Translator Disclaimer
1 January 2008 Piggyback Whorls: A New Theoretical Morphologic Model Reveals Constructional Linkages Among Morphological Characters in Ammonoids
Author Affiliations +

A new theoretical morphological model is proposed for the analysis of growth, form and morphospace of ammonoid shells. In this model, the shape of a radial cross section through the shell is simulated by “piggybacking” of successive whorls. The “piggyback whorls model” is defined in terms of the enlarging rate of the perimeter and the proportion of the dorsal wall to the whorl periphery, if an isometric relationship is assumed between perimeter and area of the cross-sectioned whorl. Allometric coefficients on these growth parameters determine how compressed and evolute shells are formed. The present model successfully reproduced some correlations among purely geometric variables that have been reported in previous works and were also observed in our biometric analyses. This model yields a hypothesis of “constructional linkages” between aperture shape and coiling geometry that might provide a functional coupling between hydrostatic and hydrodynamic characters. The model may partly explain Buckman's Law of Covariation between rib features and shell shapes.


Ammonoids have undoubtedly been the most popular targets for morphospace analysis among fossil animals. Since Raup's (1967) pioneer work, large collections of morphological data have been used to explore functional morphology, macroevolutionary trend, and/or disparity of normally coiled ammonoids (Ward 1980; Bayer and McGhee 1984; Saunders and Swan 1984; Swan and Saunders 1987; Nikolaeva and Barskov 1994; Dommergues et al. 1996; Saunders and Work, 1996, 1997; Korn 2000; Korn and Klug 2003; McGowan 2004; Saunders et al. 2004; Gottobrio and Saunders 2005). A theoretical morphological model introduced by Raup (1967) has been widely applied to investigation of ammonoid morphospace, in which gross shell geometries are defined by whorl expansion rate (W), width of umbilicus (D) and whorl shape (S). The occupation patterns of the W-D-S morphospace are commonly used for assessing diversity in basic shell geometry. In addition, Saunders and Swan (1984) and Saunders and Work (1996) introduced several different parameters representing sculpture, aperture shape and suture complexity. The Raup's W, D, S, and other metric parameters used previously are suitable for non-destructive morphometric analysis of museum collections, and also available for measurements from photographs in publications. The use of these accessible parameters allows the development of a comprehensive database that includes hundreds of ammonoid taxa. Comparison of these parameters has proven to be a fruitful way for surveying the long-term evolutionary history of morphological diversity of ammonoids (Saunders and Swan 1984; Swan and Saunders 1987; Korn and Klug 2003).

However, previous morphospace analyses were based on data sets collected from adult or submature average individuals and did not deal with ontogenetic change of shell shape, although allometry is common in ammonoids (Fig. 1; e.g., Jacobs 1992; Korn and Klug 2002). Thus, ontogenetic aspects have been incorporated tenuously in morphospace analyses of ammonoids (see Okamoto 1996). Ontogeny can not be fully understood in terms of the models that simply specify the gross geometry of shell coiling and aperture shape. Better understanding of the relationship between ontogenetic aspects and morphological diversity requires focusing on the geometric balance between size parameters that determines growth direction (Ubukata 2002). A promising approach for connecting the gross shell geometry with specific growth processes may lie through analysis of the profile of the band of newly accreted material around the aperture that is characterized as the “aperture map” (Rice 1998; Ubukata 2003) or a set of aperture growth vectors regulating shell coiling (Hammer and Bucher 2005b). However, a morphospace provided by the aperture map model is necessarily high-dimensional and is not practical for morphospace analysis.

Fig. 1.

Radial cross sections of ammonoids to illustrate allometric growth. A. Beloceras sp., Devonian; Erfoud, Morocco. B. Phylloceras consanguineum Gemmellaro, 1876, Jurassic; Sakaraha, Madagascar. C. Meekoceras gracilitatis White, 1879, Triassic; Crittenden Spring, Nevada. D. Girtyoceras meslerianum (Girty, 1909), Carboniferous; Jackforth Creek, Oklahoma.


This paper introduces a new theoretical model for shell growth of normally coiled ammonoids and applies it to morphospace analysis. This model is designed to simulate allometric growth of a cross-sectioned ammonoid shell. The basic assumption of the model is isometric relationship be- tween the circumference and area of the cross-sectioned whorl; shape of the whorl section is passively regulated so as to conform the assumption. To verify the assumptions is the primary objective of this study. For this purpose, the occupation patterns of morphospace were analyzed in more than 100 Devonian to Cretaceous ammonoid species.

Institutional abbreviation.

  • UMUT, University Museum, University of Tokyo, Japan.

Modeling of cross-sectioned ammonoid shells

Background and basic concept.—The conventional model introduced by Raup (1967) represents growth of a planispiral shell as an expansion and rotation of a generating curve, which approximates the shape of the aperture, around a fixed coiling axis. Subsequently, Okamoto (1988) and Ackerly (1989) introduced “moving frame models” in which the amount and direction of shell growth is defined with respect to the previous aperture. Both methods conceive a shell as the trajectory of successive generating curves that retain their shapes throughout growth. More recently, Stone (1995) proposed a “CerioShell model” defined by different expansion rates of the aperture in the horizontal and vertical dimensions for visualizing allometric change of the aperture shape and geometry of coiling. The shape of the aperture can also change if helicospiral growth is assumed for each point on the aperture (Bayer 1978; McGhee 1978; Savazzi 1987; Checa 1991; Checa and Aguado 1992) or multiple growth vectors installed around the previous aperture determine the shape of the next aperture (Hammer and Bucher 2005b).

In these previous models, no association was explicitly defined between the shape of the newly formed aperture and form of the preceding whorl. However, in normally coiled ammonoids, adjacent whorls overlap, and the soft part which formed the succeeding whorl must have covered the preceding whorl from outside. The outer component of the dorsal wall of the succeeding whorl is regarded as a product of organic secretion at the supracephalic mantle fold (Kulicki et al. 2001). Therefore, not only the shape of the previous aperture but also the form of the ventral surface of the preceding whorl should influence geometry of the newly formed aperture around the mantle margin. Hutchinson (1989) proposed a feedback mechanism in which the preceding whorl provides a “road” for the mantle margin and guides shell growth. His “road-holding model” may explain how a succeeding whorl maintains its position with respect to the preceding whorl to achieve planispiral geometry, but is not concerned with how to regulate the whorl shape. Morita (1991a, b) simulated behavior of an expanding elastic mantle represented by a finite element model of a double membrane tube. His “DMS-tube model” predicted that an expanding mantle tends to be elongated ventrally if the dorsal part of the mantle is fixed on the shell. Such mantle behavior may be a possible mechanism to regulate the whorl shape of gastropods in which the head-foot mass presses the mantle margin near the coiling axis when the snail clamps its body to the hard bottom (Morita 2003). However, this mechanism does not seem applicable to morphogenesis of nektonic ammonoids.

In the present study, we attempt to realize the process of piling up of the ammonoid whorls in which the shape of each whorl is determined in relation to the form of the preceding whorl. Therefore, the model introduced herein is based on a “moving reference frame” taking into account previous whorls and the present approach does not need the coiling axis except for measurements. The mode of accretionary growth seems to be influenced by the geometric balance of growth kinematics around the shell margin (Ubukata 2002). However, it is not easy to measure parameters from actual specimens defined by a three-dimensional model including these algorithms or developmental processes. Here, to simplify the problem, we developed a geometric model designed to simulate two-dimensional patterns which correspond to the radial cross sections of ammonoid shells. Since growth lines of ammonoid shells are not always linear or rectidiradiate, a radial cross section of each whorl does not exactly represent the aperture shape but is a rough approximation of a slice of newly added shell material. Usefulness of the present geometric model can be assessed if results of computer simulations are compared with biometric data.

Fig. 2.

Schematic figure of a cross section of the ammonoid shell showing how to define parameters of the piggyback whorls model. The enlarging rate of the whorl periphery (E) is defined as the ratio of the total perimeter of the newly added whorl section (Li) to that of preceding whorl (Li-1). The proportion of the dorsal whorl (R) is given by the ratio of the circumferential length along the dorsal wall (l) with respect to Li. The shape of the whorl section (S) is represented by the ratio of breadth (B) and height (H) of the whorl.


Piggyback whorls model.—In the cross section, growth of a normally coiled ammonoid shell is represented by additions of successive whorls: a succeessive whorl “piggybacks” on the immediately preceding one. Geometric properties of the cross-sectioned shell consist of size, shape and positional relationship between neighboring whorls. In a simple case, the cross section of the whorl is regarded as elliptic (Fig. 2). Since shelly material is secreted at the distal extremity of mantle, increase rate of the perimeter of the whorl stands for the growth rate of the mantle. If the size of each whorl is defined by the perimeter of the whorl, the growth rate in the radial direction is determined by the enlarging ratio of the perimeter with respect to the preceding whorl, E. That is,

where Li , and L i-1 are total perimeters of the whorl section of the succeeding and preceding whorls, respectively. Also the growth rate of body mass is reflected by the enlarging rate of the area of the cross-sectioned whorl. The ratio of the area to the perimeter influences the shape of the whorl section which is defined by the aspect ratio of the ellipse, S (see Fig. 2). In the present model, the area0.5/perimeter ratio is assumed to be constant throughout growth as mentioned below. The positional relationship between neighboring whorls is represented by the overlap of the whorl. The whorl overlap is partly influenced by the whorl shape but is mainly determined by the proportion of the circumferential length along the dorsal wall (l) to the total perimeter of the whorl section (Li ), which is expressed as R:

Fig. 3.

Different shapes of succeeding whorl (gray area) piling up on the hemispherical preceding whorl with same Li, l, and A values.


Fig. 4.

Schematic diagram of an ammonoid whorl section. The shape of the whorl (S) is represented by the closed curve illustrated here which is given by the parametric equations indicated above the diagram. Figures in the right show examples of hypothetical whorl shapes with systematically varying the S value.


Fig. 5.

Spectrum of the computer-produced ammonoids with various values of E and R when each of them is fixed throughout growth.


The shape of a newly added whorl, S, is dependently determined by E and R, if the preceding whorl is given and the area of the succeeding whorl in the section, A, is specified. In the default setting of the model, the initial shell consists of a pair of semicircular whorls (Fig. 2), and isometric growth of A with respect to Li is assumed.

S depends on E and R but is not always uniquely determined by E and R. In some cases, two totally different whorl shapes are potentially generated using the same values of E and R (Fig. 3). In such cases, we chose the whorl shape which has a closer S value to the preceding whorl than the other.

In many ammonoids, the cross-sectional shape of the whorl is oval or subquadrate rather than elliptic. In computer simulations, therefore, the following parametric equations were used for producing an oval or subquadrate closed curve as a better representation of a whorl section (Fig. 4):


These equations define x and y coordinates of a point on the curve at an angle θ. B and H express the breadth and height of the whorl, respectively, and S is defined as B/H. The equations above were arbitrary designed just for approximation of the whorl shape, and there is no biological basis of its formulation. Possible varieties of whorl shape can be generated with varying the value of S (Fig. 4).

We can now define shell growth in terms of shape parameters E and R, if an isometric relationship is assumed between size parameters A and Li, and the values of E and R are fixed throughout growth. In this model, the circumferential length along the dorsal wall of the succeeding whorl (l = RLi) cannot be larger than that of the external wall of the preceding whorl [(1-R)Li /E]. Therefore, in the case of isometry, a theoretical model can be defined only within the following range of shape parameters:


In computer simulations, vast number of computer models with various shapes (S) and positions of the whorl were generated, and we searched the computer-generated models for the desired combination of Li and A which satisfies equations (1), (2), and (3) and isometric relationship between A and Li.

Computer simulations of shell growth

Isometric growth.—The effects of E and R on the shell form in cross section are shown on computer-produced shells in Fig. 5. When E and R are small, the model generates an evolute conch in which the whorls overlap little so that there is a wide umbilicus. An increase in E or R tends to produce an involute shell with a narrow umbilicus. It is a matter of course that the whorl expansion rate increases with increasing E. If E and R are too large to satisfy the former condition, any theoretical form cannot be defined in the piggyback whorls model (“forbidden range” sensu Tyszka 2006).

Fig. 6.

Allometric growth of ammonoids. Computer models were generated with ae and ar values of 0.97, 1.0, and 1.3. Each diagram shows ontogenetic change in relationship between whorl height (H) and whorl breadth (B) that are standardized by the perimeter of the last whorl (Lmax). Dashed line in each diagram represents the B/H curve in the case of isometry of E and R (ae = ar = 1.0). Note that B/H decreases with growth in the case of ar = 0.97. In each model, e = 1.9 and r = 0.24.


Theoretical models based on parameter values that are fixed throughout growth do not realize ontogenetic change in shell form. In addition, models shown in Fig. 5 do not produce variations in whorl shape: all of them have subcircular or subquadrate whorls. A greater variety can be accomplished by simulations in which E and R change with growth.

Allometric growth.—For representing allometric growth of ammonoid shells, the following equations were employed to define ontogenetic change of E and R:


where e, r, ae and ar , are allometric coefficients. Parameters e and r have same effects as the enlarging ratio of the whorl perimeter and proportion of the dorsal wall respectively, and are exactly same as E and R in the case of isometry (ae =1 and ar =1). Then, E and R are given by fi01_113.gif and rLi ar-1 , respectively. When ae >1 or ar <1, the portion of the whorl overlap decreases with growth. This condition inevitably causes an excessive area of the whorl section and results in positive allometry of A with respect to Li , if S is fixed to 1.0 so as to maintain a subcircular or subquadrate whorl shape. In other words, isometric growth of A with respect to this condition requires reduction of S to produce a more compressed whorl, because a slender shape generally has a small A 0.5 /Li ratio. Therefore, computer simulations in the default setting produce a slender whorl and cause a decrease of S with growth when ae >1 and/or ar <1 (Fig. 6).

Fig. 7.

Relationship between whorl perimeter (Li) and square root of cross-sectional area of the whorl (A0.5) which varies with ae and ar values. Dashed line in each diagram represents the A0.5/Li curve in the case of isometry of E and R (ae = ar = 1.0). A0.5/Li is generally small when ar has a large value. In each model, e = 1.9 and r = 0.24.


On the contrary, when ae<1 and/or ar >1, the whorl overlap increases as growth proceeds, and a burly shape tends to be formed for each whorl to make up for the “deficiency” of the area (Fig. 6). However, compensation for the “deficiency” by changing the whorl shape has its limit because A 0.5/Li takes a maximum value around S = 1. If ae is small enough or ar is large enough, the default setting, i.e., isometric growth of A with respect to Li , can not be held. In this case, a negative allometry was allowed in the present simulations: the maximum value of A was searched within the theoretically possible range and was employed for each whorl. Consequently, when ae is small and ar is large, A 0.5/Li tends to decrease with increasing size so that it has a smaller value in comparison with other cases (Fig. 7). If ae and ar are both considerably larger than zero, any theoretical form cannot be defined.

Fig. 8.

Sketches of radial cross sections of ammonoids to illustrate various types of shell forms, such as planorbicone (A, B), serpenticone (C, D), oxycone (E), discocone (F), platycone (G, H), spherocone (I, J, K), and cadicone (L). A. Paraceltites elegans. B. Pseudoclymenia dillensis. C. Tropigastrites lahontanus. D. Pterolytoceras sp. E. Beloceras sp. F. Phylloceras consanguineum Gemmellaro. G. Craspedites sp. H. Tetragonites glabrus. I. Damesites sugata. J. Goniatites multiliratus Gordon. K. Latanarcestes sp. L. Cabrieroceras sp.


Fig. 9.

Theoretical morphospace composed of ae and ar displaying several examples of computer-generated ammonoids that represent observed types illustrated in Fig. 8. A. e = 1.5 and r = 0.27. B. e = 1.75 and r = 0.34. C. e = 2.0 and r = 0.17. D. e = 2.3 and r = 0.05. E. e = 2.0 and r = 0.6. F. e = 1.7 and r = 0.27. G. e = 1.8 and r = 0.24. H. e = 2.3 and r = 0.21. I. e = 2.3 and r = 0.32. J. e = 1.95 and r = 0.28. K. e = 1.8 and r = 0.10. L. e = 1.6 and r = 0.8.


If appropriate values of ae and ar are used with various combinations of e and r, many varieties of shell forms observed in actual ammonoids as shown in Fig. 8, can be produced (Fig. 9). An evolute conch with a compressed whorl, or planorbicone, is generated when ar is small and ae is around 1.0 (Fig. 9A, B). A model with a large value of ae can produce a discoidal form with a compressed whorl and a narrow umbilicus (Fig. 9F). A spheroconic shell which is characterized by a subglobular whorl and involute coiling is readily reproduced when ar is around one and ae is slightly smaller than 1.0 (Fig. 9G, I, J). If r is large and ar is small, a theoretical model possessing a large value of e tends to generate an oxyconic form with an extremely compressed whorl (Fig. 9E), while a model with a small e value can produce a cadiconic form with a depressed whorl (Fig. 9L). Other observed forms, such as are seen in Fig. 1C, D, H, K, can be generated by our model (Fig. 9C, D, H, K).

Morphospace analysis

Computer simulations.—The piggyback whorls model assumes that the shape of a whorl section depends on the balance between the perimeter and area of the whorl section. This assumption may cause biased occupation patterns of theoretically possible forms in the morphospace which defines the relationship between growth parameters and crosssectional form. If so, analysis of the distribution of actual forms in relation to the theoretical morphospace allows us to test the assumption. Since allometric growth is defined by e, r, ae and ar, the morphospace composed of these growth parameters and a dependent subsidiary parameter such as S should be high-dimensional. Since accurate estimations of e and r from specimens are not easy, the relationship between pairs of ae and ar and subsidiary parameters including Raup' s ones was analyzed to assess explicability of the piggyback whorls model for well-known correlations among Raup's parameters.

In addition to S, the following three subsidiary parameters were also assessed: whorl expansion rate, Wc; width of umbilicus, or Raup' s D (Fig. 10); and ratio of the square root of the area to the perimeter, A0.5/Li. Raup (1967) defined the whorl expansion rate as the rate of increase in the diameter of an ammonoid shell per revolution: it can be defined independent of S in the case of isometry. However, increase in compression of the whorl during growth results in increase in the whorl expansion rate defined by Raup (1967). In this study, for avoidance of algebraic dependency, Wc was defined as the ratio of the distance from the coiling axis to the center of a given whorl to that of the preceding one (Fig. 10). The center of the whorl is defined as the point on the median line equally distant from the ventral and dorsal extremes of the whorl. When ae <>1 or ar <>1, the values of S, Wc, D, and A0.5/Li generally change with growth. In the present analysis, these subsidiary parameters were estimated for the latest whorl both in theoretical models and actual specimens.

Computer simulations were performed to establish theoretical morphospaces composed of ae, ar , and S, Wc, D or A 0.5/Li with various values of e, r, ae , and ar . A total of 200 theoretical models were simulated. In each model, values of e, r, ae and ar were randomly chosen from the normally distributed populations with the following means and standard deviations: µ = 2, σ = 0.2 for e; µ = 0.25, σ = 0.05 for r, σ = 1, σ = 0.05 for ae and ar . The subsidiary parameters associated with each combination of ae and ar attain various values depending on e and r, but are obviously correlated with ae and/or ar (Fig. 11A-D). S tends to decrease as ar decreases or ae increases (Fig. 11A). In this case, ar is the predominant factor in determination of the shape parameters. On the other hand, A 0.5/Li generally increases with decreasing ar (Fig. 11B). As mentioned above, a small ar or large ae value reduces the portion of the whorl overlap with growth and potentially yields an excessive area of the whorl section to increase A 0.5/Li (Fig. 11B). In this condition, balancing between A and Li reduces the value of S to produce a compressed whorl (Fig. 11A). Wc dominantly depends on ae (Fig. 11C), and this relation is rather obvious: a large ae value readily causes a large whorl expansion rate. D has a tendency to decrease with increase either of ae or ar , but is mainly influenced by ar (Fig. 11D). As a natural consequence of comparison of morphospaces between Fig. 11A and Fig. 11C, a clear negative correlation between S and Wc was generated in our computer simulations (Fig. 11E). For a similar reason, there is a positive correlation between S and D (Fig. 11F), and consequently a negative one between S and Wc , which have been reported in many ammonoids (Bayer and McGhee 1984; Saunders and Swan 1984; Swan and Saunders 1987; Dommergues et al. 1996; Saunders and Work 1996). These correlations derived from the assumption of the piggyback whorls model: balancing between A and Li determines S.

Fig. 10.

Measurements of the whorl expansion rate (Wc) and width of umbilicus (D). Note that C1 and c2 are defined as the distances from the coiling axis to the centers of the whorls. d1 and d2 are distances of umbilical seams from the coiling axis.


Biometric analyses.—The occupation patterns of hypothetical models in the morphospaces shown in Fig. 11 were compared with the distribution of actual ammonoids to test our assumption. To assess the pattern of morphological diversity of actual ammonoids, ae, ar, S, Wc, D and A0.5/Li were estimated on 123 specimens of 115 species, belonging to 72 families (see Appendix 1). Since each species was represented by one or a few specimen, the present analyses and following discussions basically dealt with shape variation above species level. Our material covers six ammonoid orders spanning from the Devonian to the Cretaceous.

Each specimen was first cut using an electric diamond saw vertical to the median plane. The half section was polished with graded series of silicon-carbide powders along the radial plane. A sheet of cellophane tape was pressed onto the polished surface to make the structure on the surface clearly visible. The polished surface was laid onto the image scanner (Epson GT-F600), and an image of the surface was captured and saved as a computer bitmap file. Two-dimensional coordinate data along the periphery of the whorl were collected on the bitmap image using a digitizing program written in Visual Basic 6.0 (Ubukata 2004, 2005), and then, the perimeter along the sectioned whorl was calculated. We measured to the outside of the shell wall because morphogenesis of the external shell surface is not affected by shell thickening. The height, breadth and center of the whorl were determined by the coordinate data to obtain S, D and W. For measuring A, the periphery of each whorl was traced and colored differently on a personal computer using Justsystem Hanako PhotoRetouch 2004 program, and its area was measured by counting pixels (Ubukata 2001). For this counting, a program written in Visual Basic 6.0 was used. Consequently, ae and ar were obtained as slopes of reduced major axes of logLi against logLi-1, and logl against logLi, respectively (Jones 1937). For details of the regression method using the reduced major axis, refer also to Imbrie (1956). Multiple regression analyses were conducted for parameters on ae and ar.

The range of forms of actual ammonoids examined was consistent with the occupation patterns of theoretical models in morphospace shown in Fig. 11. In actual ammonoids, S appears to decrease with increase of ae or decrease of ar in Fig. 12A. A multiple regression analysis revealed a significant negative partial correlation between ae and S, and a positive correlation between ar and S, and the latter is especially prominent (Table 1). Fig. 11B indicates a negative relationship between A 0.5/Li and ar , and a multiple regression analysis provided a significant trend to decreasing A 0.5/Li as a function of ar (Table 1). The value of Wc seems to increase as ae increases (Fig. 12C), which was supported by multiple regression analysis (Table 1). A significant negative partial correlation was found between ar and D (Table 1, Fig. 12D). Biometric analyses also revealed a clear negative or inverse relationship between S and Wc (Fig. 12E) and a week positive correlation between S and D (Fig. 12F). These results are generally concordant with results of computer simulations (Fig. 11), and thus support the assumption of the present model.

Fig. 11.

Results of computer simulations. A-D, theoretical morphospaces based on the piggyback whorls model composed of ae and ar showing values of shape parameters (S, A0.5/Li, Wc, D) of a theoretical model corresponding to each combination of ae and ar E, negative correlation between Wc and S values obtained from theoretical models. F, positive correlation between D and S. The values of S, A0.5/Li, Wc or D are exhibited by the size of the plots, and × indicates that a “forbidden” combinations of ae, ar, e, and r was employed in each simulation resulting in failure of defining form (A-D).


Table 1.

Results of multiple regression analyses.



Correlations among Raup's parameters.—Saunders and Swan (1984) and Swan and Saunders (1987) analyzed Raup's W-D-S morphospace in various Paleozoic ammonoids and identified correlations among these parameters that are not explicable in terms of hydrodynamics or hydrostatics: a negative correlation between W and D, and a positive one between S and D, that have been confirmed by several authors (Bayer and McGhee 1984; Dommergues et al. 1996; Saunders and Work 1996), and were also reconfirmed in the present study. Swan and Saunders (1987) suggested that these correlations originated in restrictions imposed by the relative aperture height and aperture area with respect to shell size. They found that the major axis of plots was distributed along apertureheight curves in the W-D morphospace and S-D plots run parallel to aperture-area curves (Swan and Saunders 1987: figs. 8, 9). Swan and Saunders (1987) supposed preferences for aperture height and aperture area that make a moderately deep U-shaped cross section to avoid biomechanical disadvantages.

In the present piggyback whorls model, isometric relationship is assumed between the area of the whorl cross section and the whorl perimeter. It may be safe to say that this assumption of the model is concordant with the Swan and Saunders (1987) hypothesis of restrictions on aperture height and aperture area. The piggyback whorls model displayed how constructional bias imposed by the area of whorl section work on the occupation pattern of morphospace: changing the geometry of shell coiling inevitably results in change in the aperture shape so as to keep the constant relative area of the aperture with respect to its perimeter. Since the values of Raup's parameters depend on r and ar , correlations among Raup's parameters can be regarded as axiomatic consequences of algebraic interdependence among them, if an isometric relationship is assumed between the area and perimeter of the whorl cross section. Concerning the Raup's model, Schindel (1990) pointed out that, the dorsal shell margin of two whorls can move away from the coiling axis at the same rate but with different D values, because of an algebraic interdependence between W and D under a fixed displacement rate of the dorsal shell margin (see also Stone 1996). Origination of a correlation between parameters from an assumed condition (e.g., fixed displacement rate, or isometry between area and perimeter) can be attributed to a simple mathematical problem; however, the assumed condition should be placed in a biological context. In the case of the present study, correlations among Raup's parameters may be interpreted as “constructional linkages” associated with particular configurations of aperture height and aperture area that were assumed by Swan and Saunders (1987).

Fig. 12.

The same morphospaces as Fig. 11 showing the distribution of 115 species based on measurements.


Buckman's Law of Covariation.—The piggyback whorls model may partly explain Buckman's Law of Covariation between ornament and shell shape: strongly ornamented am- monoids tend to have an evolute conch and circular whorls (Westermann 1966). This “law” has been confirmed by the morphologic analysis of large population samples of some Triassic ceratites (Dagys and Weitschat 1993; Dagys et al. 1999; Dagys, 2001) and a Cretaceous ammonite Neogastroplites (Kennedy and Cobban, 1976), and this phenomenon has been explained in terms of a reaction diffusion model (Guex et al. 2003) and a simple principle of proportionality (Hammer and Bucher 2005a).

If an ornamented closed curve is a priori given for whorl cross section as an initial setting, same computer simulations as shown in Fig. 11D tend to generate more depressed whorls with larger S values in the piggyback whorls model (Fig. 13), where compression is a posteriori determined. In terms of the piggyback whorls model, formation of heavy ornaments makes a long periphery of the whorl and reduces the A0.5/Li ratio. In this case, employing a circular aperture maximizes the area of the whorl section and then balances between A and Li, as in the case that a burly aperture makes up for the deficiency of the area caused by increase of the whorl overlap with growth (Fig. 6). Since shell involution is practically determined by r and ar that are a priori given in the present simulations, the relationship between evolute conch and ornamentation was not reproduced.

Fig. 13.

Results of same computer simulations as shown in Fig. 11 with an ornamented whorl shape. In order to produce an ornamented closed curve, the following parametric equations were used: x = 0.5Bcosθ, y = 0.5H(sinθ + 0.3|sin4θ|Q), where Q = H/B/5 if θ < π, while Q = 5B/H if θ > π. Note that simulated S values tend to be large as compared with Fig. 11F.


Yacobucci (2004) found that Cretaceous acanthoceratid ammonites with highly variable shell shapes did not show variable patterns of ornamentation, although the Buckman's Law should expect a positive correlation of variations between shell shape and rib features. She suggested that ornament morphogenesis is genetically and/or developmentally more controlled than is shell shape. If gross shell geometry is more variable than shell sculpture, heavy ornamentation which may lead changes in whorl shape. However, this explanation based on the present model can only be applied to tubercles and spiral ribs, because radial ribbing involves little change in aperture shape, which in turn does not influence A0.5/Li.

Hydrostatics and hydrodynamics.—As a result of computer simulations based on the piggyback whorls model, we expected that positive allometry of whorl periphery causes the whorl to become more slender and more rapidly expanding (Fig. 11A, C). Morphometric analysis also found that the allometric coefficient of periphery is negatively correlated with whorl compression but is positively correlated with whorl expansion rate (Fig. 12A, C, Table 1). Positive allometry of the whorl periphery means acceleration of the lateral expansion of the soft body. If this is not attended with acceleration of increase in total soft-body mass, body-chamber length is necessarily shortened to compensate rapid expansion of the diameter. Therefore, the process of building a compressed whorl with a high expansion rate may result in formation of a short body chamber. This study did not examine the body-chamber length because most of the specimens utilized did not preserve the entire body chamber. However, a negative correlation between body-chamber length and whorl expansion rate has been observed in some taxa (Okamoto 1996; Klug and Korn 2004). From the view point of hydrostatics, this correlation is theoretically expected if neutral buoyancy is presumed (Saunders and Shapiro 1986; Ebel 1990). The piggyback whorls model may explain how ammonoids constructed neutrally buoyant shell morphology; constraints imposed by the relative area of the aperture with respect to its perimeter may cause a negative correlation between body-chamber length and whorl expansion rate.

In addition, a negative correlation between S and whorl expansion rate (Fig. 12E) strongly suggests a negative correlation between whorl compression and body chamber length. A shell with a short body chamber has high hydrostatic stability which allows the animal's fast locomotion without rotation (Trueman 1941; Raup 1967; Saunders and Shapiro 1986; Ebel 1990; Okamoto 1996). In terms of hydrodynamics, a compressed form generally has a lower drag coefficient than robust one. However, a compressed shell has a smaller amount of retractor muscle and a larger surface area which generates a larger frictional drag than a thicker form. Thus, the compressed form is effective for reducing drag only at a high Reynolds number which represents a large size and/or a fast swimming velocity (Jacobs 1992; Jacobs and Chamberlain 1996; Seki et al. 2000). Positive allometry of whorl periphery readily realizes an ideal coupling of morphological characters for a fast-swimming ammonite, because of the constructional linkage between whorl compression and whorl expansion rate. Furthermore, it also results in an ontogenetic change from a less compressed to more compressed shape: such ontogenetic trend is suitable for swimming because smaller forms that swim slowly benefit from less compressed shapes (Jacobs 1992; Jacobs and Chamberlain 1996).

In general terms, this study suggests that constructional biases as assumed in the piggyback whorls model can provide a functional coupling of morphological characters. Functional and developmental integrations of characters have been focused in the context of developmental constraints (Alberch 1980, 1982; Kauffman 1983; Schwenk and Wagner 2001; Richardson and Chipman 2003), and have been discussed from viewpoints of pleiotropy (Cheverud 1996; Raff 1996; Adams 1998; Moore et al. 2004; Klingenberg 2005) or epigenetic processes (Müller 1989; Morita 1993). Although this study did not attempt to elucidate developmental mechanism of morphogenesis of functionally coupling characters, the piggyback whorls model expects “constructional linkages” between aperture shape and coiling geometry.

Concluding remarks and prospects

The piggyback whorls model introduced for mimicking ammonoid shell growth is based on the geometric balance among perimeters of the external and internal portions and area of the whorl section. The balance determines degree of compression and involution of whorls. None of the growth parameters used in this model directly defines the shape of the aperture and geometry of shell coiling. Theoretical forms generated by using the piggyback whorls model has a potential to explain wellknown correlations among whorl expansion rate, width of umbilicus and whorl shape (Fig. 11E, F), as consequences from constructional biases imposed by isometric relationship between perimeter and area of the whorl cross section.

The piggyback whorls model is not designed to simulate openly coiled or uncoiled form, because it attributes great variation in aspect ratio of whorl section to ontogenetic change of whorl overlap and whorl expansion rate. The present model is not comprehensive enough to reproduce various teratological patterns observed in actual ammonoids. In addition, it is not sufficient for analysis of the three-dimensional properties of shell form. The model does not address the effect of pattern of relative growth rate around the aperture on the aperture shape. Combining the present model with the aperture map model (Rice 1998) or aperture growth vector model (Hammer and Bucher 2005b) may be required to show how three-dimensional shell form is determined.

Nevertheless, the present model is useful for analysis of ontogenetic variation of ammonoids, which is usually based on cross sections of shells. There is a prospect that the model may explain how to realize a functional coupling between hydrostatic and hydrodynamic characters and how to produce the Buckman's Law of Covariation between rib features and shell shapes. The approach introduced here is also available for studies on nautilids, and might be applicable to gastropod shell morphology after some necessary modifications.


We thank Neil H. Landman (American Museum of Natural History, New York, USA) for review of an early draft, Hugo Bucher (Palä ontologisches Institut und Museum, Universität Zürich, Switzerland), Øyvind Hammer (Natural History Museum, University of Oslo, Norway), Jaroslaw Tyszka (Instytut Nauk Geologicznych PAN, Kraków, Poland), and an anonymous referee for helpful comments aiding improvement of the paper. Funding of this work was provided in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 17740338).


  1. S.C. Ackerly 1989. Kinematics of accretionary shell growth, with examples from brachiopods and molluscs. Paleobiology 4: 374–378. Google Scholar
  2. R.A. Adams 1998. Evolutionary implications of developmental and functional integration in bat wings. Journal of Zoology 246: 165–174. Google Scholar
  3. P. Alberch 1980. Ontogenesis and morphological diversification. American Zoologists 20: 653–667. Google Scholar
  4. P. Alberch 1982. Developmental constraints in evolutionary processes. In : J.T. Bonner (ed.), Evolution and Development , 313–332. Springer-Verlag, Berlin. Google Scholar
  5. U. Bayer 1978. Morphologenetic programs, instabilities, and evolution-a theoretical study. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 156: 226–261. Google Scholar
  6. U. Bayer and J.R. McGhee Jr. 1984. Iterative evolution of Middle Jurassic ammonite faunas. Lethaia 17: 1–16. Google Scholar
  7. M.F. Bogoslovskaya , L.F. Kuzina , and T.B. Leonova 1999. Classification and distribution of late Paleozoic ammonoids [in Russian, with English abstract]. In : A.Û. Rozanov and A.A. Ševyrev (eds.), Iskopaemye cefalopody: novejšie dostiženiâ v ih izučenii , 89–124. Paleontologičeskij Institut, Rossijskajâ academiâ nauk, Moskva. Google Scholar
  8. A. Checa 1991. Sectorial expansion and shell morphogenesis in molluscs. Lethaia 24: 97–114. Google Scholar
  9. A. Checa and R. Aguado 1992. Sectorial-expansion analysis of irregularly coiled shells; application to the Recent gastropod Distorsio. Palaeontology 35: 913–925. Google Scholar
  10. J.M. Cheverud 1996. Developmental integration and the evolution of pleiotropy. American Zoologists 36: 44–50. Google Scholar
  11. A.S. Dagys 2001. The ammonoid family Arctohungaritidae from the Boreal Lower-Middle Anisian (Triassic) of Arctic Asia. Revue Paléobiologie, Genève 20: 543–641. Google Scholar
  12. A.S. Dagys , H. Bucher , and W. Weitschat 1999. Intraspecific variation of Parasibirites kolymensis Bychkov (Ammonoidea) from the Lower Triassic (Spathian) of Arctic Asia. Mitteilungen Geologisches-Paläontologisches Institut Universität Hamburg 83: 163–178. Google Scholar
  13. A.S. Dagys and W. Weitschat 1993. Extensive intraspecific variation in a Triassic ammonoid from Siberia. Lethaia 26: 113–121. Google Scholar
  14. J-L. Dommergues , B. Laurin , and C. Meister 1996. Evolution of ammonoid morphospace during the Early Jurrasic radiation. Paleobiology 22: 219–240. Google Scholar
  15. K. Ebel 1990. Swimming abilities in ammonites and limitations. Paläontologische Zeitschrift 64: 25–37. Google Scholar
  16. W.E. Gottobrio and W.B. Saunders 2005. The clymeniid dilemma: functional implications of the dorsal siphuncle in clymeniid ammonoids. Paleobiology 31: 233–252. Google Scholar
  17. J. Guex , A. Koch , L. O'Dogherty , and H. Bucher 2003. A morphogenetic explanation of Buckman's law of covariation. Bulletin de la Société géologiaue de France 174: 603–606. Google Scholar
  18. Ø. Hammer and H. Bucher 2005a. Buckman's first law of covariation—a case of proportionality. Lethaia 38: 67–72. Google Scholar
  19. Ø. Hammer and H. Bucher 2005b. Models for the morphogenesis of the molluscan shell. Lethaia 38: 111–122. Google Scholar
  20. J.M.C. Hutchinson 1989. Control of gastropod shell shape: the role of the preceding whorl. Journal of Theoretical Biology 140: 431–444. Google Scholar
  21. J. Imbrie 1956. Biometrical methods in the study of invertebrate fossils. Bulletin of the American Museum of Natural History 108: 217–252. Google Scholar
  22. D.K. Jacobs 1992. Shape, drag, and power in ammonoid swimming. Paleobiology 18: 203–220. Google Scholar
  23. D.K. Jacobs and J.A. Chamberlain Jr. 1996. Buoyancy and hydrodynamics in ammonoids. In : N.H. Landman , K. Tanabe , and R.A. Davis (eds.), Ammonoid Paleobiology , 169–224. Plenum, New York. Google Scholar
  24. H.E. Jones 1937. Some geometrical considerations in the general theory of fitting lines and planes. Metron 13: 21. Google Scholar
  25. S.A. Kauffman 1983. Developmental constraints: internal factors in evolution. In : B. Goodwin , N. Holder , and C.C. Wiley (eds.), Development and Evolution , 195–225. Cambridge University Press, Cambridge. Google Scholar
  26. W.J. Kennedy and W.A. Cobban 1976. Aspects of ammonite biology, biogeography, and biostratigraphy. Special Papers in Palaeontology 17: 1–94. Google Scholar
  27. C.P. Klingenberg 2005. Developmental constraints, nodules, and evolvability. In : B. Hallgrímsson and B.K. Hall (eds.), Variation: A Central Concept in Biology , 219–249. Academic Press, San Diego. Google Scholar
  28. C. Klug and D. Korn 2004. The origin of ammonoid locomotion. Acta Palaeontologica Polonica 49: 235–242. Google Scholar
  29. D. Korn 2000. Morphospace occupation of ammonoids over the Devonian—Carboniferous boundary. Paläontologische Zeitschrift 74: 247–257. Google Scholar
  30. D. Korn and C. Klug 2002. Ammoneae Devonicae. In : W. Riegraf (ed.), Fossilum Catalogus I: Animalia Pars 138. 375 pp. Backhuys, Leiden. Google Scholar
  31. D. Korn and C. Klug 2003. Morphological pathways in the evolution of Early and Middle Devonian ammonoids. Paleobiology 29: 329–348. Google Scholar
  32. C. Kulicki , K. Tanabe , N.H. Landman , and R.H. Mapes 2001. Dorsal shell wall in ammonoids. Acta Palaeontologica Polonica 46: 23–42. Google Scholar
  33. G.R. McGhee Jr. 1978. Analysis of the shell torsion phenomenon on the Bivalvia. Lethaia 11: 315–329. Google Scholar
  34. A.J. McGowan 2004. The effect of the Permo-Triassic bottleneck on Triassic ammonoid morphological evolution. Paleobiology 30: 369–395. Google Scholar
  35. R. Morita 1991a. Finite element analysis of a double membrane tube (DMS-tube) and its implication for gastropod shell morphology. Journal of Morphology 207: 81–92. Google Scholar
  36. R. Morita 1991b. Mechanical constraints on aperture form in gastropods. Journal of Morphology 207: 93–102. Google Scholar
  37. R. Morita 1993. Development mechanics of retractor muscles and the “Dead Spiral Model” in gastropod shell morphogenesis. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 190: 191–217. Google Scholar
  38. R. Morita 2003. Why do univalve shells of gastropods coil so tightly? A head-foot guidance model of shell growth and its implication on developmental constraints. In : T. Sekimura , S. Noji , N. Ueno , and P.K. Maini (eds.), Morphogenesis and Pattern Formation in Biological Systems: Experiments and Models , 345–354. Springer-Verlag, Tokyo. Google Scholar
  39. P.J. Moore , W.E. Harris , V.T. Montrose , D. Levin , and A.J. Moore 2004. Constraints on evolution and postcopulatory sexual selection: trade-offs among ejaculate characteristics. Evolution 58: 1773–1780. Google Scholar
  40. G.B. Müller 1989. Ancestral patterns in bird limb development: a new look at Hampé's experiment. Journal of Evolutionary Biology 2: 31–47. Google Scholar
  41. S.V. Nikolaeva and I.S. Barskov 1994. Morphogenetic trends in the evolution of Carboniferous ammonoids. Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 193: 401–418. Google Scholar
  42. T. Okamoto 1988. Analysis of heteromorph ammonoids by differential geometry. Palaeontology 31: 35–52. Google Scholar
  43. T. Okamoto 1996. Theoretical modeling of ammonoid morphology. In : N.H. Landman , K. Tanabe , and R.A. Davis (eds.), Ammonoid Paleobiology , 225–251. Plenum, New York. Google Scholar
  44. K.N. Page 1996. Mesozoic ammonoids in space and time. In : N.H. Landman , K. Tanabe , and R.A. Davis (eds.), Ammonoid Paleobiology , 755–794. Plenum, New York. Google Scholar
  45. R.A. Raff 1996. The Shape of Life: Genes, Development, and the Evolution of Animal Form. 544 pp. University of Chicago Press, Chicago. Google Scholar
  46. D.M. Raup 1967. Geometric analysis of shell coiling: coiling in ammonoids. Journal of Paleontology 41: 43–65. Google Scholar
  47. S.H. Rice 1998. The bio-geometry of mollusk shells. Paleobiology 24: 133–149. Google Scholar
  48. M.K. Richardson and A.D. Chipman 2003. Developmental constraints in a comparative framework: a test case using variations in phalanx number during amniote evolution. Journal of Experimental Zoology Part B: Molecular and Developmental Evolution 296: 8–22. Google Scholar
  49. W.B. Saunders and E.A. Shapiro 1986. Calculation and simulation of ammonoid hydrodynamics. Paleobiology 12: 64–79. Google Scholar
  50. W.B. Saunders and A.R.H. Swan 1984. Morphology and morphological diversity of mid-Carboniferous (Namurian) ammonoids in time and space. Paleobiology 10: 195–228. Google Scholar
  51. W.B. Saunders and D.M. Work 1996. Shell morphology and suture complexity in Upper Carboniferous ammonoids. Paleobiology 22: 189–218. Google Scholar
  52. W.B. Saunders and D.M. Work 1997. Evolution of shell morphology and suture complexity in Paleozoic prolecanitids, the rootstock of Mesozoic ammonoids. Paleobiology 23: 301–325. Google Scholar
  53. W.B. Saunders , D.M. Work , and S.V. Nikolaeva 2004. The evolutionary history of shell geometry in Paleozoic ammonoids. Paleobiology 30: 19–43. Google Scholar
  54. E. Savazzi 1987. Geometric and functional constraints on bivalve shell morphology. Lethaia 20: 293–306. Google Scholar
  55. D.E. Schindel 1990. Unoccupied morphospace and the coiled geometry of gastropods: architectural constraint or geometric covariation? In : R.M. Ross and W.D. Allmon (eds.), Causes of Evolution: A Paleontological Perspective , 270–304. University of Chicago Press, Chicago. Google Scholar
  56. K. Schwenk and G.P. Wagner 2001. Function and evolution of phenotypic stability: connecting pattern to processes. American Zoologists 41: 552–563. Google Scholar
  57. K. Seki , K. Tanabe , N.H. Landman , and D.K. Jacobs 2000. Hydrodynamic analysis of Late Cretaceous desmoceratine ammonites. Revue Paléobiologie, Genève, Volume spéciale 8: 141–155. Google Scholar
  58. J.R. Stone 1995. CerioShell: a computer program designed to simulate variation in shell form. Paleobiology 21: 509–519. Google Scholar
  59. J.R. Stone 1996. The evolution of ideas: a phylogeny of shell models. The American Naturalist , 148: 904–929. Google Scholar
  60. A.R.H. Swan and W.B. Saunders 1987. Function and shape in late Paleozoic (mid-Carboniferous) ammonoids. Paleobiology 13: 297–311. Google Scholar
  61. A.E. Trueman 1941. The ammonite body-chamber, with special reference to the buoyancy and mode of life of the living ammonite. Quaternary Journal of the Geological Society of London 96: 339–383. Google Scholar
  62. J. Tyszka 2006. Morphospace of foraminiferal shells: results from the moving reference model. Lethaia 39: 1–12. Google Scholar
  63. T. Ubukata 2001. Geometric pattern and growth rate of prismatic shell structures in Bivalvia. Paleontological Research 5: 33–44. Google Scholar
  64. T. Ubukata 2002. Stacking increments: a new model and morphospace for the analysis of bivalve shell growth. Historical Biology 15: 303–321. Google Scholar
  65. T. Ubukata 2003. Pattern of growth rate around aperture and shell form in Bivalvia: a theoretical morphological study. Paleobiology 29: 480–491. Google Scholar
  66. T. Ubukata 2004. A three-dimensional digitizing system based on triangulation using multiple viewing images [in Japanese, with English abstract]. Geoscience Reports of Shizuoka University 31: 65–72. Google Scholar
  67. T. Ubukata 2005. Theoretical morphology of bivalve shell sculptures. Paleobiology 31: 643–655. Google Scholar
  68. P. Ward 1980. Comparative shell shape distribution in Jurassic-Cretaceous ammonites and Jurassic-Tertiary nautilids. Paleobiology 6: 32–43. Google Scholar
  69. G.E.G. Westerman 1966. Covariation and taxonomy of the Jurassic ammonite Sonninia adicra(Waagen). Neues Jahrbuch für Geologie und Paläontologie, Abhandlungen 124: 289–312. Google Scholar
  70. M.M. Yacobucci 2004. Buckman's Paradox: variability and constraints on ammonoid ornament and shell shape. Lethaia 37: 57–69. Google Scholar


Appendix 1

Specimens examined and measured values of parameters. All specimens have the prefix UMUT. The higher taxonomy follows Korn and Klug (2002) for Devonian ammonoids, Bogoslovskaya et al. (1999) for Carboniferous and Perminan ammonoids, and Page (1996) for Mesozoic ammonoids.





Takao Ubukata, Kazushige Tanabe, Yasunari Shigeta, Haruyoshi Maeda and Royal H. Mapes "Piggyback Whorls: A New Theoretical Morphologic Model Reveals Constructional Linkages Among Morphological Characters in Ammonoids," Acta Palaeontologica Polonica 53(1), (1 January 2008).

Back to Top