Identification of Life-History Stages in Fusulinid Foraminifera

Marine Micropaleontology 122 (2016) 87–98

Contents lists available at ScienceDirect

Marine Micropaleontology

journal homepage: www.elsevier.com/locate/marmicro

Research paper Identiﬁcation of life-history stages in fusulinid foraminifera

Yukun Shi a,b,c,⁎, Norman MacLeod b,c,d,e a School of Earth Sciences and Engineering, Nanjing University, China b The Natural History Museum, London, UK c Nanjing Institute of Geology & Palaeontology, Chinese Academy of Sciences, China d Department of Earth Sciences, University College London, London, UK e Faculty of Life Sciences, University of Manchester, Manchester, UK article info abstract

Article history: The whorls of fusulinid foraminifera preserve a record of each individual's ontogeny, and therefore provide access Received 25 March 2015 to aspects of morphological life-history patterns. Outline-based morphometric analyses of representative fusuli- Received in revised form 1 December 2015 nid specimens assigned to Robustoschwagerina, Sphaeroschwagerina, and Schwagerina show that these genera ex- Accepted 13 December 2015 hibit characteristically different ontogenetic trajectories within a space defined by test size, test shape, and whorl Available online 17 December 2015 number. In Robustoschwagerina and Sphaeroschwagerina the shape of the test altered during development, first exhibiting a spherical prolocular character, transforming to a fusiform shape, and then reverting to a secondary Keywords: fi Fusulinid spherical form. In contrast, Schwagerina exhibits only the rst two stages. Development-based morphological Pseudoschwagerininae transitions also vary among these genera concerning the test size change. Moreover, these patterns appear to Ontogeny be taxon-specific, and so have potential utility for taxonomic identification as well as for understanding fusulinid Geometric morphometric life history. The distinct test inflation of the spherical pseudoschwagerines during their ontogeny raised the ques- Foraminifera tion whether there is a habitat shift in a certain developmental stage. Intriguingly, the patterns of their ontogenetic shape variation appear reminiscent of the morphological reversion that occurs in several small benthic genera such as Tretomphalus for which a late ontogenetic planktonic phase has been proposed. This similarity may have implications for the functional interpretation of spherical fusulinid tests. © 2015 Elsevier B.V. All rights reserved.

1. Introduction is minimal such that the test performs the function of giving shape to the body (especially in benthic species), this is not always the situation Foraminifera are a group of single-celled marine protists, most of (e.g., Hastigerina). Nevertheless, the chambers of multi-cameral forami- which grow hard, internal, single- or multi-chambered shells (referred niferal tests can be regarded as physiognomic in the sense of reflecting to as tests) throughout their life span. Fusulinid foraminifers inhabited aspects of the organism's phylogeny, current taxonomy, ecology, and shallow sea bottoms during the Late Paleozoic, exhibiting fusiform, development. spherical or lenticular tests composed of multiple involute and Generations of taxonomists have observed and commented on the planispiral chambers (Fig. 1). Because of their rapid morphological fact that the sequence of chambers comprising an individual fusulinid change and the association between morphological forms and deposi- test often undergoes various morphological transitions, with changes tional environments, fusulinids have gained a well-deserved reputation that appear to be species-specific. Indeed, ontogeny-based morpholog- as index fossils for Late Paleozoic biostratigraphic and palaeoecological ical transitions have been used as key diagnostic characters for species investigations. However, only a few studies on any aspect of fusulinid identification (e.g., Bolivia, Heterohelix, Tetrataxis, Orbulina). Moreover, palaeobiology have been published to date. the sequential and episodic nature of chamber formation can be used Foraminifer tests exhibit an enormous variety of sizes and shapes. In to subdivide – at least conceptually – the organism's life history into dis- multi-chambered species (such as fusulinids) chambers are added crete stages (Brummer et al., 1986, 1987). Our interest lies in exploring through growth and often assumed to reflect the size and shape of the how the tools of morphometric analysis can be used to test hypotheses protoplasmic body. However, foraminiferal tests are internal skeletons, related to developmental aspects of fusulinid palaeobiology. held entirely within the protoplasm during life. While it may be the case Many researchers have applied morphometric analyses to living and that, for many species, the amount of cytoplasm located outside the test fossilized foraminifera to investigate shell development dynamics, living strategies, and ontogeny (e.g. Brummer et al., 1986; Galeotti and Coccioni, 2001; Hemleben et al., 1985, 1989; Malmgren and Kennett, ⁎ Corresponding author at: School of Earth Sciences and Engineering, Nanjing University, No. 163 Xianlin Avenue, Nanjing 210023, China. 1972; Showers, 1980; Yang and Hao, 1991). Most of these investigations E-mail addresses: [email protected] (Y. Shi), [email protected] (N. MacLeod). have focused on geometric relationships between adult test forms and

Fig. 1. The whole exterior test and axial thin sections of fusulinids exhibiting variable forms. 1. Parafusulina displaying the coiling axis, from Middle Permian of Huagong, Guizhou; 2. Pseudofusulina from Early Permian of Zongdi, Guizhou; 3. Eoparafusulina from the Early Permian of Zongdi; 4. Schubertella from the Middle Permian of Zongdi; 5. Staffella from the Early Permian of Zongdi; 6. Triticites from Early Permian of Zongdi; 7. Sphaeroschwagerina from the Early Permian of Zongdi; 8. Verbeekina from the Middle Permian of Huopu; 9. Schwagerina from the Early Permian of Zongdi; 10. Monodiexodina from the Middle Permian of Xiaoxinzhai, Yunnan. Scale = 1 mm. either taxonomic or environmental parameters. Linear distance mea- history transitions in foraminifera, (2) develop methods of analysis surements (e.g., test diameter, test height) of particular developmental that can be applied quickly and easily to the characterization of forami- stages (e.g., prolocular diameter), have normally been used for this pur- niferal life-history patterns, and (3) evaluate the potential of such re- pose. Such measurements are typically made using a microscope sults to make finer and more comprehensive biologically based (Brummer et al., 1987; Galeotti and Coccioni, 2001; Hemleben et al., distinctions between foraminiferal genera and species, including the 1985; Malmgren and Kennett, 1972; Scott, 1973; Showers, 1980). Re- identification of cryptic species. cent developments in high-resolution X-ray tomography now provide researchers with access to the forms of every foraminiferal chamber, 2. Materials & methods and therefore more morphological characters among foraminiferal ontogeny were successfully captured in three dimensions and revealed 2.1. Materials more details concerning the test geometry (Briguglio and Hohenegger, 2014; Hohenegger and Briguglio, 2012; Speijer et al., 2008). The pseudoschwagerine genera, Robustoschwagerina and A few researchers have collected and/or analyzed information bear- Sphaeroschwagerina, both with spherical adult tests, were chosen in ing directly on the geometry of test or chamber form by digitizing out- this preliminary investigation along with the fusiform schwagerine lines of tests or landmark coordinate locations. To date however, genus Schwagerina representing a morphological outgroup. The bulk specific studies of foraminiferal ontogeny, including the definition of de- of our specimens were collected from the Asselian-Sakmarian strata of velopmental stages, have been made in a very approximate and qualita- Zongdi section in Guizhou Province, illustrated in part by Shi et al. tive manner, and so have led to ambiguous, difficult-to-interpret results (2012), and from the Yishan section in Guangxi Autonomous Region (e.g. Pharr and Williams, 1987; Shan et al., 2006; Yang and Hao, 1991). of South China. Several illustrations of published specimens, mostly So far as we are aware Yang and Hao (1991) constitutes the only pub- types, were also included in our dataset. Specimens with more cham- lished example of a fusulinid ontogenetic investigation that has been bers and, more importantly, clear images in published articles, were se- approached from a comprehensively geometric point-of-view. Never- lected preferentially. All specimens are listed in Table 1 theless, the abundance of fusulinid fossil materials, along with great Robustoschwagerina tumida and Sphaeroschwagerina karnica are type quantities of published images of fusulinid fossil cross-sections that ex- species of their genera. The type species of Schwagerina was, unfortu- pose the varying form of the test through developmental time, make fu- nately, excluded from this investigation owing to the poor quality of sulinids ideal subjects for ontogenetic study. its published photographs. On the subject of fusulinid palaeobiology and biostratigraphy, spe- Images of axial sections through these specimens were used to quan- cies assigned to the subfamily Pseudoschwagerininae have always re- tify form variation throughout ontogeny. These were made by grinding ceived extra attention. This is not only because of their quick the specimen parallel to the coiling axial until the middle of the proloculus diversification as an Early Permian index taxon (Shi et al., 2009; Yang was exposed. The resulting section allowed the sizes and shapes of the et al., 2005), but also for the distinctive test-shape shift, from a fusiform chambers comprising the whorl to be seen and photographed. While it to a subspherical or spherical morphology (Yang and Hao, 1991; Zhou was not possible to assess all the chambers that formed during the et al., 1997). Fusulinid researchers have long suspected that this shift individual's lifespan using these sections, axial sections are the standard signals a change from a benthic to a planktonic life habit (Dunbar, sections from which taxonomists assess the individual's internal mor- 1963; Ross, 1982; Yang and Hao, 1991). While this hypothesis cannot phology for the purpose of taxonomic identification, ecological inference, be tested directly, it can be assessed indirectly with regard to its mor- and phylogeny reconstruction. Digital images of each specimen from the phological character, and evaluated against various predictions drawn Zongdi and Yishan localities were collected using cameras attached to a from sedimentary particle physics and the life histories of putative mod- transmitted light microscope with digital resolutions of 150 dots per ern analogs. inch (dpi) and 72 dpi, respectively. Here, we address this issue within the pseudoschwagerine by conducting a geometric morphometric investigation to describe the 2.2. Morphometric methods morphological test ontogeny of three genera. Accordingly, the primary goals of this investigation are to: (1) explore the potential of modern The whorl stage number was set by counting the number of chamber morphometric approaches of morphological analysis to quantify life outlines exposed in each axial section with each chamber being Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 89

Table 1 Specimens in research sample.

Number of Species Collecting site References specimens

Robustoschwagerina 24 R. xiaodushanica Sheng et al., 1984 8 Zongdi Section, Guizhou, South China 3 (type specimens) Xiaodushan Section, Yunnan, South China Sheng et al. (1984), pl. 1, Figs. 9, 12, 14 1 Zongdi R. guangxiensis Yang and Hao, 1991 3 Yishan Section, Guangxi, South China Four published in Shi et al. (2012), pl. 20, 5 Zongdi R. yishanensis Yang and Hao, 1991 Figs. 1, 4, 5, 13 1 (holotype) Yishan Yang and Hao (1991), pl. II, Fig. 7 R. regularis Yang and Hao, 1991 1 Xiaodushan Sheng et al. (1984), pl. 1, Fig. 8 R. tumida (Likharev, 1934) 1 (type specimen) Darvas, Tajikistan Likharev (1939), pl. 4, Fig. 1 R. sp. 1 Zongdi

Sphaeroschwagerina 10 S. constans (Scherbovich, 1949) One published in Shi et al. (2012), pl. 21, 6 Zongdi (in Shamov and Scherbovich, 1949) Fig. 1 1 Zongdi S. karnica (Schellwien, 1898) Schellwien (1898), pl. XXI, Fig. 9, as 1 (type specimen) Bombaschgrabens Schwagerina princeps S. subrotunda (Ciry, 1943) 2 Zongdi

Schwagerina 8 S. furoni Thompson, 1946 1 Zongdi Shi et al. (2012), pl. 16, Fig. 2 S. jinzhongensis Liu et al., 1978 1 Zongdi Shi et al. (2012), pl. 14, Fig. 20 S. regularis (Schellwien, 1898) 1 Zongdi Shi et al. (2012), pl. 16, Fig. 3 S. rhomboides (Shamov and Scherbovich, 1949) 1 Zongdi Shi et al. (2012), pl. 16, Fig. 5 S. andresensis Thompson, 1954 1 Dona Ana, New Mexico, USA Thompson (1954), pl. 31, Fig. 3 S. emaciata (Beede), emend. Thompson, 1954 1 (neotype) Cowley Co., Kansas, USA Thompson (1954), pl. 25, Fig. 15 S. longissimoidea (Beede), emend. Thompson, 1954 1 (type specimen) Elk Co., Kansas, USA Thompson (1954), pl. 27, Fig. 10 S. aculeata Thompson and Hazzard, 1946 1 Zongdi Shi et al. (2012), pl. 14, Fig. 1 (in Thompson et al., 1946) identified as a discrete developmental event, in fact indicating the period we inverted every other outline across the coiling axis for each growing each half-whorl biologically. In all likelihood these counts do not individual's developmental sequence to remove this artificial pattern of represent equivalent developmental times. However, it provides a way shape variation. allowing us to compare the test size change according to the whorl- Following this inversion all outlines were arranged in a consistent stage count, and further to determine the greatest structural regularity orientation. For species characterized by spherical tests the coiling The full digital image of each specimen was edited to a series of out- axis — a conceptual axis about which the spiral tests of foraminifera lines that exhibit the overall form of the test at each whorl stage. are coiled — crosses the proloculus and intersects the test outlines at Planispiral-involute fusulinid chambers are added in such a way as to axial positions, which may or may not coincide with polar ends of a fu- cover those of previous whorls without leaving a cavity. As a result, in siform test (Fig. 2). Similarly, a vertical axis was also defined as the axis axial section the final chamber oscillates alternatively between the crossing the proloculus and oriented at a right angle to the coiling axis. “upper” and “lower” surfaces, resulting in a somewhat artificial aspect This vertical axis also intersects the periphery of the test at each whorl. of shape variation. [Note: if it were possible to revolve the test about the Together these axes can be used to locate four type 2 landmarks coiling axis the final chamber would always remain in the “upper” or (e.g., extremal points, maximum of curvature, see Bookstein, 1986), “lower” position.] Given the likelihood that fusulinids were oriented on that together, subdivide the peripheral outlines of the test at each the substrate in a consistent way during their lifespan (Murray, 1973), whorl-defined developmental stage into four quadrants.

Fig. 2. Digitization protocol for a specimen image. The outlines of each specimen test at every whorl stage (7th and 12th whorl stages as examples here) were digitized and 4 landmarks were located at the outline's cardinal points (see text). These landmark points were then used to subdivide the outline into quadrants such that each quadrant outline segment could be represented by 25 equally spaced semilandmark points. Outlines of the odd-numbered stages were inverted to insure a consistent orientation with the even-numbered stages in terms of the positions of the ultimate chamber throughout the developmental sequence. Once size information has been removed this procedure ensures all developmental sequence outlines have been represented in a geometrically comparable manner and that all parts of the test participate equally in the characterization of shape variation. The specimen used to illustrate this procedure here is Robustoschwagerina xiaodushanica Sheng et al., 1984, catalogued as ZF105-4-7. 90 Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98

Each whorl-stage test outline was digitized in a counterclockwise di- high-dimensional shape covariance matrix into as few variables as pos- rection from a common starting landmark location and the resulting sible (see MacLeod, 2008; MacLeod et al., 2007a, 2007b). This strategy pixel location coordinates were used to interpolate the outline to a con- was mentioned by Mitteroecker and Bookstein (2011) but these au- stant number of semilandmark coordinate points across the entire thors did not investigate its use in a morphometric context. Experi- dataset. The number of semilandmarks needed to represent each ments using this procedure indicate variable reductions of over 90% landmark-defined outline segment was determined by the shape com- can be achieved routinely with as little as a five percent loss in informa- plexity analysis with the outline fidelity index set to ≥95% (MacLeod, tion content (see above, also MacLeod, 2015; MacLeod and Steart, 1999). 2015). Second, a variety of statistical resampling procedures are avail- Generalized least-squares Procrustes superposition (Rohlf and Slice, able for testing the significance of group separations achieved in the 1990) was used to align these interpolated and resolution-standardized analysis of actual data (e.g., Monte Carlo simulation, bootstrapping, semilandmark data as well as to enable the size and shape aspects of see Manly, 2007). Use of these statistical techniques provides an effec- form variation to be treated separately for life-history characterization. tive guard against over-interpreting the results of a CVA regardless of Principal component analysis (PCA) and geometric modelling of math- the dimensionality characteristics of any dataset (see above). In this in- ematical spaces formed by the PCA axes (see MacLeod, 2009) on the vestigation both procedures were employed. In addition, CVA was also pairwise, r-mode shape covariance matrix (calculated from the shape used to compare and contrast life-history pattern differences among coordinates, see MacLeod, 2010) was used to summarize shape varia- the three genera, i.e. Robustoschwagerina, Sphaeroschwagerina and tion of the whole dataset. As a result, scores obtained from the projec- Schwagerina. Furthermore, the geometric character of the various dis- tion of each whorl stage test shape onto the first component were criminant functions was assessed via the calculation of shape variation used as a sample-specific standard for shape-variation comparisons. models along the trends of each axis through the empirical discriminant The first ten whorl stages were selected to quantify shape variation space (see MacLeod, 2009, 2012a, 2012b for details). across all specimens because, in all specimens, no major changes in Size change was defined as change in the centroid size of the fusuli- test shape were observed beyond the tenth whorl stage. Meeting this nid test during each whorl stage. Based on the assumption that quasi- requirement eliminated all but 40 specimens from our sample and no- equal developmental times are represented by each whorl stage, the tably excluded the paratypes of Robustoschwagerina xiaodushanica size growth rate exhibits an abrupt change at particular points in the each of which exhibited only nine whorl stages. life histories of most specimens, possibly indicating the onset of differ- Since the three to be genera under investigation include different ent developmental stages during ontogeny. To locate the change species, and as the number of specimens was variable for each species, point, or turning point, of the growth rate, the method of circumcircles the structure of the sample could possibly introduce bias into the com- was employed (Fig. 3). The circumcircle of size points of every three parison of inter-generic forms and ontogenetic trajectories. Therefore, successive stages was found to be the intersection of any two perpen- patterns of shape variation within each genus were examined first, dicular lines crossing the midpoints of the lines connecting any two prior to undertaking any comparisons between genera. points in the size vs developmental timing plot. The middle point con- While different patterns of shape variation and/or test sizes across structing the minimum radius among all circumcircles identifies the whorl numbers reflect the varieties of test form-change trajectories turning point, mathematically the inflection point. among specimens, distinctions and similarities between genera are ob- The classic method for locating the inflection point is through linear vious and can be recognized visually. These were explored quantitative- regression or trend analysis, to spot the intersection of two (sometimes ly using canonical variate analysis (CVA, see also Campbell and Atchley, more) straight lines fit to different parts of the ontogenetic curve 1981; MacLeod et al., 2007a). Use of CVA in morphometric analysis has (i.e., Huxley, 1932; Brummer et al., 1987; Masuda, 2009). However, been criticized recently because, when the number of variables greatly since our curves rarely could be represented by a straight line — either exceeds the number of specimens, spurious between-group discrimina- as a whole or as a set of curve segments, the circumcircle method repre- tions can be generated (Mitteroecker and Bookstein, 2011). These au- sents a more robust approach taking advantage of all the information thors suggest a simple procedure of undertaking a PCA of mean group present in the developmental size variation trajectory and can be ap- shapes be used to specify a shape ordination space with subsequent plied to the digitized representation of any curve regardless of its projection of specimen data onto this a priori-defined principal compo- complexity. nent. This procedure has been referred to as “between groups PCA”. Between groups PCA aims to maximize the significance of between- groups separation with regard to that within-group dispersion, by re- placing the total covariance matrix with the between-group covariance matrix. However, MacLeod (2015) found that between-groups PCA does not perform as well as classical CVA in terms of maximizing between-group differences. This result can be explained by noting that the latter procedure involves an eigenvalue scaling step whose purpose is to normalize between-group variance differences (see Campbell and Atchley, 1981) whereas the former does not. It is the inclusion of this scaling step that provides CVA with its superior operational performance. For some applications the straightforward geometric interpretation of between-groups PCA might be desirable (see Mitteroecker and Bookstein, 2011). Nonetheless, if the purpose of the analysis is to test group distinctions or create an identification tool — in other words where group segregation is the point of the analysis — there is reason to believe CVA will always be the better choice, even for morphometric data. The well-known “curse of dimensionality” (see Bellman, 1961) that Fig. 3. Size vs whorl number trajectory of Robustoschwagerina yishanensis.Thisistoillus- trate the manner in which the curve's inflection point was estimated, by calculating the can conspire to compromise the results of a CVA of high-dimensional radii of successive circumcircles for a 3-point moving window along the trajectory and morphometric data can be addressed in two standard ways. First, a pre- identifying the inflection point at the middle coordinate position of the circle with the liminary PCA transform can be used to repack the variance structure of a smallest radius. Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 91

3. Results extreme geometries illustrated by the ends of the PC-2 and PC-3 shape model spectrum represent extrapolations of trends existing in the larger 3.1. The pooled sample shape space dataset, but are themselves represented by only a few specimens in the sample. As noted above, a PCA of the shape covariance matrix computed from the pooled, Procrustes-aligned specimen outline shape coordi- 3.2. Intrageneric comparisons nates for the first ten whorl stages was used to establish a shape space within which informative geometric comparisons can be made. The In order to determine the nature of shape variation within each geometric characters of the first three aspects of this shape space are genus the ten whorl-stage test shapes represented by the pooled sam- shown in Fig. 4. These shape components summarize 80.77%, 5.64%, ple, PC-1 scores were used to create a fully quantified and comparable and 2.92% of the ontogenetic shape variance present in these data re- trajectory of each individual's ontogeny. As noted above, the aspect of spectively. The first principal component (PC-1) captures the distinction geometric variation expressed along this component is the transition between fusiform and spherical tests. As this axis represents the over- from fusiform (low PC-1 scores) to spherical (high PC-1 scores) shapes. whelming majority of the observed shape variation we concluded that Accordingly, ontogenetic trajectories defined by these scores describe this spherical-to-fusiform transition is the predominant mode of the shape change along the ten whorl stage referenced to this empirical- shape variation exhibited by these data. ly determined fusiform-spherical shape index. With regard to PC-2, this aspect of test shape variation reflects mod- However, before proceeding to an analysis of intergeneric ontoge- ifications of the chamber shape in the final stage, especially at the polar netic patterns we must first address the question of whether there is ends of the test body. Since it is possible for the maximum test chord significant inter-specific structure among the ontogenetic trajectories (=length) to be placed on, above, or below, the test midline (see within any of our genera. If such structure did exist species-specific dif- Fig. 4, PC-2 models) this aspect of test outline asymmetry is being ferences would preclude the pooling of specimens from obtaining a expressed along PC-2. Shape variation along PC-3 also involves changes valid genus-level characterization of developmental variation. In order in test symmetry about the central part. Low scores along PC-3 denote to make this test we employed CVA to create a geometric space within specimens characterized by a distinct invagination along the right, which inter-specific distinctions between developmental patterns lower lateral slope, while high scores along this axis denote specimens were optimized. We then tested the null hypothesis that no statistically in which the lower outline invagination switches sides to the left. As significant separation of species means relative to intraspecific variance can be seen from the eigenvalue associated with this axis (2.89%) and exists using a bootstrapped variant of the log-likelihood ratio test the scale associated with it in the plot of the shape space (Fig. 4), this (see Methods). is a very minor aspect of shape variation that probably reflects the vaga- ries of test development. 3.2.1. Robustoschwagerina As the range of PC coordinate values over which we've calculated our Fig. 5 shows the full CVA space for the Robustoschwagerina specimen theoretical shape models is greater than the ranges of most of the spec- ontogenies. The three species ontogenies comprising this genus' record imens' projected positions in the pooled sample shape space (Fig. 4), the all appear to be very different in the context of this species-optimized

Fig. 4. Patterns of fusulinid test shape variation represented by the first three principal components. The PCA based on the pooled sample shape covariance matrix for the first 10 test whorl developmental stages of all specimens. Each shape model, shown in the lower part of the figure, was calculated at equally spaced intervals along PC axes with coordinates for individual shape model calculations indicated by stars and numbers in the PC shape spaces displayed in the upper part of figure. Together this collection of shapes represents a three-dimensional space within which 89.33% of observed shape variation is represented. 92 Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 space. Nevertheless, appearances can be deceiving in a multivariate space and the more pertinent question is whether this level of apparent distinction between groups can prove that there exists a statistically low likelihood that such a result could be obtained via random sampling from a single distribution. Results from the bootstrapped log- likelihood ratio test using 1000 pseudoreplicate samples indicate that there is a 17% chance of obtaining a ϕ value as large as that calculated for the CVA results shown in Fig. 4 (ϕ = 28.34) under a random sampling model. Accordingly, the null hypothesis of no statistically significant between-species differences in ontogenetic trajectory cannot be refuted and these trajectories may be grouped together to create a genus-level model of test shape change over developmental time. Fig. 6 illustrates this curve along with genus-specific error statistics (with outliers) representing the pattern of variable about the mean.

3.2.2. Sphaeroschwagerina The ontogenetic trajectories of our three Sphaeroschwagerina species Fig. 6. Test shape developmental sequence for the ﬁrst ten whorl stages of also appear as distinct groups in the full CVA space (Fig. 7). However, re- Robustoschwagerina specimens based on the fusiform-spherical (PC-1) shape index. For sults from the bootstrap sampling likelihood ratio test using a 1000 these box and whisker plots the horizontal white line represents the sample (=whorl pseudoreplicate scheme, suggest that the chance of obtaining a ϕ stage) median, the black box represents the 75th quartile above and below the median, and the whisker represents the 95% fence above and below the median. Dots represent value of 11.7 as calculated for the CVA results under a random model, the positions of whorl-stage shape outliers that lie beyond the 95% fence limit. The line is 32.3%. Therefore, again, the null hypothesis cannot be refuted and joining the median values represents the single most representative pattern for develop- these trajectories can be combined for the purpose of genus-level char- mental test shape variation for the Robustoschwagerina sample as a whole. acterization of developmental shape transformations. The resultant genus-level ontogenetic shape trajectory, along with the genus- speciﬁc error statistics and outliers, is shown in Fig. 8. 3.3. Intergeneric comparisons

3.2.3. Schwagerina Comparing these genus-level ontogenetic trajectory plots it is clear As the outgroup, only one specimen for each species of Schwagerina that the proloculi of Robustoschwagerina and Sphaeroschwagerina are was present in our analyses in which case there is no opportunity to test both nearly spherical and then either rapidly (Robustoschwagerina)or the sample for intra-group structure using the CVA approach. Yet, we gradually (Sphaeroschwagerina), become fusiform. However, in both gen- can obtain an image of the intra-group patterns of developmental era test shape regains its overall spherical form late in ontogeny in the op- shape variation among these species using PCA. This strategy also allows posite rate-mediated manner: either gradually (Robustoschwagerina)or us to test the previous species identiﬁcations which were based mainly rapidly (Sphaeroschwagerina). Accordingly, these patterns can, in princi- on qualitative observation and semi-quantitative linear character com- ple, be used to characterize these genera. If we are willing to accept the parisons. In this instance PCA was employed to evaluate the possibility provisional assumption of a quasi-equal developmental period for each that intrageneric grouping relationships exist. Results show that all whorl stage, Robustoschwagerina would be characterized as exhibiting a eight species ontogenies are randomly scattered in the PC space much quicker developmental shape shift from prolocular spherical to its (Fig. 9), indicating no obvious intrageneric structure among the ontoge- maximal fusiform (in the vicinity of whorl stage 4) whereas the test of netic trajectories exist in this group. Accordingly, all specimens were Sphaeroschwagerina reaches its maximal fusiform life-history stage later pooled to determine a genus-level characterization of developmental in its developmental sequence (at stage 6 or 7). In Schwagerina the variation. Fig. 9 displays this schwagerine trajectory along with genus- proloculus is also nearly spherical, but soon migrates to the fusiform speciﬁc error statistics. mode (between whorl stages 3 and 4) after which this test shape is main- tained for the remainder of the ontogeny (Fig. 10).

Fig. 5. Scatterplot of projections of 10-whorl stage ontogenetic trajectories for Robustoschwagerina specimens onto the ﬁrst two canonical variate (discriminant) axes Fig. 7. Scatterplot of projections of 10-whorl stage ontogenetic trajectories for based on the outline shape variation fusiform-spherical (PC-1) shape index. Note apparent Sphaeroschwagerina specimens onto the ﬁrst two canonical variate (discriminant) axes between-species separation indicating different species groups exhibit markedly different based on the outline shape variation fusiform-spherical (PC-1) shape index. Note apparent test-shape developmental sequences. between-species separation. Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 93

Fig. 8. Test shape developmental sequence for the ﬁrst ten whorl stages of Fig. 10. Test shape developmental sequence for the ﬁrst ten whorl stages of Schwagerina Sphaeroschwagerina specimens based on the fusiform-spherical (PC-1) shape index. See specimens based on the fusiform-spherical (PC-1) shape index. See Fig. 6 caption for Fig. 6 caption for graphing conventions. graphing conventions.

Whereas the mean ontogenetic sequences of test shape variants ap- differences in the test shape ontogenetic histories of these species. pear quite distinct in Figs. 6, 8 and 10, in some cases, the wide range of Moreover, the fact that a few developmental outliers exist even within inter-specific variation that characterize these genera, plus the overtly this small sample suggests such information may be useful in examining artificial nature of the genus concept itself, raises the question of wheth- questions of species identification, generic assignment and (perhaps) er these differences are more apparent than real. In order to test this hy- ecophenotype. pothesis a CVA was used to assess the extent to which the proposed genus-level distinctions represent real differences among the develop- 3.4. Size change mental sequences among the individuals in our sample. Fig. 11 illustrates results of this CVA analysis for the ten whorl-stage Both Robustoschwagerina and Sphaeroschwagerina exhibit sudden ontogenetic trajectories exhibited by the 40 specimens comprising our inflation of the whorls, with dramatic size enlargement throughout on- sample based on the pooled shape PC-1 test-shape index. The late- togeny (Fig. 2). In contrast the rate of ontogenetic size change is rather stage fusiform Schwagerina trajectories are distinguished clearly from modest in Schwagerina. Notwithstanding these linear trends, ontoge- the late-stage spherical Robustoschwagerina and Sphaeroschwagerina netic size change in these taxa is decidedly nonlinear, and a pattern trajectories. However, the ontogenetic test shape histories of these that, again, has been interpreted in other groups to denote the presence two latter genera can also be distinguished from one another to quite of a sequential series of developmental phases (e.g. Brummer et al., a high degree of confidence (91% correct post-hoc identifications 1987; Huxley, 1932). This pattern can be quantified and the ontogenetic based on proximity of positions in the CVA plant to group centroids curve inflection points calculated for every specimen through all its de- for these two species, 92.5% overall). A bootstrapped log-likelihood velopmental stages. Most specimens are characterized by a relatively ratio test of this result identifies this result as being highly significant low rate of test size change in the early stages. Test size change then en- statistically with an associated centroid-separation probability under ters a life-stage interval of (in many cases) exponential growth. the null model of much less than 1%. If quasi-equal developmental time durations are assumed for the This result indicates that the genus-specific developmental patterns whorl stages, the presence of these different rates of size change may in- shown in Figs.6,8,and10are likely genuine reflections of fundamental dicate the relative timing of changes in the onset or offset signals that control the transition between developmental states. Differential

Fig. 9. Scatterplot of projections of 10-whorl stage ontogenetic trajectories for Schwagerina Fig. 11. Scatterplot of projections of 10-whorl stage ontogenetic trajectories for all speci- specimens onto the first two principal component axes based on the outline shape varia- mens included in the study sample onto the first two canonical variate (discriminant) tion fusiform-spherical (PC-1) shape index. Note scattered positioning of specimens axes based on the outline shape variation fusiform-spherical (PC-1) shape index. Note ap- assigned to different species indicating that each represents a comparably distinct test parent between-species separation indicating that different species groups may exhibit shape ontogenetic trajectory. markedly different test-shape developmental sequences. 94 Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 growth rates of body mass, skeleton size or shape change, have been mutualistic association. If we consider the fusulinid tests as approximat- used to identify ontogenetic stages in crabs, trilobites, fishes, dinosaurs, ing a prolate spheroid, their surface-to-volume ratio will always decline birds, mammals, planktonic and benthic foraminifera (Brummer et al., with an increase in test volume. However, the spherical shell has a larg- 1987; Erickson et al., 2001; Hughes et al., 2006; Huxley, 1932; Jarman, er surface-to-volume ratio than the fusiform of the identical radius, 1983; Masuda, 2009; Starck and Ricklefs, 1998; Yang and Hao, 1991). though this difference may not be significant at the larger sizes Here, the lower rate may indicate retention of the juvenile stage while (Groves et al., 2012). Regardless, with test growth, transforming the the higher rate may signify acquisition of sexual maturity. The onset test from a fusiform to a spheroidal shape, of the same radius presum- timing of the inflection or turning point, defined as the ratio of test ably, represents a plausibly effective strategy to simultaneously host size of each specimen at the turning point to its final adult size, repre- more symbionts and lighten the shell. sents the point of (probable) life-history stage transition. The larval or juvenile stage, probably only consisting of the spherical Onset timing of the turning points among our three genera are quite proloculus, of benthic foraminifera is suggested to be planktonic for different (Fig. 12). While the position of developmental stage turning transport and dispersion (Alve and Goldstein, 2003, 2010; Murray, points overlap for Robustoschwagerina and Schwagerina, the range of 2006). However, during the rest of their lives, fusulinid individuals set- the latter forms a restricted subset of the former confined to the small tleandliveontheseafloor, though water currents, wave-generated tur- (= early stage) end of the distribution. Sphaeroschwagerina exhibits bulence, attachments to other creatures and/or other factors may still the distinctive earlier onset timing than either of the other two genera. entrain their tests and carry them long distances. To transport or dis- Noticeably, adult individuals of Sphaeroschwagerina developed more perse with the water motion, fusulinids need to overcome the gravity whorls than most adult Robustoschwagerina individuals, with larger ac- vertically and the shear stress horizontally, or so-called “settling companying final adult test sizes (Fig. 13). Despite their clear and well- velocity” and “shear velocity” (Jorry et al., 2006; Yordanova and supported character, these results should be regarded as preliminary Hohenegger, 2007; Hohenegger, 2009; Hohenegger and Briguglio, since more extensive research using larger and more fully representa- 2012). Of these, settling velocity is the most important factor determin- tive samples is required to confirm them. ing the suspended condition before transport, and is related closely to the shape, size and density of the foraminiferal test (Hohenegger and 4. Discussion Briguglio, 2012). In terms of density, fusulinids didn't develop a thick wall as some of the Recent homomorphic alveolinids or nummulitids. Abundant fusulinid faunas have been reported worldwide, but are Therefore the folding of their walls in most Permian fusulinid taxa is be- especially common in China, North America and Russia. Within the lieved to result in the enhancement of test density. However, spherical voluminous literature available on virtually every aspect of fusulinid tests of pseudoschwagerine taxa such as Robustoschwagerina and morphology from the 19th century to the present, only a few compre- Sphaeroschwagerina lack these test-wall folds and retained the thin hensive morphometric studies have been published. These have focused wall (c. 0.01 mm in thickness) during most of their ontogenies. There- alternatively on the investigation of ecological adaptations in different fore, the exponential size increase in the test development of these spe- morphological groups (Wang et al., 1982; Zhang and Payne, 2012)or cies would have produced an explicit increase in test size but a intra-specific variations within similar morphological groups (Colpaert continuous decrease in test density. Estimating the magnitude of this et al., 2014; Huang, 2011), and have mainly employed studies of the density reduction in fusulinids is difficult because we do not know adult test. However, the morphological transformation during fusulinid what proportion of organic matter, air or sea water was present in ontogeny, especially among pseudoschwagerine species, contains their chambers. However, assuming the contents of the test chambers abundant information on their development histories that not only had similar physical properties in both pseudoschwagerine and non- can be used to understand their palaeobiology, but also has implications pseudoschwagerine species, the overall reduction of the test density, for their taxonomy and classification. Geometric morphometric analysis and so the settling velocity, could benefit the former by making it rela- offers an efficient approach to the collection and analysis of this infor- tively easier for current actions and/or turbulence to erode and trans- mation so that explorations of further aspects of fusulinid palaeobiology port living individuals. (e.g., function, ontogeny, phylogeny), can be addressed. Dimensionless size of equivalent sphere and shape entropy are pa- Fusulinids are suspected to have hosted photosymbionts as modern rameters proposed to evaluate the influence of test size and shape on larger benthic foraminifera (Ross, 1972; Shi, 2008; Groves et al., 2012), the settling velocity for foraminifera (Hofmann, 1994; Jorry et al., and their shells may have been used functionally as the greenhouses 2006; Hohenegger, 2009, 2011). According to these calculations, both with the keriotheca honeycomb in the interior wall serving as “pore of the test size increase and the shape shift towards the sphericity in cups” to hold these symbionts (Shi, 2008). Accordingly, the large the ontogeny of Robustoschwagerina and Sphaeroschwagerina increase surface-to-volume ratio of the fusulinid tests could benefitthis the settling velocity. This means their adults are more difficult to float

Fig. 12. Onset timing of the turning point of all specimens. Onset timing is deﬁned as the ratio of test size at the turning point to the ﬁnal adult size of each specimen. Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 95

Fig. 13. Adult size (filled rectangle) and whorl stage number (hollow rectangle) of specimens. Adult size is represented by test length (mm). mostly due to the increase of the maximum interface area of the spher- solely as an intriguing working hypothesis. According to the fossil re- ical test and sea water. cord, Robustoschwagerina/Sphaeroschwagerina did not spread as exten- With the method developed by Hohenegger and Briguglio (2012),it sively as fusiform, cosmopolitan contemporaneous schwagerinine is possible to estimate the hydrodynamic condition of fusulinid tests species, but did achieve impressive diversity in most areas of the Early even within the protoplasm and gas or lipids filled, if the three dimen- Permian Tethys, Uralian-Franklinian, Midcontinent realms (Leven, sional measurements of the tests could be acquired. This is for the 2004; Ross, 1995). time being out of our capability. There is another complication arising The spherical pseudoschwagerines, including Robustoschwagerina, from the pseudopods and adhesive matter while considering the living Sphaeroschwagerina and Zellia, is an easily recognized group in the fusulinids. They probably help the fusulinids to fix to the substrate and field indicating the start of the Early Permian, which is another reason it's much more difficult to evaluate this fixation power. However, why these species are important indices for stratigraphy. However, it's there are still modern analogs for this process that we could consider. difficult to pinpoint their distinctions among each other, and disagree- Several benthic foraminiferal genera exhibit a meroplanktonic life ments exist regarding their taxonomy and phylogeny (Davydov, 1984; stage (e.g., Neoconorbina, Rosalina, Cymbaloporeta, Tretomphalus, Miklukho-Maklay, 1959; Sheng et al., 1988; Zhou et al., 1997). It has Tretomphaloides). One adaptation to this life style is the late-stage onto- been suggested that Robustoschwagerina and Sphaeroschwagerina be- genetic development of a special float chamber (Banner et al., 1985; long to different evolutionary lineages because of the distinct character Murray, 2006; Myers, 1943; Todd, 1971). These chambers exhibit either of their inner whorls. One of the distinctions is the outer wall spherical or hemispherical shapes and are normally filled with gas (spirotheca). Spirotheca of the inner whorls in Sphaeroschwagerina (Banner et al., 1985). By successfully diminishing the body density were thought to lack the honeycomb-like keriotheca structure whereas with this large, spherical, gas-filled chamber, the benthic individuals keriothecal structure was thought to be present in Robustoschwagerina adopt a planktonic life state during their last developmental stage. (Davydov, 1984; Zhou et al., 1997). However, since both of these genera No one can entertain seriously the idea that the large, massive exhibit noticeable keriotheca in the outer whorls it unlikely that pseudoschwagerine could float in the water column even though the Sphaeroschwagerina developed different spirotheca during its lifespan, tests of pseudoschwagerine species could have been much lighter especially when the type species — Schwagerina princeps Schellwien, than their non-pseudoschwagerine counterparts. However, so long as 1898 (pl. XXI, Fig. 9)andSchwagerina moelleri Rauser-Chernousova, the fusulinid test remains fusiform it would lie close to the sediment- 1936 (pl. 9, Fig. 1a, b) — exhibit detectable keriotheca in their inner water interface and be acted on by the predominantly laminar flow of whorls. Another distinction noted by Sheng et al. (1988) is the presence currents in that region. If a species was able to build its test as large as of faint description of Triticites-like inner whorls in Robustoschwagerina, to cause the turbulent flow by exceeding the certain Reynolds number characterized by a larger proloculus and robust chomata. These brought (Yordanova and Hohenegger, 2007), it might be differentially plucked about the question that to what extent these two modifiers “larger” and away from the sediment-water interface by turbulent water flow and “robust” indicate, since the size of the proloculus and robustness of set down again at a new location, perhaps far away from its original chomata in species of both genera vary greatly. resting place. Ontogenetic shape changes illustrate a variety of developmental The evolutionary advantages of wide dispersal are well known, differences between these two genera. As shown in Fig. 14,ontogenytra- based on well-documented examples from across all organismal king- jectory variation models for Robustoschwagerina and Sphaeroschwagerina doms. What we might be seeing in the case of late-stage spherical fusu- can be constructed based on the geometric character of the discrim- linid species is an evolutionary convergence on a developmental inant functions specified through CVA. Distinctions between strategy that enhances opportunities for transportation and dispersion, Robustoschwagerina and Sphaeroschwagerina are mainly organized though we hasten to add at this point that we offer this suggestion along CV-2. Along axis models of this aspect of test-shape variation 96 Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 reveals a pattern of marked shape change from Robustoschwagerina to and more whorls than Robustoschwagerina. These distinctions reveal Sphaeroschwagerina. Compared with the early occurrence of the maxi- the different ontogeny processes between the two genera, and propose mumfusiformshapeofRobustoschwagerina, e.g., near whorl stage 4 in more details on their taxonomy and palaeoecology. models 1 to 2, this shift occurs later in Sphaeroschwagerina, e.g. around The morphometric approach to morphological analysis can success- whorl stage 8 in models 4 to 5. This is consistent with the previous trend fully characterize, summarize and facilitate comparisons between spec- identified by the ontogenetic trajectory plots of our samples. Moreover, imens, species, genera, etc., living or fossilized, and is especially well- the later onset of fusiform stage in Sphaeroschwagerina may imply its suited to developmental/ontogenetic investigations. Analyses based on longer residency in the water column during its early development combined outline and landmark datasets explicate both the size and stages. This presents an interesting potential explanation for the wider shape information of the specimens, and pinpoint both similarities geographic distribution of this genus that can be explored further. and differences quantitatively far beyond the normal visual inspection. Shape variation modelling further illustrates and clarifies distinctions 5. Conclusions between groups. Since a large amount of information concerning ontogeny, taxonomy, phylogeny, ecology and biogeography are recorded Different development stages are recognized in the ontogeny of within the forms of specimens — especially fossils — morphometric pro- spherical pseudoschwagerine. Regarding size, juvenile and adult stages cedures should form fundamental tools that can be accessed and used are indicated by the distinct shift in test size change. The shape shift of by all morphologists. spherical-fusiform-spherical modes suggests the presence of different and diagnosable growth stages, and possibly implies the presence of Acknowledgements functional adaptations responsible for the wide distribution. Robustoschwagerina and Sphaeroschwagerina exhibit different form- This work was done during the stay of the senior author at The Nat- change trends: the former exhibits a developmentally quicker shift from ural History Museum, London as an academic visitor, supported by the proloculus spherical to fusiform, and a shorter period of fusiform State Program of Oversea Study for Outstanding Young Scholars of retainment; the latter is relatively early in its acquisition of adult stage Jiangsu. This research is also funded by the National Natural Science morphology. Moreover, Sphaeroschwagerina develops larger adult size Foundation of China (No. 41732008), State Key Laboratory of

Fig. 14. Pattern of ontogeny trajectory variation between Robustoschwagerina and Sphaeroschwagerina. Models 1 to 5 (lower) are calculated with evenly spaced coordinates, denoted by corresponding numbers in the CVA space, along CV2 axis, and principally describe the ontogeny trajectory variation from Robustoschwagerina to Sphaeroschwagerina. Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98 97

Palaeobiology and Stratigraphy (Nanjing Institute of Geology and MacLeod, N., 2008. Understanding morphology in systematic contexts: 3D specimen ordination and 3D specimen recognition. In: Wheeler, Q. (Ed.), The New Taxonomy. CRC Palaeontology, CAS) (No. 133110), and the Fundamental Research Press, Taylor & Francis Group, London, pp. 143–210. Funds for the Central Universities (No. 20620140390). The authors are MacLeod, N., 2009. Form & shape models. Palaeontological Association Newsletter 72, grateful to Dr. John Groves and Prof. Johann Hohenegger for their valu- pp. 14–27. MacLeod, N., 2010. Principal warps, relative warps and procrustes PCA. Palaeontological able suggestions and comments, Prof. Pamela Hallock Muller for the in- Association Newsletter 75, pp. 22–33. formative discussions, and they also thank the editor, Dr. Richard Jordan MacLeod, N., 2012a. Going round the bend: eigenshape analysis I. Palaeontological Asso- for improving the manuscript. ciation Newsletter 80, pp. 32–48. MacLeod, N., 2012b. Going round the bend: eigenshape analysis II. Palaeontological Asso- ciation Newsletter 81, pp. 23–39. Appendix A. Supplementary data MacLeod, N., 2015. The direct analysis of digital images (eigenimage) with a comment on the use of discriminant analysis in morphometricis. In: Lestrel, P.E. (Ed.), Proceedings of the Third International Symposium on Biological Shape Analysis. World Scientific, Supplementary data to this article can be found online at http://dx. Singapore, pp. 156–182. doi.org/10.1016/j.marmicro.2015.12.002. MacLeod, N., Steart, D., 2015. Automated leaf physiognomic character identification from digital images. Paleobiology 41 (4), 528–553. MacLeod, N., M., O.N., Walsh, S., 2007a. Automated tools for the identification of taxa from References morphological data: face recognition in wasps. In: Macleod, N. (Ed.), Automated Taxon Recognition in Systematics: Theory, Approaches and Applications. CRC Press, Alve, E., Goldstein, S.T., 2003. Propagule transport as a key method of dispersal in benthic Taylor & Francis Group, Boca Raton, Florida, pp. 153–188. foraminifera (Protista). Limnol. Oceanogr. 48 (6), 2163–2170. MacLeod, N., M., O.N., Walsh, S., 2007b. A comparison between morphometric and artifi- Alve, E., Goldstein, S.T., 2010. Dispersal, survival and delayed growth of benthic foraminif- cial neural network approaches to the automated species recognition problem in sys- eral propagules. J. Sea Res. 63, 36–51. tematics. In: Curry, G., Humphries, C. (Eds.), Biodiversity Database: From Cottage Banner, F.T., Pereira, C.P.G., Desai, D., 1985. Tretomphaloid float chambers in the Industry to Industrial Network. Taylor & Francis, London, pp. 37–62. Discorbidae and Cymbaloporidae. J. Foraminifer. Res. 15 (3), 159–174. Malmgren, B., Kennett, J.P., 1972. Biometric analysis of phenotypic variation: Globigerina Bellman, R., 1961. Adaptive Control Processes: A Guided Tour. Princeton University Press, pachyderma (Ehrenberg) in the South Pacific Ocean. Micropalaeontology 18 (2), Princeton. 241–248. Bookstein, F.L., 1986. Size and shape spaces for landmark data in two dimensions. Stat. Sci. Manly, B.F.J., 2007. Randomization, Bootstrap and Monte Carlo Methods in Biology. third 1(2),181–222. ed. Chapman &Hall/CRC Press, Taylor & Francis Group, Boca Raton, Florida. Briguglio, A., Hohenegger, J., 2014. Growth oscillation in larger foraminifera. Paleobiology Masuda, R., 2009. Behavioral ontogeny of marine pelagic fishes with the implications for 40 (3), 494–509. the sustainable management of fisheries resources. Aqua-BioSci. Monogr. 2 (2), 1–56. Brummer, G.-J.A., Hemleben, C., Spindler, M., 1986. Planktonic foraminiferal ontogeny and Miklukho-Maklay, 1959. System and phylogeny of Fusulinids (genus Triticites and its al- new perspectives for micropalaeontology. Nature 319, 50–52. lied genera). Vestnik LGY 6 (1), pp. 6–21 (in Russian). Brummer, G.-J.A., Hemleben, C., Spindler, M., 1987. Ontogeny of extant spinose planktonic Mitteroecker, P., Bookstein, F., 2011. Linear discrimination, ordination, and the visualiza- foraminifera (Globigerinidae): a concept exemplified by Globigerinoides sacculifer tion of selection gradients in modern morphometrics. Evol. Biol. 38, 100–114. (Brady) and G. ruber (D'Orbigny). Mar. Micropaleontol. 12, 357–381. Murray, J.W., 1973. Distribution and Ecology of Living Benthic Foraminiferids. Heinemann Campbell, N., Atchley, W.R., 1981. The geometry of canonical variate analysis. Syst. Biol. Educational Books, London. 30, 268–280. Murray, J.W., 2006. Ecology and Applications of Benthic Foraminifera. Cambridge Univer- Ciry, R., 1943. Les fusulinidés de Turquie. Ann. Paleontol. 29, 53–78. sity Press, Cambridge. Colpaert, C., Monnet, C., Vachard, D., 2014. Eopolydiexodina (Middle Permian giant fusuli- Myers, E.H., 1943. Biology, Ecology and Morphogenesis of a Pelagic Foraminifer. Stanford nids) from Afghanistan: biometry, morphometry, paleobiogeography, and end- University Press, California. Guadalupian events. J. Asian Earth Sci. 102, 127–145. Pharr, R.B., Williams, D.F., 1987. Shape changes in Globorotalia truncatulinoides as a Davydov, V., 1984. Zonational subdivision of the Upper Carboniferous in Darvaz. Byl. function of ontogeny and palaeobiogeography in the Southern Ocean. Mar. MOIP Otd. Geol. 59, pp. 41–57 (in Russian) Micropaleontol. 12, 343–355. Dunbar, C.O., 1963. Trends of evolution in American fusulines. In: Von Koenigswald, Rauser-Chernousova, D., 1936. On the renaming of the genus Schwagerina and G.H.R., Emeis, J.D., Buning, W.L., Wagner, C.W. (Eds.), Evolutionary trends in Forami- Pseudofusulina proposed by Dunbar and Skinner. Academy of Sciences of the URSS, nifera. Elsevier, Amsterdam, pp. 25–44. B., Ser Geol 4, pp. 573–584. Erickson, G.M., Rogers, K.C., Yerby, S.A., 2001. Dinosaurian growth patterns and rapid Rohlf, F.J., Slice, D., 1990. Extensions of the Procrustes method for the optimal superimpo- avian growth rates. Nature 412, 429–433. sition of landmarks. Syst. Biol. 39 (1), 40–59. Galeotti, S., Coccioni, R., 2001. Morphometric analysis of selected benthic foraminifera Ross, C.A., 1972. Paleobiological analysis of Fusulinacean (Foraminiferida) shell morphol- from Cape Roberts Project (CRP) and CIROS-1 Cores, Victoria Land Basin, Antarctica. ogy. J. Paleontol. 46, 719–728. Terra Antarct. 8, 359–368. Ross, C., 1982. Paleobiology of fusulinaceans. Third North American Paleontological Con- Groves, J.R., Pike, M., Westley, K., 2012. A test for the possibility of photosymbiosis in ex- vention, Proceedings 2, pp. 441–445. tinct fusuline Foraminifera: size and shape related to depth of habitat. Palaios 27, Ross, C.A., 1995. Permian fusulinaceans. In: Scholle, P.A., Peryt, T.M., Ulmer-Scholle, D.S. 738–751. (Eds.), The Permian of Northern Pangea. Springer-Verlag, Berlin, pp. 167–185. Hemleben, C., Spindler, M., Breitinger, I., Deuser, W.G., 1985. Field and laboratory studies Schellwien, E., 1898. Die Fauna des Karnischen Fusulinenkalks. Palaeontographica 24, on the ontogeny and ecology of some globorotaliid species from the Sargasso Sea off 237 –282. Bermuda. J. Foramineferal Res. 15 (4), 254–272. Scott, G., 1973. Ontogeny and shape in Globorotalia menardii. J. Foraminifer. Res. 3 (2), Hemleben, C., Spindler, M., Erson, O., 1989. Modern Planktonic Foraminifera. Springer, 142–146. Berlin. Shamov, D.F., Scherbovich, S.F., 1949. Some pseudofusulines from the Schwagerina hori- Hofmann, H., 1994. Grain-shape indices and isometric graphs. J. Sediment. Res. 64, zon of Bashkortostan. Tr. GIN AN SSSR, Series of Geology 105, pp. 163–170 916–920. (in Russian). Hohenegger, J., 2009. Functional shell geometry of symbiont-bearing benthic Foraminif- Shan, H.-Y., Wei, K.-Y., Huang, C.-Y., 2006. Morphometric study of the benthic foraminifera, Galaxea. J. Coral Reef Stud. 11, 81–89. eral clade Gaudryina in an Oligocene–Miocene series of the Kouhsing Area in Central Hohenegger, J., 2011. Growth-invariant meristic characters tools to reveal phylogenetic Taiwan. Collect. Res. 19, 35–47. relationships in Nummulitidae (Foraminifera). Turk. J. Earth Sci. 20, 655–681. Sheng, J.Z., Wang, Y.J., Zhong, B.Z., 1984. Some species of the genus Robustoschwagerina Hohenegger, J., Briguglio, A., 2012. Axially oriented sections of nummulitids: a tool to in- from Eastern Yunnan. Acta Palaeontol. Sin. 23 (5), 523–530 (in Chinese with English terpret larger benthic foraminiferal deposits. J. Foramineferal Res. 42, 134–142. abstract). Huang, H., 2011. Fusulinid species determination based on population variation: an ex- Sheng, J.Z., Zhang, L.X., Wang, J.H., 1988. Fusulinids. Science Press, Beijing (in Chinese). ample from Eopolydiexodina. Sci. China Earth Sci. 54 (3), 380–387. Shi, Y.K., 2008. Verification on the Evolutionary Pattern of Late Carboniferous–Early Perm- Hughes, N.C., Minelli, A., Fusco, G., 2006. The ontogeny of trilobite segmentation: a com- ian Fusulinid Fauna in South China — An Example from the Zongdi Section, Ziyun parative approach. Paleobiology 32 (4), 602–627. County, Guizhou Province Ph.D. Thesis Nanjing University, Nanjing, China (in Chinese Huxley, J., 1932. Problems of Relative Growth. Methuen & Co. Ltd., London. with English abstract). Jarman, P., 1983. Mating system and sexual dimorphism in large terrestrial, mammalian Shi, Y.K., Yang, X.N., Huang, H., 2009. Differences between subfamilies in diversification herbivores. Biol. Rev. 58, 485–520. process of the Early and Middle Permian fusulinid fauna in South China. Palaeoworld Jorry, S.J., Hasler, C.-A., Davaud, E., 2006. Hydrodynamic behaviour of Nummulites: 18, 34–40. implications for depositional models. Facies 52, 221–235. Shi, Y.K., Yang, X.N., Liu, J.R., 2012. Early Carboniferous to Early Permian Fusulinids from Leven, E.J., 2004. Fusulinids and Permian scale of the Tethys. Stratigr. Geol. Correl. 12, Zongdi Section in Southern Guizhou. Science Press, Beijing (in Chinese with English 33–47. summary and systematic description). Likharev, B., 1939. An Atlas of the Leading Forms of the Fossil Fauna of the U.S.S.R. 6: Showers, W.J., 1980. Biometry of the Foraminifera Rosalina globularis (d'Orbigny) in Ant- Permian. Central Geological and Prospecting Institute, Leningrad (in Russian). arctic Environments. J. Foraminifer. Res. 10 (1), 61–74. Liu, C., Xiao, X., Dong, W., 1978. Fusulinida, Paleontological Atlas of Southwest China. Speijer, R.P., Van Loo, D., Masschaele, B., Vlassenbroeck, J., Cnudde, V., Jacobs, P., 2008. Geological Publishing House, Beijing (in Chinese). Quantifying foraminiferal growth with high-resolution X-ray computed tomography: MacLeod, N., 1999. Generalizing and extending the eigenshape method of shape space vi- new opportunities in foraminiferal ontogeny, phylogeny, and paleoceanographic ap- sualization and analysis. Paleobiology 25, 107–138. plications. Geosphere 4 (4), 760–763. 98 Y. Shi, N. MacLeod / Marine Micropaleontology 122 (2016) 87–98

Starck, J.M., Ricklefs, R.E., 1998. Avian Growth and Development: evolution Within the Al- Yang, X.N., Liu, J.R., Zhu, L.M., Shi, G.J., 2005. Early Permian bioevent in the fusulinacean tricial–Precocial Spectrum. Oxford UNIVERSITY Press, New York. fauna of South China. Lethaia 38 (1), 1–16. Thompson, M.L., 1946. Permian fusulinids from Afghanistan. J. Paleontol. 20, 140–157. Yordanova, E.K., Hohenegger, J., 2007. Studies on settling, traction and entrainment of Thompson, M.L., 1954. American Wolfcampian Fusulinids. University of Kansas Paleonto- larger benthic foraminiferal tests: implications for accumulation in shallow marine logical Contributions, pp. 1–226, 5. sediments. Sedimentology 54, 1273–1306. Thompson, M.L., Wheeler, H.E., Hazzard, J.C., 1946. Permian fusulinids of California. Zhang, Y., Payne, J.L., 2012. Size-frequency distributions along a latitudinal gradient in Geological Society of America Memoirs 17, pp. 1–72. Middle Permian fusulinoideans. PLoS One 7 (6), e38603. Todd, R., 1971. Tretomphalus (Foraminifera) from Midway. J. Foraminifer. Res. 1 (4), Zhou, Z.R., Yang, Q., Sheng, J.Z., 1997. Fusulinid genera Pseudoschwagerina and 162–169. Sphaeroschwagerina. Acta Micropalaeontol. Sin. 14 (2), 111–128 (in Chinese with Wang, L.T., Dong, W.L., Ye, N.Z., Chen, W.Y., 1982. A discussion of the relationship between English abstract). fusulinid ecology and hydrodynamic environment. Oil Gas J. 3, 231–239 (in Chinese with English abstract). Yang, X.N., Hao, Y.C., 1991. A study on ontogeny and evolution of Robustoschwagerinids (Permian fusulinids). Acta Palaeontol. Sin. 30 (3), 277–306 (in Chinese with English abstract).