Combining ontogenetic and evolutionary scales ... - Wiley Online Library

EVOLUTION & DEVELOPMENT

9:5, 472 –482 (2007)

Combining ontogenetic and evolutionary scales of morphological disparity: a study of early Jurassic ammonites Sylvain Gerber, Pascal Neige, and Gunther J. Eble Universite´ de Bourgogne, UMR CNRS 5561 Bioge´osciences, 6 bd Gabriel, 21000 Dijon, France Author for correspondence (email: [email protected])

SUMMARY Two major research themes in Evolutionary Developmental Biology and in Paleobiology, respectively, have each become central for the analysis and interpretation of morphological changes in evolution: the study of ontogeny/ phylogeny connections, mainly within the widespread and controversial framework of heterochrony; and the study of morphological disparity, the morphological signal of biodiversity, describing secular changes in morphospace occupation during the history of any given clade. Although enriching in their respective fields, these two themes have remained rather isolated to date, despite the potential value of integrating them as some recent studies begin to suggest. Here, we explore the recent notion of developmental morphospaceFmorphospace carrying ontogenetic informationFas

a potential tool for bridging the gap between disparity dynamics and developmental dynamics. We elaborate this approach with a case study of Early Jurassic ammonite family Hildoceratidae (Mollusca, Cephalopoda). Morphometric analyses of the shell shape of 20 species spanning the morphological spectrum of the family are used to quantify and contrast juvenile and adult disparity levels. Adult disparity is significantly greater than juvenile disparity at the family level; yet, some subclades also display different patterns. In addition, comparisons of ontogenetic trajectories underline the prevalence of heterochrony-based evolutionary modifications within subfamilies (via ontogenetic scaling); they also point to the probable existence of pervasive developmental constraints structuring inhomogeneous morphospace occupation.

INTRODUCTION

Conceptual background The role of developmental changes as mechanisms in morphological macroevolution has been investigated for a long time under the framework of heterochrony (Gould 1977; Alberch et al. 1979; McKinney and McNamara 1991). Over the last few decades, heterochrony has undergone several conceptual and operational reformulations, and controversial broadening in scope by extension to the different levels of the biological hierarchy (Shea 1983; Raff and Wray 1989; McKinney and McNamara 1991; Gould 1992, 2000; Godfrey and Sutherland 1995; Raff 1996; Rice 1997; Klingenberg 1998; McKinney 1999; Hall 2003). At the same time, several related concepts have been revisited or devised to attempt to study more rigorously the interplay between ontogeny and phylogeny (e.g., heterotopy, Zelditch and Fink 1996). It is now widely acknowledged (Klingenberg 1998; McKinney 1999; Gould 2000; Webster and Zelditch 2005) that the extensive broadening of terminology and diversification of formalisms have brought confusion and conflicting interpretation in the analysis of empirical data. Discord over the value of the different formalisms led to mixed conclusions about the types of processes and the relative frequency of various types 472

of evolutionary change in development. Nevertheless, there seems to be a general agreement on the putative explanatory power of developmental phenomena on the rates and magnitudes of evolutionary changes, as attested by the emergence of the field of Evolutionary Developmental Biology (e.g., Maderson et al. 1982; Raff 1996; Hall 1999; Arthur 2002) after the return of ontogeny as a relevant causal vector in evolution (Gould 1977). Another major active research field concerned with the evolution of form is the study of morphological disparity (Gould 1989, 1991; Foote 1993a, 1996, 1997; Wills et al. 1994; Roy and Foote 1997). DisparityFthe quantitative description of the empirical distribution of taxa in morphospace (Foote 1991)Fsupplies an explicitly morphological component to biodiversity, and, complementing taxonomic data, has increased our knowledge and comprehension of the large-scale dynamics of biodiversity, mainly in Evolutionary Paleobiology (Foote 1997 and references herein), although it is proving to be informative as well in neontological studies (Roy et al. 2001; Neige 2003; McClain et al. 2004). Surprisingly, integrating both approaches has received relatively little attention until recent work (see Wagner 1995; & 2007 The Author(s) Journal compilation & 2007 Blackwell Publishing Ltd.

Gerber et al. Eble 1998, 2000, 2002, 2003; Ciampaglio 2002; Zelditch et al. 2003; McNamara and McKinney 2005 for explicit attempts). Such an integrative view might be illustrated by the following example: ontogenetic changes entail shifts in morphospace occupation, which in turn may alter disparity patterns. The variational properties of ontogeny are likely to constrain the frequency and magnitude of types of developmental changes. Thus, clade-wide disparity patterns might reflect an important contribution of developmental causes underlying lineage evolution. Such a scenario, using disparity as a quantitative measure, offers an opportunity to test hypotheses, which effectively render rigorous statements about the importance of developmental allowance vis-a`-vis different types of selection (Alberch 1982; Maynard-Smith et al. 1985).

Developmental morphospace and developmental disparity The notion of developmental morphospaceFmorphospace carrying ontogenetic informationFshould help find causal connections between developmental and disparity dynamics. Conversely, common morphospaces depicting adult morphologies only are referred to as nondevelopmental morphospaces. The simplest type of developmental morphospace refers to empirical morphospace, within which both juvenile and adult formsFdescribed under the same morphometric schemeFcan be displayed. In this context, ‘‘adult’’ mor-

Morphological disparity and developmental dynamics

473

phospace may be considered to be the last snapshot of a developmental morphospace partitioned into successive ontogenetic subspaces. Eble (1998, 2003) described some simple patterns that we may expect from comparisons of developmental versus nondevelopmental morphospaces in a clade (Fig. 1): on the one hand, juveniles and adults can display comparable levels of morphological disparity. It may result (1) from a strict one-to-one correspondence of juvenile and adult subspaces, which corresponds to a set of species having parallel ontogenetic (allometric) trajectories in the morphospace (Fig. 1a), or (2) from a set of species having nonparallel trajectories but occupying at the ontogenetic stages chosen for comparisons a comparable amount of morphospace (e.g., using range-based estimates of disparity). The latter implies a more or less pronounced low level of disparity (nonlinear ontogeny of disparity; Fig. 1b) occurring between the stages compared, even in the simplest case of linear allometric trajectories in shape space (see ‘‘Discussion’’). On the other hand, juveniles and adults may be characterized by different patterns of morphospace occupation: adults can be more disparate than juveniles (Fig. 1c), or the converse (Fig. 1d). Those cases are intuitively more inclined to occur. In both cases, consideration of subclades may also be of interest (Foote 1993b; Zelditch et al. 2003): it has been shown that a clade disparity pattern can be largely driven by some of its subclades. Hence, a global pattern can be the expression of the additive contributions of congruent subpatterns as well as

Fig. 1. Modified from Eble (2003). A typology of developmental versus nondevelopmental (adult) morphospace comparisons. Four taxa, A, B, C, and D, are represented in juvenile and adult morphospaces. Expected disparity contrasts fall into four categories: (a) similar levels of juvenile and adult disparity with one-to-one correspondence of juvenile and adult subspaces; (b) similar levels of juvenile and adult disparity with nonparallel ontogenetic trajectories; (c) discordant pattern with a higher adult disparity (divergent trajectories); (d) discordant pattern with a higher juvenile disparity (convergent trajectories). See text for further explanations.

474

EVOLUTION & DEVELOPMENT

Vol. 9, No. 5, September^October 2007

a consequence of the prevalent dynamics of one or some of the subclades.

Table 1. Number of specimens for each species and average specific adult size reported in the literature (septal approximation criterion)

Lower Jurassic ammonites as a case study We chose to illustrate this new approach with the early Jurassic ammonite family Hildoceratidae through a morphometric analysis of shell shape as a basis for quantifying and contrasting juvenile and adult disparity. Ammonites are a well-known group of externally shelled cephalopods. The quality of their fossil record, both rich in abundance and stratigraphically well defined, has made them a valuable group when dealing with large-scale trends across space and time. Interestingly, ammonites have been used as a model in both disparity and heterochronic perspectives (Saunders and Swan 1984; Dommergues et al. 1986; Landman 1987; Dommergues and Meister 1989; Neige et al. 2001; Saunders et al. 2004; Navarro et al. 2005; Moyne and Neige 2007). In the context of heterochronic studies, this focus is partly related to the fact that ammonites preserve in their shells a record of their growth (Landman 1988). Four stages can be generally recognized during ammonite growth with more or less marked morphological changes (see Bucher et al. 1996 for a review, and references herein): embryonic stage (Ammonitella), neanic stage, juvenile stage, and mature (adult) stage. The latter three define the postembryonic growth. Growth is determinate in ammonoids, and maturity occurs at a more or less definite size for each species (Bucher et al. 1996; Dommergues et al. 2002). Septal sutures, ribbing patterns, and shell shapes have been the main characters used for tracking evolutionary modifications of ontogeny among related species. Here, we focus on shell shape and shell coiling encompassing juvenile and adult stages. Approaches focusing on the ontogenetic variation and covariation of other shell characters would be possible as well, and would valuably enrich the description of shell developmental disparity (see for instance the ‘‘ornamental morphospace’’ of ribbing patterns in Dommergues et al. (2006), which could easily incorporate developmental information).

MATERIALS AND METHODS Dataset The present study is based on Lower Jurassic ammonites (Mollusca, Cephalopoda). Selected species belong to the Hildoceratidae, known from the Pliensbachian to the Bajocian (Lower and Middle Jurassic). Here, we used a subsample of this family ranging from Upper Pliensbachian to Upper Toarcian (
190 up to
180 Ma, Lower Jurassic). Selected species are well defined regarding their morphology and their intraspecific variation, and have been extensively used for biostratigraphic applications (e.g., Dean et al. 1961; Elmi et al. 1997). Our dataset consists of 370 specimens belonging to 20 species and 15 different genera, offering a reasonable sampling of the spectrum of shell shape diversity of the Hildoce-

Species Harpoceratinae Protogammoceras celebratum marianii Fuciniceras cornacaldense Matteiceras geometricum Tiltoniceras antiquum Eleganticeras elegantulum Clevieras elegans exaratum Polyplectus discoides Harpoceras serpentinum falciferum subplanatum Osperlioceras bicarinatum Pseudolioceras lythense Arieticeratinae Emaciaticeras emaciatum Arieticeras algovianum amalthei Hildoceratinae Hildaites murleyi Orthildaites douvillei Hildoceras sublevisoni

N

Adult size (mm)

18 14 14 10 11 29 29 39 17 23 24 17 14 20

75 42 78 60 66 64 110 80 75 122 228 101 58 109

9 23 9

33 50 32

14 10 26

85 108 92

Specimens measured provide a broad coverage of size and shape variations occurring during ontogeny (cross-sectional data; Cock 1966).

ratidae family (Table 1). No clear phylogenetic hypothesis exists in the literature for the selected species. However, a large consensus exists (Gabilly; 1976; Donovan et al. 1981; Howarth 1992) for gathering selected species into five monophyletic subsets (Fig. 2). Adult stage is determined using septal approximation, which underlines the end of shell growth, a classical procedure within ammonoids (see Bucher et al. 1996). Some have proposed that sexual dimorphism may occur in some species, based on the recognition of two co-occurring morphotypes (microconchs and macroconchs) sharing a similar morphology at early stages of development. This dimorphism hypothesis has been challenged by Davis et al. (1996), who claimed to go beyond simple ‘‘matchmaking’’ and to accumulate many instances before to propose any microconch–macroconch association. In the clade studied herein, only a few cases of species that could display dimorphism have been reported. These cases are far from being properly demonstrated (under rigorous assumptions listed by Davis et al. 1996), and thus we have preferred to use only macroconchs for species in which sexual dimorphism has been proposed.

From measurements to morphospaces Morphometric scheme and Raup’s parameters Given the relative simplicity of the coiling process in ammonoids, three linear measurements suffice to provide a comprehensive

Gerber et al.

Morphological disparity and developmental dynamics

475

Fig. 3. Morphometry of ammonoid shell. Linear measurements (diameter D, radius R, Whorl height Wh) from which Raup’s parameters are computed are shown on the left. The four landmarks selected for the geometric morphometric (GM) approach are shown on the right. The GM approach offers a valuable notion of shape distance, while retaining close relationships with the Raup model (see text).

Fig. 2. Hypothesized phylogenetic relationships among Hildoceratid genera. Five monophyletic subfamilies are recognized: Harpoceratinae 1 (1), Harpoceratinae 2 (2), Harpoceratinae 3 (3), Arieticeratinae (4), and Hildoceratinae (5).

capture of shell shape in lateral view: diameter length D, radius length R, and whorl height Wh (Fig. 3). Indeed, Raup’s geometric model (Raup 1966, 1967) allows variously coiled shells to be described by only four parameters. Following Raup’s model, many articles have explored coiling in ammonoids, demonstrating this model to be an operational way to investigate the morphological spectrum of coiled shells across various space and time scales (see e.g., Ward 1980; Saunders and Swan 1984; Neige et al. 1997). Two of the four Raup’s parameters are used here: the whorl expansion rate (WR) and the distance of the generating curve from coiling axis (DR). They are computed as follows: WR ¼ ðR=ðD
RÞÞ2 DR ¼ ðR
WhÞ=R The two other parameters are not informative in the context of the study, because Hildoceratidae are planispiral forms (translation rate TR vanishes) and the section shape (as described by SR) brings no additional information for characterizing coiling aspects of the shell (Raup 1966, 1967).

Raup’s space and shape space Although it offers an interesting portrayal of ammonoid lateral shell shape, the quantitative assessment of disparity in the

WR
DR plane is not straightforward. WR and DR dramatically differ both in terms of unit and magnitude. Computing disparity would then be biased in favor of the parameter having the highest magnitude, and similar biases would appear whatever the standardization chosen (e.g., reducing variables to unit variance). To circumvent this limitation, we applied landmark-based geometric morphometrics to our data (Bookstein 1991; Marcus et al. 1996; Dryden and Mardia 1998; Zelditch et al. 2004). Raw measurements are easily converted into one-dimensional coordinates. This gives four landmarks having [0, Wh, R, D] as coordinates. The procedure is then comparable with generalized Procrustes analysis (Rohlf and Slice 1990), except that the rotational fit is not required (see Reyment and Kennedy 1998 for a similar approach). For organisms described by p collinear landmarks, centering and unit centroid size scaling make all specimens lie on the p-2 surface of a unit radius hypersphere (Small 1996; Dryden and Mardia 1998). As in most morphometric studies, the amount of shape variation in the sample is sufficiently small for the linear tangent space to be a reasonable representation of intershape distances in the nonlinear shape space (r40.99) (Rohlf 1999; Slice 2001). The Euclidean metric is spherical under the null model of random digitization (isotropic) error and supports the interpretation of principal components (Rohlf 2000). As p equals 4 in our case, the shape space is two-dimensional and its tangent approximation can thus be entirely displayed on a sheet of paper. The approach applied here provides a meaningful notion of shape distance for subsequent statistical and disparity analyses performed.

Comparing ontogenetic stages Shape prediction The approach we adopted to describe ontogenetic trajectories is close to that described in Zelditch et al. (2003). Parameters from multivariate regression of log-transformed measurements against

476

EVOLUTION & DEVELOPMENT

Vol. 9, No. 5, September^October 2007

log(D) are used to predict average shape at a given size (diameter) along the trajectory. We slid residuals to estimate dispersion around the predicted shape. This assumes equal variability at every ontogenetic stage for a given species; this may sound biologically unrealistic but it allows taking into account intraspecific variation when computing and comparing disparity estimates.

Pattern of morphospace occupation and disparity measures Classical procedures devised for disparity analyses (see Foote 1997 for a review) are used here, although applied in a different context. Numerous disparity proxies have been proposed to quantify the spread and spacing of forms in morphospaces. Some of them are more or less correlated (Ciampaglio et al. 2001), and we have thus selected two of them that reflect two different aspects of disparity: (1) total variance (sum of univariate variances), which is proportional to the average interspecimen distance (Procrustes distance here), and computed as the trace of the covariance matrix of shape coordinates, and (2) total range (sum of univariate ranges), which reflects the amount of morphospace occupied. For subclade comparisons (subfamily level), summed ranges are measured along the major axes of each subset (see Foote 1992). Disparity values and confidence intervals are computed using a bootstrap resampling procedure at the specimen level (2000 replicates).

Comparing ontogenetic trajectories As a complement to the foregoing approach outlined, which contrasts juvenile and adult shapes, we also compare ontogenetic trajectories among species. This is performed by computing and testing for significant differences between ontogenetic pathways of the species in size–shape space and shape space (i.e., the morphospace). In the context of this study, which uses continuous morphometric data to describe the shell as a unique, integrated morphological structure, evolutionary modifications of ontogenies may be ascribed to heterochrony or allometric repatterning. Referring to recent disambiguation of the ontogeny/phylogeny relationship and how a given pattern may be empirically tractable, we recognize heterochrony-based evolutionary modification as implying overlapping of compared trajectories in shape space, but these trajectories need not be overlapping in size–shape space (e.g., see

Webster and Zelditch 2005; Mitteroecker et al. 2005 for recent conceptual and methodological clarifications, but see McNamara 2002). In contrast, allometric repatterningFallometric reassociation between traitsFis detected when a significant nonzero angle occurs in the morphospace between two species allometric trajectories. Three complementary tests are used here to distinguish between these two types of ontogenetic changes (Fig. 4): only the first two identify heterochrony under the narrow definition adopted.

Test 1 (Fig. 4a) Permutation test based on within-species multivariate regression of the shape variables on size (Mitteroecker et al. 2005). The statistics used for this test are the within-species summed squared residuals. This is a test for elongation or truncation of an ancestral allometric trajectory (ontogenetic scaling as null hypothesis) when the size and shape remain associated (no dissociation of size and shape in size– shape space).

Test 2 (Fig. 4b) Permutation test based on within-species model II multivariate regression in shape space (Mitteroecker et al. 2005). In a similar way, although not in the same space, the statistic used is the sum of within-species squared residuals from regression. Rejection of the null hypothesis implies a difference between trajectories of shape change (angle and/or location in the morphospace between compared species).

Test 3 (Fig. 4c) Permutation test based on within-species multivariate regression of the shape variables on size (see Test 1 herein), but data are centered before computation. In a case where both previous tests are rejected (arguing for allometric repatterning), this one tests for parallelism of trajectories in shape space (i.e., descendant trajectory corresponds to a translation from the ancestral one). Statistical significance is assessed by random reassignment of specimens to species (9999 permutations) with Bonferroni’s adjustment of the level of significance for multiple comparisons (Bonferroni-type procedures allow correction of a-inflation when carrying out simultaneous statistical inferences; Rice 1989; Garcia 2004). All morphometric, disparity, and statistical analyses have been programmed in R (Ihaka and Gentleman 1996) (scripts available from S. G. on request).

Fig. 4. Comparison of ontogenetic trajectories. Label s stands for size; s1 and s2 define the shape space (usually multivariate). Two ontogenetic trajectories are compared in size–shape space, as well as the resulting trajectories of shape change in shape space (s1–s2). (a) Overlapping trajectories in both size–shape space and shape space (heterochrony via ontogenetic scaling); (b) trajectories identical in shape space but differing in size–shape space (heterochrony with dissociation of size and shape); (c) trajectories differing in both size–shape space and shape space (allometric repatterning).

Gerber et al. RESULTS The linear model holds well for multivariate regression of logtransformed measurements (Wh and R; see before) on logdiameter (R240.9 for all species). The size in the adult stage of a species corresponds to the average specific diameter measured at the end of the adult phragmocone (septal approximation as a homologous proxy for maturity; Table 1). The size in the juvenile stage has been set to 30 mm because it corresponds to the common smallest size for which specimens have been measured in all species under study, thus avoiding extrapolation beyond the ranges of variation observed. Figure 5 displays the spectrum of morphological variation exhibited by the 20 hildoceratid species at juvenile and adult stages in Raup DR
WR morphospace. These species occupy a rather large amount of morphospace compared with the spread reached by all Jurassic ammonites (Ward 1980). Juveniles occupy a more central location in the morphospace, whereas adults have a highly anisotropic distribution, all lying within an almost linear range of variation between two ex-

Morphological disparity and developmental dynamics

477

treme shapes: low DR values (approximately 0.10) associated with high WR values (approximately 2.75) and high DR values (approximately 0.55) associated with low WR values (approximately 1.75).

Juvenile/adult disparity levels Disparity comparisons have been performed in tangent shape space. Variance-based measures are proportional to the average interspecimen distance (Procrustes distance). At the family level, adults exhibit a significantly higher disparity level than their related juveniles (Table 2 and Fig. 6). This pattern is also found in the Harpoceratinae 3 subfamily. Harpoceratinae 1 provides an opposite pattern, whereas juvenile and adult levels of disparity are similar in Harpoceratinae 2, Arieticeratinae, and Hildoceratinae. Range-based measures of disparity, although describing a different aspect of morphospace occupation, yield concordant results (note that no rarefaction analysis is required in the context of juvenile–adult comparisons; Foote 1992).

Fig. 5. Distribution of hildoceratid juveniles and adults in the DR
WR plane of Raup’s morphospace (Raup 1967). Gray dots correspond to raw data. Dashed line corresponds to the distribution of Jurassic ammonites (Ward 1980; note that Ward’s contour encloses 85% of his sample [587 Jurassic ammonite species]). Open circles are the juvenile average shapes of each of the 20 species investigated (Table 1). Black squares represent their related adult average shapes. Error bars correspond to bootstrap standard errors (see ‘‘Materials and methods’’).

478

EVOLUTION & DEVELOPMENT

Vol. 9, No. 5, September^October 2007

Table 2. Results of morphological disparity analyses for juvenile–adult comparisons using variance- and range-based metrics Disparity level (total variance) Clade

Juvenile

Adult

Difference in means

Disparity level (total range)

Bootstrap 95% CI of difference in means

Hildoceratidae 0.006528 0.012500 0.005972 [0.004514, Harpoceratinae 1 0.002663 0.001671
0.000992 [
0.001839, Harpoceratinae 2 0.004174 0.005072 0.000898 [
0.000507, Harpoceratinae 3 0.003459 0.009046 0.005587 [0.004053, Arieticeratinae 0.004178 0.005087 0.000909 [
0.000496, Hildoceratinae 0.000877 0.001033 0.000156 [
0.000452,

0.007449]
0.000108] 0.002167] 0.007194] 0.002378] 0.000765]

Juvenile 0.454452 0.219743 0.350851 0.308243 0.175449 0.186887

Adult

Difference in means

Bootstrap 95% CI of difference in means

0.490616 0.036164 [0.018430, 0.186373
0.033370 [
0.051907, 0.328135
0.022716 [
0.048690, 0.360050 0.051807 [0.021582, 0.179929 0.004480 [
0.049255, 0.194422 0.007535 [
0.060035,

0.053565]
0.013880] 0.007074] 0.080192] 0.056889] 0.075863]

Comparisons have been performed at family and subfamily levels (see Figs. 2 and 6 and Table 1). Ninety-five percent CI for differences in juvenile and adult average disparity obtained by bootstrap resampling at the specimen level (2000 replicates). CI, confidence intervals.

Comparison of trajectories Eight species grow isometrically (insignificant multivariate regression of shape variables on size): Fuciniceras cornacaldense and Matteiceras geometricum (Harpoceratinae 1), Tiltoniceras antiquum, Eleganticeras elegantulum, and Polyplectus discoides (Harpoceratinae 2), Harpoceras falciferum and Pseudolioceras lythense (Harpoceratinae 3), and Arieticeras amalthei (Arieticeratinae). The 12 other species display allometric growth with more or less pronounced changes throughout ontogeny. This is manifested as ontogenetic trajectories of various length in shape space. Within each subclade (subfamily), the null hypothesis of heterochrony as a description of evolutionary modifications among related species cannot be rejected (except for Harpoceras serpentinum) even using the sharpened sequential Bonferroni’s method (e.g., Hochberg and Benjamini 1990) having an increased statistical power (Table 3). Heterochronic changes via ontogenetic scaling (OS; test 1) occurred within all

subfamilies. Heterochronic changes including dissociation of size and shape, but overlapping trajectories of shape change (HD; test 2), are only present among subfamilies (Table 3). Interestingly, allometric repatterning is observed in several interclade comparisons, but trajectories are nevertheless all parallel in shape space (test 3). However, we can distinguish two sets of species, pointing in two opposite directions (the ‘‘extreme shapes’’ described above). This suggests a marked constraint in the direction of ontogenetic shape changes, restricted to a narrow bundle of alternative pathways. Hence, the two observed opposite directions of ontogenetic trajectories may be related to a ‘‘passive trend’’ resulting from the central location of the juvenile distribution on the morphospace.

DISCUSSION Disparity analysis of the family Hildoceratidae at the ontogenetic scale reveals that, for shell coiling at least, adult

Fig. 6. Bar plot of the results of morphological disparity analyses for juvenile–adult comparisons using variance- and range-based metrics. Comparisons have been performed at family and subfamily levels (see Fig. 2, Tables 1 and 2). Significant differences between juvenile and adult disparity levels.

Gerber et al.

479

Morphological disparity and developmental dynamics Table 3. Results of pairwise comparison of trajectories

Harpoceratinae 1

P. celeb. P. mari. Cl. eleg. Cl. exar. H. serp. H. subp. O. Bicar. E. emac. A. algo. H. mur. O. douv. H. subl.

P. celeb.

P. mari.

– NS //NS //NS NS //NS //NS /NS /NS NS NS /NS

OS – //NS //NS NS //NS //NS /NS /NS NS NS /NS

Harpoceratinae 2 Cl. eleg.

Cl. exar.

Harpoceratinae 3 H. serp.

H. subp.

Arieticeratinae

O. Bicar.

OS OS – NS //NS NS NS //NS //NS //NS //NS //NS

OS – //NS NS /NS //NS //NS //NS //NS //NS

OS OS

OS HD

– NS //NS //NS //NS //NS //NS

OS – //NS //NS //NS //NS //NS

–

//NS //NS /NS /NS /NS /NS /NS

Hildoceratinae

E. emac.

A. algo.

H. mur.

O. douv.

H. subl.

HD HD

HD HD

OS OS

OS OS

HD HD

HD

HD

HD

HD

HD

– NS NS NS NS

OS – NS NS NS

OS OS – NS NS

OS OS OS – NS

OS OS OS OS –

Three tests are sequentially performed for inferring the types of evolutionary modifications occurring. Test 1 tests for heterochrony via ontogenetic scaling (OS, extension/truncation of trajectories). If the null hypothesis of ontogenetic scaling is rejected, test 2 tests for heterochrony with dissociation of size and shape (HD). If both hypotheses are rejected, it implies allometric repatterning and test 3 tests for a possible parallelism of trajectories in shape space. Significance () and nonsignificance (NS) are reported below the diagonal (permutation test). Evolutionary interpretations (OS, HD) are above the diagonal. P. celeb., Protogrammoceras celebratum; P. mari., Protogrammoceras marianii; Cl. Eleg., Cleviceras elegans; Cl. exar., Cleviceras exaratum; H. serp., Harpoceras serpentinum; H. subp., Harpoceras subplanatum; O. Bicar., Osperleioceras bicarinatum; E. emac., Emaciaticeras emaciatum; A. algo., Arieticeras algovianum; H. mur., Hildaites murleyi; O. douv., Orthildaites douvillei; and H. subl., Hildoceras sublevisoni. Within-subfamily comparisons are delimited by rectangles.

forms display a greater variety of shapes (in terms of both variance and range) than their corresponding juvenile forms. It indicates a higher degree of similarity in early ontogeny, followed by a significant spreading of taxa within the morphospace as growth proceeds. This may seem counterintuitive at first glance when considering that flexibility is usually thought to be higher early in ontogeny. Nevertheless, our dataset only deals with postembryonic growth and it is highly plausible that embryonic stages (ammonitella and neanoconch)Fprovided they could be commensurably portrayed beside postembryonic stages within a common morphospaceFmay display a different pattern. Whereas in some cases the recognition of growth stages in ammonites is difficult, transitions between stages are sometimes accompanied by abrupt changes (critical points or ‘‘Knickpunkte’’; Kullmann and Sheuch 1972). Such abrupt changes have been reported particularly between embryonic and postembryonic stages (slope changes in allometric plots; e.g., Landman 1987; Neige 1997 for examples on Jurassic and Cretaceous ammonites). For such cases, developmental disparity patterns are likely to be polyphasic as well. Most organisms have complex, nonlinear ontogenies. Hence, two-stage comparisons (juvenile/adult) of disparity level may not necessarily offer a complete rendition of the ‘‘ontogeny’’ of disparity (e.g., Fig. 1B where the decrease of disparity in between stages cannot be confidently inferred by interpolation of adult and juvenile disparity levels). A possible solution consists in taking into account a larger number of homologous developmental stages if these can be easily

identified (for instance, successive developmental instars in arthropods, or teeth stages in mammals; see Eble 2002) and to follow disparity fluctuations across these stages. Alternatively, an allometric approach may be used as exemplified here for the Hildoceratidae ammonites in Fig. 7. This is a more convenient approach for most empirical cases and in the context of paleontological studies, where biological age data of individuals are often unavailable (but see Jones and Gould 1999). This may be viewed as a ‘‘multistage’’ extension of Eble’s (2003) ‘‘allometric partitioning’’ (whereas the homologous developmental stage approach is named ‘‘maturational partitioning’’). Allometric partitioning corresponds to a partition of the developmental morphospace into successive ontogenetic subspaces. Several predicted shapes are computed along the trajectories for different sizes (based on regression parameters), allowing a comparison of the disparity level at given sizes to be made. Disparity curves then describe the continuous diffusion of taxa within the morphospace throughout ontogeny and may reveal strong nonlinearities (Figs. 1B and 7). Interestingly, this allometric investigation of morphospace occupation establishes links between disparity variation and size variation. Allometric and size evolutionary trends have been widely documented as well as their consecutive size-related shape alterations in the paleontological literature (e.g., Gould 1966). Likewise, as a corollary, this approach should help in emphasizing changes in morphospace occupation driven by evolutionary modification of size. For instance, a decrease of disparity (adult disparity, as

480

EVOLUTION & DEVELOPMENT

Vol. 9, No. 5, September^October 2007

Fig. 7. Multistage extension of ‘‘allometric partitioning’’ of developmental morphospace (the ‘‘ontogeny’’ of disparity). Example with the Hildoceratidae and two of its subclades. Disparity is measured here as the total range (error bars are bootstrapped standard errors). Size (mm) is in abscissa. The first column shows the variation in the total range (in ordinate) as growth proceeds. Columns 2 and 3 present the inspection of disparity pattern along the axes of the shape space. Letters x and y stand for the principal components of Procrustes shape variables (i.e., relative warps). The remaining columns illustrate the same range variations, but here as shifts of minima and maxima of specimen distributions along shape space axes (in ordinate), thus depicting a shift of clouds within morphospace. For instance, in Harpoceratinae 1 (second row), the disparity curve shows a marked decrease along the y-axis of the shape space (third column), with a minimum around 50–60 mm.

in most disparity articles) can result from a decrease in size (e.g., selective pressure acting on size) and such a pattern may not be fully explained only from the exploration of adult morphospace (Fig. 7). Combined analyses of ontogenetic trajectories in shape space (developmental morphospace) and size–shape space allow a comprehensive description of the evolutionary modifications of ontogenies among related taxa. Noticeably, in the Hildoceratidae studied here, evolutionary changes in ontogeny within subfamilies (except for isometrically growing taxa) all match the criteria of heterochrony (common ontogenetic trajectories with or without dissociation in size–shape space), even though the absence of a resolved phylogenetic framework precludes further investigations at this time: indeed, a deeper phylogenetic knowledge at the species and genera levels is still required to ensure correct ancestor-descendant polarity and to allow robust heterochronic inferences (Fink 1982, 1988). Among-subfamily comparisons reveal parallelism of trajectories in shape space and two sets

can be distinguished, pointing toward opposite directions. This fact is responsible for the adult distribution in Raup’s space and the increase in disparity, as growth proceeds and taxa move away from their central juvenile distribution. Relating heterochrony and other evolutionary modifications of ontogeny to disparity dynamics (McNamara and McKinney 2005), and assessing how a given ontogenetic change impacts disparity level is still exploratory and goes beyond the scope of this article. Nevertheless, needless to say that this is of primary interest for inferring explanatory processes of disparity patterns that have been documented. Which kind of evolutionary ontogenetic changes are the most frequent? Which one potentially has a greater impact on morphospace occupation and structuring? Heterogeneity in morphospace is acknowledged to reflect contingency and/or constraints (e.g., Raup 1967, MaynardSmith et al. 1985, for discussions on Raup’s space). In our case, the retaining of ancestral trajectories (‘‘phylogenetic legacy’’) seems to be widespread in the Hildoceratidae

Gerber et al. family. Within subfamilies (and sets of subfamilies [Harpoceratinae 1/Arieticeratinae/Hildoceratinae], [Harpoceratinae 2 and 3] pro parte), differences between trajectories, when they exist, reduce to simple translations in shape space (parallelism is retained). These results strongly argue in favor of evolutionary changes occurring earlier in ontogeny (embryonic stage) and outline a more constrained variability during postembryonic stages. For interclade comparisons revealing allometric repatterning, the maintenance of parallel trajectories in shape space, even if pointing toward opposite directions, is striking. It suggests pervasive developmental constraints channeling the potentialities of morphometric trait covariation (without regard to the polarity of the sequence of morphological changes occurring during ontogeny), leading to the patterned, inhomogeneous occupation of the morphospace we have described. This article demonstrates the advantages of taking into account developmental information in analyses of morphological disparity. Developing tools that allow to assess putative interactions between disparity dynamics and developmental dynamics promises fruitful investigations and valuable inferences when tracking explanatory processes that have entailed the variety of macroevolutionary patterns documented to date (e.g., in terms of morphospace occupation). Developmental morphospace is one of these tools, and one of the easiest to settle for empirical purposes. It offers a characterization of morphological disparity in a developmental perspective (plot of disparity curve against ontogenetic time) and extends its range of interpretation by the incorporation of major ontogenetic evolutionary changes as testable hypotheses. Acknowledgments We gratefully acknowledge Chuck Ciampaglio, Alistair J. McGowan, and Rudolf A. Raff for constructive comments on an earlier version of the manuscript. This article is a contribution to the team FED (Forme, Evolution, Diversite´) of the UMR CNRS 5561 Bioge´osciences, the GDR MEF (Morphome´trie et Evolution des Formes), and the ECLIPSE II (CNRS) research program.

REFERENCES Alberch, P. 1982. Developmental constraints in evolutionary processes. In J. T. Bonner (ed.). Evolution and Development. Springer-Verlag, Berlin, pp. 313–332. Alberch, P., Gould, S. J., Oster, G. F., and Wake, D. B. 1979. Size and shape in ontogeny and phylogeny. Paleobiology 5: 296–317. Arthur, W. 2002. The emerging conceptual framework of evolutionary developmental Biology. Nature 415: 757–764. Bookstein, F. L. 1991. Morphometric Tools for Landmark Data: Geometry and Biology. Cambridge University Press, New York. Bucher, H., Landman, N. H., Klofak, S. M., and Guex, J. 1996. Mode and rate of growth in ammonoids. In N. H. Landman, K. Tanabe, and R. A. Davis (eds.). Ammonoid Paleobiology. Plenum Press, New York, pp. 407– 461. Ciampaglio, C. N. 2002. Determining the role that ecological and developmental constraints play in controlling disparity: examples from the crinoid and blastozoan fossil record. Evol. Dev. 4: 170–188.

Morphological disparity and developmental dynamics

481

Ciampaglio, C. N., Kemp, M., and McShea, D. W. 2001. Detecting changes in morphospace occupation patterns in the fossil record: characterization and analysis of measures of disparity. Paleobiology 27: 695–715. Cock, A. G. 1966. Genetical aspects of metrical growth and form in animals. Q. Rev. Biol. 41: 131–190. Davis, R. A., Landman, N. H., Dommergues, J. L., Marchand, D., and Bucher, H. 1996. Mature modifications and dimorphism in ammonoid cephalopods. In N. H. Landman, K. Tanabe, and R. A. Davis (eds.). Ammonoid Paleobiology. Plenum Press, New York, pp. 463–539. Dean, W. T., Donovan, D. T., and Howarth, M. K. 1961. The Liassic ammonite zones and subzones of the North-West European province. Bull. Br. Mus. Nat. Hist. Geol. 4: 435–505. Dommergues, J.-L., David, B., and Marchand, D. 1986. Les relations ontogene`se–phylogene`se: applications pale´ontologiques. Geobios 19: 335–356. Dommergues, J.-L., Dommergues, C.-H., and Meister, C. 2006. Exploration of the Oxynoticeratidae ornamental morphospace using the discrete cosine transform (DCT) to analyze rib patterns. Paleobiology 32: 628–651. Dommergues, J.-L., and Meister, C. 1989. Trajectoire ontoge´ne´tiques et he´te´rochronies complexes chez des ammonites (Harpoceratinae) du Jurassique infe´rieur (Dome´rien). Geobios 12 (Me´m. spec.): 157–166. Dommergues, J.-L., Montuire, S., and Neige, P. 2002. Size patterns through time: the case of the Early Jurassic ammonite radiation. Paleobiology 28: 423–434. Donovan, D. T., Callomon, J. H., and Howarth, M. K. 1981. Classification of the Jurassic Ammonitina. In M. R. House and J. R. Senior (eds.). The Ammonoidea. The Evolution, Classification, Mode of Life and Geological Usefulness of a Major Fossil Group. Academic Press, London, pp. 101– 155. Dryden, I. L., and Mardia, K. V. 1998. Statistical Shape Analysis. John Wiley and Sons, New York. Eble, G. J. 1998. The role of development in evolutionary radiations. In M. L. McKinney and J. A. Drake (eds.). Biodiversity Dynamics: Turnover of Populations, Taxa and Communities. Columbia University Press, New York, pp. 132–161. Eble, G. J. 2000. Contrasting evolutionary flexibility in sister groups: disparity and diversity in Mesozoic atelostomate echinoids. Paleobiology 26: 56–79. Eble, G. J. 2002. Multivariate approaches to development and evolution. In N. Minus-Purvis and K. J. McNamara (eds.). Human Evolution through Developmental Change. Johns Hopkins University Press, Baltimore, pp. 51–78. Eble, G. J. 2003. Developmental morphospaces and evolution. In J. P. Crutchfield and P. Schuster (eds.). Evolutionary Dynamics. Oxford University Press, Oxford, pp. 35–65. Elmi, S., Rulleau, L., Gabilly, J., and Mouterde, R. 1997. Toarcien. In E. Cariou and P. Hantzpergue (eds.). Biostratigraphie du Jurassique OuestEurope´en et Me´diterrane´en : Zonations Paralle`les et Distribution des Inverte´bre´s et Microfossiles. Bull. Centre Rech. Elf Explor. Prod., Me´m. 17, Pau, pp. 25–36. Fink, W. L. 1982. The conceptual relationship between ontogeny and phylogeny. Paleobiology 8: 254–264. Fink, W. L. 1988. Phylogenetic analysis and the detection of ontogenetic patterns. In M. L. McKinney (ed.). Heterochrony in Evolution: A Multidisciplinary Approach. Plenum Press, New York, pp. 71–91. Foote, M. 1991. Analysis of morphological data. In N. L. Gilinsky and P. W. Signor (eds.). Analytical Paleobiology. The Paleontological Society, Knoxville, pp. 59–86. Foote, M. 1992. Rarefaction analysis of morphological and taxonomic diversity. Paleobiology 18: 1–16. Foote, M. 1993a. Discordance and concordance between morphological and taxonomic diversity. Paleobiology 19: 185–204. Foote, M. 1993b. Contributions of individual taxa to overall morphological disparity. Paleobiology 19: 403–419. Foote, M. 1996. Models of morphological diversification. In D. Jablonski, D. H. Erwin and J. H. Lipps (eds.). Evolutionary Paleobiology. The University of Chicago Press, Chicago, pp. 62–86. Foote, M. 1997. The evolution of morphological diversity. Annu. Rev. Ecol. Syst. 28: 129–152. Gabilly, J. 1976. Le Toarcien a` Thouars et dans le centre-ouest de la France. Editions du CNRS, Paris.

482

EVOLUTION & DEVELOPMENT

Vol. 9, No. 5, September^October 2007

Garcia, L. V. 2004. Escaping the Bonferroni iron claw in ecological studies. Oikos 105: 657–663. Godfrey, L. R., and Sutherland, M. R. 1995. Flawed inference: why sizebased tests of heterochronic processes do not work. J. Theor. Biol. 172: 43–61. Gould, S. J. 1966. Allometry and size in ontogeny and phylogeny. Biol. Rev. 41: 587–640. Gould, S. J. 1977. Ontogeny and Phylogeny. Harvard University Press, Cambridge. Gould, S. J. 1989. Wonderful Life: The Burgess Shale and the Nature of History. Norton, New York. Gould, S. J. 1991. The disparity of the Burgess Shale arthropod fauna and the limits of cladistic analysis: why we must strive to quantify morphospace. Paleobiology 17: 411–423. Gould, S. J. 1992. Heterochrony. In E. F. Keller and E. A. Lloyd (eds.). Keywords in Evolutionary Biology. Harvard University Press, Cambridge, pp. 158–165. Gould, S. J. 2000. Of coiled oysters and big brains: how to rescue the terminology of heterochrony, now gone astray. Evol. Dev. 2: 241–248. Hall, B. K. 1999. Evolutionary Developmental Biology. Kluwer, Dordrecht. Hall, B. K. 2003. Evo-Devo: evolutionary developmental mechanisms. Int. J. Dev. Biol. 47: 491–495. Hochberg, Y., and Benjamini, Y. 1990. More powerful procedures for multiple significance testing. Stat. Med. 9: 811–818. Howarth, M. K. 1992. The Ammonite Family Hildoceratidae in the Lower Jurassic of Britain. The Palaeontographical Society, London. Ihaka, R., and Gentleman, R. 1996. R: a language for data analysis and graphics. J. Comput. Graph. Stat. 5: 299–314. (http://www.r-project.org). Jones, D. S., and Gould, S. J. 1999. Direct measurement of age in fossil Gryphaea: the solution to a classical problem in heterochrony. Paleobiology 25: 158–187. Klingenberg, C. P. 1998. Heterochrony and allometry: the analysis of evolutionary change in ontogeny. Biol. Rev. 73: 79–123. Kullmann, J., and Scheuch, J. 1972. Absolutes und relatives Wachstum bei Ammonoideen. Lethaia 5: 129–146. Landman, N. H. 1987. Ontogeny of upper cretaceous (Turonian–Santonian) scaphitid ammonites from the western interior of North America: systematics, developmental patterns, and life history. Bull. Am. Mus. Nat. Hist. 185: 117–241. Landman, N. H. 1988. Early ontogeny of Mesozoic ammonites and nautilids. In J. Weidmann and J. Kullmann (eds.). CephalopodsFPresent and Past. Sweitzerbart’sche Verlagsbuchhandlung, Stuttgart, pp. 215–228. Maderson, P. F. A., et al. 1982. The role of development in macroevolutionary change. In J. T. Bonner (ed.). Evolution and Development. Springer-Verlag, Berlin, pp. 279–312. Marcus, L. F., Corti, M., Loy, A., Naylor, G. J. P., and Slice, D. E. 1996. Advances in Morphometrics. Plenum Press, New York. Maynard-Smith, J., Burian, R., Kauffman, S., Alberch, P., Campbell, J., Goodwin, B., Lande, R., Raup, D., and Wolpert, L. 1985. Developmental constraints and evolution. Q. Rev. Biol. 60: 265–287. McClain, C. R., Johnson, N. A., and Rex, M. A. 2004. Morphological disparity as a biodiversity metric in lower bathyal and abyssal gastropod assemblages. Evolution 58: 338–348. McKinney, M. L. 1999. Heterochrony: beyond words. Paleobiology 25: 149–153. McKinney, M. L., and McNamara, K. J. 1991. Heterochrony: The Evolution of Ontogeny. Plenum Press, New York. McNamara, K. J. 2002. Changing times: changing places: heterochrony and heterotopy. Paleobiology 31 (suppl.): 17–26. McNamara, K. J., and McKinney, M. L. 2005. Heterochrony, disparity, and macroevolution. Paleobiology 31 (suppl.): 17–26. Mitteroecker, P., Gunz, P., and Bookstein, F. L. 2005. Heterochrony and geometric morphometrics: a comparison of cranial growth in Pan paniscus versus Pan troglodites. Evol. Dev. 7: 244–258. Moyne, S., and Neige, P. 2007. The space–time relationship of taxonomic diversity and morphological disparity in the Middle Jurassic ammonite radiation. Palaeogeography, Palaeoclimatology, Palaeoecology 248: 82– 95. Navarro, N., Neige, P., and Marchand, D. 2005. Faunal invasions as a source of morphological constraints and innovations? The diversification

of the early Cardioceratidae (Ammonoidea; Middle Jurassic). Paleobiology 31: 98–116. Neige, P. 1997. Ontogeny of the Oxfordian ammonite Creniceras renggeri from the Jura of France. Eclogae Geol. Helv. 90: 605–616. Neige, P. 2003. Spatial patterns of disparity and diversity of the recent cuttlefishes (Cephalopoda) across the Old World. J. Biogeogr. 30: 1125– 1137. Neige, P., Elmi, S., and Rulleau, L. 2001. Existe-t’il une crise au passage Lias-Dogger chez les ammonites? Approche morphome´trique par quantification de la disparite´ morphologique. Bull. Soc. Ge´ol. France 172: 257–264. Neige, P., Marchand, D., and Laurin, B. 1997. Heterochronic differentiation of sexual dimorphs among Jurassic ammonite species. Lethaia 30: 145–155. Raff, R. 1996. The Shape of Life: Genes, Development, and the Evolution of Animal Form. University of Chicago Press, Chicago. Raff, R. A., and Wray, G. A. 1989. Heterochrony: developmental processes and evolutionary results. J. Evol. Biol. 2: 409–434. Raup, D. M. 1966. Geometric analysis of shell coiling: general problems. J. Paleontol. 40: 1178–1190. Raup, D. M. 1967. Geometric analysis of shell coiling: coiling in ammonoids. J. Paleontol. 41: 43–65. Reyment, R. A., and Kennedy, W. J. 1998. Taxonomic recognition of species of Neogastroplites (Ammonoidea, Cenomanian) by geometric morphometric methods. Cretaceous Res. 19: 25–42. Rice, S. H. 1997. The analysis of ontogenetic trajectories: when a change in size or shape is not heterochrony. Proc. Natl. Acad. Sci. USA 94: 907–912. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223–225. Rohlf, F. J. 1999. Shape statistics: Procrustes superimpositions and tangent spaces. J. Classif. 16: 197–223. Rohlf, F. J. 2000. On the use of shape spaces to compare morphometric methods. Hystrix (n.s.) 11: 9–25. Rohlf, F. J., and Slice, D. E. 1990. Extensions of the Procrustes method for the optimal superimposition of landmarks. Syst. Zool. 39: 40–59. Roy, K., Balch, D. P., and Hellberg, M. E. 2001. Spatial patterns of morphological diversity across the Indo-Pacific: analyses using strombid gastropods. Proc. R. Soc. London B 268: 1–6. Roy, K., and Foote, M. 1997. Morphological approaches to measuring biodiversity. Trends Ecol. Evol. 12: 277–281. Saunders, W. B., and Swan, A. R. H. 1984. Morphology and morphologic diversity of mid-Carboniferous ammonoids. Paleobiology 10: 195–228. Saunders, W. B., Work, D. M., and Nokolaeva, S. V. 2004. The evolutionary history of shell geometry in Paleozoic ammonoids. Paleobiology 30: 19–43. Shea, B. T. 1983. Allometry and heterochrony in the African apes. Am. J. Phys. Anthropol. 62: 275–289. Slice, D. E. 2001. Landmark coordinates aligned by procrustes analysis do not lie in Kendall’s shape space. Syst. Biol. 50: 141–149. Small, C. G. 1996. The Statistical Theory of Shape. Springer, New York. Wagner, P. J. 1995. Testing evolutionary constraint hypotheses with Early Paleozoic gastropods. Paleobiology 21: 248–272. Ward, P. D. 1980. Comparative shell shape distributions in Jurassic– Cretaceous ammonites and Jurassic–Tertiary nautilids. Paleobiology 6: 32–43. Webster, M., and Zelditch, M. L. 2005. Evolutionary modifications of ontogeny: heterochrony and beyond. Paleobiology 31: 354–372. Wills, M. A., Briggs, D. E. G., and Fortey, R. A. 1994. Disparity as an evolutionary index: a comparison of Cambrian and recent arthropods. Paleobiology 20: 93–130. Zelditch, M. L., and Fink, W. L. 1996. Heterochrony and heterotopy: stability and innovation in the evolution of form. Paleobiology 22: 241– 254. Zelditch, M. L., Sheets, H. D., and Fink, W. L. 2003. The ontogenetic dynamics of shape disparity. Paleobiology 29: 139–156. Zelditch, M. L., Swiderski, D. L., Sheets, H. D., and Fink, W. L. 2004. Geometric Morphometrics for Biologists: A Primer. Elsevier Academic Press, New York.