Polycystine radiolarians constitute a diverse group of marine planktic Rhizaria that secrete a skeleton made of amorphous silica. Molecular studies suggest the presence of skeletonless radiolarians since the Late Neoproterozoic (Sandin et al. 2025); however, the presence of polycystine radiolarians in the fossil record has only been confirmed since the Cambrian Series 2, Stage 4 (Obut and Iwata 2000; Pouille et al. 2011; Sennikov et al. 2017).

The Cambrian and Ordovician periods are of key significance for the early evolutionary history of marine organisms, including heterotrophic plankton clades, such as polycystine radiolarians. Indeed, the siliceous skeletons of early Palaeozoic radiolarians testify to their significant diversifications and evolutionary changes (Danelian and Monnet 2021, 2026).

During the last 35 years, significant progress has been achieved in understanding the taxonomy and biostratigraphy of early Palaeozoic radiolarians (Aitchison et al. 2017; Caridroit et al. 2017; Danelian et al. 2017; Noble et al. 2017). Success has relied on the discovery of well-preserved faunas (e.g. Won and Below (1999)), which have been carefully described and detailed with 2D imaging techniques, such as scanning electron microscopy (SEM). However, the recent application of micro-computed tomography (micro-CT) on radiolarians (Matsuoka and Teshima 2009; Matsuoka et al. 2012; Ishida et al. 2015; Kachovich et al. 2019) allows the capture of the full 3D complexity of their skeleton development, especially of its most internal parts. The latter are of higher taxonomic significance and, thus, open new avenues for improved understanding of their internal architectures (Kachovich and Aitchison 2020, 2021; Perera et al. 2020; Sheng et al. 2020, 2026a, 2026b; Perera and Aitchison 2022, 2023).

Based on Tremadocian (Lower Ordovician) material from Nevada, the species Protoentactinia gracilispinosa was first introduced by Kozur et al. (1996, p. 253), as the type species of the newly-defined genus Protoentactinia Kozur, Mostler & Repetski, 1996 and type-genus of the newly-defined family Protoentactiniidae Kozur, Mostler & Repetski, 1996. P. gracilispinosa was described as a radiolarian bearing a spicular system initiated from “a bar-centred entactinarian spicule with 3 terminal rays” at each end; the subglobular skeleton was described as formed of a repeatedly branching pattern of spicules, which accommodated a nearly spherical inner empty space. The species was later reported from upper Cambrian and lowermost Ordovician strata of western Newfoundland (Won and Iams 2002; Won et al. 2005; Pouille et al. 2014). Here, using micro-CT, we provide for the first time, details of the different stages of spicular branching re-iterations and the overall skeletal growth pattern and dynamics.

The single specimen studied here (Fig. 1) comes from a basal Tremadocian radiolarian assemblage, extracted by Lauren Pouille (University of Lille) during her PhD work, after hydrochloric acid leaching from a limestone sample of the Green Point section (level 23) of the Cow Head Group of western Newfoundland, Canada (Pouille et al. 2014). Protoentactinia gracilispinosa is abundant in this assemblage (ca. 15% of over 2,000 identified specimens) and has been well documented, with many different specimens, by Pouille et al. (2014). The reader is referred to this paper for further taxonomic and biostratigraphic details.

Skeleton analysis of the radiolarian specimen was conducted at Ghent University’s Centre of X-ray Tomography (UGCT) in Belgium. The specimen was inadvertently lost during transport after all imaging had been completed. Although the physical specimen is no longer available, the description presented here is based on comprehensive microscopic documentation, acquired prior to its loss. The obtained scans were performed using the custom-built Nanowood micro-CT scanner at 70 kV, 4W and a voxel size of 900 nm. A total of 2322 projections were acquired at an exposure time of 1000 ms/projection, resulting in a total scan time of around 40 min. The projections were reconstructed using Octopus Reconstruction (XRE), following the in-house protocol, during which ring filtering and beam hardening corrections were applied. In this study, stacks of 2D images were processed using the 3D Slicer software. This open-source software is commonly used to extract structural models from images primarily used for clinical and biomedical applications. Image segmentation delineates regions within an image, in this case, structures of the siliceous skeleton. It is essential for visualising certain structures or for quantifying them (measuring volume, surface area and shape properties). The segmentation was performed semi-automatically. By manually scanning all the slices (hundreds of 2D images) in three dimensions (X, Y, Z) and using the segmentation tools provided by the 3D Slicer software, we were able to delineate the areas of interest in the radiolarian specimen by selecting the corresponding grey levels. Each part or segment could then be coloured, facilitating the overall interpretation of each branch within the 3D volume.

Colouring dissection allowed us to recognise eight successive stages of spicular branchings that result in the formation of a subspherical spicular test displaying a practically empty inner space (Fig. 2). Table 1 lists the different types and number of new branchings, as well as the total number of new rays generated at each growth stage. The vast majority of branchings is of trifurcating type; however, some rare new branchings are single-rayed or bifurcating, quadrifurcating, pentafurcating and hexafurcating (see also Suppl. material 1: fig. S1). Only one single case of a heptafurcating branching was observed.

The first stage of successive branchings is represented by the formation of the initial bar-centred entactinarian spicule displaying two terminal trifurcating branchings developed at the two extremities of the median bar (Fig. 2A, B). The total number of both new branchings and rays increases rapidly up to stage 5, after which they both decrease. It is also worth noting that, after stage 2, only a fraction of existing rays leads to the development of new branchings. For example, out of the 18 new rays developed at stage 2, only 17 of them produce new branchings at stage 3, as a combination of two bifurcating, 14 trifurcating and one quadrifurcating branchings that result to a total of fifty new rays. Out of these 50 new rays (formed during stage 3), only 35 of them produce new branchings during stage 4. More remarkably, only a third of the rays that are newly formed during stage 4 branch out during stage 5. Therefore, it can be argued that the dynamic of skeletal growth decreases after stage 4, even though stage 5 records the maximum number of rays ever produced for this specimen.

In terms of spatial development, nearly half of the final subspherical spicular test is formed at the end of stage 3. Starting with stage 4, shell growth is clearly unidirectionally and asymmetrically orientated, away from the initial spicule, establishing a clear heteropolarity in the skeleton growth and structural organisation, forming at the same time a more complete subspherical test. The distally orientated development of the shell is more obvious during stage 5. Most of the branches developed during stage 6 are generated on the opposite side of the initial spicule and in such a way as to form a subspherical shell. The few new branchings/rays generated during stages 7 and 8 are also developed so as to increase the subspherical aspect of the test.

Based on Tremadocian (Lower Ordovician) material from Nevada, the genus Protoentactinia was first introduced by Kozur et al. (1996), as a spicular test initiated by a median bar that bears three rays at each end. The external subglobular test was described as being formed by a repeatedly branching pattern of spicules, which accommodated a nearly spherical inner empty space. Our observations confirm the assessment of Kozur et al. (1996) and further describe precisely the number of iterative branching stages, the number of new rays formed at each successive stage, as well as the pattern of their spatial distribution.

It is noteworthy that, during the first two stages of skeletal growth, the type of branchings is always trifurcating and, as this type represents the vast majority of total branchings formed during the skeleton formation, it may very well be that the trifurcating type is the standard and that all the other types of branching are possible malformations or simply part of the infraspecific variability of the species. However, in order to assess this, additional material needs to be analysed in the future. Although three specimens of Protoentactinia gracilispinosa were illustrated recently in Jiani Sheng’s PhD thesis (Sheng 2024), their analysis is still pending.

Heteropolar development of a polycystine radiolarian skeleton has been recognised until now only for the radiolarians of the order Nassellaria. Here, we document for the first time the heteropolar growth of the test in a primitive entactinarian species. Indeed, the skeleton growth of Protoentactinia gracilispinosa recalls the heteropolar and unidirectional skeleton development described in Nassellaria, but without the axial symmetry characteristic of nassellarians, as the end product of skeleton growth of Protoentactinia gracilispinosa is a subspherical form.

Few studies provide insights into the function of polycystine radiolarian skeletal morphology (i.e. Matsuoka (2007)). We suggest that their siliceous skeleton might have served as an anchoring structure for a better function of the axoflagellates and cytoplasmic extensions. We also speculate that P. gracilispinosa had a cell structure similar to the periaxoplastitid-type, as described by Hollande and Enjumet (1960, p. 21, figs I.3 and I.4) from cellular observations of living radiolarians from the western Mediterranean. If such was indeed the case, we may imagine that the axoplast of P. gracilispinosa was likely positioned close to the initial spicule and that axoflagellates were all orientated antapically, away from the axoplast. Such an axoflagellate organisation would be consistent with the unidirectional and heteropolar skeletal growth of P. gracilispinosa, in spite of its final subspherical form.

Our analysis confirms the initial description provided by Kozur et al. (1996) for Protoentactinia gracilispinosa; indeed, the subglobular “spongy” test of this entactinarian species is formed by repeated branchings of rays starting from the two trifurcating branchings of the initial spicule formed out of each termination of the median bar. The skeleton is formed subsequently through seven additional stages of reiterated branchings, but the subglobular form of the test is essentially developed during stages 3, 4 and 5. The maximum number of branchings is manifested in stage 5. The total number of new rays reaches also its maximum at stage 5, after which the total number of new rays decreases during stages 6, 7 and 8. Unidirectional and heteropolar skeletal growth (similar to the one observed in nassellarians) is established here for the first time using micro-CT analysis. Heteropolarity is expressed by the directional asymmetry in the skeleton growth and architecture, starting with the very eccentric position of the initial spicule at the edge of an empty spherical space, as described by Kozur et al. (1996).