Polycystine radiolarians are marine pelagic heterotrophic unicellular microorganisms secreting an aesthetically pleasing siliceous skeleton, made initially of amorphous silica (De Wever et al. Reference De Wever, Dumitrica, Caulet, Nigrini and Caridroit2001). Living at different depths of the oceanic water column, including the deep-sea realm, they are present today in all modern oceans, with a diversity estimated at ca. 400–500 different living species (Lazarus Reference Lazarus2005). Polycystine radiolarians constitute one of the oldest lineages of the rhizarian supergroup of eukaryotes, which comprises ecologically diverse amoeboid protists, including foraminifers and cercozoans (Sierra et al. Reference Sierra, Matz, Aglyamova, Pillet, Decelle, Not, de Vargas and Pawlowski2013). Molecular dating estimates place the origin of skeletonless radiolarians in the Neoproterozoic (ca. 760 Ma; Sandin et al. Reference Sandin, Renaudie, Suzuki and Not2025), but the earliest fossil record of polycystine radiolarians can be traced with confidence back to the Cambrian Series 2 (Obut & Iwata Reference Obut and Iwata2000; Pouille et al. Reference Pouille, Obut, Danelian and Sennikov2011; Korovnikov et al. Reference Korovnikov, Sennikov, Danelian, Obut and Pouille2013; Aitchison et al. Reference Aitchison, Suzuki, Caridroit, Danelian and Noble2017; Sennikov et al. 2017; Obut Reference Obut2023), slightly later than the earliest record of hexactinellid sponges (Terreneuvian; Chang et al. Reference Chang, Feng, Clausen and Zhang2017). Thus, the emergence of silica biomineralisation in these two eukaryotic, but phylogenetically distinct, lineages (Danelian & Moreira Reference Danelian and Moreira2004) had profound implications since the early Cambrian, on the oceanic silica cycle and chert accumulation in early Palaeozoic oceans (Tolmacheva et al. Reference Tolmacheva, Danelian and Popov2001; Zhang et al. Reference Zhang, Danelian, Feng, Servais, Tribovillard and Caridroit2013; Chang et al. Reference Chang, Zhang, Clausen and Feng2020).
Polycystine radiolarians have long been proven to be useful in numerous geological and palaeoenvironmental studies (i.e., Danelian et al. Reference Danelian, Asatryan, Galoyan, Sosson, Sahakyan, Caridroit and Avagyan2012; Matsuzaki & Itaki Reference Matsuzaki and Itaki2017) and more recently to decipher plankton palaeobiodiversity dynamics in relation to past palaeoceanographic/palaeoclimatic events (i.e., Danelian & Johnson Reference Danelian and Johnson2001; Kiessling & Danelian Reference Kiessling and Danelian2011; Renaudie & Lazarus, Reference Renaudie and Lazarus2013). Renewed interest in Palaeozoic radiolarians since the early 1980s and the extensive use of the scanning electron microscope allowed substantial progress in their taxonomic and biostratigraphic understanding (see Danelian et al. Reference Danelian, Aitchison, Noble, Caridroit, Suzuki and O’Dogherty2017 for a historical overview). Regarding the early Palaeozoic, interest has focused on understanding radiolarian palaeobiodiversity dynamics in relation to the Great Ordovician Biodiversification Event (GOBE; Noble & Danelian Reference Noble, Danelian, Webby, Droser and Paris2004) or Silurian palaeoceanography (Tetard et al. Reference Tetard, Monnet, Noble and Danelian2017), or to explore patterns in their biogeographic distribution (Danelian et al. Reference Danelian, Noble, Pouille, Maletz, Harper and Servais2013). Finally, Danelian & Monnet (Reference Danelian and Monnet2021) analysed an exhaustive sample-based dataset of middle Cambrian to Silurian radiolarian occurrences to document changes in taxonomic diversity over a 90-million-years’ time interval, and more importantly to evaluate, with modern quantitative techniques, possible biases (i.e., sampling effort) that may affect the observed patterns in diversity changes. We here continue further the previous palaeobiological effort and present results of a factor analysis that was applied to an updated version of the above-mentioned dataset. Our results establish the presence of three evolutionary radiolarian faunas within early Palaeozoic radiolarians that are composed of radiolarian families with simultaneous dominance through their increases and decreases in their species richness over geological time.
1. Material and methods
1.1. Occurrence data
The analyses presented in this study are based on an exhaustive global dataset of all known Cambrian to Silurian radiolarian occurrences in temporally well-dated samples (at the stage resolution) and for which the age is based on evidence independent to radiolarians (i.e., conodonts, graptolites or radiometric ages; see supplementary material available at https://doi.org/10.1017/S1755691025100947; Appendix A). The compilation constitutes an update (both in terms of radiolarian samples/occurrences and taxonomy) of the dataset used by Danelian & Monnet (Reference Danelian and Monnet2021); it includes 55 publications, six of which concern the Cambrian (Dong et al. Reference Dong, Knoll and Lipps1997; Won & Below Reference Won and Below1999; Won & Iams Reference Won and Iams2002; Won et al. Reference Won, Iams and Reed2005; Pouille et al. Reference Pouille, Danelian and Maletz2014a; Sheng et al. Reference Sheng, Kachovich and Aitchison2020), nine the Early Ordovician (Kozur et al. Reference Kozur, Mostler and Repetski1996; Aitchison Reference Aitchison1998; Won et al. Reference Won, Iams and Reed2005, Reference Won, Iams and Reed2007; Maletz & Bruton Reference Maletz and Bruton2007; Won & Iams Reference Won and Iams2011, Reference Won and Iams2013, Reference Won and Iams2015; Pouille et al. Reference Pouille, Danelian and Maletz2014a), 12 the Middle Ordovician (Nazarov & Popov Reference Nazarov and Popov1980; Maletz & Bruton Reference Maletz and Bruton2007, Reference Maletz and Bruton2008; Maletz et al. Reference Maletz, Albanesi and Voldman2009; Pouille et al. Reference Pouille, Danelian, Pour and Popov2013, Reference Pouille, Danelian and Popov2014b; Yi et al. Reference Yi, Yuan, Aitchison and Feng2018; Kachovich & Aitchison Reference Kachovich and Aitchison2020, Reference Kachovich and Aitchison2021; Perera et al. Reference Perera, Aitchison and Northdurft2020; Wu et al. Reference Wu, Huang, Luo, Cai, Guo, Liu, Wang, Aitchison and Zhang2024, Reference Wu, Song, Luo, Liu, Aitchison, Wang, Yan, Chen, Chen and Zhang2025), seven the Late Ordovician (Nazarov & Popov Reference Nazarov and Popov1980; Renz Reference Renz1990; Wang Reference Wang1993; Noble & Webby Reference Noble and Webby2009; Perera & Aitchison Reference Perera and Aitchison2022, Reference Perera and Aitchison2023; Obut Reference Obut2023), and 21 the Silurian (Goodbody Reference Goodbody1986; Renz Reference Renz1988; Furutani Reference Furutani1990; Noble Reference Noble1994; Amon et al. Reference Amon, Braun and Ivanov1995; MacDonald Reference MacDonald1998, Reference MacDonald2004, Reference MacDonald2006a, Reference MacDonald2006b; Noble et al. Reference Noble, Braun and Mc Clellan1998; Kurihara & Sashida Reference Kurihara and Sashida2000a; Noble & Maletz Reference Noble and Maletz2000; Umeda & Suzuki Reference Umeda and Suzuki2005; Jones & Noble Reference Jones and Noble2006; Kurihara Reference Kurihara2007; Noble & Lenz Reference Noble and Lenz2007; Siveter et al. Reference Siveter, Aitchison, Siveter and Sutton2007; Nuramkhaan et al. Reference Nuramkhaan, Horie and Tsukada2013; Tetard et al. Reference Tetard, Danelian, Ghienne and Dabard2014a, Reference Tetard, Danelian and Noble2014b, Reference Tetard, Noble, Danelian, Monnet and Lenz2015). Radiolarian assemblages that were not accurately dated to the stage level were not included in the analysis (e.g., Nazarov Reference Nazarov1975; Goto et al. Reference Goto, Umeda and Ishiga1992; Li Reference Li1995; Aitchison et al. Reference Aitchison, Flood and Malpas1998; Danelian & Clarkson Reference Danelian and Clarkson1998; Danelian Reference Danelian1999; Kurihara & Sashida Reference Kurihara and Sashida2000b; Buckman & Aitchison Reference Buckman and Aitchison2001; Danelian & Floyd Reference Danelian and Floyd2001; Won et al. Reference Won, Blodgett and Nestor2002; Danelian & Popov Reference Danelian and Popov2003; Danelian et al. Reference Danelian, Popov, Tolmacheva, Pour, Neyevin and Mikolaichuk2011; Yan et al. Reference Yan, Fu, Aitchison, Zheng, Niu, Zhou, Xiao and Buckman2025). The constructed dataset contains 16 families, 89 genera and 446 species through the 21 time-bins covering the Miaolingian (Cambrian) to Pridoli (Silurian) interval (ca. 509 to 419 Ma). Some of the analysed species are left in open nomenclature (i.e., ‘sp. A’) or are of doubtful species identification (i.e., ‘cf’). The taxonomy of the dataset follows essentially the recent updates by Caridroit et al. (Reference Caridroit, Danelian, O’Dogherty, Cuvelier, Aitchison, Pouille, Noble, Dumitrica, Suzuki, Kuwahara, Maletz and Feng2017) and Noble et al. (Reference Noble, Aitchison, Danelian, Dumitrica, Maletz, Suzuki, Cuvelier, Caridroit and O’Dogherty2017).
1.2. Palaeobiodiversity
The global biodiversity through time of the studied radiolarians is estimated primarily at the species taxonomic rank by using two traditional incidence-based binning approaches and two modern occurrence-based resampling methods (Fig. 1). First, species richness is calculated as total diversity, defined as face-value range-through diversity, including single-interval taxa (for details see Alroy Reference Alroy2010; Foote Reference Foote2000); it corresponds to the count of all documented taxa per time-bin, in addition to interpolated discontinuous taxon ranges. Total diversity is here complemented with the diversity of boundary-crossers, which counts the number of common taxa between two successive time-bins. Second, to overcome uneven sampling and preservation among the considered time intervals (Close et al. Reference Close, Evers, Alroy and Butler2018), species richness is calculated by shareholder quorum subsampling (SQS; Alroy Reference Alroy2010; here, results are illustrated at a quorum/coverage of 50) and by corrected first-order jackknife (cJ1; Alroy Reference Alroy2020). Each of these methods reduces the effect of sample size on diversity estimates relative to using simple counts of species (Kerr & Alroy Reference Kerr and Alroy2021).
Palaeobiodiversity (species richness) of early to mid-Palaeozoic radiolarians according to face-value counting (total diversity and boundary-crosser diversity) and sample-standardised estimates (coverage-based rarefaction with a quorum of 50, SQS; and corrected first-order jackknife, cJ1).
Furthermore, the fossil record completeness of each geological stage considered is evaluated by means of a classical size-based rarefaction analysis (supplementary Fig. 1; Gotelli & Colwell Reference Gotelli and Colwell2001; Chao et al. Reference Chao, Gotelli, Hsieh, Sander, Ma, Colwell and Ellison2014). Also, insufficient sampling is highlighted by comparing the actual coverage to the maximum observed within-stage total diversity (see Marcot et al. Reference Marcot, Fox and Niebuhr2016). Early Palaeozoic radiolarian diversity is further investigated by evaluating taxonomic changes through time (i.e., origination and extinction components) with a poly-cohort analysis (supplementary Fig. 2), which enables to graphically investigate and compare survivorship and prenascence of taxonomic assemblages through time (Raup Reference Raup1978; Van Valen Reference Van Valen1979; Foote Reference Foote2001; Monnet et al. Reference Monnet, Bucher, Escarguel and Guex2003). By measuring the remaining percentage of taxa through time within each cohort separately (i.e., the pool of species present in a given stratigraphic interval), this analysis allows to localise potential time breakpoints in the dynamics of origination or extinction: typically, in the interpolated contour graph of survivorship and prenascence log values, contour lines are expected to be parallel to the diagonal only if the extinction and origination rates are constant through time (for further details see Monnet et al. Reference Monnet, Bucher, Escarguel and Guex2003; Brayard et al. Reference Brayard, Escarguel, Bucher, Monnet, Brühwiler, Goudemand, Galfetti and Guex2009).
1.3. Evolutionary faunas
To further investigate diversity dynamics of early Palaeozoic radiolarians, the total diversity (species richness) within families (Fig. 2) is processed to evaluate the possible presence of so-called ‘evolutionary faunas’ (Sepkoski Reference Sepkoski1981; see also Adrain et al. Reference Adrain, Fortey and Westrop1998; Figueirido et al. Reference Figueirido, Janis, Pérez-Claros, De Renzi and Palmqvist2012; Cleal & Cascales-Miñana Reference Cleal and Cascales-Miñana2014; Bault et al. Reference Bault, Balseiro, Monnet and Crônier2022; Rojas Reference Rojas2022; Allaire et al. Reference Allaire, Korn, Balseiro, Monnet and Crônier2025). These evolutionary faunas correspond here to radiolarian families that record simultaneously high species richness through successive stages. The presence of such long-term shifts in taxonomic composition holds the potential to highlight periods of consistent functional and biotic/environmental conditions or of differential responses to their changes. Like Sepkoski (Reference Sepkoski1981), radiolarian evolutionary faunas are here identified by a factor analysis (FA). Resulting factors (supplementary Fig. 3) define the evolutionary faunas, the historical diversity of which is rescaled to the global total diversity (Fig. 3) according to their proportion of explained variance in the FA model. Furthermore, the pervasiveness of the FA-based evolutionary faunas is evaluated by a hierarchical cluster analysis (HCA with Pearson’s correlation and complete agglomeration method; Legendre & Legendre Reference Legendre and Legendre2012) of the within-family total species richness through the successive analysed time-bins (supplementary Fig. 4).
Spindle diagram illustrating the variations in within-family radiolarian total diversity (species richness) across the mid-Cambrian to Silurian interval. Colours refer to the different radiolarian taxonomic orders: purple = Archaeospicularia; orange = Entactinaria; brown = Spumellaria; pink = Albaillelaria; grey = incertae sedis.
Palaeobiodiversity of the three early to mid-Palaeozoic radiolarian evolutionary faunas (pink: Cambrian–Early Ordovician fauna; blue = Darriwilian–Sandbian fauna; green = Katian-Silurian fauna) extracted from the factor analysis of the within-family species richness through geological stages. Palaeobiodiversity of the faunas is rescaled to the total diversity (thick black line) based on the relative proportion of explained variance of their factors.
Interestingly, despite relying on species richness, evolutionary faunas enable to highlight large-scale turnovers in taxonomic composition (rather than fluctuations in global diversity) because they focus on the concomitant and relative dominance of subclades through their richness. These two aspects of biodiversity are relatively independent, and changes in evolutionary faunas (taxonomic composition) may occur without changes in global taxonomic richness. Furthermore, the recognition of evolutionary faunas appears to be hardly sensitive to low sampling. Indeed, time intervals with low diversity due to insufficient data (e.g., the Dapingian; see Danelian & Monnet Reference Danelian and Monnet2021) contribute to the residuals of the factor analysis, and thus they do not contribute to the reconstructed evolutionary faunas (see supplementary Fig. 3). Although insufficient sampling may impact the global diversity, it does not necessarily create turnovers in the relative within-family richness through time: insufficient sampling does not prevent macroevolutionary adaptive innovations in some families leading them to diversify and dominate over other families through time (see supplementary Fig. 4).
2. Results
2.1. Changes in palaeobiodiversity
Early Palaeozoic radiolarian diversity trajectories at the species level and at the stage resolution are presented in Figure 1, both at face-value and sample-standardised taxonomic richness metrics. In general, they confirm the trends recognised by Danelian & Monnet (Reference Danelian and Monnet2021). More specifically, the curves of the sample-standardised taxonomic richness (SQS50) and of the corrected first-order jackknife (cJ1) establish that the obtained diversity patterns are robust, except for the Guzhangian, the Paibian, the Dapingian and the Hirnantian, which still remain undersampled in terms of radiolarian occurrences in the studied samples (see also supplementary Fig. 1).
In short, the analysed data point to: (i) a substantial and protracted diversity increase during the late Cambrian–Early Ordovician interval, peaking in the Darriwilian (Middle Ordovician); the Dapingian drop in total diversity values is due to the paucity of data for this stage; (ii) a continuous diversity decrease after the Darriwilian up to the end of the Ordovician, although no data exist for the Hirnantian; and (iii) a rapid recovery and increase in diversity during the Llandovery (early Silurian), followed by two diversity highs expressed in the Aeronian–Telychian and Gorstian–Ludfordian intervals.
2.2. Evolutionary faunas within early Palaeozoic radiolarians
Factor analysis on species richness within the 16 radiolarian families encountered over the 21 time-bins (geological stages) considered for the Cambrian–Silurian interval (ca. 509 to 419 Ma) produced six major eigenvectors, which explain almost 90 % of the original variance (supplementary Fig. 3). These factors clearly aggregate in clusters of adjacent time units and, therefore, constitute three major radiolarian evolutionary faunas, which follow one another over time (Fig. 3): (i) a Cambrian–Early Ordovician fauna (EF1); (ii) a Darriwilian–Sandbian fauna (EF2); and finally (iii) a Katian to Silurian fauna (EF3). It is noteworthy that these three evolutionary faunas are clearly reflected in their origination/extinction rates, which also cluster in three distinct successive intervals as highlighted by the polycohort analysis (supplementary Fig. 2). Finally, these three evolutionary faunas are also recovered by the cluster analysis on within-family species richness, both among families and among stages (supplementary Fig. 4).
The oldest evolutionary fauna EF1 is composed of Cambrian to Early Ordovician radiolarians assigned to the families Echidninidae and Palaeospiculidae (order Archaeospicularia) and the family Protoentactiniidae (order Entactinaria). This evolutionary fauna is complemented in the Early Ordovician by radiolarians assigned to the family Antygoporidae, which includes the oldest representatives of the order Spumellaria, as well as radiolarians assigned to the families Aspiculidae and Proventocitidae (order incertae sedis). Nearly all these families appear to be still present in the Middle Ordovician, although with a much-reduced diversity.
A new evolutionary fauna EF2 emerged during the Darriwilian (Middle Ordovician) and lasted clearly during the Sandbian (Late Ordovician); it is dominated by the spumellarian family Inaniguttidae, and to a lesser extent by the entactinarian families Haplentactiniidae, Entactiniidae and Pylentonemidae. The spumellarian family Haplotaeniatidae is well present in this fauna (EF2), but it contributes more to the next one (EF3).
The Late Ordovician Katian stage records clearly the dominance of the third evolutionary fauna EF3, which emerged timidly in the Early Ordovician and continued throughout the entire Silurian. Although the EF3 evolutionary fauna is composed taxonomically overall by the same families as EF2, it is mostly characterised by the archaeospicularian family Secuicollactidae, the entactinarian family Paleoscenidiidae, the spumellarian family Haplotaeniatidae and the deep-dwelling albaillellarian family Ceratoiciscidae.
It is worth noting that some families may be present beyond the time interval of the evolutionary fauna that they characterise. For example, the family Protoentactiniidae is characteristic of the Cambrian–Early Ordovician EF1, but some of its representatives persisted until the mid-Silurian. Likewise, the family Ceratoikiscidae (order Albaillellaria), although it probably first appeared in the latest Darriwilian (i.e., Protoceratoikiscum clarksoni Buckman & Aitchison Reference Buckman and Aitchison2001; Danelian & Floyd Reference Danelian and Floyd2001), is a main biotic component of the youngest evolutionary fauna EF3, since it is more diverse during the Silurian. A third example is the spumellarian family Haplotaeniatidae, which although it first appeared in the Darriwilian and contributes to the diversity of this interval, is a characteristic component of the EF3 fauna.
3. Discussion
The publication of putative lower Cambrian radiolarians has been recently discussed by Maletz (Reference Maletz2017) who commented on early reports by Braun et al. (Reference Braun, Chen, Waloszek and Maas2005, Reference Braun, Chen, Waloszek and Maas2007) and Cao et al. (Reference Cao, Feng, Feng and Ling2014) and has rightfully considered them as being contaminations with younger radiolarian material. Maletz stressed on the fact that well-preserved middle Cambrian radiolarian fauna establish that the skeleton of Cambrian radiolarians was formed by a spicular skeleton that was not necessarily spherical in shape. A number of recent reports documenting poorly preserved spherical or ellipsoidal microfossils from China, most of them displaying lattice shells, are equally problematic, and may be considered as contaminations from younger material (Chang et al. Reference Chang, Feng and Zhang2018; Ma et al. Reference Ma, Feng, Cao, Zhang, Ye and Gu2019; Zhang & Feng Reference Zhang and Feng2019; Zhang et al. 2021; Huang et al. Reference Huang, Ma, Dong, Xu and Yi2025).
Analysis of the updated radiolarian occurrence dataset confirms overall the previously observed trends in radiolarian palaeobiodiversity (Danelian & Monnet Reference Danelian and Monnet2021). However, the contrast thought to exist between the Late Ordovician and Silurian evolutionary history and dynamics is not confirmed here (see discussion below).
Regarding the Ordovician, the sample-standardised curves suggest a protracted rise in species diversity throughout the late Cambrian Stage 10 up to the Darriwilian, if the Dapingian drop in apparent total radiolarian diversity is due to the paucity of data for this stage. Indeed, radiolarian data for this stage come from a single study (Maletz & Bruton Reference Maletz and Bruton2007) and the rarefaction curve suggests a huge potential to increase the radiolarian diversity for this stage by adding new samples.
Radiolarian assemblages participating in the Cambrian Stage 10 _ Dapingian phase of this radiolarian diversity increase are all part of the evolutionary fauna EF1, which is dominated essentially by the archaeospicularian families Echidninidae and Palaeospiculidae and the entactinarian family Protoentactinidae. Archaeospicularians are characterised by hollow shells made of spicular elements, while Entactinarians initiate their test development from a small initial spicule made of a bar and rays trifurcating from each end of this median bar. These are the radiolarian groups that participated in the ‘Ordovician plankton revolution’ (sensu Servais et al. Reference Servais, Perrier, Danelian, Klug, Martin, Munnecke, Nowak, Nützel, Vandenbroucke, Williams and Rasmussen2016). The subsequent participation in the EF1 evolutionary fauna of the Order incertae sedis families Aspiculidae and Proventocitidae (since the Tremadocian) and of the spumellarian family Antygoporidae (since the Floian) contributed to the sharp continuous diversity increase recorded during the Tremadocian and the Floian.
Our new analysis confirms and further supports the ‘game-changing’ character of the Darriwilian for the evolutionary history of polycystine radiolarians. Indeed, the performed factor analysis highlights the uniqueness of the Darriwilian–Sandbian evolutionary fauna (EF2), which stands out clearly, with sharp transitions, compared to the other two evolutionary faunas (see Fig. 3). Our analysis highlights also the profound significance of the Darriwilian diversification of the spumellarian family Inaniguttidae, made of large porous spherical shells bearing rod-like spines initiated from an internal microsphere, but also of the diverse entactinarians assigned to the families Haplentactiniidae, Entactiniidae and Pylentonemidae. As representatives of the EF1 are still present in the Darriwilian, this stage records the highest sample-standardised diversity for the entire analytical interval. As mentioned already by Danelian & Monnet (Reference Danelian and Monnet2021), it is likely that a major faunal turnover took place within the Darriwilian, with as a result an accumulated species richness at the level of this stage, an inevitable consequence of our analytical level, which is performed at the stage level. It is now clear to us that the Darriwilian evolutionary fauna EF2 emerged in relation to the global cooling and associated oceanographic changes that took place during this stage. Indeed, Rasmussen et al. (Reference Rasmussen, Ullmann, Jakobsen, Lindskog, Hansen, Hansen, Eriksson, Dronov, Frei, Korte, Nielsen and Harper2016) established that a sudden greenhouse to icehouse climatic shift took place during the early Darriwilian, which would have initiated an oceanic ‘conveyor belt’ and polar front shifts (Vandenbroucke et al. Reference Vandenbroucke, Armstrong, Williams, Florentin, Zalasiewicz, Sabbe, Nõlvakh, Challands, Verniers and Servais2010) in the global ocean, a global climatic context that coincides with the main phase of the GOBE (Rasmussen et al. Reference Rasmussen, Ullmann, Jakobsen, Lindskog, Hansen, Hansen, Eriksson, Dronov, Frei, Korte, Nielsen and Harper2016; Kröger et al. Reference Kröger, Franeck and Rasmussen2019; Stigall et al. Reference Stigall, Edwards, Freeman and Rasmussen2019).
Thanks to the additional recent data obtained from Perrera & Aitchison (Reference Perera and Aitchison2022, Reference Perera and Aitchison2023) and Obut (Reference Obut2023) from the Sandbian and the Katian stages, our updated dataset confirms the continuous and rather sharp decline in radiolarian diversity that is recorded in the sample-standardised curves of species richness (SQS, cJ1; Fig. 1) between the Darriwilian and the Katian; regrettably there are still no radiolarian data available for the Hirnantian. This is especially true since the Sandbian has still a great potential to reveal a higher diversity with the addition of new samples (see supplementary Fig. 1).
Although it appeared timidly since the Early Ordovician, the third evolutionary fauna (EF3) became prominent since the Katian, with a clear separation from the previous EF2 fauna. This was an unexpected result of our analysis, which suggests an evolutionary affinity of the Katian fauna with the Silurian one. This new perception contrasts with the one reached a few years earlier (Danelian & Monnet Reference Danelian and Monnet2021), suggesting that the evolutionary dynamics of Silurian radiolarian was distinct compared to the Late Ordovician ones. The emergence in the Katian of the EF3 coincides with the climate cooling recorded since this stage, ahead of the Hirnantian glaciation, and the initiation of the Late Ordovician Mass Extinction (LOME; Rasmussen et al. Reference Rasmussen, Vandenbroucke, Nogues-Bravo and Finnegan2023).
As previously concluded by Danelian & Monnet (Reference Danelian and Monnet2021), the analysed dataset points to a rapid recovery and increase in diversity during the Llandovery (early Silurian), followed by two diversity highs expressed in the Aeronian–Telychian and Gorstian–Ludfordian intervals.
4. Conclusions
Analysis of the updated early Palaeozoic radiolarian occurrence dataset confirms the previously observed trends in radiolarian palaeobiodiversity dynamics: a protracted rise in species richness throughout the late Cambrian Stage 10 to the Darriwilian, followed by a progressive, protracted drop during the entire Late Ordovician and a post-extinction rebound since the earliest Silurian.
Factor analysis of the dataset allows to establish the presence of three radiolarian evolutionary faunas reflecting the concomitant dominance of some families, and provide complementary insights to early Palaeozoic radiolarian macroevolution:
-
• It confirms the evolutionary affinity of the Early Ordovician radiolarian fauna with the one present in the late Cambrian oceans; however, although the oldest evolutionary fauna EF1 is dominated by spicular radiolarians (Archaeospicularia), it also includes entactinarian families (since the Cambrian Stage 10) and the oldest spumellarians (since the Floian).
-
• It highlights the uniqueness of the Darriwilian–Sandbian evolutionary fauna (EF2), the origination of which coincides with the early Darriwilian climate shift and the main phase of the GOBE; the EF2 is clearly characterised by the spumellarian family Inaniguttidae and to a lesser extent by the entactinarian families Haplentactiniidae, Entactiniidae and Pylentonemidae.
-
• The third evolutionary fauna (EF3), although composed taxonomically of the same families as EF2, is characterised more specifically by representatives of the archaeospicularian family Secuicollactidae, the entactinarian family Paleoscenidiidae, the spumellarian family Haplotaeniatidae and the deep-dwelling albaillellarian family Ceratoiciscidae. It is the first time that existing data point to a rather sharp shift in radiolarian assemblage composition during the Late Ordovician (Sanbian to Katian transition) and an evolutionary affinity of the Katian fauna with the Silurian one. This faunal change coincides with the Katian climate cooling and initiation of the Late Ordovician mass extinction.
5. Supplementary material
Supplementary material is available online at https://doi.org/10.1017/S1755691025100947.
6. Acknowledgements
The senior author of this paper is immensely grateful to Euan Clarkson who has been at the origin of his interest in early Palaeozoic radiolarians. Indeed, since the very first day of his arrival as a post-doc at the Department of Geology & Geophysics of the University of Edinburgh, Euan came to welcome him and introduced him to thin sections of cherts from the Southern Uplands (Scotland, UK), in which one could see the large Inaniguttid, and other Ordovician radiolarians, preserved in the cherts. This paper is dedicated to his memory; it is also a contribution to the IGCP project 735 ‘Rocks and the Rise of Ordovician Life (Rocks n’ ROL)’. Constructive remarks from Jonathan Aitchison and an anonymous reviewer helped improve the initial version of the manuscript. Thanks to Yves Candela for his editorial remarks and to Jörg Maletz for discussion on early Cambrian radiolarians.
7. Competing interests
The authors declare none.
