COMPARISON OF THE ONTOGENY OF CHESTERIAN PENTREMITID BLASTOIDS


James W. Atwood, Troy Fadiga, and Colin D. Sumrall
Department of Earth and Planetary Sciences, University of Tennessee, Knoxville 37996, jatwood2@utk.edu

ABSTRACT— This study utilizes geometric morphometrics to examine the ontogeny of the blastoid genus Pentremites. Specimens of Pentremites fredericki, P. godoni, P. pyriformis, P. tulipaformis, P. symmetricus, and P. spicatus were collected from four different Upper Mississippian localities in Kentucky, Indiana, Tennessee, and Illinois, USA. 3D images were acquired for randomly selected specimens across the ontogeny of each species and coordinates were collected for a series of homologous landmarks that fully describe blastoid morphology through ontogeny. Data were analyzed using the R (language and environment for statistical computing and graphics) package GEOMORPH. We were able to show that all species in this study share a common ontogenetic pathway and shape varies significantly with size and by species.

INTRODUCTION
Geometric morphometric techniques have previously shown that our knowledge of blastoids paleobiology can be greatly expanded through the use of three-dimensional (3D) modeling and mathematical methodology (Foote 1991; Atwood and Sumrall 2012). These studies showed the utility of 3D geometric morphometric methods for understanding shape disparity and species discrimination respectively.  This paper investigates blastoids ontogeny using 3D geometric morphometrics to look for commonality and differences in the growth trajectories for closely related blastoids species. 
	Pentremites is the most abundant blastoid in North American Mississippian faunas, at times being the dominant echinoderm. In many localities, specimens of Pentremites can be collected by the hundreds providing large collections from which study specimens can be drawn (Waters et al. 1985; Dexter et al. 2009; Atwood and Sumrall 2012). Blastoids collected for this study range from 100μm to 4cm in thecal height and localities were selected that provided specimens with exquisite preservation of detail (Fig. 1).  While many blastoids collected from these localities are taphonomically damaged by crushing, breakage or compaction, a substantial number of fully three dimensional and undistorted specimens were recovered for each species.
Most organisms exhibit large morphologic change during growth. Whether the change is metamorphosis, allometric scaling, or any other growth mechanism, the morphological changes during ontogeny can be remarkable. Furthermore, environmental pressures can change the timing of ontogenetic events and how an organism grows drastically changing morphology between closely related species or within a single species (Zelditch, 1989).  Since determining ages of individuals is often problematic, care has to be taken to not confuse genera/species with ontogenetic variation. 
Allometric change can evolve through heterochrony (McKinney and MacNamara 1991) or heterotopy (Zelditch and Fink 1996). McKinney and MacNamara define heterochrony as developmental change in the timing of events while Zelditch and Fink define heterotopy as evolutionary changes in spatial patterning of development. It is reasonable, given the data, to argue that we are seeing the affects of heterochrony since we do see the same changes across species but to differing amounts. To properly assess this hypothesis, a phylogenetic analysis would be necessary. 
This study utilizes 3D shape analysis on blastoid thecae to comparatively assess the ontogenies of closely related species of Pentremites, using laser imaging to collect landmark data and geometric morphometrics to analyze data. This approach provides a better understanding of blastoid ontogeny than classic linear measurements with calipers (Macurda 1975; Waters 1977; Waters et al. 1985). These methods will utilize the preserved three-dimensional thecae which is frequently lost in invertebrate paleontology. These data will be used to better explore the ontogeny of Pentremites.

PREVIOUS INVESTIGATIONS AND STRATIGRAPHY
	The most significant work was conducted on Pentremitid blastoid ontogeny from 1940 to 2005 (Moore 1940; Galloway and Kaska 1957; Waters 1977; Waters et al. 1985; Sevastopulo 2005; Dexter 2009).  Recent work has preliminarily quantified blastoid ontogeny though traditional morphologic measurements based on allometry, and examined immature blastoid forms (Passalocrinus) with speculations about the developmental sequence including the possible heterochrony in blastoids evolution (Moore 1940; Galloway & Kaska 1957; Waters 1977; Waters et al. 1985; Sevastopulo 2005). 
	Beaver et al. (1967) identified four stages in blastoid ontogeny: larval stage (Passalocrinus), juvenile stage, adult stage, and gerontic stage (most mature form).  The stages were poorly defined so there was much ambiguity in defining life stages and forms.  In 2005, Sevastopulo argues that larval stage blastoids were free swimming (none have been recognized) and that Passalocrinus is the earliest juvenile stage. Sevastopulo also argues that the biggest difference delineating juvenile from adult are the plating on the summit of Passalocrinus which are not possessed by the adult forms. That leaves only size to delineate adult and gerontic blastoids as defined by Beaver et al. (1967).
	By utilizing newer techniques and geometric morphometrics, individual plate growth patterns can be reassessed and growth patterns can be differentiated among species and by thecal region. 

SPECIES SELECTION AND STRATIGRAPHY
	The Upper Mississippian (Chesterian) Formations of the North American mid-continent yield large numbers of extraordinarily well-preserved marine fossils (Butts, 1917; Nelson et al., 2002). This study focuses on blastoid collections derived from four different sections of Upper Mississippian (Chesterian) localities yielding six separate species of the blastoid genus Pentremites. Although there are nine recognized species of Pentremites only six were used in this study (Figure 1). Species included were selected based on the availability of fossil localities that could produce large numbers of well preserved, undeformed specimens. Specimens were collected from four different localities Pennyrile Parkway, KY, USA; Floraville, IL, USA; Sparta TN, USA; and Sulphur, IN, USA.  All of these localities contained numerous multiple species of Pentremites except for the Sparta TN locality.
	Pennyrile Parkway, Christian Co. Kentucky, USA—This locality exposes strata of the Lower Chesterian Glen Dean Formation in a road cut along the Pennyrile Parkway in Christian County, southwestern Kentucky. The lowermost 1-2m of the exposed section are highly weathered, fissile, fossiliferous, shale with minor limestone interbeds. The limestone is grayish brown on fresh faces and weathers light gray. The shale is light olive gray on fresh faces and weathers dark gray. The blastoids fauna includes four species of Pentremites including: P. meganae, P. fredericki (Fig. 1-16-1.20), P. pyriformis (Fig 1.11-1.15), P. tulipiformis (Fig. 1.6-1.10). Specimens of P. meganae were rare and did not preserve enough of an ontogeny to be used in the study. The other three Pentremites species were present in sufficient numbers of well preserved specimens to be used in the study and easily separated along morphological (Atwood and Sumrall, 2012). Few other echinoderms were collected during this study but include cladid crinoids, discocystinid edrioasteroids, and single specimen of the blastoids Diploblastus fadigai.
	Floraville, St. Clair Co. Illinois, USA —outcrops of the Lower Chesterian, Ridenhower Formation of the Paint Creek Group along the banks of a streambed located at either side of the Prairie Du Long Creek about 1.6 km notrth of Floraville, IL.  The outcrops were composed of highly weathered, fissile, fossiliferous, light green shale with minor limestone interbeds. Blastoids found at this locality were primarily P. godoni (Fig 1.1-1.5) with fewer P. pyriformis. For this study, only specimens of P. godoni were includinde from this locality because not enough well preserved specimens of P. pyriformis were collected to be utilized.  The two species are easily distinguished by simple vault pelvis ratio (Waters et al., 1985). The Floraville, IL locality has a rich and diverse echinoderm fauna including Crinoids, Blastoids, asteroids, and edrioasteroids (Sumrall 1996).  
Sulphur, Crawford Co. IN, USA—The Sulphur Indiana locality of the Indian Springs Shale Member of the Big Clifty Formation in the Stephensport Group at Sulpur, Indiana is exposed along road cuts around the intersection of Interstate 64 and Indiana Highway 37. The outcrops are composed of fossiliferous grey shale interbedded with limestone. Samples of P. symmetricus (Fig, 1.21-1.25) and P. godoni were present at the locality, but only P. symmetricus were present in sufficient numbers to be used in the study.  The two species are easily separated using vault pelvis ratios.  The locality has a diverse crinoid and asteroid fauna (Horowitz 1965; Blake and Elliott 2003)
Sparta, White Co. TN, USA—The Sparta Tennessee locality exposes the Upper Chesterian Pennington Formation. Locality was approximately 1.6 km off the main road on a little used hiking trail. The locality is highly weathered and no fresh strata is visible. Collections were made primarily from float material that from fossiliferous brown fissile shale.  This locality only had one blastoid species P. spicatus (Fig. 1.26-1.30).  Few other echinoderms were noted except for tegmal wing plates of Pterotocrinus.  

METHODOLOGY
Collection.— Museum collections were not used in this study because they are often biased samples of well-preserved and large specimens with few or no small sized blastoids (juvenile). Often they are culled of unusual specimens.  There is also no guarantee that all specimens in a sample lot are from the same bed at the same locality.  Consequently, new collections of all species were assembled to assure that temporal and environmental mixing, paleoenvironment, and shape could be controlled and sampled randomly.
	The focus during collection of material was to assure that we did not impose bias on the distribution of shape within the species populations. At all localities, all blastoid specimens (typically larger specimens) were surface collected and additional (typically smaller specimens) were collected in bulk samples of weathered shale. This bulk material was dry sieved into five 20 L buckets to remove any larger rock fragments and larger organic material from the samples. 
Typically five buckets of bulk material were collected at each locality with the exception of the Pennyrile Parkway locality, where approximately 30 - 40 buckets of bulk material were collected for use in a previous study (Atwood and Sumrall 2012).
Bulk samples were washed in warm water followed by hydrogen peroxide to break down the shale matrix and organic material releasing fossil residue from the shale. Blastoids were then retrieved from these residues via handpicking. This method generates large sample population collections of blastoids that included individuals from all sizes present at each locality. All localities except for the Sparta, TN possessed at least one blastoid species that numbered too few to use in this study. With heavier collecting the underrepresented species could be utilized. By using this method, our samples were biased according to size frequency, but not biased according to shape. Similar methods were used to collect blastoids ontogenies for other studies (Dexter et al., 2009; Atwood and Sumrall, 2012).
	For each sample locality, individual blastoids were sorted by species and screened for taphonomic deformation. We visually inspected specimens by looking for deviation from the regular pentameral cross section and other obvious signs of deformation. If a blastoid showed nearly perfect radial symmetry, it was considered to be usable. Teratological specimens were removed, including specimens with four ambulacra, or ambulacra that were truncated with respect to other ambulacra on the same specimen. Rejecting specimens in this method left only phenotypic variation, sexual dimorphism and ontogenetic variation as the primary sources of variance in the six populations. Although external sexual dimorphism is not known in Pentremites, it has been documented internally in the CD side hydrospires of Pentremites rusticus (Katz and Sprinkle, 1976).
From all usable specimens, a study sample was selected for measurement. 
Specimens of each species were divided into twenty groups based on size. Perhaps better to say - From the set of usable specimens, a study sample was selected.  Specimens of each species were sorted by height into several size categories.  From each size category a specimens were chosen using a random number generator that was set to the population size of the group. Selected specimens were inspected under a binocular microscope to insure that all of the landmarks could be identified. If a specimen was damaged so that a landmark was unusable, it was replaced by another specimen using the random number generator.  This method assured a full range of sizes were used in the study. Because of the limits of the laser scanner, specimens 4mm and smaller in height were not included in the sample populations.
A total of 120 specimens were analyzed in this study, but because of scanning errors, some of the specimens were not used.  The anaylsis which follows, utilized 112 of the 120 scans (20 of P. symmetricus and P. tulipaformis, 19 P. godoni, 18 P. spicatus and P. pyriformis, and 17 P. fredericki) (Table ___). 

Analysis.— To aid in the recording of data, landmarks were marked with a micro pen on specimens under high magnification prior to scanning to allow landmark positions to be recorded on the 3D specimen images. A 3D image of each specimen was attained using the Nextengine 3D Laser Scanner at the University of Tennessee. Specimens were scanned at the highest possible resolution, of 0.127mm in an A-ray orientation. Each image was composed of three scans that were 60 degrees apart rotated about an axis that was perpendicular to both the polar and A/CD axes of the specimens to insure total coverage of the A-ray. This provided a 3D image that included landmarks located along the peristomial border and stem facet. From these 3D images X, Y, Z coordinates of landmarks (modified from Foote, 1991) were recorded. These landmarks are morphologically common to all known species of blastoids, and fully describe the morphology of the specimens in 3D (Fig. 2). All landmarks were type I (sensu stricto Bookstein, 1991), being positioned at three plate junctions of thecal plates except landmark 6 which was type 2 landmark.
Landmark data were collected from the models using the imaging program MESHLAB (Cignoni et al., 2008) and analyzed using R (language and environment for statistical computing and graphics). Landmarks were aligned and analyzed using the R package GEOMORPH (Adams and Otarola-Castillo, 2013; R Core Team, 2013). Landmarks were aligned using a generalized Procrustes superimposition (Bookstein, 1991; Zelditch et al., 2004; Dryden, 2007) that removed differences in location, scale, and orientation. The residuals were then projected into a tangent Euclidean space, and these projected coordinates were used as shape variables. Ontogenetic allometry and group differences were assessed with a permutational Procrustes analysis of variance (ANOVA) (Goodall, 1991) using 9,999 replicates. This analysis differs from traditional ANOVA by using Procrustes distances to assess goodness-of-fit instead of the sum of squares and by using permutations to approximate a null distribution instead of using a parametric F-distribution. 
	Multivariate allometry was visually inspected by plotting regression scores against the log of the centroid sizes (Drake and Klingenberg, 2008).  The regression scores indicate the amount of variation predicted by a regression plus the residual variation along the same direction as the regression.  The regression coefficients used were taken from a regression model that included centroid size as a covariate and species identifier as a factor.




RESULTS AND DISCUSSION
The ANOVA results (Table 1) indicate that Pentremites shape varies significantly with size and by species, but there is not a significant interaction term between these two variables.  It appears that all six of the Pentremites species share an indistinguishable ontogeny over the size span sampled in this analysis. Figure 3 shows the regression scores plotted against the log centroid size, visually demonstrating that these Pentremites species have a shared, significant relationship to size. Even though they cannot be significantly separated as multiple ontogenies it appears visually that the pryiforms (P. pyriformis and P. symmetricus) and godoniforms (all other species) have different y-intercepts. Meaning that pyriforms (long pelvis) and godoniforms (wide with short pelvis) most likely have different starting shapes. In the future it may be possible to separate the ontogeny of the two forms but many more blastoids would be needed To further add to this argument, we have found two basic “Passalocrinus” forms at the Glen Dean Formation where both a pyriform (P. pyriformis) and a godoniform (P. tulipaformis) are present. The first form as seen in figure 4.1, resembles a pyriform while figure 4.2 resembles a godoniform. 
	Predicted ontogenetic series.— Figure 5 shows three different perspectives of the predicted ontogenetic series landmark mean: slightly dipping summit view (Fig. 5.1), A-ray (Fig 5.2), and orthogonal to A-ray (Fig. 5.3). The small configuration (juvenile) plotted as black, while large (adult) is plotted as red.
	Interpretation.— looking at landmarks five, six, and seven  we can see that the ambulacra get narrower with the respect to the total width of the blastoid while getting longer. We can also see that the pelvis gets shorter with respect to larger specimen sizes as the stem facet (Fig. 5.13) appears to get shorter. Also, the mouth appears to get smaller with respect to the thecal size. Lastly, the radial plates appear to grow much more proximally than distally. This can be seen by comparing proximal radial landmarks (Figs. 5.4-5.8) to distal radial landmarks (Figs. 5.10-5.12). 

CONCLUSIONS AND FUTURE STEPS
Even though this study was ultimately limited by the resolution of the laser scanner, we have demonstrated that six different species of Pentremites from four different localities have a common growth pattern that may be able to divided into the two morphotypes. It is our contention that these preliminary findings are accurate but to test this hypothesis, we will need to add more species, collect higher numbers, and have ability to image the smallest of specimens. 
Future studies could take a variety of paths. One step would be to build a more extensive collection (number of specimens and higher species diversity).  Collecting will be will be very time consuming but could potentially be very informative. Another direction that we could continue would to be a phylogenetic analysis on the species of Pentremites with the considerations paleobiogeography and well as heterochrony. Future ontogenies could be used to test the current hypothesis as well as compare growth patterns between families. 


ACKNOWLEDGEMENTS
We thank Daniel Frederick and Larry Knox for introducing us to two of the localities and aiding in sample collection. We also thank Tim Paton for spending many hours scanning blastoids. Assistance in the field was provided by B. Ford, R. Shroat-Lewis, G. Gilleaudeau, M. Schnuck, and E. Hogan. Assistance with museum collections was provided by Brenda Hunda, Cincinnati Museum Center. Thank you to ____ and ____ for reviewing this manuscript. The laser scanner was acquired through NSF grant EAR-0745918. Financial assistance for this study was provided by the NSF - Assembling the Echinoderm Tree of Life Grant, DEB-1036260.














REFERENCES
ADAMS, D., AND A. NISTRI. 2010. Ontogenetic convergence and evolution of foot morphology in European cave salamanders (Family: Plethodontidae). Bmc Evolutionary Biology, 10(1):1-10.
ADAMS, D. C., AND E. OTÁROLA-CASTILLO. 2013. geomorph: an r package for the collection and analysis of geometric morphometric shape data. Methods in Ecology and Evolution, 4(4):393-399.
ATWOOD, J. W., AND C. D. SUMRALL. 2012. Morphometric Investigation of the Pentremites Fauna from the Glen Dean Formation, Kentucky. Journal of Paleontology, 86(5):813-828.
BEAVER, H. H., K. E. CASTER, D. J. W., R. O. FAY, H. B. FELL, R. V. KESLING, D. B. MACURDA, R. C. MOORE, G. UBAGHS, AND J. WANNER. 1967. Treatise on Invertebrate Paleontology (S)Echinodermata(1). The University of Kansas and The Geological Society of America, Boulder, 2.
BEAVER, H. H., AND A. J. FABIAN. 1998. Color Patterns in Mississippian (Chesterian) Blastoids. Journal of Paleontology, 72(2):332-338.
BEAVER, H. H., A. J. FABIAN, AND M. PALATAS. 2000. Summit structures in Mississippian blastoids. Journal of Paleontology, 74(2):247-253.
BLAKE, D. B., AND D. R. ELLIOTT. 2003. OSSICULAR HOMOLOGIES, SYSTEMATICS, AND PHYLOGENETIC IMPLICATIONS OF CERTAIN NORTH AMERICAN CARBONIFEROUS ASTEROIDS (ECHINODERMATA). Journal of Paleontology, 77(3):476-489.
BOOKSTEIN, F. L. 1991. Morphometric tools for landmark data: geometry and biology. Cambridge University Press, Cambridge, New York etc, i-xvii, 1-435 p.
BUTTS, C., S. KENTUCKY GEOLOGICAL, AND S. GEOLOGICAL. 1917. Mississippian formations of western Kentucky. State Journal Co., Frankfort.
CIGNONI, P., M. CORSINI, AND G. RANZUGLIA. 2008. MeshLab: an open-source 3D mesh processing system. ERCIM News, 73:47-48.
DEXTER, T. A., C. D. SUMRALL, AND M. L. MCKINNEY. 2009. Allometric strategies for increasing respiratory surface area in the Mississippian blastoid Pentremites. Lethaia, 42(2):127-137.
DRAKE, A. G., AND C. P. KLINGENBERG. 2008. The pace of morphological change: historical transformation of skull shape in St Bernard dogs. Proceedings of the Royal Society B: Biological Sciences, 275(1630):71-76.
DRYDEN, I. L. 2007. Shapes: Statistical Shape Analysis. R Package Version 1.1-1., http://www.maths.nott.ac.uk/~ild/shapes.
FOOTE, M. 1991. Morphological and Taxonomic Diversity in a Clade's History: The Blastoid Record and Stochastic Simulations. Contributions from the Museum of Paleontology, 28(6):101-140.
GALLOWAY, J. J., AND H. V. KASKA. 1957. Genus Pentremites and its Species. The Geological Society of America Memoir, 69:1-104.
GOODALL, C. 1991. PROCRUSTES METHODS IN THE STATISTICAL-ANALYSIS OF SHAPE. Journal of the Royal Statistical Society Series B-Methodological, 53(2):285-339.
HOROWITZ, A. S. 1965. Crinoids from the Glen Dean Limestone (Middle Chester) of Southern Indiana and Kentucky. Indiana Geological Survey Bulletin 34:52.
KATZ, S. G., AND J. SPRINKLE. 1976. Fossilized Eggs in a Pennsylvanian Blastoid. Science, 192(4244):1137-1139.
MACURDA, D. B. 1975. Pentremites (Blastoidea) of Burlington Limestone (Mississippian). Journal of Paleontology, 49(2):346-373.
MCKINNEY, M. L., AND K. J. MACNAMARA. 1991. Heterochrony : the evolution of ontogeny. Plenum, New York [etc.].
MOORE, R. C. 1940. Early Growth Stages of Carboniferous Microcrinoids and Blastoids. Journal of Paleontology, 14(6).
NELSON, W. J., AND S. ILLINOIS STATE GEOLOGICAL. 2002. Sequence stratigraphy of the lower Chesterian (Mississippian) strata of the Illinois Basin. Illinois State Geological Survey, Champaign, IL.
SEVASTOPULO, G. D. 2005. The early ontogeny of blastoids. Geological Journal, 40(3):351-362.
SUMRALL, C. D. 1996. Late Paleozoic edrioasteroids (Echinodermata) from the North American Midcontinent. Journal of Paleontology, 70(6):969-985.
TEAM, R. C. 2013. R: A language and environment for statistical computing. R
  Foundation for Statistical Computing, http://www.R-project.org/, Vienna, Austria.
WATERS, J. A. 1977. Quantification of shape by use of Fourier analysis; the Mississippian blastoid genus Pentremites. Paleobiology, 3(3):288-299.
WATERS, J. A., A. S. HOROWITZ, AND D. B. MACURDA. 1985. Ontogeny and Phylogeny of the Carboniferous Blastoid Pentremites. Journal of Paleontology, 59(3):701-712.
ZELDITCH, M. L., AND A. C. CARMICHAEL. 1989. Ontogenetic Variation in Patterns of Developmental and Functional-Integration in Skulls of Sigmodon-Fulviventer. Evolution, 43(4):814-824.
ZELDITCH, M. L., AND W. L. FINK. 1996. Heterochrony and heterotopy: Stability and innovation in the evolution of form. Paleobiology, 22(2):241-254.
ZELDITCH, M. L., D. L. SWIDERSKI, D. H. SHEETS, AND W. L. FINK. 2004. Geometric Morphometrics for Biologists. Elsevier Academic Press (USA), San Diego.