
We used MacClade v. 4.07  to trace the evolution of morphological characters according to our phylogenetic hypothesis from the Bayesian analysis with the 7A model.
Testing phylogenetic hypotheses
To test whether the tree topologies obtained with our data were significantly better than other phylogenetic hypotheses, we re-analyzed the dataset with RAxML and the 7A-model of nucleotide evolution under specific topology constraints of the tested taxa.
In Calcaronea, we constrained the following monophyletic taxa: (a) a clade containing Lithonida, monophyletic Baerida and monophyletic Leucosolenida (following ); (b) Leucosolenida Hartman 1958 and Sycettida Hartman 1958 with the modification that members of Lithonida were considered as Sycettida following Hartman's definition of the order ; (c) the families Amphoriscidae, Grantiidae, Heteropiidae and Jenkinidae being all monophyletic; (d–g) constraining each of these families as monophyletic: (d) Amphoriscidae; (e) Grantiidae; (f) Heteropiidae; (g) Jenkinidae.
In Calcinea, we constrained the following taxa as monophyletic: (a) monophyletic orders Clathrinida and Murrayonida ; (b) Murrayonida, Leucaltidae, (Clathrinidae+Leucascidae+Leucettidae), a scenario presented in ; (c) the order Clathrinida sensu Hartman ; and the families (d) Leucaltidae, (e) Leucascidae and (f) Clathrinidae.
ML trees under each topological constraint were obtained using RAxML (model S7A for stem regions) and the previously described settings. The resulting trees were calculated and combined with the unconstrained ML tree (S7A-model) in one file for each subclass (Calcaronea and Calcinea). RAxML was used to calculate site-specific likelihood values for these two sets of trees. Using these files, an approximately unbiased (AU) test  was performed in Consel  following the program's manual.
Topologies from ML and BI under different models
Following suggestions made by Savill, et al.  we present the tree topologies obtained with model 7A for stem regions in Fig. 3. In Calcarea, some minor differences between this model and other 7-state, 6-state and 16-state models mostly occurred in nodes without strong support in the presented topology (see Figs. S4, S5). Topologies obtained with ML under model 7A slightly differed from the Bayesian inference (insert, Fig. 3). Differences in the posterior probabilities (PP) and bootstrap support (BS) can be appreciated in Fig. 3. When standard GTR models were applied to the dataset partitioned in stem and loop, we observed a different position for Leucosolenia (occurring basal to clade LEUC I, see below), which, in the case of the Bayesian analysis (but not under ML), results in a phylogeny with a weakly supported (PP: 55) sister group relationship of Baerida and Leucosolenida (Fig. S6). When using a partitioning scheme by gene (SSU and LSU) with GTR models, the position of Leucosolenia is recovered as by the analyses with doublet models (Fig. S6).
Figure 3. Bayesian phylogeny of Calcarea inferred with the RNA7A model for partition stem.
Outgroup taxa not shown (compare Figs. S4, S5). Support values are given at the nodes (PP/BS of ML analyses under the same model). Clades are shaded and numbered for taxa that are not monophyletic. Order names are abbreviated: BAE = Baerida; CLAT = Clathrinida; LEUC = Leucosolenida; LITH = Lithonida; MUR = Murrayonida. Insert: ML topologies of two clades that differ from Bayesian inference (with BS values).doi:10.1371/journal.pone.0033417.g003
Monophyly of Calcarea and relationship to outgroup taxa
All analyses resulted in a strongly supported monophyletic Calcarea, with a subdivision into Calcinea and Calcaronea (also with high support values, Fig. 3 and Figs. S4, S5, S6). The position of Calcarea with respect to the outgroup taxa differed with the applied models and between ML and Bayesian inference; however, Calcarea was never found in a monophyletic clade Porifera. Instead, monophyletic Demospongiae and Hexactinellida were sister taxa ( = Silicea) with the homoscleomorph sponge Oscarella as sister taxon (with high support values in most cases). Several relationships of other outgroup taxa were strongly supported by PP and BS values and were found in all the analyses regardless of the applied model (e.g., the monophyly of the phyla Placozoa, Cnidaria and Ctenophora). Cnidaria and Placozoa were sister taxa. Otherwise the relationships among these phyla and their relationships to the sponge clades are not strongly supported and varied in the different analyses.
Relationships of Calcarea
In most cases, our phylogeny is compatible with the results of previous rDNA analyses , . Likewise, we found strong support for the two monophyletic subclasses Calcinea and Calcaronea. Below the subclass level, we confirm the non-monophyly of several taxa that had been previously reported as such : (i) in Calcaronea, the order Leucosolenida, the families Heteropiidae, Grantiidae, Jenkinidae Sycettidae, and the genera Sycon, Sycettusa, Leucandra; (ii) in Calcinea, the orders Clathrinida and Murrayonida, the families Clathrinidae, Leucaltidae, Leucascidae and the genus Clathrina.
Our topology could resolve some relationships that were only recovered as polytomies by Dohrmann, et al.  e.g., within Leucosolenida (Calcaronea) and Leucettidae (Calcinea). In addition, several clades found in the former study were not recovered in the analyses of our extended taxon and character set. For instance, our topology does not contain Clade H1 and clade H2 in Calcaronea nor Clade K in Calcinea (Fig. 3 in ). With the extended taxon set, new species could be placed into the phylogeny, and we uncovered additional contradictions to the classification of some taxa.
Relationships within Calcaronea
In Calcaronea, the only sampled species of Lithonida, Plectroninia neocaledoniense, is the sister taxon to a clade that comprises the other sampled Calcaronea, in which Leucosolenia is basally diverging, as previously reported . The order Baerida (clade BAER, Fig. 3) is nested within Leucosolenida, with a sister clade relationship (PP: 93, BS: 53) to a clade of several Leucosolenida (LEUC I, Fig. 3).
The clade LEUC I comprises all sampled members of Heteropiidae (occurring in two clades Heteropiidae I and II, Fig. 3), several Grantiidae (Ute aff. syconoides, Synute and Aphroceras, which are also not recovered as a clade), and two Sycon species, Sycon ciliatum and Sycon capricorn. The clade Heteropiidae I contains Sycettusa tenuis and Sycettusa cf. simplex, which form a sister group to Syconessa panicula. Two Grantessa specimens (most likely conspecific) are the sister group to all of the latter species. Aphroceras sp. (family Grantiidae) is the sister group to Heteropiidae I with very low PP support.
In Heteropiidae II, Sycettusa aff. hastifera is more closely related to Vosmaeropsis than to Sycettusa sp. Sycon ciliatum (Sycettidae) is the sister taxon to Heteropiidae II. This topology has high PP support (97–100) but is not recovered in the same way in the 7A ML analysis. In this latter analysis, the species of Heteropiidae II and Sycon ciliatum also form a highly supported clade, but here relationships among the species are recovered differently, with Sycon ciliatum nested inside a clade of Heteropiidae (Fig. 3, insert). The topology recovered with ML also finds high BS support (96–99). The relationships of the species of Heteropiidae II and Sycon ciliatum are therefore dependent on the method employed. However, both hypotheses are consistent with a close relationship between these heteropiid species and Sycon ciliatum and agree that Sycettusa sp. and Sycettusa cf. simplex do not form a monophyletic group. Because two other included species of Sycettusa are also present in the clade Heteropiidae I, the non-monophyly of the genus is out of the question, regardless of the relationships within Heteropiidae II.
The clade Grantiidae I contains a clade of three specimens of Ute aff. syconoides with Synute pulchella as sister species, but it does not include Aphroceras sp. Interestingly, all of these grantiid genera in clade LEUC I have giant longitudinal diactines in their cortex (Fig. 4). However, one additional species with this feature, Ute ampullacea, is found in LEUC II and is not closely related to Ute aff. syconoides. Grantiidae I and Sycon capricorn (Sycettidae) form a clade (with high PP but low BS support), which itself is sister group to (Heteropiidae I+Sycon ciliatum), but with low support (PP: 77; BS: 25).
Figure 4. Skeletal organization of Grantiidae of clade LEUC I.
A,B: Ute aff. syconoides (GW975) in cross section (A) and longitudinal section (B); C: Cross section of Synute pulchella; D: Cross section of Aphroceras sp. Arrows point to the giant longitudinal diactines. atr = atrium; ch = choanosome; cx = cortex, ext = exterior of the sponge.doi:10.1371/journal.pone.0033417.g004
In Baerida, Eilhardia schulzei is the sister taxon to (Petrobiona massiliana+Leuconia nivea), which results in Baeridae as a non-monophyletic clade.
In LEUC II, Grantiidae II and Lelapiidae form a clade with high support. Grantiidae II comprises Teichonopsis, Ute ampullacea and Leucandra sp. Considering that Ute aff. syconoides falls in clade LEUC I, the genus Ute is clearly not monophyletic. The genus Leucandra is also paraphyletic because Leucandra nicolae and Leucandra aspera are neither in a close relationship to each other nor to Leucandra sp. Within the remaining taxa of clade LEUC II, Jenkinidae (Anamixilla and Leucascandra), Amphoriscidae (Leucilla and Paraleucilla) and additional taxa of Sycettidae (genus Sycon) and Grantiidae (Leucandra, Grantia) are clearly all non-monophyletic. While Sycon raphanus and Sycon cf. villosum are sister taxa, Sycon carteri is most closely related to Leucascandra caveolata form the family Jenkinidae. However, Anamixilla toressi, the only other included species of Jenkinidae in the dataset, is more closely related to Leucilla (Amphoriscidae), Leucandra aspera (Grantiidae) and the previously mentioned Sycon raphanus and Sycon cf. villosum. Grantia compressa is the sister taxon to the clade including the latter species (PP 98, BS: 57). Leucandra nicolae and Paraleucilla sp. form a highly supported clade, but the position of this clade, as shown in Fig. 3, finds only low support from the data (PP: 59, BS: 27).
With the presented relationships, many classically recognized taxa of Calcaronea are not monophyletic: the order Leucosolenida; the families Heteropiidae, Grantiidae, Jenkinidae Sycettidae, Amphoriscidae and Baeridae; and the genera Sycon, Sycettusa, Leucandra and Ute.
Testing phylogenetic hypotheses in Calcaronea.
Our phylogenetic test (Table 3) shows that the classification of Calcaronea into the three monophyletic orders Lithonida, Baerida and Leucosolenida was not supported by our data. The same occurred with the tested families Amphoriscidae, Grantiidae, Heteropiidae and Jenkinidae. The only hypothesis that could not be rejected was the taxonomic scheme of Hartman 1958, which separates homocoel and heterocoel Calcaronea into his orders Leucosolenida and Sycettida.
Table 3. p-values for the approximately unbiased test  for different topological constrains in Calcaronea.doi:10.1371/journal.pone.0033417.t003
Relationships within Calcinea
In the subclass Calcinea, the order Murrayonida, represented by Murrayona phanolepis and Lelapiella incrustans, is not monophyletic and both species are nested in Clathrinida (Murrayona phanolepis forms a low supported clade with Leucascus sp., and Lelapiella is the sister group to clade CLAT I, see below).
At the base of Calcinea, the relationships presented in Fig. 3 did not find high support (PP of 74 and BS <50). As such, the position of the root within Calcinea remains uncertain. However, Bayesian and ML trees obtained with the 7A model resulted in the same topology. Accordingly, two Clathrina species branched off subsequently (Clathrina sp. GW975 and Clathrina adusta), followed by clade Clathrinidae I (or CLAT II, Fig. 3), which comprises four additional Clathrina species (C. helveola, C. wistariensis, C. luteoculcitella, C. cerebrum) and Guancha sp. Next, a clade containing an additional Clathrina species and members of Ascandra, Levinella and Soleneiscus branched off. All of these taxa share an asconoid aquiferous system, i.e., they are homocoel, and lack a cortex.
More resolution is present in the remaining Calcinea, which are all characterized by the possession of a cortex and, with the exception of Ascaltis sp., they are heterocoel with syconoid, leuconoid or solenoid aquiferous systems. These species form a strongly supported clade, including the mentioned members of Murrayonida. In this clade, Leucettidae is monophyletic, while the genus Leucetta is not. Leucetta microraphis and Leucetta sp. are more closely related to Pericharax heteroraphis than to the clade of Leucetta chagosensis and Leucetta villosa. This relationship finds high PP and BS support.
Compared to previous studies, we included additional taxa from two genera of the family Leucaltidae (Order Clathrinida): Ascandra sp. and three specimens representing two undetermined species of the genus Leucettusa. None of the genera are closely related to each other or to the other included species of Leucaltidae, Leucaltis clathria; therefore, the family is polyphyletic. Ascandra is associated with Soleneiscus (Soleneiscidae) and Levinella (Levinellidae); thus, it is closely related to other taxa with an asconoid grade of organization. In contrast, the Leucettusa species form a monophyletic sister group to Leucettidae and, together with this latter family, a sister clade to another clade formed by Leucaltis clathria with Ascaltis sp. (Leucascidae). The position of Ascaltis and Leucascus (with Murrayona, see above) suggests that Leucascidae is not monophyletic.
Our phylogeny shows several non-monophyletic taxa: the orders Clathrinida and Murrayonida, the families Clathrinidae, Leucaltidae, Leucascidae and the genera Clathrina and Leucetta.
Testing phylogenetic hypotheses in Calcinea.
The results of the AU test are presented in Table 4. The separation of Calcinea into the orders Clathrinida and Murrayonida was rejected according our data. Moreover, a subdivision into three monophyletic lineages (i) Murrayonida, (ii) Leucaltidae and (iii) (Clathrinidae+Leucascidae+Leucettidae), needed to be disregarded. These lineages were supposed to have independently gained a more complex aquiferous system from asconoid ancestors according to Borojevic, et al. . Likewise, the monophyly of Leucaltidae alone had to be rejected. The contrasting scheme of Hartman , which classified homocoel Calcinea into one order (Clathrinida sensu Hartman), was not recovered in our ML and Bayesian analyses but could not be excluded as a possible scenario from our dataset. Similarly, a tree topology with monophyletic Leucascidae or monophyletic Clathrinidae cannot be omitted according to our AU test.
Table 4. p-values for the approximately unbiased test  for different topological constrains in Calcinea.doi:10.1371/journal.pone.0033417.t004
Evolution of morphological characters.
Figure 5. Evolution of morphological characters.
A: Organization of the aquiferous system; B: Evolution of a cortex. Tree topology identical to Fig. 3.doi:10.1371/journal.pone.0033417.g005
In Calcaronea, leuconoid aquiferous systems have evolved several times independently from ancestral syconoid stages. The cortex was lost several times in the polyphyletic Sycon species and Syconessa. A subcortical or cortical layer of pseudosagittal spicules, a diagnostic character for the family Heteropiidae, was reconstructed to have evolved two times independently from articulated ancestors (Fig. S7). However, after collapsing nodes with less than 90% PP support, it is also possible that it evolved once and in this case was lost in Sycon ciliatum (Fig. S7, A). Inarticulated choanoskeletons evolved several times independently from ancestors with articulated choanoskeletons in this subclass (Fig. S7, B).
In Calcinea, the basally diverging clades are asconoid and lack a cortex. Interestingly, the acquisition of the leuconoid aquiferous systems and a cortex occurred only once at the same node according to our reconstruction (Fig. 5). The asconoid aquiferous system of Ascaltis has to be interpreted as a secondary reorganization of the aquiferous system. Furthermore, the solenoid Leucascus seems to descend from leuconoid ancestors.
Polyphyly of Leucaltidae and Murrayonida and implications for morphological evolution in Calcinea
Our obtained phylogeny and our phylogenetic tests contradict the classification and scenarios of the evolution of morphological characters in Calcinea, which have been suggested before by Borojevic and coworkers . In our phylogenetic tree, the polyphyly of Leucaltidae has broad implications for the classification of Calcinea. Borojevic, et al.  rejected Hartman's subclass-level subdivision into homocoel, cortex-lacking Calcinea (his Clathrinida) and Leucettida (heterocoel Calcinea with a cortex or dermal membrane) and suggested that a cortex and the heterocoel organization of the aquiferous system evolved independently in different lineages. One of these lineages was Leucaltidae, in which, according to these scenarios, a more complex aquiferous system evolved by infolding of the choanoderm (as observed in Ascandra) and a cortex developed by formation of a secondary atrial skeleton (as present in Leucaltis and Leucettusa) . Accordingly, a different lineage of Calcinea evolved a cortex and a complex aquiferous system from homocoel ancestors with a clathrinoid organization (cormus of branching and anastomosed tubes as the ones observed in Clathrina and Guancha), through the formation of a cortex (organization as in Ascaltis) to heterocoel sponges with solenoid to leuconoid aquiferous systems (i.e., Leucascidae and Leucettidae) . A third independent gain of the cortex and heterocoel organization was suggested in the order Murrayonida .
Our data exclude the monophyly of each of these groups. We found homocoel species branching off first in Calcinea. In addition, Ascandra was closely related to the other homocoel and cortex-lacking genera, Clathrina, Levinella and Soleneiscus, rather than to the heterocoel species of Leucettusa and Leucascus. Our phylogenetic tree contains a highly supported clade of cortex-bearing Calcinea, which are also heterocoel with the exception to Ascaltis. Here, Leucascus shows affinities to Murrayona, while Leucettusa is the sister group to Leucettidae. The tracing of character evolution suggests that a cortex and a heterocoel water system were gained once in this subclass in an ancestor of the extant cortex-bearing Calcinea and that the asconoid water system of Ascaltis is the result of a secondary modification (Fig. 5). This clade of Calcinea with a cortex includes Murrayona phanolepis and Lelapiella sp. (non-monophyletic Murrayonida), but otherwise it is largely congruent with Hartman's Leucettida . Only the inclusion of Ascaltis would require a modification to his definition of this order. Accordingly, we found that Leucettida sensu lato could be defined as follows:
Order Leucettida Hartmann 1958 emended.
Calcinea with a cortex.
Species previously placed in Murrayonida are included in Leucettida. Leucettida contains almost exclusively heterocoel Calcinea, with Ascaltis being the only known exception. The asconoid aquiferous system of Ascaltis is interpreted as resulting from a secondary modification rather than being a primitive state.
Unfortunately, the relationships among homocoel Calcinea are not highly supported despite the extended character set compared to previous analyses. Clathrinida sensu Hartman is not monophyletic in our phylogeny. However, it cannot be rejected from our data according to the AU-tests. This uncertainty hampers a comprehensive revision of the order-level classification in Calcinea.
Recently, a phylogenetic study suggested that several morphological characters, such as color and presence/absence of tetractines or spines on actines, carry phylogenetic signals in Clathrinida . Certainly, several of the mentioned characters can be interpreted as diagnostic synapomorphies in the phylogeny presented by the authors. However, the study largely focused on Clathrina species and did not include Murrayonida, which our and a previous study  found to be nested within the Clathrinida. In addition to this restricted taxon sampling, the position of the root in Calcinea in this study was not highly supported, similar to the results obtained with our data. A different rooting could result in a different interpretation of the evolution of these morphological characters. It will require a larger dataset (character and taxon sampling) to test the new and valuable hypotheses of character evolution proposed by Rossi, et al. , but it seems that morphological characters still can provide more information than what was expected from the strong conflicts of molecular phylogenies and the current classification shown in our study. At least for certain highly supported clades, some morphological features will probably be useful to indicate close phylogenetic relationships and to serve as diagnostic synapomorphies for revised taxa.
Order- and family-level classification in Calcaronea
Our data rejects the subdivision of Calcaronea into the three monophyletic orders Lithonida, Leucosolenida and Baerida in the currently accepted classification . According to our phylogeny, Baerida is nested in the paraphyletic Leucosolenida, which is consistent with previous results , , . For Lithonida, the monophyly remains to be tested by including additional taxa of this order. In contrast, an alternative order-level classification of Calcaronea suggested by Hartmann , with the separation of Calcaronea into Leucosolenida sensu Hartman and Sycettida cannot be completely excluded. However, Leucosolenida sensu Hartmann in our dataset is only represented by one Leucosolenia specimen. In conclusion, we cannot yet provide a satisfactory alternative classification of calcaronean orders, mainly because of the non-monophyly at the family and genus levels, which prevent the generalization of our findings for a single species to its genus or family.
Due to the non-monophyly of several families, the taxonomic value of the diagnostic characters of these families has to be doubted, e.g., the pseudosagittal spicule layer in skeletons of Heteropiidae (see below), the inarticulated choanoskeleton of Jenkinidae and the tangential tetractines supporting the cortex in Amphoriscidae.
Close relationships of Grantiidae with giant diactines and Heteropiidae
The close relationship of Grantiidae with giant diactines in their cortex to the non-monophyletic Heteropiidae was a new finding from our data. Heteropiidae are characterized by a distal layer of a special spicule type, the pseudosagittal spicules (tri-or tetractines, Fig. 6) . The polyphyly of this family implies that this character has evolved convergently at least twice or that pseudosagittal spicules were lost in other closely related species (Fig. S7, A). At first sight, pseudosagittal spicules resemble sagittal spicules (tri -or tetractines with two equal or ‘paired’ angles and one dissimilar, ‘unpaired’ angle) but differ in that one of the paired actines and the unpaired actine instead of the two paired actines are of similar length. In Heteropiidae, these spicules have a specific orientation in the skeleton, in which the equally sized unpaired and paired actines are parallel to the sponge outer surface, while the second paired actine points inwards toward the atrium (Fig. 2,A; Fig. 6). Some Heteropiidae also have longitudinal large diactines, and, in some cases, show an ‘analogous’ organization to certain genera of Grantiidae, the only difference is the possession of a layer of pseudosagittal spicules . Such analogous genera of the two families are the pairs Ute (Grantiidae)-Heteropia (Heteropiidae) and Amphiute (Grantiidae)-Paraheteropia (Heteropiidae). Even in some of these Grantiidae, pseudosagittal spicules occasionally occur (e.g., in Amphiute ), but they were interpreted as the result of restricted growth of ‘normal’ spicules caused by the presence of a strong cortex . In contrast to this idea, our results provide evidence that these occasional pseudosagittal spicules might indeed be homologous to the pseudosagittal spicules of Heteropiidae and that the similar skeletal organization in the mentioned pairs of genera actually reflects phylogenetic relationships. The inclusion of specimens of Heteropiidae with large diactines in molecular phylogenies would provide further insights into this question. Clearly, several genera of Heteropiidae and Grantiidae require further attention, especially the genera Ute and Sycettusa, which are polyphyletic assemblages.
Figure 6. A pseudosagittal spicule and a subatrial sagittal spicule, shown in their orientation in the skeleton of Heteropiidae.
The arrows point at the unpaired angles. pa: paired actine; ua: unpaired actine. Pseudosagittal spicules have an appearance similar to sagittal spicules but their paired actines are of different sizes, with one being more similar to the unpaired actine than to the other paired actine. The latter points towards the atrium.doi:10.1371/journal.pone.0033417.g006
Leucascandra and Sycon carteri – implications for the evolution of inarticulated choanoskeletons
The sister group relationship of Leucascandra caveoltata (Jenkinidae) and Sycon carteri (Sycettidae) was unexpected because several key features appear to be obviously different in both species. For instance, the skeleton of Leucascandra consists of a cortex of triactines, and the inarticulated choanoskeleton contains subatrial triactines that support an irregular alveolar leuconoid choanosome (Fig. 7). In contrast, Sycon carteri is lacks a cortex, and its choanoskeleton forms short radial tubes with few rows of articulated triactines in the distal cones, each containing a choanocyte chamber of the syconoid aquiferous system (Fig. 7).
Figure 7. Comparison of Sycon carteri (A,C) and Leucascandra caveolata (QM G316146) (B,D).
A, B: Skeletal arrangement; the atrial skeleton at the lower side, the distal cones or the cortical skeleton respectively on top, C,D: syconoid and alveolar leuconoid aquiferous system.doi:10.1371/journal.pone.0033417.g007
However, both species share characteristics, as e.g., the growth form. Sycon carteri is built from tubes ‘united in a copiously branching, bushy mass’ (, p. 79; see also Fig. S2, A), and specimens of Leucascandra caveolata are formed by ‘copiously branched and anastomosed tubes’ (, p.199). In addition, the spiculation of both species is similar (aside from the spicule size and the occurrence of diactines in the distal cones of Sycon carteri). In both species, the sponge wall is thin and supported almost entirely by the subatrial triactines, whose unpaired actine crosses the complete sponge wall in Leucascandra and reaches the distal cones in Sycon carteri.
The inarticulated choanoskeleton of Jenkinidae was interpreted as a primitive state rather than evolved by reduction of sponges with articulated skeletons . The polyphyly of Jenkinidae (see also ) questions these interpretations and our data suggest that the inarticulated choanoskeleton of included species of Jenkinidae was developed twice from articulated ancestors. The close resemblance of Sycon carteri and Leucascandra sp. can be used to illustrate how ‘easily’ such transitions might be possible. One might consider a hypothetical evolution from a Sycon-like organization (as in Sycon carteri) to an inarticulated, leuconoid organization (as in Leucascandra caveolata) by the flattening of the distal cones so that the triactines form a cortical layer.
Starting with Haeckel, the morphological diversity of different grades of complexity in extant Calcarea has repeatedly misled biologists to presume one or several evolutionary lines leading from simple to more complex forms. Our results show that the evolution of Calcarea does not follow such clear trajectories and, instead, is characterized by frequent secondary loss and convergent evolution.
The classification of Calcaronea, as understood today, is highly artificial. The fact that most orders, families and several genera are paraphyletic or polyphyletic assemblages suggests that classical revisions of such taxa (e.g., for Clathrina, ) will almost certainly exclude ‘unexpected relatives’ and, therefore, will not result in a phylogenetic classification. Yet, a basic phylogenetic framework to understand the evolution of characters in this sponge class is not available, and until it is established, any taxonomic revision should include DNA data and consider all available taxa from the given subclass. Future works should include much more species in molecular phylogenies, but not only those from missing families and genera. In addition, the use of independent molecular markers, such as mitochondrial genes, would be desirable. In Calcarea, mitochondrial sequences seem to evolve relatively fast compared to other sponges, making the genes hard to amplify with standard primers ; however, they would probably provide a good phylogenetic signal to resolve the nodes in our phylogeny with weak support, especially at the base of Calcinea.
At present, a revision of the higher classification is only possible for some clades because the evolution of the different organization forms is far from understood and recognizing potential diagnostic characters remains impossible. We are confident that thorough taxon sampling and DNA analysis will provide such characters in many cases, at least at shallower taxonomic levels, as indicated in a previous study focusing on Clathrina-species . Molecular data could help to evaluate competing hypotheses and, in our case, lead to the recognition of a previously proposed order Leucettida sensu lato. In contrast, our study illustrates also how several relationships that were previously not conceived (such as the close relationship of Sycon carteri and Leucascandra caveolata or the sister group relationship of Leucettusa and Leucettidae) could be brought forward by molecular studies. Extending the available molecular and morphological dataset is crucial to finally providing a classification that is congruent with the phylogeny of this sponge group.
Included new specimens of Calcinea (habitus and transversial sections). A–H: Leucaltidae; A, B: Ascandra sp.; C, D: Leucettusa sp.1. Note the scattered small tetractines in the choanosome (D, insert); E–H: Leucettusa sp.2. E, F: QM 323253; G,H: QM 323283. Note the scattered v-shaped triactines in the choanosome (F, H, insert); H: arrow points at the apical ray of a large tetractine, which supports the choanosome; I–J: Ascaltis sp. (Leucascidae). J, insert: overview of section; K: Clathrina sp. (Clathrinidae) GW957. atr: atrium; chc: choanocyte chamber; cx: cortex; ext: exterior of the sponge; eh: exhalant channel; ih: inhalant channel.
Included new specimens of Calcaronea (habitus and transversial sections). A: Sycon carteri (in phylogeny: SAM PS0142, a conspecific specimen); B: Synute pulchella (drawn from photography); D: Leucandra sp. (transversial section); D: Teichonopsis labyrinthica (transversial section). Due to the growth-form the upper surface corresponds to the atrium. E: Sycettusa cf. simplex (transversial section). The arrow points at the unpaired angle of a pseudosagittal triactine. F: Sycettusa aff. hastifera in-situ; G,H: Grantessa sp. GW974; I: Leucilla sp. (transversial section); J: Grantiopsis cylindrica. atr: atrium; cx: cortex; ext: exterior of the sponge.
LSU secondary structure ( Leucetta microraphis ).
Strict consensus of Bayesian phylogenies obtained with different doublet models of each model family. 6-state: 6A–6F; 7-state: 7A–7F; 16-state: 16A–16F). Polytomies indicate model specific differences in tree topologies.
Strict consensus of ML phylogenies obtained with different doublet models of each model family. 6-state: 6A–6E; 7-state: 7A–7E; 16-state: 16A, 16B). Polytomies indicate model specific differences in tree topologies.
Phylogenies obtained under different partitioning schemes (stem+loop, SSU+LSU) with standard, non-doublet models. Support values (ML:BS, BI:PP) are given at the nodes. With a partitioning of stem+loop, the position of Leucosolenia differs from trees inferred with doublet model (Figures S4, S5), while with a partitioning into SSU+LSU, the position is as presented in Figure 3. Note also that all analyses result in a different topology at the base of Calcinea compared to our preferred doublet-inferred phylogenies (see main text).
Evolution of pseudosagittal spicules and of the organization of the choanoskeleton. Tree topology identical to Fig.3, only class Calcaronea is shown. A. Presence of a continuous layer of pseudosagittal spicules. left: obtained phylogeny, right: Excerpt of clade LEUC I with nodes of PP-support below 90% collapsed. B. Morphology of the choanoskeleton (characters modified from ). Note that inarticulated choanoskeletons evolved at least four times from ancestors with articulated choanoskeletons.
LSU rRNA primer sequences.
GenBank accession numbers of outgroup taxa.
For letting us subsample specimens of Calcarea included in this study, we thank John Hooper and Monika Schlacher-Hoenlinger from the Queensland Museum in Brisbane, Jane Fromont from the Western Australian Museum in Perth, Thierry Laperousaz and Shirley Sorokin from the South Australian Museum in Adelaide and Rob van Soest from the Zoölogisch Museum, Universiteit van Amsterdam, Amsterdam, Netherlands. The Great Barrier Reef Marine Park Authority is gratefully acknowledged for permitting our fieldwork in the GBR (Permit No G98/142, G98/022, G00/638, G06/16547.1). We are also grateful to the Egyptian Environmental Affairs Agency (EEAA), especially Mohammed Fouda for allowing our fieldwork in Egypt and to Alexander Keck and Christian Alter for their support during fieldwork in Egypt. We thankfully acknowledge the detailed comments of two anonymous reviewers, which helped to improve the manuscript.
Conceived and designed the experiments: OV GW. Performed the experiments: OV EW. Analyzed the data: OV. Contributed reagents/materials/analysis tools: OV EW GW. Wrote the paper: OV. Revised the manuscript: OV GW.
- 1. Gavaze E, Lapébie P, Ereskovsky AV, Vacelet J, Renard E, et al. (2011) No longer Demospongiae: Homoscleromorpha formal nomination as a fourth class of Porifera. Hydrobiologia. doi: 10.1007/s10750-011-0842-x.
- 2. Van Soest RWM, Boury-Esnault N, Hooper JNA, Rützler K, de Voogd NJ, et al. The World Porifera Database. Available: http://www.marinespecies.org/porifera. Accessed 2012 Feb 1.
- 3. Philippe H, Derelle R, Lopez P, Pick K, Borchiellini C, et al. (2009) Phylogenomics revives traditional views on deep animal relationships. Curr Biol 19: 706–712.
- 4. Pick KS, Philippe H, Schreiber F, Erpenbeck D, Jackson DJ, et al. (2010) Improved phylogenomic taxon sampling noticeably affects nonbilaterian relationships. Mol Biol Evol 27: 1983–1987.
- 5. Rapp HT, Janussen D, Tendal OS (2011) Calcareous sponges from abyssal and bathyal depths in the Weddell Sea, Antarctica. Deep-Sea Res Pt Ii 58: 58–67.
- 6. Manuel M, Borchiellini C, Alivon E, Boury-Esnault N (2004) Molecular phylogeny of calcareous sponges using 18S rRNA and 28S rRNA sequences. Boll Mus Ist Biol Univ Genova 68: 449–461.
- 7. Haeckel E (1872) Die Kalkschwämme, Band 3: Atlas der Kalkschwämme (Calcispongien oder Grantien). Berlin: Georg Reimer.
- 8. Haeckel E (1872) Die Kalkschwämme, Band 1: Biologie der Kalkschwämme (Calcispongien oder Grantien). 484 p. Georg Reimer.
- 9. Haeckel E (1872) Die Kalkschwämme, Band 2: System der Kalkschwämme (Calcispongien oder Grantien). Berlin: Georg Reimer. 416 p.
- 10. Cavalcanti FF, Klautau M (2011) Solenoid: a new aquiferous system to Porifera. Zoomorphology 130: 255–260.
- 11. Dohrmann M, Janussen D, Reitner J, Collins AG, Wörheide G (2008) Phylogeny and evolution of glass sponges (Porifera, Hexactinellida). Syst Biol 57: 388–405.
- 12. Rossi AL, Farina M, Borojevic R, Klautau M (2006) Occurrence of five-rayed spicules in a calcareous sponge: Sycon pentactinalis sp. nov. (Porifera: Calcarea). Cah Biol Mar 47: 261–270.
- 13. Manuel M (2006) Phylogeny and evolution of calcareous sponges. Can J Zool 84: 225–241.
- 14. Bidder GP (1898) The skeleton and classification of calcareous sponges. Proceedings of the Royal Society, London 6: 61–76.
- 15. Manuel M, Borojevic R, Boury-Esnault N, Vacelet J (2002) Class Calcarea Bowerbank 1864. In: Hooper JNA, van Soest RWM, editors. Systema Porifera: A Guide to the Classification of Sponges. New York: Kluwer Academic/Plenum Publishers. pp. 1103–1110.
- 16. Minchin EA (1896) Suggestions for a Natural Classification of the Asconidae. Annals and Magazine of Natural History (6) 18: 349–362.
- 17. Hartman WD (1958) A re-examination of Bidder's classification of the Calcarea. Syst Biol 7: 97–110.
- 18. Wörheide G, Hooper JNA (1999) Calcarea from the Great Barrier Reef. 1: Cryptic Calcinea from Heron Island and Wistari Reef (Capricorn-Bunker Group). Mem Queensl Mus 43: 859–891.
- 19. Manuel M, Borchiellini C, Alivon E, Le Parco Y, Vacelet J, et al. (2003) Phylogeny and evolution of calcareous sponges: Monophyly of Calcinea and Calcaronea, high level of morphological homoplasy, and the primitive nature of axial symmetry. Syst Biol 52: 311–333.
- 20. Dohrmann M, Voigt O, Erpenbeck D, Wörheide G (2006) Non-monophyly of most supraspecific taxa of calcareous sponges (Porifera, Calcarea) revealed by increased taxon sampling and partitioned Bayesian analysis of ribosomal DNA. Mol Phylogenet Evol 40: 830–843.
- 21. Vacelet J (1961) The order Pharetronida in Hartman's classification of the Calcarea. Syst Zool 10: 45–47.
- 22. Vacelet J (1981) Éponges hypercalcifées (‘Pharétronides’‘Sclérosponges’) des cavités des récifs coralliens de Nouvelle-Calédonie. Bull Mus Nat Hist Natur (Paris) 3: 313–351.
- 23. Borojevic R, Boury-Esnault N, Vacelet J (1990) A revision of the supraspecific classification of the subclass Calcinea (Porifera, class Calcarea). Bull Mus Nat Hist Natur (Paris), Sec A 12: 243–276.
- 24. Borojevic R, Boury-Esnault N, Vacelet J (2000) A revision of the supraspecific classification of the subclass Calcaronea (Porifera, class Calcarea). Zoosystema 22: 203–263.
- 25. Hooper JNA, van Soest RWM (2002) Systema Porifera: a guide to the classification of sponges. New York: Kluwer Academic/Plenum Publishers. 1810 p.
- 26. Borojevic R, Boury-Esnault N, Manuel M, Vacelet J (2002) Order Leucosolenida Hartman, 1958. In: Hooper JNA, Van Soest RWM, editors. Systema Porifera. A Guide to the Classification of Sponges. New York: Kluwer Academic/Plenum. pp. 1157–1184.
- 27. Galtier N, Gouy M, Gautier C (1996) SEAVIEW and PHYLO_WIN: two graphic tools for sequence alignment and molecular phylogeny. Comput Appl Biosci 12: 543–548.
- 28. Voigt O, Erpenbeck D, Wörheide G (2008) Molecular evolution of rDNA in early diverging Metazoa: First comparative analysis and phylogenetic application of complete SSU rRNA secondary structures in Porifera. BMC Evol Biol 8: 69.
- 29. Schnare MN, Damberger SH, Gray MW, Gutell RR (1996) Comprehensive comparison of structural characteristics in eukaryotic cytoplasmic large subunit (23 S-like) ribosomal RNA. J Mol Biol 256: 701–719.
- 30. Jow H, Hudelot C, Rattray M, Higgs PG (2002) Bayesian phylogenetics using an RNA substitution model applied to early mammalian evolution. Mol Biol Evol 19: 1591–1601.
- 31. Hudelot C, Gowri-Shankar V, Jow H, Rattray M, Higgs PG (2003) RNA-based phylogenetic methods: application to mammalian mitochondrial RNA sequences. Mol Phylogenet Evol 28: 241–252.
- 32. Stamatakis A (2006) RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22: 2688–2690.
- 33. Wheeler WC, Honeycutt RL (1988) Paired sequence difference in ribosomal RNAs: evolutionary and phylogenetic implications. Mol Biol Evol 5: 90–96.
- 34. Hancock JM, Tautz D, Dover GA (1988) Evolution of the secondary structures and compensatory mutations of the ribosomal RNAs of Drosophila melanogaster. Mol Biol Evol 5: 393–414.
- 35. Hillis DM, Dixon MT (1991) Ribosomal DNA: molecular evolution and phylogenetic inference. Q Rev Biol 66: 411–453.
- 36. Higgs PG (2000) RNA secondary structure: physical and computational aspects. Q Rev Biophys 33: 199–253.
- 37. Savill NJ, Hoyle DC, Higgs PG (2001) RNA sequence evolution with secondary structure constraints: comparison of substitution rate models using maximum-likelihood methods. Genetics 157: 399–411.
- 38. Dixon MT, Hillis DM (1993) Ribosomal RNA secondary structure: compensatory mutations and implications for phylogenetic analysis. Mol Biol Evol 10: 256–267.
- 39. Telford MJ, Wise MJ, Gowri-Shankar V (2005) Consideration of RNA secondary structure significantly improves likelihood-based estimates of phylogeny: examples from the Bilateria. Mol Biol Evol 22: 1129–1136.
- 40. Erpenbeck D, Nichols SA, Voigt O, Dohrmann M, Degnan BM, et al. (2007) Phylogenetic analyses under secondary structure-specific substitution models outperform traditional approaches: Case studies with diploblast LSU. J Mol Evol 64: 543–557.
- 41. Schöniger M, von Haeseler A (1994) A stochastic model for the evolution of autocorrelated DNA sequences. Mol Phylogenet Evol 3: 240–247.
- 42. Muse SV (1995) Evolutionary analyses of DNA sequences subject to constraints on secondary structure. Genetics 139: 1429–1439.
- 43. Tillier ERM, Collins RA (1995) Neighbor Joining and Maximum-Likelihood with RNA sequences - addressing the interdependence of sites. Biochem Syst Ecol 12: 7–15.
- 44. Tillier ERM, Collins RA (1998) High apparent rate of simultaneous compensatory base-pair substitutions in ribosomal RNA. Genetics 148: 1993–2002.
- 45. Swofford DL (2003) PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sunderland, MA: Sinauer Associates.
- 46. Maddison WP, Maddison DR (2002) MacClade. Sunderland, MA: Sinauer Associates.
- 47. Shimodaira H (2002) An approximately unbiased test of phylogenetic tree selection. Syst Biol 51: 492–508.
- 48. Shimodaira H, Hasegawa M (2001) CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17: 1246–1247.
- 49. Rossi AL, De Moraes Russo CA, Solé-Cava AM, Rapp HT, Klautau M (2011) Phylogenetic signal in the evolution of body colour and spicule skeleton in calcareous sponges. Zool J Linn Soc 163: 1026–1034.
- 50. Borojevic R (1965) Éponges Calcaires des Côtes de France I. Amphiute paulini HANTISCH: les generes Amphiute HANITSCH et Paraheteropia n. gen. Arch Zool Exp Gen 106: 665–670.
- 51. Dendy A (1893) Synopsis of the Australian Calcarea Heterocoela; with a proposed classification of the group and descriptions of some new genera and species. Proceedings of the Royal Society of Victoria 5: 69–116.
- 52. Borojevic R, Klautau M (2000) Calcareous sponges from New Caledonia. Zoosystema 22: 187–201.
- 53. Klautau M, Valentine C (2003) Revision of the genus Clathrina (Porifera, Calcarea). Zool J Linn Soc 139: 1–62.
- 54. Voigt O, Eichmann V, Wörheide G (2011) First evaluation of mitochondrial DNA as a marker for phylogeographic studies of Calcarea: a case study from Leucetta chagosensis. Hydrobiologia. doi: 10.1007/s10750-011-0800-7.