Ancient deep ocean as a harbor of biotic innovation revealed by Carboniferous ophiuroid microfossils

Determining the age of biotic innovation is one of the most debated topics in evolutionary biology. Whereas it has long been the purview of paleontology to assess the temporal scale of evolutionary change through the study of fossil evidence, the emergence of molecular clock inference (Zuckerkandl and Pauling, 1965) has revolutionized the dating of clades. During the past few decades, paleontologists and molecular tree analysts have joined forces to produce meaningful dated trees (e.g., Donoghue and Benton, 2007). However, the accuracy of these trees and, consequently, their potential for temporalizing biotic evolution and Earth history depends directly on the quality of the underlying molecular and fossil evidence.

The fossil record is inherently incomplete. Nevertheless, many gaps in our knowledge result from overly selective sampling and ignorance of fossils and fossil-bearing facies outside of our a priori expectations. All too often, the available fossil record is only superficially explored, leaving many potentially informative sources of evidence untapped. This is especially true for organisms with a multi-element skeleton, such as vertebrates and echinoderms that are notoriously undersampled because only rare finds of intact skeletons are considered, although fragmented or dissociated skeletal remains are significantly more abundant and widespread geographically, temporally, and paleoenvironmentally. Not only does this compromise the completeness of a group’s fossil record per se, but it also eliminates entire paleoenvironments, such as deep-water settings (e.g., Thuy et al., 2014), where preservation of intact skeletons is extremely unlikely. Therefore, a better understanding of how isolated skeletal remains can be confidently identified has the potential to unlock a substantial and previously ignored fossil record and, thus, contribute to betterinformed evolutionary trees.

An organismal group where the completeness gap between fossil and molecular data is particularly evident are the ophiuroids (brittle stars), the most speciose of the five extant echinoderm classes. They are among the marine invertebrate clades with the best-resolved molecular phylogeny (Christodoulou et al., 2019), yet knowledge of their fossil record is still sparse. This results largely from a strong taphonomic overprinting of the fossil record of their multi-element skeletons. The ophiuroid skeleton is composed of a multitude of calcitic plates that disaggregate into sand-sized microfossils within hours to days after death as the soft tissues decay (Meyer, 1971). As a result, the bulk of the ophiuroid fossil record consists of these dissociated, microscopic skeletal plates. Detailed morphological studies and comparison with recent relatives have shown that these ophiuroid microfossils, in particular the spine-bearing lateral plates of the arm, are morphologically complex, identifiable to species level, and provide informative suites of character data for phylogenetic analyses even when other characters are unknown (Thuy and Stöhr, 2011, 2016; O’Hara et al., 2014). These insights have been successfully applied to selected Mesozoic and Cenozoic ophiuroid faunas, but the bulk of the ophiuroid microfossil record, in particular in the Paleozoic, remains unexplored (Boczarowski, 2001).

Recently, phylogenomic studies have shown that ophiuroids can be excellent tools for exploring large-scale trends in biogeography and evolutionary history (Woolley et al., 2016; Bribiesca-Contreras et al., 2017). Combining these phylogenomic insights with a well-known, stratigraphically constrained fossil record holds significant potential for establishing Ophiuroidea as a powerful model organism for studying macroevolution correlated with Earth history by using direct fossil evidence.

Here, we contribute to filling one of the most critical gaps in the ophiuroid fossil record using new fossils from Carboniferous deep-water sediments of Oklahoma (USA). Although these fossils are preserved only as fully disarticulated skeletal plates, the pristine skeletal microstructure preservation allows for definitive identification and inclusion into a morphology-based phylogeny. The new fossils provide compelling evidence that the basal diversification of crowngroup ophiuroids took place in deep-water settings some 50 m.y. earlier than predicted by molecular estimates.

We systematically screened 81 micropaleontological samples, consisting of sieve-washed residues of bulk sediment samples, taken at Dutton Ranch, northeast of Ardmore, central southern Oklahoma (Fig. 1; 34.3057810°N, 96.9837620°W). The most abundant and bestpreserved remains (described herein) were retrieved from a sample taken from blue to olive-gray shales of Tennant et al.’s (1982) unit 9BC, ~5.5 m above the base of the unit, corresponding to the Bostwick Member of the Lake Murray Formation and dated to the latest Bashkirian (Atokan, Lower Pennsylvanian, Upper Carboniferous, 313 Ma). The unit was deposited in the northern Ardmore Basin of the Southern Oklahoma aulacogen, a northwestern extension of the eastern subequatorial Panthalassic deep ocean (Fig. 1). Stratigraphic sequence, clay mineralogy, sediment grain size and color, organic carbon content, and the taxonomic composition of the co-occurring ostracod fauna all suggest deposition at deep shelf to upper slope depths resulting from a rapid transgression (Knox, 1990; Boardman et al., 1995).

Ophiuroid microfossils (~400 in total) retrieved from the Dutton Ranch samples include various types of arm plates (single ambulacrals and pairs fused into vertebrae; and lateral, ventral, and dorsal arm plates) as well as disc plates (radial shields and oral plates) (Fig. 2). In agreement with previous studies (Thuy and Stöhr, 2016), the lateral arm plates have the greatest morphological complexity and suggest the presence of five types corresponding to at least five species. Two types of lateral arm plates can be identified as being from crowngroup ophiuroids, i.e., the clade including all extant ophiuroids and their direct stem members, because they (1) lack groove spine articulations, (2) bear a well-defined tentacle notch instead, (3) have an overall shape that suggests the vertebrae are largely encompassed, and (4) bear ridges and knobs connecting lateral arm plates and vertebrae instead of protruding pegs (Thuy and Stöhr, 2011). A formal taxonomic assessment of these two lateral arm plate types exceeds the scope of this paper and is not required to support our conclusions. Therefore, we preliminarily call them lateral arm plate types A and B.

Type A (Fig. 2H) consists of elongate lateral arm plates with coarsely reticulate stereom on the outer surface and a vertical row of large, freestanding spine articulations on an elevated distal edge. The spine articulations are composed of a single opening encompassed by a pair of arched dorsal and ventral lobes forming a lens-shaped elevation. The inner side has a low and poorly defined vertebral articular knob and a large tentacle notch. The spine articulation shape of type A lateral arm plates as well as their outline and coarse reticulation on the outer surface are typically present in members of the extant ophiuroid order Ophioscolecida, in particular Ophioscolex Müller and Troschel, 1842 (Fig. 3) (Thuy and Stöhr, 2011).

Type B lateral arm plates (Figs. 2A and 2B) are significantly more common than type A, accounting for >90% of the entire lateral arm plate assemblage. They are relatively robust and strongly arched, suggesting a complete enclosing of the vertebrae by opposing pairs of lateral arm plates. The outer surface ornamentation comprises a fine tuberculation, proximally bordered by a band of more coarsely meshed stereom including a central area with a fine horizontal striation and two spurs that establish the position of articulation with the overlapping adjacent lateral arm plate. The inner side of the lateral arm plate has a single, vertical ridgeshaped vertebral articular structure and a deep tentacle notch. The outer distal edge of the lateral arm plates bears a vertical row of small, freestanding spine articulations, each comprising a pair of parallel, horizontal dorsal and ventral lobes encompassing a small nerve opening and a slightly larger muscle opening. This particular type of spine articulation is exclusively present in members of the extant ophiuroid order Amphilepidida; e.g., in Ophiolepis Müller and Troschel, 1840, and Amphilepis Ljungman, 1867 (Thuy and Stöhr, 2016).

Other skeletal remains retrieved from the same samples can be associated with type B lateral arm plates because of similarities in relative abundance, plate size, outer surface ornamentation in the case of ventral arm plates (Fig. 2E) and radial shields (Fig. 2F) and matching respective articular surfaces in case of the vertebrae (Fig. 2D). Although the ventral arm plates and vertebrae are consistent with amphilepidid affinities of type B lateral arm plates, their morphology is not unique to that order. The radial shields, in contrast, have the highly characteristic scalene triangular shape, exclusively present in the Amphilepidida (Thuy and Stöhr, 2016; see the Supplemental Material¹).

To test the phylogenetic position of types A and B within the crown-group ophiuroids, we performed a Bayesian inference analysis using the matrix by Thuy and Stöhr (2016, 2018), including the same list of characters used by those authors and Aganaster gregarius (Meek and Worthen, 1869) as an outgroup taxon. Bayesian inference analysis was performed with the software package MrBayes (Huelsen-beck and Ronquist, 2001), using a modified version of the Juke-Cantor model for morphological data (Lewis, 2001) with variable character states from 2 to 10 (Wright and Hillis, 2014), and using the Mkpars model (Wright and Hillis, 2014), assuming equal frequency for all character states, equal prior probabilities for all trees, and evolutionary rates that vary between sites according to a discrete gamma distribution. We used a compound Dirichlet prior distribution on branch lengths (Rannala et al., 2012; Zhang et al., 2012). Average standard deviations of split frequencies stabilized at ~0.007 after 3 × 10⁶ generations, sampled every 1000 generations. The first 25% of the trees was discarded as burn-in. The consensus tree was examined with the software FigTree version 1.4.4 by Andrew Rambaut (http://tree.bio.ed.ac.uk/software/figtree/). The nexus file used for the Bayesian analysis is available in the Supplemental Material.

The resulting tree confirms that type A is close to the ophioscolecids Ophioscolex and Ophiolycus Mortensen, 1933, and that type B is at the very base of the clade corresponding to the Amphilepidida (see the Supplemental Material). Interestingly, the inclusion of the two Carboniferous ophiuroids improved the level of resolution of the basal crown-group divergences compared to previous attempts (Thuy and Stöhr, 2016, 2018), thus corroborating the pivotal role of basal crown members in understanding deep divergences (Thuy et al., 2021).

The phylogenetic position of lateral arm plate types A and B confirms that they represent the oldest known members of the Ophioscolecida and Amphilepidida, respectively, and of the superorder Ophintegrida collectively (Fig. 3). Our analysis constrains a minimum age for the origin of the two orders and thereby imposes a minimum age for the basal divergence of the ophiuroid crown group, i.e., the split between the superorders Ophintegrida and Euryophiurida, which must have taken place before the Early Pennsylvanian (i.e., before 313 Ma). Earlier studies hypothesized a major turnover of the Ophiuroidea at the end-Permian mass extinction followed by a Triassic radiation of the crown groups (Smith et al., 1995; Twitchett and Oji, 2005), in line with the evolution of other echinoderm classes (e.g., Kroh and Smith, 2010; Gale, 2011). Recent studies, in contrast, suggest a pre-Triassic radiation of the ophiuroid crown group (O’Hara et al., 2014, 2021) and furthermore challenge the concept of a major turnover coinciding with the end-Permian mass extinction (Thuy et al., 2017). Our findings provide the first direct and phylogenetically informative evidence that the ophiuroid crown-group diversification was well under way long before the end-Permian mass extinction. The minimum age for the basal crown-group divergence imposed by our data implies that the early diversification of the modern ophiuroids took place much earlier than previously suggested by molecular clock estimates (O’Hara et al., 2014, 2021) (Fig. 3). This conclusion aligns remarkably well with fossil evidence for a Silurian origin of the modern ophiuroids (Thuy et al., 2022).

In addition to the unexpectedly old divergence time of crown-group ophiuroids supported by the taxa described herein, these fossils are also significant for their depositional setting. Coming from deep sublittoral to shallow bathyal paleo-depths, the new ophiuroid fossils challenge the widely accepted onshore-offshore paradigm according to which major biotic innovation originates in shallow waters and later expands or retreats into greater depths (e.g., Jacobs and Lindberg, 1998). In contrast, they add to the growing evidence that the deep sublittoral and shallow bathyal environment has been a key area for biotic innovation throughout Earth history (Lindner et al., 2008; Pante et al., 2012; Thuy, 2013; Thuy et al., 2014; Forel et al., 2019). More specifically, our conclusions align with the deep-water origin for the Ophioscolecida and Amphilepidida predicted using molecular evidence (Bribiesca-Contreras et al., 2017).

Unaltered deep-water deposits are rarely accessible and notoriously undersampled, in part because of their inconspicuous fossil content. The near-absence of storm-induced burial in deep-water deposits make them much less likely to preserve multi-element skeletons intact (Ausich and Sevastopulo, 1994) and has led them to be overlooked in most echinoderm studies. As a consequence of sampling bias toward more fossiliferous shallow-water deposits, groups that originated in the deep shelf or the upper slope are systematically underrepresented in the fossil record and commonly have extremely long ghost ranges and stark discrepancies between the oldest known fossil occurrences and divergence estimates based on molecular evidence. Our study demonstrates that these issues can be overcome at least in part when (1) deepwater deposits, such as the blastoid-yielding deep-water formations mentioned by Ausich et al. (1988), are systematically screened, and (2) remains of large benthos are sought in the microfossil record, taking into account the full disarticulation of multi-element skeletons under the taphonomic conditions prevailing in most deep-water settings.

We thank the reviewers Bill Ausich, Andy Gale, and David Gladwell for their constructive comments that helped improve an earlier version of this article.