Convergent evolution of spherical shells in Miocene planktonic foraminifera documents the parallel emergence of a complex character in response to environmental forcing

Abstract The spherical encompassing final chamber of the planktonic foraminifera Orbulina universa is a prime example of a complex character whose evolution has been documented by a sequence of intermediate forms. However, the mechanism that induced evolution of the spherical chamber remain unclear. Here we show that shortly after the emergence of Orbulina, documented throughout the oceans, a convergent evolutionary transition occurred in the semi-isolated Paratethys, leading to the emergence of the endemic Velapertina, which occupied a similar niche to Orbulina in the surface waters. Using X-ray computed tomography scanning, we show that the evolution of the encompassing final chamber involved the same sequence of steps in both lineages, combining a progressively spherical shell shape with changes in the position, number, and sizes of apertures. The similarity in the sequence of character acquisitions suggests structural determinism in the way foraminiferal shells are constructed and the presence of natural selection favoring a spherical morphology. Collectively, the emergence of spherical chambers in the two lineages at a similar time suggests that the evolution of this spectacular complex character occurred in response to a singular environmental driver.


Introduction
The evolution of complex characters has been contested since the first formulation of the theory of evolution by natural selection by Darwin (1859). In theory, the fossil record should allow direct assessment of the sequence of changes leading to the emergence of such traits, but because speciation may frequently occur in small, peripherally isolated populations that are rarely preserved but usually accompanied by rapid morphological change Eldredge 1977, 1993;Kelley 1983;Spanbauer et al. 2018), the fossil record often lacks the necessary resolution. A notable exception is the fossil record of marine plankton, such as planktonic foraminifera, which allows species' transformations to be traced through time and in space with unparalleled continuity (Malmgren et al. 1983;Coxall et al. 2007;Pearson and Ezard 2014;Bicknell et al. 2018).
An iconic example of gradual morphological transformation leading to the emergence of a complex character is the evolution of the spherical encompassing final chamber in the planktonic foraminifera Orbulina universa d'Orbigny, 1839, which is completely documented by a series of transitional forms leading from the ancestral Trilobatus (Spezzaferri et al. 2015) through the intermediate Praeorbulina (Olsson 1964) to the descendant Orbulina (Blow 1956;Jenkins 1968;Pearson et al. 1997). This transition occurred in the open ocean, and the transitional forms emerged throughout the cosmopolitan warm-water habitat of the evolving lineage within a very short time, providing several key biostratigraphic data (Kennett and Srinivasan 1983;Wade et al. 2011). The transition does not appear to be associated with a shift in the habitat of the evolving lineage (Pearson et al. 1997) and does not occur in association with any distinct global climatic event, making it difficult to speculate about the trigger for the emergence of the idiosyncratic shell form. However, a similar character evolved independently in an unrelated lineage of Paleogene foraminifera, culminating in the morphologically similar but substantially older Orbulinoides (Cordey 1968), implying that the spherical shape, minimizing the surface-to-volume ratio of the adult shell, may represent a response to a specific lifestyle or a morphological expression of a specific habitat among planktonic foraminifera.
Indeed, there is abundant morphological (Norris 1991), genetic (Weiner et al. 2015), and isotopic (Coxall et al. 2007) evidence for parallel or repeated evolution of specific chamber shapes and shell elements in planktonic foraminifera. Notable examples are the iterative evolution of compressed chambers with a keel (Norris 1991) or of digitate or radially elongated chambers (Coxall et al. 2007;Weiner et al. 2015). The prevalence of iterative evolution among planktonic foraminifera can be explained by the presence of strong constructional constraints, imposing functional limits on the geometric variability of shells constructed by sequential addition of interconnected chambers (Raup 1966). Alternatively, the repeated evolution of similar traits may reflect phenotypic integration, resulting from the existence of a developmental and genetic network controlling the emergence of morphological traits in an organism (Pigliucci 2003). Functional and developmental integration may be both heritable and interconnected, jointly shaping (or rather channeling) the phenotypic landscape of an evolving clade (Müller and Wagner 1996).
In the case of the emergence of the spherical shell shape in planktonic foraminifera, it appears that the Paleogene Orbulinoides (Decima and Bolli 1970) and the Neogene Orbulina (Blow 1956;Pearson et al. 1997) followed a similar sequence of steps during their evolution. However, it was only in Orbulina that the evolutionary trend culminated in the emergence of a complex character, involving at the same time the enlargement of the final chamber and the migration of the sutural apertures over the entire chamber surface in the form of areal apertures. This makes it difficult to speculate about the evolutionary mechanisms and potential drivers of the evolution of the complex character in Orbulina. However, next to these two well-known and globally distributed lineages, there also exists an additional case of the evolution of a spherical shell shape in planktonic foraminifera: the enigmatic Velapertina. The genus Velapertina is endemic to the Central Paratethys, where it appears to have repeated the same sequence of transitional steps as Orbulina, culminating in the acceleration of chamber growth rate combined with apertural displacements. The existence of a potential endemic form with shell morphology similar to the Orbulina lineage was first formally acknowledged by Łuczkowska (1955), who described these forms from the Miocene of the Carpathian Foredeep as Globigerinoides indigena. The independent origin of this lineage from the Orbulina lineage was highlighted by Popescu (1969), who assigned this species to a new genus, Velapertina. Popescu (1969) described two new species of the genus, but these appear to be extreme morphologies with aberrant (kummerform) final chambers and apertural bulla covering the sutural apertures, otherwise identical with Velapertina indigena. A fourth species, Velapertina sphaerica Popescu, 1987, has a morphology consistent with the Orbulina lineage. The origin of the idiosyncratic Velapertina remains unclear, but ever since its discovery, it has been reported only from Miocene deposits in the Paratethys.
Remarkably, the dating of the Miocene formations where the enigmatic Velapertina appears (NN6 Zone; e.g., Filipescu 1996;Hohenegger et al. 2014) implies that the evolution of the final encompassing chamber in this lineage took place very shortly after the emergence of Orbulina (NN4 Zone). This coincidence cast doubts on the nature of Velapertina, because there are only a few well-documented cases of endemic evolution in planktonic foraminifera (e.g., Rögl 1994;Darling et al. 2007;Aurahs et al. 2009;Huber et al. 2020), and because it proved difficult to establish by external morphology alone whether or not it represents a variant of Orbulina (Łuczkowska 1971;Szczechura 1984). Thus, the existence of Velapertina could so far not be used in arguments on evolutionary processes in planktonic foraminifera.
Here, we resolve the nature of Velapertina by revealing the interior shell architecture through X-ray computed tomography scanning of exceptionally well-preserved specimens and analysis of the full ontogenetic sequence from the first chamber (proloculus), through the juvenile, neanic, and adult stages (sensu Brummer et al. 1987), leading to the development of its final encompassing chamber during the terminal stage. We compare the ontogenetic trajectory of Velapertina with Orbulina and constrain the habitat of the taxa by stable isotopic investigation of the shells and discuss the evolutionary implications of the resulting findings.

Materials and Methods
In order to constrain the spatiotemporal distribution of Velapertina throughout the Paratethys and to resolve the degree of its cooccurrence with the Orbulina lineage, we carried out an extensive literature review, compiling all localities where either lineage has been reported ( Fig. 1, Supplementary Table 1). Only localities that provide sufficient stratigraphic constraints to resolve the occurrences to the level of planktonic foraminifera Subzone M5b (Praeorbulina glomerosa Lowest-occurrence Subzone), Zone M5 (Praeorbulina sicana Lowest-occurrence Zone), Zone M6 (Orbulina suturalis Lowest-occurrence Zone), and Zone M7 (Fohsella peripheroacuta Lowest-occurrence Zone) were considered (Wade et al. 2011). The search was carried out by querying the literature through Google Scholar, using combinations of taxonomic (Velapertina, Orbulina, Praeorbulina, indigena, suturalis, glomerosa, circularis) and geographic (Paratethys) keywords. The taxonomy has been harmonized across the papers to the genus level, and in each case the occurrence and biozone has been recorded. In all cases where Velapertina was recorded as co-occurring with one of the other taxa, we made sure that the co-occurrence was reported from within the same sample in the sediment sequences described in each paper. Next to the canonical Velapertina indigena Łuczkowska, 1955, three more species of the genus have been formally described (Popescu 1969(Popescu , 1973(Popescu , 1987, but these have been so far only recorded from single localities, and in our opinion it remains questionable whether the described morphologies represent distinct taxa or extreme forms within a variable species. The terminal-stage morphology is similar between the genera Velapertina and Orbulina, with the spherical encompassing final chamber having multiple small areal apertures (Fig. 2). Thus, we decided to study their relatedness by reconstructing the ontogenetic trajectory preserved in the sequence of chambers preceding the terminal spherical chamber. Velapertina is endemic to the Paratethys (Łuczkowska 1971) and can be studied only using material from this realm. Thus, to resolve the relatedness of Velapertina and Orbulina, we concentrated on specimens of the Orbulina lineage from the Paratethys, assuming that these are representative of populations that could be the nearest relatives of Velapertina. In addition, we have also analyzed one specimen of Praeorbulina from the Pacific, to ensure the morphology of the Paratethyan specimens is representative for the lineage at large. In the Paratethys, we concentrated on three localities ( Fig. 1) of middle Miocene age (Badenian stage in the regional Paratethyan stratigraphy; Kováč et al. 2018), representing the time shortly after the origin of both lineages and exhibiting the best preservation of planktonic foraminifera, making it likely that the initial whorls inside the spherical shells of the studied species are preserved and suitable for X-ray computed tomography scanning and 3D rendering.
At localities Jevíčko and Chiuza and from the Modrany-1 core sample, about 200 g of sediment was collected. Samples were crushed into fragments of about 0.5 to 1 cm 3 size, soaked in tap water, diluted in 3% hydrogen peroxide (H 2 O 2 ) until fully disintegrated, and wet sieved over 63 μm and 150 μm mesh sieves. The residues were then dried at 40°C for 24 hours and split. Planktonic foraminifera were manually picked and identified from the >150 μm size fractions. The sample treatment of material obtained from Hole 872C, leg 144, is given by Pearson (1995). Planktonic foraminifera taxonomy follows the concepts of Blow (1956), Łuczkowska (1971), and Pearson et al. (1997). Initially, 80 individuals of O. suturalis were isolated from the dried residue from Modrany-1, 80 individuals of P. gl. circularis and of O. suturalis from Jevíčko, 80 individuals of P. gl. circularis from the ODP Hole 872C sample, and 80 individuals of V. indigena from Chiuza. These specimens were subsequently transferred to a petri dish filled with water to separate out tests that were not filled with sediment. Specimens with sediment-filled chambers sank to the bottom of the petri dish, while those with empty shell interiors remained floating. From among the presumably unfilled tests, we selected two individuals per species (four individuals for the morphologically more variable Velapertina) with particularly good preservation and well-developed terminal-stage morphology for X-ray computed tomography scanning.
The X-ray computed tomography scanning was performed with the SkyScan 1172 highresolution micro-computed tomography device at the Natural Museum in Prague (Czech Republic). Tube voltage was set to 40 kV, and the current source was 250 μA. No filter was used. Random movement was set to 5. Data were acquired with an angle slope of 0.2°and 180°rotation. The acquired data were processed using flat-field correction and reconstructed by the supplied software NRecon (Bruker), which resulted in an isotropic voxel size of 0.54 μm for the <0.5 mm large specimens. Each X-ray computed tomography scan was visualized and morphometrically analyzed with the software Amira ZIBE edition v. 2019.04 (Stalling et al. 2005; http://amira.zib.de). The shells were segmented into the sequential chambers, and any sediment remains within the chambers were manually removed with the Segmentation editor. The chambers were segmented by using the AmbientOcclusionField module and following the approach of Titschack et al. (2018) and Baum and Titschack (2016) (settings: number of rays: 156; ray length: ranging from 1 to 0.2 mm for every specimen and exceeding the cavity diameter in all other specimens). To separate the individual chambers within the shell, a DistanceMap of the intraspace was calculated as the basis for a CountourTreeSegmentation (persistence value: 0.05; see Titschack et al. 2018). The chamber separation was checked and manually corrected with the Segmentation editor. After this, the size extents (width, length, and height), volume, flatness, and elongation of the chambers, and the size extents (width, length, and height) and volume of the shell at different stages of growth were extracted with the ShapeAnalysis module, following Westin et al. (1997). The segmentation revealed that in all of the scanned specimens of Orbulina and Praeorbulina, and one of the Velapertina specimens, the full series of chambers was preserved and could be resolved. To characterize their coiling geometry, we extracted the centroids of all chambers in these specimens. In addition, we were also able to resolve the coiling geometry in V. indigena specimen 2, where the preservation allowed us to determine the total number of the chambers and to identify the starting point of their spiral. We used the chamber centroids to calculate the growth parameters introduced by Raup (1966) and Caromel et al. (2017): (1) (4) the distance between the chamber's centroid from the coiling axis in an x,y plane (R). For each specimen, we also recorded the total number of chambers and the number of whorls (#W) in the trochospire (Table 1).
We were able to characterize the pre-terminal morphology of all eight scanned specimens, including all four Velapertina specimens, where the earliest part of the chamber sequence of specimens 2, 3, and 4 was not preserved. Based on parameters extracted from only five chambers preceding the final chambers, we performed cluster analyses. In addition, by quantifying the shape of the chambers, the shape of the shell, the expansion rate of the chambers and the expansion rate of the shell for the five chambers before the final chamber (Table 1), we visualized the morphospace using nonmetric multidimensional scaling.
Exceptionally well-preserved specimens of planktonic foraminifera from Chiuza and Jevíčko were used for stable isotopic characterization of the habitat of the studied lineages. Despite a number of studies on the geochemistry of fossil planktonic foraminifera from the Central Paratethys (e.g., Báldi 2006;Kováčová et al. 2009;Scheiner et al. 2018), the only analysis including Velapertina is the study by Durakiewicz et al. (1997) (Locality 6 in Fig. 1). This study indicates a shallow habitat for V. indigena, overlapping with co-occurring O. suturalis and Globigerinoides sp., but the study has been conducted in an interval affected by the occurrence of evaporites (Durakiewicz et al. 1997), indicating that the oxygen isotopic composition of the local seawater may have been affected by isolation and evaporation, potentially biasing the oxygen isotopic results. Therefore, the stable isotopic habitat of the lineage relative to other planktonic foraminifera requires confirmation. The new measurements conducted here involved 15-20 specimens per species (approximately 100 μg), including individuals of P. gl. circularis, Globigerina bulloides, sinistrally and dextrally coiled Neogloboquadrina sp., and the benthic Hansenisca soldanii from Jevíčko, and V. indigena, Globigerina bulloides, Globoturborotalita sp., and Globigerinoides sp. from Chiuza. To account for the known size effect on stable isotopic composition of planktonic foraminifera (e.g., Ezard et al. TABLE 1. Morphological variables used in the cluster analysis (variables marked with an asterisk symbol) and nonmetric multidimensional scaling (Fig. 4) , individuals of P. gl. circularis were further divided into small (<200 μm) and large (>200 μm) specimens, while due to the observed large intraspecific shell size variability, small (<200 μm), moderate (200-300 μm), and large (>300μm) individuals were separated for V. indigena. The stable oxygen and carbon isotopic composition of the picked foraminifera were measured at MARUM, University of Bremen (Germany), with a Finnigan MAT 251 gas isotope ratio mass spectrometer connected to a Kiel 1 automated carbonate preparation device. The instrument was calibrated against the in-house standard (ground Solnhofen limestone), which in turn was calibrated against the NBS 19 standard reference material. Over the measurement period, the standard deviations of the in-house standard were 0.04‰ for δ 13 C and 0.06‰ for δ 18 O. Data are reported in the delta-notation versus VPDB.

Results
Our synthesis of literature occurrences reveals that the Orbulina lineage and Velapertina lineage were widespread throughout the middle Miocene Central Paratethys (Fig. 1). Because most of the records derive from outcrops, which only cover a small part of the middle Miocene sedimentary record, it was not possible to reconstruct the temporal occurrence with a precision higher than foraminiferal Zone M5, Zone M6, and Zone M7 (Wade et al. 2011). At this resolution, the data synthesis indicates that Praeorbulina gl. circularis appeared in the Mediterranean region in the Langhian (Lirer et al. 2019 Following the first appearance of Orbulina, Velapertina indigena evolved in the Central Paratethys in the Serravallian (M7 or NN6 Zone, late Badenian; Hohenegger et al. 2014;Kováč et al. 2018). We identified 17 Serravallian localities where Velapertina was found distributed throughout the Central Paratethys (Fig. 1). Of these, Velapertina co-occurred with Orbulina at 10 localities (co-occurrence implying the species were reported in the same sample), and there are only 9 localities, mainly in the Vienna Basin, where Orbulina has been reported but Velapertina has not. Within the stratigraphic resolution of our literature data synthesis, it is impossible to interpret whether the lack of Velapertina at some localities indicates an older age of those deposits. We can thus only conclude that at the given resolution of the Paratethyan stratigraphy, Orbulina and Velapertina had broadly overlapping distributions, with welldocumented co-occurrences, but we cannot identify where in Paratethys Velapertina originated.
After establishing the biogeography and the pattern of co-occurrence of the two lineages in the Central Paratethys, we analyzed their shell geometry by X-ray computed tomography scanning. Five of the eight analyzed specimens showed well-preserved internal shell features with minimal sediment infill, allowing us to manually reconstruct the pattern of chamber addition from the proloculus. The Praeorbulina specimens had 15 and 16 chambers, the Orbulina specimens had 12 and 13 chambers, and one of the Velapertina specimens had 14 chambers (Figs. 2, 3). The initial whorl of the remaining Velapertina specimens proved to be damaged, with septae missing, and the shell architecture could therefore be reconstructed for specimen 2 only for the last 10 chambers and for specimens 3 and 4 for the last 8 chambers. The segmentation of the shell interior into the successive chambers reveals a pattern with rather constant and similar chamber growth rates throughout the ontogeny for all three taxa (Fig. 3A-C,E,F), with a conspicuous acceleration of growth rate for the final chamber occurring only in Praeorbulina and Orbulina. The average volume-based growth rate for the five successive chambers preceding the final chamber (forming an entire whorl and corresponding FIGURE 3. Ontogenetic trajectories of Praeorbulina glomerosa circularis, Orbulina suturalis, and Velapertina indigena extracted from the X-ray computed tomography scanning data. Different colors and lines denote different specimens. The growth trajectories of the analyzed species are plotted backward from the final chamber to the proloculus: A, growth rate based on the chamber volume; B, growth rate based on the chamber size extents (length, width, and height); C, chamber length and height ratio; D, chamber flatness and elongation ratio; E, cumulative growth rate based on chamber size extents (length, width, and height); F, cumulative growth rate based on chamber volume. For the two incompletely resolved specimens of Velapertina, we assumed for the purpose of data visualization that they also had 14 chambers, but values of chamber shape and size are only shown for chambers that could be fully resolved. The growth trajectories of the analyzed species are indeed similar; however, the shape of the final chamber and the lack of growth rate acceleration for the final ultimate chamber in Velapertina indicate a different growth pattern from the Praeorbulina-Orbulina lineage. to the adult stage; sensu Brummer et al. 1987) was in Praeorbulina, 122% and 126%; in Orbulina, 141% and 155%; and in Velapertina, 127%, 200%, 149%, and 139% (Fig. 3A). Whereas in Praeorbulina and Orbulina, the final chamber became more spherical (similar length and height; Fig. 3C,D), in Velapertina, the final chamber became conspicuously flatter, especially in specimens 2 and 4 (Fig. 3C,D).
The lack of growth acceleration toward the final chamber indicates that Velapertina must have achieved the spherical shell shape in a different manner than Orbulina and Praeorbulina. To assess the ontogenetic trajectory preceding the final chamber, we extracted the coordinates of the geometric centers of all chambers added before the final chamber and used these to describe the shape of the logarithmic spire. We used cluster analysis to visualize the similarity among the specimens. The outcome of this analysis shows higher similarity in the preadult ontogenetic trajectory and shell morphology between the two specimens of Praeorbulina, two specimens of Orbulina, and two specimens of Velapertina, indicating that the analysis likely captures a consistent aspect of the species' pre-adult shell architecture. Next, the analysis reveals that the analyzed shells of P. gl. circularis and O. suturalis follow a more similar ontogeny, whereas the two specimens of V. indigena are different (Fig. 4A). To account for the large morphological variability in Velapertina, also indicated by the larger distance between the two specimens seen in the cluster analysis of the spiral growth parameters (Fig. 4A), we carried out nonmetric multidimensional scaling to visualize the morphospace (Fig. 4B). The results confirm that Velapertina is more variable, but indicate that all four Velapertina specimens differ from Praeorbulina and Orbulina. Thus, both the analysis of the full ontogenetic trajectory (Fig. 4A) and the analysis of the five penultimate chambers (Fig. 4B) suggest that the similar adult shell shape with encompassing final chamber in Velapertina and in the Orbulina lineage conceals a morphologically different pre-adult ontogenetic trajectory.
After analyzing the shell architecture of the two lineages, we constrained the habitat of the lineages by comparing new stable isotopic analyses of their shells from Jevíčko and Chiuza with data from Durakiewicz et al. (1997). Our results at Jevíčko revealed a δ 18 O offset of up to 1.5‰ between planktonic and benthic taxa. The benthic fauna at this site is represented by Hansenisca soldanii, which reveals the most positive (i.e., "coldest") δ 18 O signature. Among the analyzed planktonic taxa, the specimens from the Praeorbulina-Orbulina and Velapertina lineages show more positive (i.e., "colder and thus deeper") δ 18 O values, but these species have a greater enrichment of δ 13 C as an indicator of the presence of symbionts (Birch et al. 2013; Fig. 5A). The Chiuza site, the best locality in terms of shell preservation of V. indigena, lacks benthic foraminifera, but the overlap in the δ 18 O values among all of the analyzed planktonic species indicates that they likely all inhabited essentially the same surface layer. Here, we also found a conspicuous δ 13 C enrichment in V. indigena, especially among the larger specimens, which is of a similar magnitude as in the co-occurring Globigerinoides sp. These results are congruent with the earlier study by Durakiewicz et al. (1997), who recorded a similar δ 13 C enrichment in V. indigena, which is thus isotopically similar to O. suturalis and Globigerinoides sp. Nevertheless, those authors recorded overall more positive δ 18 O values (Fig. 5B), which may reflect the more northerly location of their material or an isotopic enrichment due to evaporation, as evidenced by the presence of evaporites in their section. Collectively, these results indicate that V. indigena likely possessed symbionts, and its stable isotopic habitat cannot be distinguished from that of the Orbulina lineage.

Discussion
Even though the external shell morphologies of the Orbulina lineage and Velapertina indigena are similar (Fig. 2), with spherical shell shape and sutural and areal apertures, the X-ray computed tomography scanning data highlight a different internal shell architecture and a distinct growth pattern in Velapertina (Figs. 3, 4, 6). The high degree of overlap between the successive chambers and the negligible growth rate acceleration for the final chamber found in Velapertina (Fig. 3A,B,E,F) indicate that this form achieved its terminal spherical shell morphology by progressively more overlapping chambers of similar size, whereas in Orbulina, the spherical shell emerges only in the final growth stage as a result of massive growth acceleration of the final chamber that envelops the entire shell (Fig. 6).
Indeed, next to the lack of growth acceleration, the final chamber in Velapertina is less globular (Fig. 3C,D) and does not completely encompass the earlier parts of the shell, as is the case in Praeorbulina and Orbulina (Figs. 2,  6). Instead, the analyzed individuals of Velapertina bear consistent differences in shell architecture before the development of the final chamber (Fig. 4), implying that their pre-adult morphology is not consistent with that of Praeorbulina and Orbulina. In other words, the final chamber is not required to differentiate between the two lineages, because Velapertina is characterized by more overlap between the successive chambers, with centroids localized closer to the coiling axis (R = 87.36), throughout the early ontogeny, whereas in Praeorbulina-Orbulina, the pre-adult part of the shell reveals a more loosely coiled and more evolute growth geometry with less overlap between successive chambers, occupying more whorls and with chamber centroids disposed far from the coiling axes (R = 121.28; Fig. 6). Thus, next to the Paleogene Orbulinoides (Cordey 1968), the Neogene Orbulina (d'Orbigny 1839), and Globigerinatella (Cushman and Stainforth 1945), Velapertina very likely represents the fourth example of the evolution of spherical shell shape among planktonic foraminifera. However, unlike the three other examples, in the case of Velapertina, the ancestor remains unclear. Sediment layers recording transitional forms leading to Velapertina have not yet been described, which likely indicates rapid FIGURE 4. Cluster analysis (based on Euclidean distance and paired group clustering algorithm) of 11 variables describing shell coiling and architecture of two specimens of each of the three analyzed species (A) and a nonmetric multidimensional scaling (based on Euclidean distance) visualization of morphospace occupation including two additional Velapertina specimens (B). The analysis in B is based on parameters describing the ontogeny of chamber and shell shape and volume expansion determined from X-ray computed tomography scanning data of the five chambers added before the final chamber. All analyzed variables are listed in Table 1. The analysis indicates that the convergent adult shape of Velapertina indigena is the result of a different ontogenetic sequence than in the Praeorbulina-Orbulina lineage. speciation within a geographically limited population, akin to the concept of punctuated equilibria Eldredge 1977, 1993). On the other hand, the ancestor ought to be sourced from within the globigerinid taxa that were present in the Paratethys Sea before the emergence of Velapertina. Although the emergence of supplementary apertures in planktonic foraminifera also occurred in parallel many times (Wade et al. 2018), it appears more parsimonious to speculate that this trait would have already been present in the ancestor, as was the case in the emergence of the Orbulina lineage (Blow 1956;Jenkins 1968;Pearson et al. 1997). This would imply an ancestry of Velapertina in either the Trilobatus or Globigerinoides lineages, which were both present in the Paratethys and possess supplementary apertures. In our opinion, there is no way to pinpoint the ancestor any closer at this time, because the only remaining character, shell wall texture, is inconclusive. Therefore, in the absence of knowledge of the exact ancestor of Velapertina, we retain the classification of the lineage in a separate genus.
The existence of a fourth independent case of the evolution of spherical shell shape in planktonic foraminifera, in Velapertina, allows us to assess the degree of developmental integration during the emergence of this character. When evolution is directed toward spherical shell geometry, the foraminifera are "faced" with the problem of retaining communication with the exterior through an aperture without compromising the spherical shape. The morphological transitions in Orbulinoides, Orbulina, and Globigerinatella exhibit a trend, where from a certain point along the transition toward a spherical shell, the large primary aperture transforms into a series of small sutural apertures. Remarkably, the same is observed in Velapertina, indicating that this may be the only solution available because of constructional constraints (Raup 1966) during the morphogenesis of subsequent chambers. In Orbulina and Globigerinatella, where the terminal chamber encompasses the  Durakiewicz et al. (1997). FIGURE 6. X-ray computed tomography scanning reconstructions visualizing the shell ontogeny of all analyzed specimens by holding the external shell transparent. Specimens are labeled as in Fig. 4B. The successive positions of chamber centroids are shown in the top row specimens to visualize the shape of the logarithmic spiral from which the Raupian parameters shown next to the analyzed specimens have been extracted. This comparison show that the Velapertina has more overlapping chambers of similar size with chamber centroids disposed near to the coiling axes, whereas in Praeorbulina and Orbulina, the coiling is more evolute with a lower degree of overlap between the successive chambers that occupy more whorls and are disposed far from the coiling axes. The Raupian parameters of the logarithmic spiral (Raup 1966) are provided for each specimen. entire shell, this sequence progressed further, whereby next to the sutural apertures multiple areal apertures appeared throughout the final chamber (Blow 1956). The Paratethyan Velapertina appears to have followed the same path of transformation; because the terminal chamber is not completely encompassing, there are still many sutural apertures, but it also possessed the first clearly identifiable areal apertures appearing in the final chamber (Popescu 1976; see also Fig. 2). This indicates a consistent connection between shell architecture and aperture types during the evolution of a spherical shell, with the emergence of areal apertures as the prerequisite for maintaining the trophic behavior of the evolving lineage while possessing a spherical shell shape.
The independent origin of spherical shells in V. indigena implies that the similar morphological traits of Velapertina and Orbulina are the result of parallel evolution, a common phenomenon in the history of planktonic foraminifera (Norris 1991;Coxall et al. 2007;Weiner et al. 2015). Coxall et al. (2007) documented the emergence of digitate chambers and concluded that there must have been an environmental driver associated with life in a deeper environment that favored this chamber morphology. Our stable isotopic analysis of V. indigena indicates essentially the same habitat and presence of symbionts as in the coexisting Orbulina lineage and the likely ancestral Globigerinoides or Trilobatus (Pearson et al. 1997;Fig. 5). This in line with previous observations reporting no depth parapatry in the evolution of Orbulina (Pearson et al. 1997) and implies that the emergence of the spherical shell shape was not associated with a change in depth habitat. On the other hand, the independent evolution of the same trait in a coexisting lineage in the Central Paratethys indicates that there likely was a strong functional advantage of this shape at that time. This prompts the question of why the spherical shell shape is so advantageous for planktonic foraminifera? As already recognized by Pearson et al. (1997), a spherical shell shape creates the minimum possible surface-to-volume ratio. Surface-to-volume ratio is an important parameter for cell physiology, determining the rate of key processes like gas exchange. Minimizing this ratio implies a lifestyle where gas exchange is not limiting (no oxygen depletion; Burke et al. 2018), and its potential advantage could be that the spherical shell shape requires the least amount of material to achieve the maximum volume.
Having established that Velapertina evolved independently from Orbulina, it also has to be accepted that this peculiar form was endemic throughout its existence to the semi-isolated system of partly interconnected basins of the Central Paratethys (Fig. 1). The middle Miocene planktonic foraminifera composition in this complex marine realm was affected by tectonics and sea-level changes, creating marine gateways that facilitated population exchange between the Mediterranean and the Paratethys. The most prominent gateway during the Langhian was the Trans-Tethyan Trench Corridor, east of the Alps, which remained open until the Serravallian (Rögl 1998;Kováč et al. 2007Kováč et al. , 2017 Fig. 1). The timing of its restriction appears to coincide with the emergence of Velapertina, and because Velapertina has never been reported outside the Central Paratethys, the speciation of Velapertina likely took place in the then semi-isolated Central Paratethys. Here, the species clearly evolved in the presence of the Orbulina lineage, with which it co-occurred throughout the Central Paratethys (Fig. 1) and with which it shared its habitat (Fig. 5). It is unclear from the stratigraphic resolution of our data if V. indigena evolved in the absence of Orbulina in some limited basins of the Paratethys, but certainly Velapertina co-occurred with Orbulina shortly after its emergence (Fig. 1). Assessing the spatial and temporal origin of Velapertina would require more detailed stratigraphic analysis of the basins where the co-occurrence of the two lineages is documented. However, likely because of the restricted connection of the Central Paratethys in the Serravallian, the endemic Velapertina never succeeded in invading the world ocean or the Mediterranean. Like all other planktonic foraminifera inhabiting the Paratethyan basins, it fell victim to the ongoing environmental transformation due to further restriction of the Central Paratethys, culminating in the Tortonian with the Pannonian Lake (Kováč et al. 2007(Kováč et al. , 2017. The isolated nature of the Paratethys, and frequent changes in the regional paleogeography ( Fig. 1) very likely affected the nature of the regional water-column stratification. We note that changes in stratification have been considered important drivers of size disparity (Schmidt et al. 2004) as well as diversity (Lowery et al. 2020) of planktonic foraminifera, and we speculate that it may have been this aspect of the regional environment that led to the emergence as well as the extinction of Velapertina.
As a result of its peculiar biogeography, Velapertina is one of only a few species of planktonic foraminifera with a known place of origin and an endemic distribution. Unlike the bipolar or Indopacific-Atlantic isolation known for some cryptic (Darling et al. 2003(Darling et al. , 2007Morard et al. 2011Morard et al. , 2019Quillévéré et al. 2013;Weiner et al. 2015) or even morphologically distinguishable species (Lazarus et al. 1995;Kučera and Kennett 2000), the distribution area of Velapertina was strikingly small and restricted. It is possible that this unusual degree of restriction facilitated speciation among the enclosed foraminifera, as evidenced by the existence of other endemic species that appear restricted to the Paratethys (Rögl 1994).

Conclusions
The evolution of the spherical shell shape of Orbulina universa is a textbook example of the emergence of a complex character documented by a series of transitional steps, combining chamber growth rate and aperture modifications. Here we show that shortly after the emergence of Orbulina, the enigmatic lineage of Velapertina from the Central Paratethys iterated the same steps of morphological integrations toward spherical shell shape. Through detailed X-ray computed tomography scanning analyses of Praeorbulina glomerosa circularis, Orbulina suturalis, and Velapertina indigena, we revealed a consistent difference in the adult and pre-adult growth patterns of the two lineages, indicating that the Velapertina lineage had a different origin. This indicates that the final spherical morphology with areal apertures in V. indigena results from parallel evolution with the Orbulina lineage. Because of increasing restriction of the biogeographic province, Velapertina remained endemic to the Paratethys throughout its short existence. The fact that it evolved shortly after Orbulina and shared the same habitat indicates a common environmental driver favoring spherical shells in planktonic foraminifera, implying that this complex character evolved in response to a specific environmental stimulus, such as the emergence of a specific habitat.

Acknowledgments
We would like to thank the editor and two anonymous reviewers for their constructive comments on earlier versions of this article. We thank K. Holcová for valuable discussions, M. J. Sabo for proofreading the manuscript, and S. Filipescu for providing material from Chiuza (Romania). The present study used a sample obtained during the Ocean Drilling Program (Leg 144). P.K. was supported by the Collegium Talentum Programme of Hungary, the Slovak Research and Development Agency under the contracts APVV-20-0079 and APVV 17-0555, the Grants for Young Scientists of the Comenius University in Bratislava (UK/214/2019), the German Academic Exchange Service (DAAD) Research Grant for Doctoral Candidates and Young Academic Scientists, the German Academic Exchange Service (DAAD) STIBET Scholarship, the National Scholarship Programme of the Slovak Republic, and the Erasmus+ Programme of the European Commission. We also acknowledge funding through the Cluster of Excellence "The Ocean Floor-Earth's Uncharted Interface."

Declaration of Competing Interests
The authors declare no competing interests.