A new sponge genus from the Chengjiang biota with an intriguing combination of skeletal characters

The development of fossil sponge systematics is hindered by factors such as their low anatomical complexity and the low preservation potential of many characters used in the definition of extant sponge clades. Here we describe Calliospongia chunchengia gen. et sp. nov., a new sponge from the Lower Cambrian Chengjiang biota, which possesses a unique combination of characters. The skeleton of C. chunchengia is similar to those of leptomitid protomonaxonids in being composed of large spicules arranged longitudinally. However, it is distinct enough to belong in its own genus and even question a leptomitid affinity, because, unlike the members of this group, which usually have two categories of diactine spicules, its skeleton is entirely composed of triactines. The new taxon shares the presence of large megascleres with demosponges, hexactinellids and the ascosponges, suggesting an affinity to the total group of Silicea. It is also reminiscent of the hexactinellid-like reticulosans, showing a potential connection between two major types of Early Palaeozoic sponge body plans. A more accurate definition of the Cambro-Ordovician groups and the creation of a total evidence framework that integrates extinct and extant sponge morphology into the phylogenetic scenario derived from phylogenomics seem to be essential steps for the improvement of our understanding of early sponge evolution.


Introduction
The phylogenetic affinities of sponges have been the subject of major controversy in recent years (Li et al. 2021;Redmond and McLysaght 2021). Both their traditional position as the earliest offshoot of Metazoa (Telford et al. 2016) and their monophyly (Borchiellini et al. 2001;Wörheide et al. 2012) have been challenged. These debates are among the most transcendental in the study of early animal evolution because together with the Placozoa, they are the two simplest animal phyla lacking both a nervous and a muscular system (Nielsen 2008).
The early fossil record of several phyla has helped elucidate important questions in animal evolution [e.g.: (Legg et al. 2013)]. However, identifying the oldest sponges is difficult because there is no evidence that spicules were part of the bauplan of the last common ancestor of all animals (Nielsen 2008), and therefore the earliest stem-group of sponges likely lacked spicules altogether. Molecular clocks (Dohrmann and Wörheide 2017) and fossil biomarkers (Love et al. 2009) suggest a deep, pre-Ediacaran, origin of sponges, but the fossil evidence is lacking (Sperling et al. 2010). There are numerous enigmatic fossils described as  (Turner 2021), but none of them have been universally accepted (Antcliffe et al. 2014). Some have been alternatively interpreted as parts of Ediacaran fronds (Serezhnikova 2007), and others as colonial protists or even as non-biological sedimentary structures (Mehra et al. 2020).
In contrast with the apparent lack of bona fide sponges in Precambrian strata, the early Cambrian presents a wealth of undisputed sponges (Pisera 2006). The crown group of some classes had already diversified by the Lower Cambrian as evidenced by numerous complete body fossils (Rigby 1986), and spicule assemblages (Zhang and Pratt 1994). However, the majority of early sponges can not be easily assigned to any of the four extant classes. One of the most diverse groups among them is the Protomonaxonida, a polyphyletic assemblage characterised by the possession of needle-like spicules . Some of them, like the hazeliids, are likely to be demosponges . Others such as choiids and leptomitids have been recently recognised as their own class called Ascospongiae (Botting 2021). Another diverse group from the Cambrian is the Reticulosa, which get their name from their highly organised, reticulated, skeletons. They are interpreted as either the stem group of Hexactinellida (Antcliffe et al. 2014), or Porifera as a whole (Botting and Muir 2018). They are also suspected to be polyphyletic because their anatomy, other than their basic skeleton organisation, is very disparate (Botting and Muir 2018). For example, Cyathophycus loydelli shares characters with hexactinellids, demosponges, and homoscleromorphs .
The latest review about the early fossil record of sponges underlined the suspected polyphyly of these groups and the need for reassessment of a number of early Paleozoic groups (Botting and Muir 2018). The systematics of fossil sponges are intrinsically difficult to study because they are anatomically simple and multiple major clades are defined based on characters that have a very low preservation potential. For example, the subclasses of Calcarea, the Calcinea and Calcaronea, are defined by the position of the nucleus in the choanocytes and the anatomy of their larvae (Wörheide et al. 2012). Microscleres, the smallest type of spicules in demosponges and hexactinellids, are hardly ever preserved but they have been used historically to define large clades such as the two hexactinellid subclasses, and orders of demosponges (Van Soest et al. 2012). The paucity of characters has led sponge palaeontologists to propose phylogenetic scenarios for the phylum without explicit phylogenetic analyses (e.g. : Pisera 2006;. The internal relationships of sponges have gone through a number of different hypotheses through the years (Reitner and Mehl 1996), but the development of molecular phylogenetics and phylogenomics has led to a broad agreement about the relationships of the four extant classes (Pick et al. 2010;Wörheide et al. 2012). Consensus is lacking, however, on lower taxonomic levels. Traditional genera such as Sycon have been found to be polyphyletic (Dohrmann et al. 2006), suggesting a significant amount of homoplasy in sponge evolution. The description of taxa with new character combinations can alleviate the uncertainty around the morphology, linking groups and establishing the polarity of character acquisition. Furthermore, describing new evolutionary continua can also lead to the revision of fossil clades [e.g.: unusual mineralogies in reticulosans (Botting and Butterfield 2005) and the unexpected presence of hexactine spicules on a number of fossil taxa (Botting and Muir 2018)]. The Lower Cambrian Chengjiang biota of China is one of the best sources of fossil sponges. It contains a rich assortment of species, including many described ones, some of which are exclusively found in this assemblage, and multiple others that await description (Wu et al. 2014). Calliospongia chunchengia gen. nov and sp. nov, is a sponge with a unique skeletal morphology of large T-shaped spicules that expands the known morphospace of the phylum and could serve to connect leptomitid-like taxa to other fossil groups of the Cambrian.

Materials and methods
The fossils were sampled from the Lower Cambrian Yu'anshan Member of the Chiungchussu Formation at the Ercaicun sections (Cambrian Series 2, Stage 3), in Haikou, Yunnan, China. This site is part of the extremely diverse Chengjiang biota, which features a large number of sponges and arthropods as well as many other animal phyla (Hou et al. 2017). It features Burgess shale-style preservation with many specimens preserving exquisite detail and soft anatomy. Sponges are usually preserved in a characteristic orange colouration due to the replacement of the original spicule mineralogy with various minerals during diagenesis. The two specimens of Calliospongia chunchengia described here, holotype and paratype with counterparts, are deposited at the Yunnan Key Laboratory for Palaeobiology (YKLP) at Yunnan University in China. They were exposed after preparation with steel needles under a Leica M205C stereomicroscope and a Keyence VHX5000 digital microscope, then examined and photographed using a Nikon D3X camera under the stereomicroscope. Figures 1 and 2 were assembled with GIMP (v.2.10) and Fig. 3 was drawn using Inkscape (v.1.0.2).

Type species. Calliospongia chunchengia (by monotypy).
Diagnosis. Small thin-walled sponge with lenticular to subglobular shape. The skeletal net is composed of large curved T-shaped triactines in an interlocked arrangement. The two longest of the three rays are slightly curved and parallel to the longitudinal axis. The longitudinal rays of adjacent spicules overlap each other. The shorter ray is straight and perpendicular to the other two, pointing towards the centre of the sponge. The perpendicular rays sometimes form parallel rows with the perpendicular rays of the adjacent spicules.
Description. The holotype, YKLP 14065a and b (Fig. 1a), consists of a seemingly complete specimen of 40 mm that branches into two bodies of roughly the same size. The whole specimen, including the two branches, has an almost spherical outline. The individual branches are lenticular and seem to split at the midpoint of the height of the specimen (21 mm). These two bodies can be recognised as separate branches because they both have defined skeletal walls on either side and a section of matrix without any observable spicules between them. The orientation of the spicules on both branches points towards the midline of the individual branches instead of the middle of the specimen as would be expected of a single sponge body that was altered taphonomically. The right branch is 9 mm at the widest point. The top has a straight edge, which could be an open osculum (Fig. 1b). The branch on the left is slightly thinner at 7 mm of maximum width and it tapers into a subtriangular tip. The skeleton consists of large curved T-shaped triactines arranged in a dense mesh (Fig. 1c). The spicules have two curved longitudinal rays and a straight perpendicular one (Fig. 1d). The perpendicular rays are arranged horizontally and parallel to the rays of the adjacent rows of spicules, sometimes preserved in bundles probably due to compression. The two longitudinal rays of one spicule intersect perpendicularly with the horizontal rays of four other spicules. The curvature of the spicules in the left branch is smaller than on the right branch, probably due to a slight difference in compressive stresses. The longitudinal spicules on the right branch consist of approximately 22-24 rows, while the left branch has somewhere between 19-21 rows. The longitudinal spacing between spicules is slightly wider in the middle and closer towards the base and the top. The length of the curved rays of these spicules on the top part is 3.2-4.0 mm on the right branch, and 3.8-4.5 mm on the left branch. The diameter near the joint of the rays of the spicules is up to 0.035-0.045 mm on the right branch, and 0.04-0.05 mm on the left one.
There are small differences in the size of the spicules in the two branches. The length of the curved rays on the middle part is 4.0-4.5 mm on the right branch and 4.4-5.3 mm on the left branch, and the diameter of the proximal part of the rays is 0.045-0.055 mm on the right and 0.058-0.062 mm on the left. The horizontal rays are hard to measure because they are usually stacked together, but they are at least 2-3 mm long. The arches formed by the longitudinal rays of the spicules are 2 mm in width. At the base, spicule orientation is no longer parallel to the long axis of the sponge, being instead inclined by about 45 degrees, and the spicules have a higher degree of overlap with each other.
The paratype, YKLP 14145 a and b (Fig. 2a), is partially disarticulated and has a lens-shaped outline. It measures 22 mm long and 7.8 mm wide. The primary skeleton is composed of a network formed by the adjacent rays of the large T-shaped spicules. The length of the longitudinal rays of the spicules on the middle part is 2.5-3.0 mm, and the diameter of the proximal part of the rays is 0.040-0.045 mm. Some spicules have a more pronounced curvature. The perpendicular rays are horizontally arranged in rows parallel to the perpendicular rays of the adjacent spicules.
Remarks. Calliospongia chunchengia is described as a new genus because there are currently no other comparable fossil sponges known. It is superficially similar to Paraleptomitella (Chen et al. 1989) from the Chengjiang biota, but the spicules of the latter are curved monaxons and the ones of C. chunchengia have a unique curved T-shape with an additional perpendicular ray, making them easily distinguishable from one another.

Discussion
The skeleton of C. chunchengia is unique in that it is entirely composed of one type of large triactines. This spicule morphology could be alternatively interpreted as the combination of two single rayed spicules intersecting, but this is clearly not the case because a well defined centrum can be seen in the case of spicules preserved without overlap (Figs. 1d and 2b). None of the spicules visible in the two specimens show evidence of biminerallic composition or internal details of the axial canal.
The morphology of the megascleres is reminiscent of various spicules of extant sponges, but most of the similarities are only superficial. Microscleres have very varied morphologies, including some T-shaped ones, with equivalent or unequal rays (Mehl 1998), and even S-shaped ones (Xiao et al. 2005). However, the spicules of C. chunchengia are clearly within the megasclere size range. Y-shaped triactines with regular angles are present in multiple groups (Van Soest et al. 2012), but the curved rays of the spicules of C. chunchengia, form a 180º angle, and the internal ray is perpendicular to them, forming a T shape. Many calcareous sponges have various triactine spicules, but the size and presumed mineralogy are completely different. Those calcareous triactines have also been compared to the skeleton of Vauxia, another Cambrian sponge (Bengtson 1986). However, the skeleton of this taxon has been reinterpreted as being mostly made out of fibres (Rigby 1986;Finks et al. 2003).
Some examples of sponges that have comparable spicules are hexactinellids such as the family Pheronematidae (Tabachnick and Menshenina 2002) which have triactines and large anchor spicules. However, neither the size, the proportion of the rays, nor the position in the skeleton are similar. The anchor spicules are located at the base of the sponge instead of forming the choanosomal skeleton. In addition, those types of spicules have a long straight ray and usually two short longitudinal ones, which is the opposite condition to the spicules of C. chunchengia. There are other arguments in favour of a hexactinellid affinity. The spicules of C. chunchengia could be derived from the paratangential layer of pentactines seen in many fossil hexactinellids (e.g.:. Li et al. 2019) through the reduction of two rays, but this scenario would also require an explanation for the lack of other spicules types. There is, however, a paraphyletic group of fossil hexactinellids, the "rossellimorpha", which have simpler skeletons composed of reduced triaxons such as diactines and triactines (Reitner and Mehl 1995). An affinity to that group would be an interesting possibility. If any support for the hypothesis that the spicules of C. chunchengia could be reduced triaxonal spicules is found, C. chunchengia would be a hexactinellid with a radically simplified skeleton. The two arguments, absence due to secondary simplication or due to taphonomic loss, are based on lack of evidence and are therefore inconclusive until there is more information.
The spicules of Calliospongia chunchengia are clearly megascleres, strongly suggesting a silicean affinity of some sort, but, in terms of size, they are particularly similar to those of the newly erected clade Ascospongiae, a subgroup of the Protomonaxonida (Botting 2021). Although the ends of the spicules of C. chunchengia are often obscured due to their length, even a conservative estimate places them in the category of very large spicules, reaching lengths upwards of 9 mm. Regular sized megascleres, like those of demosponges, mostly range between 1 and 3 mm ). Among protomonaxonids, Calliospongia chunchengia is particularly similar to leptomitids. Their spicules are large and sub-longitudinal, arranged parallel to the body, but they are not monaxonic. Triaxonal spicules can be sometimes seen in leptomitids, but those are much smaller and not a major component of the skeleton (Botting 2021). Additionally, there is no evidence in the fossils of hexactines or microscleres. Within leptomitids, C. chunchengia is most reminiscent of Paraleptomitella dictyodrama (Chen et al. 1989). However, they are different in the number of rays on the spicules, the lack of small oxeas and the orientation of its spicules, which are slightly oblique instead of parallel to the axis of the sponge. They also differ from leptomitids in aspect ratio. C. chunchengia is less elongated than leptomitids, but these specimens are small in size compared to leptomitids, so this difference could be due to these fossils being young specimens. The number of rays on the spicules are the main difference: leptomitids have two rays and C. chunchengia has three. Leptomitids also generally have two distinct types of spicules: small oxeas, and large monaxons. It is worth mentioning that a leptomitid affinity does not necessarily refute a hexactinellid affinity, because leptomitids are considered to be hexactinellids by some authors (Reitner and Mehl 1995) and, furthermore, some of the arguments presented against a hexactinellid affinity are equally valid against an affinity to that group of protomonaxonids.
Despite the differences, C. chunchengia fits the classical definition of Protomonaxonida and it shares a number of characters present in the definition of the new class Ascospongiae (Botting 2021). In addition to the size of the spicules, their skeleton is single-layered and radially arranged. The class was described as being mostly solitary, which is not the case for the holotype, but the definition of the group also includes non-solitary taxa such as Pirania muricata (Rigby 1986). The body shapes within this group are very varied, including lenticular and subcylindrical taxa. The anatomy of the osculum of C. chunchengia is not entirely clear, because the ends of the branches of the holotype are different from each other, but the frayed appearance of the end of the right branch (Fig. 1b) may suggest that the tapering morphology reminiscent of the osculum of leptomitids of the left branch was the original anatomy.
If Calliospongia chunchengia is related to the leptomitids, that would place it within the newly erected Ascospongiae (Botting 2021). The Ascospongiae is placed as either the sister group of the crown of Porifera or the stem group of Silicea (Botting 2021). However, the internal relationships within this group are unresolved and the three-rayed nature of its spicules makes it different enough to be cautious about including it in the new group. Because the spicules are unique to this taxon we cannot be completely sure about their origin and homology to other types. Currently there is not enough evidence available to discern the origin of these spicules. Two examples of possible origins is that they could be homologous to some of the monaxons that make up the majority component of the skeleton of the ascosponges with an extra ray resulting from the fusion of simpler spicule types, a phenomenon that has been posited as the source of complex spicules by some authors (e.g.: McMenamin 2008). Another possibility is that they were derived from triaxonic hexactines or stauractines, like those of reticulosans and other hexactinellids, that lost the lateral rays but kept the longitudinal and internal ones. The latter seems more plausible and it may explain the unusual morphology of its spicules.
The most recent hypothesis of fossil sponge relationships (Botting and Muir 2018) shows Reticulosa as the paraphyletic stem group from which various major sponge clades originated, but the traditional interpretation is that they are the stem group of Hexactinellida. An additional morphological link between taxa close to the origin of Hexactinellida and the Ascospongiae which is composed of taxa that has been traditionally considered to be on the stem group of Demospongiae, would have an important role in elucidating the sequence of character acquisition of sponges. A potential connection between those groups has been previously suggested. The large monaxons characteristic of the group would be derived from the large anchor spicules of reticulosans ). In the scenario proposed for the origin of the Ascospongiae, leptomitids are close to the base of the group and the presence of triaxonic spicules is plesiomorphic (Botting 2021). Interpreting the spicules of C. chunchengia as triaxonic spicules with only three rays makes both and affinity to leptomitids and stem hexactinellids plausible. However, without proper phylogenetic analyses these hypotheses will remain as suggestions for future lines of enquiry.

Concluding remarks
The description of fossil taxa with new character combinations such as Calliospongia chunchengia can provide important insight for early sponge systematics. Its unique combination of features may help by bridging the gap between different types of skeletons. Calliospongia chunchengia fits into the traditional Protomonaxonida and is particularly similar to taxa that has been reassigned to the Ascospongiae. C. chunchengia could be an ascosponge with an autapomorphic type of spicule or, since leptomitids have also been interpreted as hexactinellids by some authors, an intermediate form between stem hexactinellids and protomonaxonids or ascosponges (Fig. 3). We refrained from assigning the new taxon to the Ascospongiae because this group is not firmly established yet and the unclear affinities of C.chunchengia could undermine the definition of this new group, going against the necessary reinforcement of the diagnoses of the fossil groups we advocate for. Instead, we decided that assigning it to the total group of Silicea, which also includes all the other fossil groups that share features with the new taxon, was safer since its skeleton is structured and its spicules are large and probably made of silica.
The affinities of C. chunchengia could be resolved by coding it into a morphological matrix, but we consider that there are not enough characters to score for it, and no appropriate preexisting morphological dataset to fit it into. Currently, neither the fossil groups, nor the approach used to evaluate the affinities are reliable enough to reconstruct the early evolution of sponges. Resolving the issues of fossil sponge systematics requires a two-pronged approach. First, the nonmonophyletic early fossil groups need to be redefined. The definition of the clade Ascospongiae could be a step in the right direction, but it has not been validated by other fossil sponge experts yet. The second aspect is the establishment of a framework for the fossils that integrates the morphology of extant and extinct taxa with genomic data. In the next few years, further revisions of the systematics of fossil sponges and an integrative framework of the phylum will open the door to a more stable and reliable tree of sponges. Calliospongia chunchengia, is yet another remarkable example of the unusual morphologies of the Cambrian that will find its precise place in the sponge tree of life as their phylogeny becomes better understood.