Tiny, ornamented eggs and eggshell from the Upper Cretaceous of Utah represent a new ootaxon with theropod affinities

A new Cretaceous ootaxon (eggshell type) from the Kaiparowits Formation of Grand Staircase-Escalante National Monument is among a growing number of very small eggs described from the Mesozoic. Analyses of two partial eggs (~ 17.7 mm in diameter) and 29 eggshell fragments reveal that this new ootaxon exhibits nodose ornamentation with distinctive branching pore canals that open atop the nodes. Its two-layered microstructure consists of a mammillary layer and a continuous layer with rugged grain boundaries between calcite grains. Although the exact identity of the egg producer is unknown, the eggshell microstructure and small size is consistent with a small-bodied avian or non-avian theropod. The specific combination of small egg size, branching pores, two-layered microstructure, and dispersituberculate ornamentation preserved in this new ootaxon is unique among theropod eggs. This underscores that both eggshell and skeletal fossils of Cretaceous theropods can display a mosaic of transitional morphological and behavioural features characteristic of both avian and non-avian taxa. As such, this new ootaxon increases the diversity of Cretaceous eggs and informs our understanding of the evolution of theropod eggshell microstructure and morphology.

www.nature.com/scientificreports/ Eggshell consists of two structural layers of calcite separated by a discrete boundary (Figs. 3, 4). The inner mammillary layer (ML) is characterized by tabular ultrastructure with tightly spaced mammillae (Figs. 3d,e). The continuous layer (CL) includes horizontal accretion lines and displays irregular grain boundaries extending to the eggshell surface with a columnar extinction pattern under cross-polarized light. Eggshell is 0.31-0.40 mm-thick (0.31-0.59 mm including ornamentation) with a ML:CL thickness ratio of 1:1.1-1:1.5 (excluding ornamentation). Accretion lines are not traceable near nodes and do not arch below nodes to follow surface of eggshell (Fig. 3b). Electron backscatter diffraction (EBSD) Inverse Pole Figure (IPF) maps demonstrate that the nodes are crystallographically discontinuous from the underlying eggshell; crystals comprising the node originate within the CL and splay outward in a wedge shape to form the node (Fig. 4b). Electron microprobe analysis (EMPA) elemental mapping indicates that the concentration of magnesium is higher along the inner and outer surfaces, and reaches a minimum between the middle and outer edge of the ML (Fig. 4a). EMPA analyses of calcium, phosphorus, and sulphur did not reveal any patterns.
Pore openings are located on the nodes (Fig. 5a) and are not readily visible on the surface due to their small size (0.03-0.04 mm diameter). Pore structure in S. storrsi is difficult to characterize owing to the three-dimensional nature of the pore canals, which make them difficult to observe in two-dimensional views, including  www.nature.com/scientificreports/ traditional thin sections (Fig. 5b). Cross sections of pores in radial thin sections appear as cavities under the nodes (Figs. 3f, 5b). Accretion lines dip at these cavities, indicating that they are biogenic structures that were part of the pore canal system (Fig. 3b). Multiple thin sections through the same pore show that canals are sinuous, increasing in diameter away from the mammillary layer and branching towards the surface within the nodes. X-ray computed tomography (CT) scans of the partial egg specimen DMNH EPV.128286 allows for threedimensional visualization of pore spaces, revealing that the pore canals branch in a complex anastomosing system within each node (Fig. 5c). Branches show a high degree of interconnectivity. Many of the branches are difficult to trace as they approach the outer surface of the eggshell and appear to pinch out before they reach the surface.

Discussion
Eggshell has been traditionally organized into structural morphotypes; however, this system is being largely supplanted by the use of parataxonomic binomial nomenclature 5,20 . Preliminary description of S. storrsi placed it within the oofamily Laevisoolithidae based upon eggshell thickness, ornamentation, and egg size, as well as its close microstructural similarity to the laevisoolithid taxon Subtiliolithus kachchhensis 4 . Additional imagery has allowed better characterization of the unique pore morphology of S. storrsi, further distinguishing the oospecies from other laevisoolithid taxa and warranting exclusion from the oofamily as it is presently defined. Further, certain taxa placed within Laevisoolithidae (discussed briefly below) currently strain the diagnosis of the oofamily such that a re-evaluation of the taxonomic group utilizing updated methods is in order, though outside the scope of this treatment. However, because of the similarity of S. storrsi to this oofamily, we include brief descriptions and comparisons of laevisoolithid ootaxa. Ootaxa placed within Laevisoolithidae have previously been described from the Upper Cretaceous of India, Mongolia, Morocco, as well as the Lower Cretaceous of Japan and include Laevisoolithus sochavi, Subtiliolithus microtuberculatus, Subtiliolithus kachchhensis, Subtiliolithus hyogoensis, and Tipoolithus achloujensis (Table 1) [21][22][23][24][25] . Unassigned laevisoolithid eggshell has also been described from the Upper Cretaceous of France 26 . In the traditional classification system, the oofamily Laevisoolithidae was considered part of the Ornithoid Basic (Ratite) morphotype which included eggshells with two microstructural layers (mammillary and continuous) 21,22 . It is characterized by small, ellipsoid eggs with smooth, thin eggshell (0.3-0.6 mm), a mammillary layer one half to two thirds the eggshell thickness (ML:CL thickness ratio of 1:1-2:1), and angusticanaliculate pores. At the time of its characterization, Mikhailov 21 included branching morphology in the definition of 'angusticanaliculate' . In subsequent publications, however, 'angusticanaliculate' refers to straight, unbranching pores 22 . In addition, when Subtiliolithidae became a junior synonym to Laevisoolithidae, it included taxa which did not fit within the revised diagnosis 22 . For example, the diagnosis specifies smooth eggshell but includes the ornamented Subtiliolithus www.nature.com/scientificreports/ microtuberculatus and Subtiliolithus kachchhensis. In addition, the ML:CL ratios of the included taxa exceed the diagnosed range of 1:1-2:1 (Table 1). Stillatuberoolithus storrsi is most similar to eggshell from the Maastrichtian Lameta Formation of India which was initially referred to as "Ornithoid type" 19 . Khosla and Sahni 23 subsequently assigned this eggshell to Subtiliolithus kachchhensis within the oofamily Subtiliolithidae (= Laevisoolithidae 22 ). Subtiliolithus kachchhensis eggshell is two-layered with a distinct boundary between mammillary and continuous layers, 0.35-0.50 mm thick (including ornamentation), displays low, subcircular, irregularly spaced dispersituberculate nodes, shows columnar extinction in the CL, displays tightly spaced mammillae, and has a ML:CL thickness ratio of 1:1-1:2. Like S. storrsi, pore openings are so small that they are not readily visible on the eggshell surface and cross sections of the nodes reveal cavities similar to pores of S. storrsi, though full pore morphology is hard to discern (see Fig. 13.10c in Sahni et al. 19 ). Stillatuberoolithus storrsi can be distinguished from Subtiliolithus kachchhensis and the oogenus Subtiliolithus as a whole on the basis of node morphology (circular and oval nodes vs. subcircular only) and complex, branching pore structure.
The unassigned laevisoolithid eggshell from France and L. sochavai differ from S. storrsi in that they do not possess ornamentation and have different ML:CL ratios (1:2.5-1:3 and 2:1-1:1, respectively). Tipoolithus achloujensis has angusticanaliculate pores with wide pore openings both atop the nodes (similar to Porituberoolithus warnerensis) and on the eggshell surface between nodes. Subtiliolithus microtuberculatus has a much larger ML:CL thickness ratio (3:1-2:1) than S. storrsi. Subtiliolithus hyogoensis has conical, tightly spaced, and often coalescing nodes in contrast to the discrete, flattened nodes of S. storrsi.
Ootaxa outside Laevisoolithidae bear similarities to S. storrsi as well. Pseudogeckoolithus nodosus from the Maastrichtian Tremp Formation of Spain also has dispersituberculate ornamentation with pore openings situated atop the nodes and has an overlapping eggshell thickness (0.30-0.35 mm); however, it is unlike Stillatuberoolithus storrsi in having a prismatic ultrastructure and a thinner mammillary layer relative to the rest of the eggshell (ML:PL thickness ratio of 1:7-1:9) 33 . Pseudogeckoolithus aff. tirboulensis from the Late Cretaceous of Hungary and Romania differs from S. storrsi in having thinner eggshell and a thinner mammillary layer ratio. In addition, EBSD IPF mapping demonstrates that ornamentation of Pseudogeckoolithus eggshell is crystallographically continuous with the underlying eggshell 34 .
Eggs of similar size and shape to Stillatuberoolithus storrsi have been described from the Lower Cretaceous of Thailand 36,37 . These 18 × 11 mm eggs were initially attributed to theropods but were subsequently identified as anguimorph lizard based upon embryonic remains within the eggs. The thickness of the eggshell averages 0.35 mm (0.45-0.54 mm including dispersituberculate ornamentation) and has funnel-shaped pore canals opening at the nodes. However, the anguimorph eggshell differs from S. storrsi in overall microstructure (a single structural layer), pore geometry (unbranched with the wide end of the funnel at the inner surface of the eggshell, resulting in depressions on the inner surface), and node morphology (nodes are conical).
Gekkoolithid eggs are typically small (8-20 mm diameter), spheroidal to ellipsoidal, and can display dispersituberculate ornamentation associated with complex pore structures, similar to S. storrsi 38 . However, rigid gekkotan eggshell is composed of a single layer of compact columnar calcite crystals ranging from 0.04 to 0.20 mm thick, whereas S. storrsi eggshell is thicker and displays both ML and CL layers. Pore canals in modern Gekko gecko eggshells are retecanaliculate, consisting of a series of labyrinthine channels associated with wedge-shaped concretions that may detach from the inner eggshell surface leaving pits 39 . Moreover, descriptions of branching gekkoolithid pore canals are inversely oriented compared to S. storrsi. Non-branching funnel-shaped gekkoolithid pore canals are also oriented in the opposite direction (wider at the inner surface). Where nodes are present in gekkoolithid eggshell, they are composed of randomly oriented calcite grains continuous with the underlying eggshell (contrary to S. storrsi) 34,39,40 . No descriptions of gekkoolithid eggshell indicate funnel-shaped pore canals which are thinnest at the inner surface, branching into a crystallographically distinct wedge composed of radiating calcite (characteristic of S. storrsi).
Dimorphoolithus bennetti (oofamily Tubercuoolithidae) from the Maastrichtian Hell Creek Formation of Montana also includes pore openings atop nodes. It differs from Stillatuberoolithus storrsi in that it is thicker (0.56-0.85 mm, including ornamentation), displays bimodal dispersituberculate ornamentation consisting of broad tubercles as well as smaller conical nodes, has funnel-shaped pore openings that can occur between nodes, and has a smaller ML:CL thickness ratio (1:2.5) 35 .
Previous studies of Campanian eggshell from the Western Interior have referred eggshell similar to S. storrsi to Porituberoolithus warnerensis 18,29,32 . Zelenitsky et al. 28 described the oospecies Porituberoolithus warnerensis (incertae sedis) from the Campanian Oldman Formation of southern Alberta, Canada; this oospecies is diagnosed as possessing an ornithoid ratite morphotype, with an unbranching angusticanaliculate pore system, pore openings situated on nodes, dispersituberculate ornamentation, eggshell thickness 0.50-0.65 mm, and an ML:CL thickness ratio of 1:2. Zelenitsky and Sloboda 29 later amended the original diagnosis of P. warnerensis to also include "distinctly flattened nodes," but the paper does not include images of thin sections that allow comparisons of microstructure between the eggshells with flattened nodes to the initial description of P. warnerensis with conical nodes. Subsequent publications have grouped eggshell with pore openings atop flattened nodes into Porituberoolithus despite the fact that the eggshell differs microstructurally from the oogenus. For example, Tanaka et al. 18 describe a suite of eggshell identified as Porituberoolithus sp. from the Campanian Fruitland Formation of north-western New Mexico that appears to include both P. warnerensis (NMMNH P-38474a, see Tanaka et al. 18 Fig. 2c) and S. storrsi (NMMNH P-38025a, see Tanaka et al. 18 Fig. 2e). In addition, Welsh and Sankey 32 describe cf. Porituberoolithus eggshell from the Campanian Aguja Formation of Texas that is comparable to S. storrsi. Eggshell with flattened nodes from the Kaiparowits Formation, here assigned to Stillatuberoolithus storrsi oogen. et oosp. nov., differs from the original diagnosis of P. warnerensis in overall thickness (0.31-0.41 mm compared to 0.50-0.65 mm), the ML:CL thickness ratio (1:1.1-1:1.5 vs. 1:2), pore structure and size, and the morphology and microstructure of the nodes (Fig. 6 www.nature.com/scientificreports/ been described for P. warnerensis that can be compared. In contrast to S. storrsi, the pores of P. warnerensis are unbranching and have larger (0.06-0.13 mm in diameter) pore openings. The nodes of most ornamented eggshell (e.g. Elongatoolithidae, Spheroolithidae, P. warnerensis) are extensions of the eggshell beneath, demonstrating crystallographic continuity with accretion lines paralleling the surface of the eggshell 34,41 . In contrast, the nodes of S. storrsi interrupt accretion lines and are crystallographically distinct from the underlying eggshell (Fig. 4), originating within the CL and splaying outward in a wedge shape to form the node (similar to Dromaius). Choi et al. 34 suggest that distinct mechanisms of ornamentation construction account for these differences in crystal structure. The nodes of S. storrsi are distinct both morphologically and microstructurally from P. warnerensis (Fig. 6). It is unlikely that these differences are attributable to diagenesis alone, since the features outlined above recur across multiple specimens, localities, and (likely) geologic formations. It is more parsimonious to assign the Kaiparowits eggshell to a new oogenus and oospecies than it would be to amend the oogenus Porituberoolithus in almost every aspect to accommodate the Kaiparowits eggshell. Based upon published images, eggshell comparable to S. storrsi likely also occur in the Campanian Fruitland, Aguja, and Dinosaur Park formations of New Mexico, Texas, and Alberta, respectively 18,29,32 . However, we refrain from formally reassigning other described eggshell to S. storrsi without examining thin sections.
Pore canal morphology is a key diagnostic feature in fossil eggshell classification, and is primarily characterized utilizing thin sections 21 . While two-dimensional analysis is adequate for determining the shape of most www.nature.com/scientificreports/ pore canals, it offers limited information when applied to more complex pore systems. Thin sections of S. storrsi hint at the intricate pore structure within each node, and display disconnected cavities that suggest a branching morphology. However, the thin sections do not show the complex, anastomosing morphology evident in CT scans of S. storrsi. X-ray computed tomography scanning has been increasingly utilized as a high-resolution, non-destructive method for eggshell characterization with a number of applications 42,43 . Branching pores of varying morphologies occur in the eggshell of several extant and extinct avian taxa, especially within palaeognath birds (Fig. 7). The pore canals of Rhea, Dinornis, and Aepyornis eggshells display relatively simple morphologies that branch within one plane, appearing at the surface as linear arrangements of pore openings, similar to Ornitholithus and Incognitoolithus 44−47 . The branches of pores canals within megapode eggs (Leipoa ocellata and Alectura lathami) are laterally connected by channels running parallel to the eggshell surface 48 . Casuarius and Dromaius pore canals are either unbranched or split into a limited number of branches that terminate in complex reticulate structures at the eggshell surface 44,49 . In contrast, Struthio displays complex, interconnected pore canal systems with some branches terminating before reaching the surface, similar to S. storrsi 44,50 .
Several functional explanations have been suggested for the different branching pore morphologies outlined here. Considering that eggshell thickness is inversely proportional to gas conductance 51 , Tullet and Board 56 and Tullet 57 suggested that the branching pores of Struthio eggs allow for increased diffusion of respiratory gases, noting that the large eggs of palaeognath birds may require branching pore canals to compensate for longer pore length through thicker eggshell and to overcome the larger surface area to volume ratio. It seems unlikely that this functionality would apply to S. storrsi given the small size of the egg; however, when compared to regression lines of eggshell thickness versus egg mass in avian and non-avian theropods, the eggshell thickness of S. storrsi is considerable for such a small egg and is even thicker through the nodes where the pores are located (Julia Clarke & Lucas Legendre pers. comm.). Thus, in S. storrsi, the branching pores may facilitate diffusion across its (relatively) thick eggshell (Fig. 7). Grellet-Tinner et al. 48 suggested that the interconnected canals of megapode birds would allow for lateral diffusion of gases if a pore opening became obstructed, a useful adaptation for eggs that are incubated within buried nests. Determination of the eggshell gas conductance of S. storrsi would allow for interpretations of nesting environment, but accurate assessment requires more complete specimens and awaits additional finds.
The small size and two-layered microstructure of S. storrsi eggshell is consistent with eggs laid by smallbodied avian or non-avian theropods. The continuous layer displays columnar extinction with irregular crystal boundaries typical of theropod eggshell 58 . The mammillary layer can be distinguished in thin section, SEM Figure 7. Comparison of branching pore canal morphology with pore length (equivalent to eggshell thickness) and egg mass plotted on logarithmic scales. In the case of S. storrsi, pore length is not equivalent to average eggshell thickness since the pores occur within nodes (point P is the eggshell thickness at the pores/nodes). Green dashed line is avian regression from Ar et al. 51  www.nature.com/scientificreports/ imagery, and EBSD crystallographic orientation mapping. Notably, a high proportion of calcite grains in S. storrsi have a horizontal c-axis (indicated by blue or green colours in the IPF Y map; Fig. 4b), which is typical of neognath eggshells 34,41,58 . In contrast, other analysed Cretaceous theropod eggshells show calcite grains with a vertically-laid c-axis, and appear reddish in IPF Y maps 34,41,58 . EMPA elemental mapping reveals higher concentrations of magnesium along the inner and outer eggshell surfaces with a minimum concentration near the ML/CL transition, a pattern observed in avian eggshell 59 . When compared to regression lines of eggshell thickness versus egg mass, the values for S. storrsi are more similar to non-avian theropod than to avian eggs (Fig. 7); however, deducing phylogenetic signals in these egg characters is a matter of current study (Julia Clarke & Lucas Legendre pers. comm.). Small-bodied theropods and enantiornithine birds are preserved within the Kaiparowits Formation and represent potential egg-laying taxa; however, assignment of S. storrsi to a skeletal taxon awaits discovery of embryonic remains. The Kaiparowits Formation partial eggs have an estimated diameter of approximately 17.7 mm, assuming a spherical shape. The eggs may have been spherical or oblong in morphology, but regardless of shape, they are among the smallest Mesozoic eggs described. The closest comparably small fossil vertebrate eggs include a Campanian anguimorph lizard at 18 × 11 mm 36,37 , Gobioolithus minor eggs attributed to an enantiornithine at 20-24 × 30-46 mm 60,61 , an avian egg from Mongolia at 15.88 × 25.8 mm 62 , and an avian egg from Brazil at 19.5 × 31.4 mm 63 . The relative rarity of small eggs from other formations may reflect the low preservation potential of small eggs with comparatively thin eggshell. Preservation of eggshell is affected by many factors, including nest site selection (nesting in areas conducive to fossilization), nesting behaviour (open vs. enclosed nests), neonate development (precocial vs. altricial), eggshell structure (soft vs. hard shelled; thin vs. thick eggshell), depositional environment (fine vs. coarse grained sediments), and paleoenvironment (humid/reducing vs. dry/oxidizing) [64][65][66][67][68][69] . The high sedimentation rate (estimated 41 cm/ka 7 ) of the Kaiparowits Formation may have facilitated preservation of the small S. storrsi eggs.
Stillatuberoolithus storrsi is a remarkably small oospecies that displays a unique suite of features common in both avian and non-avian egg taxa. Jackson et al. 5 note that no single egg feature unambiguously separates avian from non-avian theropods. Rather, eggs preserve a suite of characters that mirrors the mosaic of behavioural and morphological adaptations characterizing the period of transition which saw the rise of the avian theropod crown group (Table 2). Previous studies have highlighted typically-avian characters and reproductive behaviours observed in some maniraptoran theropods (e.g., asymmetrical eggs, smooth eggshell, 3-layer microstructure, brooding, monoautochronic ovulation) as well as typically-maniraptoran theropod features observed in some early avian eggs (e.g., ornamentation, two-layer microstructure) and note that some features arise and are lost multiple times 5,20,70-73 . The specific combination of small egg size, branching pores, two-layered microstructure, www.nature.com/scientificreports/ and dispersituberculate ornamentation in the partial eggs described in this study is unique. The tiny, exquisitely preserved, ornamented partial eggs from the Kaiparowits Formation of Grand Staircase-Escalante National Monument not only increase the rich diversity of North American fossil ootaxa but add to our understanding of evolutionary changes reflected in theropod eggs.

Methods
The twenty S. storrsi eggshell fragments housed within the Karl Hirsch Eggshell Collection at the University of Colorado Museum of Natural History were collected by Jeff Eaton of the Utah Museum of Natural History in 1986 from six localities 267-322 m from the base of the Kaiparowits Formation. The nine Denver Museum of Nature and Science specimens of this ootaxon (including both partial eggs) were collected more recently (2011-2014) from six localities. Of the collected specimens, eleven were selected for analyses including imaging, thin sectioning, scanning electron microscopy (SEM), electron microprobe analysis (EMPA), electron backscatter diffraction (EBSD) imaging, and computed tomography (CT) scanning. Prior to analysis, all specimens were serially imaged using a Canon 5D Mark II digital camera with a 65 mm MP-E macro lens mounted on a Visionary Digital P-51 CamLift. Resulting images were combined using Helicon Focus stacking software to generate composite images for each specimen with greater depths of field. Radial thin sections were prepared in-house by embedding specimens in Struers Epofix epoxy and sectioning cured pucks with a Buehler Isomet low-speed saw with a diamond wafering blade. Thin slices were then polished with a Buehler Ecomet II polisher with 600 grit silicon carbide grinding paper and affixed to frosted petrographic slides with Devcon Two Ton epoxy. Each slide was then ground to optical thickness (~ 40 µm) using a Struers RotoPol-35 and Bortys grinding system. Thin sections were examined with a Leica DMR petrographic microscope (with cover slips and Resolve microscope immersion oil) and imaged with an attached Canon 5D Mark II digital camera. A thin section of UCM 1139 was polished with 5 μm silicon dioxide powder followed by a Buehler 0.05 μm alumina suspension for EMPA and EBSD imaging. The specimen was carbon coated before elemental distributions of magnesium, phosphorus, calcium, and sulphur were mapped with a JEOL JXA-8230 electron probe microanalyzer. The maps were made at 15 keV accelerating voltage, a beam current of 10 nA, 2 × 2 μm pixel resolution, and a dwell time of 20 ms per pixel. The carbon coating was polished off the thin section before it was imaged with electron backscatter diffraction (EBSD) on a Hitachi SU3500 SEM at the Colorado Shared Instrumentation in Nanofabrication and Characterization (COSINC) facility in the College of Engineering and Applied Science at the University of Colorado. The EBSD analyses were run with an accelerating voltage of 15.00 kV, a specimen tilt of 70.00°, a hit rate of 70.02%, 12.95 Hz of acquisition, and 2.5 × 2.5 μm pixel resolution. The specimen selected for SEM imaging was manually split in order to provide a freshly fractured surface before being coated with 4 nm of platinum using a sputter coater and imaged with a Hitachi SU3500 SEM at the COSINC facility. ImageJ software was utilized to measure eggshell characters. Specimen DMNH EPV.128286 was also analysed with a North Star Imaging CT scanner at the University of Texas High-Resolution X-ray Computed Tomography Facility. The specimen was imaged at 110 kV for 100 min, generating 1349 slices. Images were rendered in 3D and pore spaces were manually segmented using Dragonfly three-dimensional visualizer software (Object Research Systems, Inc.). The resulting mesh was then exported into MeshLab open-source software for image capture.