information to users

INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type o f computer printer. The quality of this reproduction is dependent upon the quality o f the copy submitted.

Broken or indistinct print, colored or poor quality

illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted.

Also, if

unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back o f the book. Photographs included in the original manuscript have been reproduced xerographically in this copy.

Higher quality 6” x 9” black and white

photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.


THE UNIVERSITY OF CHICAGO

THE MACROEVOLUTIONARY HISTORY OF DIVERSITY AND DISPARITY IN DISASTEROID, HOLASTEROID AND SPATANGOID HEART URCHINS VOLUME ONE

A DISSERTATION SUBMITTED TO THE FACULTY OF THE DIVISION OF THE BIOLOGICAL SCIENCES AND THE PRTTZKER SCHOOL OF MEDICINE IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY COMMITTEE ON EVOLUTIONARY BIOLOGY

BY GUNTHER JENSEN EBLE

CHICAGO, ILLINOIS AUGUST 1997


UMI Number: 9800594

UMI Microform 9800594 Copyright 1997, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI

300 North Zeeb Road Ann Arbor, MI 48103


Copyright © 1997 by Gunther Jensen Eble All rights reserved


To my parents


TABLE OF CONTENTS VOLUME ONE LIST OF FIGURES LIST OF TABLES ACKNOWLEDGMENTS

vii x xii

GENERAL INTRODUCTION

1

I. THE ROLE OF DEVELOPMENT IN EVOLUTIONARY RADIATIONS Introduction Evolutionary Radiations, Development, and Paleontology Data The Nature of the Questions General Approaches Study of origination and extinction Analysis o f morphological disparity Test of model predictions Specific Approaches Description and study of heterochrony Study of fluctuating asymmetry Inferences from mode of larval development Mapping o f developmental information onto cladograms Morphospace comparisons A Case Study: Testing Higher-Taxon Innovation in Rugged Fitness Landscapes Explanations for Temporal Asymmetry Data Evaluating the Pattern o f Temporal Asymmetry Kauffman's Model A Test of the Model of Rugged Fitness Landscapes Discussion An Alternative Procedure to Test the Model of Rugged Fitness Landscapes Coda Conclusions

7 7 9 11 13 14 14 20 22 23 23 24 24 24 26

0. DIVERSIFICATION OF DISASTEROIDS, HOLASTEROIDS AND SPATANGOIDS IN THE MESOZOIC Introduction Materials and Methods Taxonomic Diversity Morphological Disparity Results and Discussion Taxonomic Diversity Patterns and Models Morphological Disparity through Time Conclusions iv


29 29 30 30 33 35 44 47 55 57 59 59 61 61 63 66 66 74 87

m . CONTRASTING MORPHOLOGICAL DIVERSIFICATION IN SISTER CLADES HOLASTEROID AND SPATANGOID ECHINOIDS IN THE MESOZOIC 91 Introduction 91 Materials and Methods 99 Morphological Disparity 99 Sampling and Stratigraphic Resolution 105 Taxonomic Diversity 108 Comments on the Use o f Phytogeny and the Notion o f Constraint 108 Morphological Disparity and Taxonomic Diversity through Time 113 Atelostomata Disasteroids 126 Holasteroids 129 Spatangoids 131 Range of Morphospace 135 The Origin of Orders 138 Partitioning Disparity into Character Complexes 145 Disparity of Originations: Disparity, Versatility, and Evolutionary Potential 159 A Comment on Disparity and Sorting 165 Conclusions 168

VOLUME TWO IV. THE RECOVERY OF DISPARITY IN SISTER CLADES: SPATANGOIDS AND HOLASTEROIDS AFTER THE K-T Introduction Materials and Methods Morphology Sampling and Stratigraphic Resolution Taxonomic Diversity A Comment on Diversity and Disparity Disparity and Diversity in the Paleocene: Initial Recovery, Delayed Recovery or Revisiting of Extinction? Long-Term Consequences of Recovery in Spatangoids Testing for the Effect of Covariation of Characters on Disparity Patterns Disparity of Originations and the Accumulation of Disparity through Time Disparity of Originations The Accumulation of Disparity through Time The Origin of Oddities: Holasteroid and Spatangoid Morphological "Extremes" Holasteroid Failure to Recover Approach to Outliers How Bizarre? Sensitivity of Outlier Determination to Choice of Scaling Baseline Conclusions

237 237 239 239 247 250

V. ON THE DUAL NATURE OF CHANCE IN EVOLUTIONARY BIOLOGY Introduction Darwin on Chance

257 257 258


173 177 177 180 181 181 184 213 222 230 230 233

vi

On the Use of Chance in Evolutionary Biology The Simultaneous Use of Both Meanings o f Chance: Examples Extinction Studies Origins of Higher Taxa Determinism in Ecology Hierarchy Theory Evolutionary Genetics Evolutionary Developmental Biology Chance and Necessity Conclusions REFERENCES A PPEN D IX I APPENDIX H APPENDIX EH

261 265 265 268 268 269 270 271 273 276 278

Stratigraphic Ranges List of Taxa Sampled and Sources Morphometric Data

REFERENCES FOR APPENDIX I REFERENCES FOR APPENDIX H


301 308 312 336 339

LIST OF FIGURES

CHAPTER ONE

FIGURE I.

I Originations o f phyla, classes and orders through time

31

1.2

Bootstrap distribution of cumulative originations

32

1.3

Theoretical prediction of long-jump adaptation

36

1.4

Cumulative phyla vs. cumulative genera

39

1.5

Cumulative classes vs. cumulative genera

40

1.6

Cumualtive orders vs. cumulative genera

42

1.7

Alternative test o f long-jump adaptation -- phyla

49

1.8

Alternative test o f long-jump adaptation- - classes

50

1.9

Alternative test of long-jump adaptation —orders

51

1.10

Random simulation o f a declining origination process

53

CHAPTER TWO

FIGURE 2.1

Morphometric scheme

64

2.2

Analysis of taxonomic diversity

67

2.3

Morphospace occupation- - principal components

75

2.4

Morphological disparity and taxonomic diversity contrasted

80

2.5

Holasteroid and spatangoid disparity contrasted

88

vii


v iii

CHAPTER THREE

FIGURE 3.1

Stratigraphic and phylogenetic framework

94

3.2

Representatives of each group

96

3.3

Morphometric scheme

100

3.4

Morphospace occupation -- principal components

114

3.5

Disparity and diversity —Atelostomata

124

3.6

Disparity and diversity —Disasteroida

127

3.7

Disparity and diversity —Holasteroida

130

3.8

Disparity and diversity —Spatangoida

132

3.9

Changes in total range —Atelostomata

136

3.10

Changes in total range —Disasteroida

137

3.11

Changes in total range —Holasteroida

139

3.12

Changes in total range -- Spatangoida

140

3.13

The origin of orders

142

3.14

Oral and aboral disparity —Atelostomata

148

3.15

Oral and aboral disparity - Disasteroida

150

3.16

Oral and aboral disparity —Holasteroida

151

3.17

Oral and aboral disparity —Spatangoida

152

3.18

Disparity o f originations -- Atelostomata

161

3.19

Disparity of originations —Disasteroida

163

3.20

Disparity of originations - Holasteroida

164

3.21

Disparity of originations —Spatangoida

166


CHAPTER FOUR

FIGURE 4.1

Generic diversity in holasteroids and spatangoids

176

4.2

Diversity and disparity -- Campanian to Thanetian

186

4.3

Diversity and total range —Campanian to Thanetian

190

4 .4

Oral and aboral disparity —Campanian to Thanetian

194

4.5

Morphospace occupation -- principal components

198

4.6

Cenozoic disparity and diversity —Spatangoida

215

4.7

Cenozoic disparity and total range —Spatangoida

2 17

4.8

Cenozoic oral and aboral disparity - Spatangoida

2 18

4.9

PC A and disparity —Atelostomata

224

4.10 PCA and disparity- - Holasteroida

226

4.11

PCA and disparity —Spatangoida

228

4.12

Cenozoic disparity o f originations —Spatangoida

232

4.13

Cumulative disparity of originations

235

4.14

The origin of oddities

240

4.15

Disparity and phylogeny —holasteroids

245

4.16

Effect of scaling on the disparity of outliers

248

4.17

Regression of disparities implied by different baselines

252

4.18

Disparity and phylogeny —alternative baseline

CHAPTER FIVE

FIGURE 5.1.

Logical scheme for chance in evolutionary biology


274

LIST OF TABLES

CHAPTER ONE

TABLE 1.1

Major hypotheses for evolutionary radiation

15

1.2

The conceptual structure of the rugged fitness landscapes model

56

CHAPTER TWO

TABLE 2.1

Diversification rate vs. end-Cretaceous extinction percentage

72

CHAPTER THREE

TABLE 3.1

Time scale, generic diversity, and sample size

107

3.2

Categories of constraint

111

3.3

Summary of empirical results

170

CHAPTER FOUR

TABLE 4.1

Time scale, generic diversity, and sample size x


182

XI

4.2

Outlying urchins and their disparity

242

Outlying urchins and their disparity —alternative baseline

251

TABLE 4.3

CHAPTER FIVE

TABLE 5.1

Simultaneous use of two meanings o f chance


266

ACKNOWLEDGMENTS

This has been quite a journey. My years at the University of Chicago will remain unforgettable. My path through the evolutionary biology community at Chicago gave me the analytic tools to succeed in empirical research, the background needed to understand and engage in the generation o f theory through mathematical and computational means, and the freedom and feedback needed to advance conceptual research. First of all, I wish to thank Dave Jablonski, my advisor, for his enthusiasm, encouragement and guidance from the very beginning. His mind is truly multivariate, and the plethora o f insights and suggestions that he provided are too numerous to count at this stage. I was fortunate to have my intellectual interests overlapping with his own -- his ideas and research influenced me all along, in particular those pertaining to origination, extinction and developmental flexibility. My choice of study group was also made easier by his previous work on urchins and his astounding knowledge o f the relevant literature. Throughout my stay, a door was constantly open for communication, and I often found that I was treated more like a scholar, a colleague, than as a graduate student. In these and many other ways, my career was forever shaped by Dave. Mike Foote joined the paleobiology and macroevolution group at Chicago in 1994. He magnified my interest in disparity, and helped me immensely by sharing his incredible analytic skills through conversations, teaching and advice. He encouraged my work on echinoids as well as on origination, and was always receptive to my ideas and manuscripts, being very thoughtful in his suggestions. I also benefited from his concern for precision and clarity in the presentation o f empirical results, theoretical assumptions, and conceptual topics. His exciting research program has influenced all of us at Chicago, and has impacted my own future plans o f evolutionary research. xii


Sue Kidwell's knowledge about taphonomy and stratigraphy (as well as urchins) was a good reminder that, despite my biological background, I needed to acquire good geological background in order for the collection o f my data to be properly planned and the results of my research interpreted within the limits o f paleontological sampling. She gave me the necessary support in the endeavour. On a different front, her energy, enthusiasm and humor were in constant supply at the second floor o f Hinds —its contagiousness positively affected all of its denizens. Mike LaBarbera and his laboratorial mind, informed by an impressive intuition for the autoecology of organisms, guided me through many aspects o f data gathering and interpretation ~ from the mechanical and logistic to the geometrical. His concern for scaling and allometry kept me on the right track with a side project, juvenile variation in spatangoids, and our conversations have given me valuable ideas for future research. He also taught me that function and design, whatever their sources, are at least as interesting as structure per se. Jack Sepkoski was an important source of advice all along my studies. His interest in large-scale patterns influenced me very early on, and through him I was able to appreciate the beauty and strength of taxonomic data. His familial and generic databases were invaluable in my research, and I thank him for producing them and allowing access to unpublished data. At an early stage, Jack taught me the analytic skills needed in diversity research and gave me much encouragement and valuable suggestions in my own research on sea urchin diversification and on origination at higher taxonomic levels. His canonical introduction to ordination methods got me off the ground with morphometric research, and taught me the power of multivariate statistics. I also thank Jack for introducing me to the New York Dolls and the Velvet Underground. I am very thankful to Andrew Smith to have agreed to become part o f my advisory committee. His knowledge of virtually everything pertaining to urchins greatly improved


x iv

my research. He gave me morphometric guidance at an early stage, and supported my work throughout. His convictions about the importance o f phylogeny in paleobiological research educated me, encouraged use o f available phylogenetic schemes, and supplied many ideas for future research. His excitement about disparity and diversity also furthered my studies. Andrew's intuition for echinoid variation led to important suggestions, improving the presentation o f results. I am also fortunate that he was able to visit Chicago three times during my studies, which in addition to my month-long stay at the Natural History Museum in London provided many opportunities for advice and greatly enhanced our appreciation for each other's work. Outside my advisory committee, I wish to thank Barry Chemoff, M atthew Leibold, Olivier Rieppel, Leigh Van Valen and Bill Wimsatt for knowledge and insight provided by classes, conversations, and publications. John Alroy, Greg Mikkelson, John Huss and Peter J. Wagner were also great colleagues through many animated discussions. Outside Chicago, I thank Bruno David, Michael McKinney, Dan McShea and Rich M ooi for feedback, advice and encouragement. Dave Raup, now retired, was an important influence early on in my days at Chicago. His knowledge of echinoids and his interest in the structure o f morphological variation guided me through initial stages o f research planning. He also introduced me to some classical topics in particle paleontology as well as to the marginal fringe o f theoretical research through his classes. His receptiveness and advice during my early training will be warmly remembered. If one day I come to teach macroevolution and paleontology, I will no doubt remember the style, quality and content o f the classes taught by my professors — unforgettable are Dave Jablonski’s "Macroevolution", Mike Foote's "Analytical Paleobiology", Sue Kidwell’s "Principles o f Stratigraphy", Mike LaBarbera's "Biomechanics", Dave Raup's "Evolutionary Paleobiology" and Jack Sepkoski's


XV

"Historical Paleontology" and "Quantitative Paleobiology". If my teaching comes to mirror these in subtle ways, it will not be m ere coincidence. For Chapter I, I thank M. Foote, D. Jablonski, M. LaBarbera, M. L. McKinney, D. W. McShea, D. M. Raup, J.J. Sepkoski, Jr., L. M. Van Valen and P. J. W agner for discussion and reviews of different versions of the manuscript, C. R. M arshall, J. Maynard Smith, P. N. Pearson and G. A. Wray for discussion, and J.J. Sepkoski, Jr. for kindly providing access to his unpublished generic database. The research benefited from CNPq grant 201542/91 -9. For Chapters II, III, and IV, I thank M. Foote, D. Jablonski, S. Kidwell, M. LaBarbera, J.J. Sepkoski, and A. Smith, for discussion, comments and advice. I also thank S. Suter for commenting on Chapter II, and P.J. Wagner for discussion and comments pertaining to Chapters II and m . For assistance in various m useum collections, I thank D. Lindberg and K. Wetmore (UCMP, Berkeley), B. David and C. M adon (Universite de Bourgogne), G. Buckley and S. Lidgard (FMNH, Chicago), M. Jensen (Zoologisk Museum, Copenhagen), A. Smith (NHM, London), A. Prieur (Universite de Lyon), N. Ameziane and D. Neraudeau (MNHN, Paris), R. Mooi (CAS, San Francisco) and C. Aheam, D. Erwin, D. Pawson, and J. Thompson (NMNH, W ashington D.C.). Research was supported by CNPq grant 201542/91-9; the Hinds Fund, Obering Fund, a Nierman Award and a William Rainey Harper Fellowship (University o f Chicago); the Lemer Gray Fund for Marine Research (American Museum of Natural History); Sigma Xi; the National Academy of Sciences (through Sigma Xi); the Paleontological Society; and National Science Foundation grant EA R -84-177011 to D. Jablonski. For Chapter V, I thank J. Alroy, M. Foote, J. Huss, D. Jablonski, G. Mikkelson, D. McShea, R. Lupia, K. Saalfeld, L. Van Valen and W. Wimsatt for discussion and/or reviews of different versions o f the manuscript. The research benefited from CNPq grant 201542/91-9.


xvi On a more personal level, I wish to thank all of the members o f my advisory committee and all o f my colleagues in the second floor o f Hinds for making my days in Chicago such a joyous experience. This applies to day to day life, Thai dinners, scientific meetings, and nocturnal events. Outside o f Hinds, I also thank Jeanne Altmann and Carolyn Johnson for constant support and efficiency. Michael Cytrynowicz, Kay Saalfeld, Chandal Nasser and W alter Neves gave me early encouragement. Alexandra, Alroy, Ana Maria, Beatriz, Chandal, Daniel, Daniela, Francisco, Gilson, Greg, Huss, Josane, Jose Olfmpio, Julio, Kay, Karen, Marcia, Marco, Michael, Mono, Pete and Sidney were always good friends no matter how far away they were. From a vividly intellectual, yet charmingly non-academic standpoint, my years in Chicago were forever changed after I met Nancy. Her rare wit, her drive and intellect and her charming manners kept me in touch with reality beyond evolutionary biology. She also provided ceaseless encouragement in my academic pursuit. I thank her immensely. Finally, in the genealogical tradition, I thank my mother, my father and my sisters for everything.


GENERAL INTRODUCTION

Large-scale evolutionary patterns are the result o f a complex combination of constraints and opportunities to macroevolution. Diversification is the underlying motif, disparity and diversity the products. The interplay of repeatability and contingency in the generation and fate o f diversity and disparity over millions o f years can be approached by studying "replicates" in time. One can either study the long-term evolution o f a group through landmark events, like mass extinctions, or the evolution o f subgroups viewed as multiple natural experiments. This thesis represents the use of both approaches in the study of sea urchin diversification and its causes through time. It focuses on temporal patterns of diversity and disparity in spatangoid, holasteroid and disasteroid irregular echinoids (Superorder Atelostomata), and relies primarily on a mixture of morphometric and taxonomic data, complemented by available phylogenetic studies. The main theme of the thesis is evolutionary dynamics and the assessment of possible intrinsic and extrinsic constraints to it. Thus, aside from the more empirical work on echinoids (chapters II-IV), work on the role of development in evolutionary radiations (chapter I) and on chance in evolutionary biology (Chapter V) is also included. It complements the empirical contribution by providing a theoretical framework for discussion of constraints, flexibility, and their meaning. Each section corresponds to a partially independent manuscript which is either in press or otherwise written in a format suitable for publication. A short description o f each paper follows. Chapter I, "The role o f development in evolutionary radiations", discusses in depth what has been and can be learned from explicit consideration of development in

1


evolutionary studies, especially in the context o f biodiversity and one o f its main sources, evolutionary radiations. The polarization o f ecological and developmental explanations for observed patterns of stability and change entails a need to properly tease apart their respective contributions. Although the ecology/development (or extrinsic/intrinsic) dichotomy is not irreconcilable, it does encapsulate the seed o f dissent in many evolutionary debates, and has enormous heuristic value. Nevertheless, the dominance of the Darwinian view in evolutionary biology has usually led to the assumption that extrinsic constraints must a priori be more important and deserving o f primary investigation. Thus, evolution has usually been studied from an ecological and functional perspective. However, a recent burgeoning of developmental studies has strengthened the internalist view. In line with recent advances, and as an experiment in heuristics, I reversed the usual methodological outlook in evolutionary studies and considered intrinsic constraints to evolution from the outset, in terms of both the nature of the data and o f the questions. A case study is presented: Kauffman’s model o f rugged fitness landscapes, which incorporates the effects of both development and ecology. The model is tested against the fossil record to evaluate a particular prediction concerning the Cambro-Ordovician radiation and later quiescence in origination of phyla, classes and orders. The initial impetus and encouragement for my application of the rugged fitness landscapes model to paleobiological data was provided by David Raup, as part o f my term paper "Adaptive landscapes and higher taxa" for his course "Evolutionary Paleobiology", taught in the Fall o f 1992. Chapter I is in press as Eble (1998), a chapter in Biodiversity Dynamics: Turnover o f Populations, Taxa and Communities, a book edited by Michael L. McKinney for Columbia University Press. Chapter II, "Diversification of disasteroids, holasteroids and spatangoids in the Mesozoic", is a somewhat descriptive paper that introduces these echinoid groups and their Mesozoic disparity histories in conjuction with a more detailed study o f taxonomic


3 diversity. The logistic and exponential models o f diversification are tested against secular diversity patterns of genera. Holasteroids and spatangoids diversify taxonomically at virtually the same rate and in exponential fashion during the Mesozoic, but respond very differently to the end-Cretaceous extinction event. Holasteroids were hard hit, and the preferential survival of spatangoids is in retrospect well-grounded in ecological attributes. There is some concordance between morphological and taxonomic diversification through the geological history o f the Atelostomata as a whole (disasteroids, holasteroids and spatangoids), as well as in each of the component orders, but many discordances arise from disproportionate changes in diversity and disparity. Morphological selectivity through major extinction events is discussed for each group, as well as the general character of morphological diversification. Chapter II was originally produced for a symposium on "Major events in the evolution of echinoderms", organized by Gregory A. Wray for the 9th International Echinoderm Conference, San Francisco, August 1996. I thank Greg for the invitation to the symposium. Chapter II is in press as Eble (1997), in Echinoderms San Francisco, a book edited by Rich Mooi and Malcolm Telford for A. A. Balkema Press. Chapter ID, "Contrasting morphological diversification in sister clades: holasteroids and spatangoid echinoids in the Mesozoic" is a very expanded and more interpretive discussion, focused on disparity (as total variance and total range), o f the same time frame covered in chapter I. Important events, like the origin o f spatangoids and holasteroids (despite a drop in taxonomic diversity of ancestral disasteroids), are treated thoroughly, in an attempt to identify correlates. The contrast of spatangoid and holasteroid morphological diversification is interpreted in light o f the historical framework provided by the three-taxon statement that includes disasteroids. Patterns of disparity change and the pace of morphological diversification are discussed from several different angles. I partition total disparity into more and less functional components (differing mainly by the respective


4 exclusion and inclusion o f plastron landmarks, which are clearly much less functionally important than landmarks associated with burrowing and respiratory regions o f the test) and evaluate their congruence. The relationship between disparity and disparity of originations, and their analogy to variability, versatility and related concepts is also discussed. Finally, I comment on disparity and its possible role in macroevolutionary sorting. A version of this chapter is intended for submission to Paleobiology. Chapter IV, "The recovery of disparity in sister clades: spatangoids and holasteroids in the Cenozoic", naturally follows the preceding, but is logically independent since disasteroids are extinct and holasteroids and spatangoids do not have comparable diversity histories as a result of the end-Cretaceous mass extinction. The K-T and the recovery (or lack thereof) in terms of diversity and disparity (as well as total range and disparity associated with oral and aboral landmarks) constitutes the starting point for presentation of results. The case for prolonged environmental deterioration into the Paleocene is evaluated. The long-term consequences o f recovery are discussed with exclusive attention to spatangoids by virtue of a better fossil record (holasteroids are virtually absent from the record after the Paleocene, signalling partly a restriction to the deep-sea). The perennial holasteroid failure to recover diversity and possible causes are also considered, however. The Cenozoic generation and elimination o f disparity in spatangoids is compared with the Mesozoic, and originations are in addition treated in terms of the accumulation o f disparity through time. By including data on Recent holasteroid and spatangoid atypical forms and assessing the inflation in disparity that outliers produce, an assessment is made of how "bizarre" some holasteroids and spatangoids (especially deep-sea ones) really are in total disparity terms relative to various subsamples of the Atelostomata. The origin of holasteroid oddities is also analyzed with the benefit of recent detailed phylogenies, against which the disparity implied by successive


5 levels o f inclusiveness is mapped and interpreted. Chapter IV is also intended for submission to Paleobiology. Finally, Chapter V, "On the dual nature o f chance in evolutionary biology", is a modest contribution toward purification o f the (often muddled) conceptual landscape of evolutionary biology and paleobiology. Evolutionists are increasingly rising to the challenge of scrutinizing and refining central concepts in evolutionary theory. The logical and definitional status of certain concepts can have a profound impact on the practice of science, affecting both data interpretation and the generation o f theory. How natural selection is defined, for example, determines whether all o f evolution occurs by selection or whether logical alternatives, like random accidents, can even be considered. M uch in the same way as in empirical evolutionary biology, conceptual work has for too long focused on eminently Darwinian or externalist concepts like adaptation, natural selection and fitness. This has led to a quite stable, healthy discourse among evolutionists, contrasting with much fluidity in meaning and usage of concepts like constraint (discussed to some extent in Chapter I and IE). Chapter V evaluates chance and the curious duality that has become entrenched in evolutionary biology as a result o f the use of a statistical meaning (usually considered primary) and of a more purely evolutionary meaning (usually vilified), in which the frame o f reference is not just order, but adaptive order expressed as organic design. I present several examples in which use o f both meanings can actually generate insight into empirical phenomena. A clear account o f the status of chance and necessity in evolutionary biology is a prerequisite if the relative importance of different constraints in evolution is ever to be determined. While this chapter was originally intended for the pages of Paleobiology, it may well turn out to appear in Biology and Philosophy. Even so, conceptual problems in the science of evolutionary biology should whenever possible be discussed as much by evolutionists as by philosophers and historians.


6

The field of macroevolution has much to offer for empirical, theoretical and conceptual research. By combining all of these approaches in this dissertation, I hope to have provided a glimpse o f the richness of large-scale patterns o f evolution while at the same time suggesting that the unraveling of processes over long spans o f time, however difficult, is tractable. Understanding the diversity and disparity o f life is the challenge that faces all macroevolutionists.


CHAPTER I

THE ROLE OF DEVELOPMENT IN EVOLUTIONARY RADIATIONS

Introduction

Development and ecology are integral to biodiversity dynamics. Developmental hypotheses, however, have not yet been fully integrated into biodiversity studies at various hierarchical levels and temporal scales. Patterns o f diversity and associated process theories have conventionally been treated in extrinsic, particularly ecological, terms (e.g., Rosenzweig 1995). This may be due, in part, to a reluctance o f many students o f development or ecology to delve into the interplay between allegedly ahistorical principles o f form generation and their historical realization in the process of evolution. The result is a bias in evolutionary explanation. Evolutionary radiations, for example, are a m ajor locus for diversity change, but they have usually been declared "adaptive radiations" without sufficient attention to alternatives to adaptation as a motor of evolutionary diversification. Logically, the role of development is important whenever evolutionary radiations demand consideration o f morphogenetic, rather than strictly ecological, opportunities. In practice, it was not until recently, when the reality of the Cambrian explosion o f metazoan body plans became clear, that developmental explanations have been called on to account for the apparent temporal asymmetry in the turnover o f higher taxa (and associated morphological innovations). 7


8

How are biodiversity dynamics developmentally constrained at various hierarchical levels? Changes in developmental integration and regulation are now an interesting alternative hypothesis for the usual ecological explanations (e.g., Erwin and Valentine 1984; Jablonski and Bottjer 1990a; Hall 1992; Erwin 1993, 1994; Valentine 1995; Raff 1996). An array of macroevolutionary problems associated with changes in diversity through time has not yet been fully explored in explicitly developmental terms. Is diversification at lower levels ever developmentally constrained? W hat is the developmental context of key innovations? Can developmental constraints underlie failure to radiate and macroevolutionary lags (low-diversity delays to diversification —Jablonski and Bottjer 1990b)? Are diversity-dependent phenomena always played out in ecospace? Null hypotheses for diversification have usually been based on the assumption of randomness, but given nonrandom pattern we must logically pose development-based explanations as alternative hypotheses to ecology-based ones. Testing such explanations is a basic challenge faced by neontologists and paleobiologists alike. I address the role o f development in the context of evolutionary radiations, not only because they have traditionally attracted the attention of both paleobiologists and neontologists, but also because problems o f testability become more evident. However, much o f the discussion is potentially applicable to background diversification times as well. I discuss the role of development in terms o f the evolutionary scale or level of morphological distinctness considered, and in terms of different approaches that have been or can be used. Additionally, I present a case study on a recent hypothesis for the early introduction of evolutionary innovations (as in the Cambrian explosion) that purportedly includes contributions from both ecology and development: Kauffman’s model o f rugged fitness landscapes (Kauffman 1989, 1993, 1995). The model is tested in the context of the Cambro-Ordovician radiation and later quiescence in origination of phyla, classes and orders.


9

Evolutionary R adiations, Developm ent, and Paleontology

Evolutionary radiations constitute a very important aspect o f evolution because they generate much of the biosphere’s taxonomic and morphological diversity. We identify evolutionary radiations as interesting subjects for study by comparison with previous and later quiescence (background diversification). Thus, temporal asymmetries, i.e., unusually early or late radiation in the history o f a clade or clades, will attract the most attention. However, exponential branching processes (Stanley 1975, 1979; Sepkoski 1978,1991; Hey 1992; Patzkowsky 1995) will produce apparent radiations even under constant diversification rates, and ultimate decline in standing diversity is the expectation over very long periods o f time (Raup 1985). These null hypotheses should be rejected before looking for legitimate processes underlying asymmetry. Evolutionary radiations can either be adaptive or exaptive (after terminology of Gould and Vrba — 1982). Following Stanley (1990), adaptive radiations should bear a direct relationship with beneficial ecological traits (true key innovations) that enhance occupation of empty ecospace. But radiations can also be a result of an inherent propensity to generate novelties and taxa that is not in itself related to the radiation (see Vrba 1983; Stanley 1990), via incidental cooptation of previous or novel traits. Here we can talk about exaptive radiations, and this is an important realm in which to evaluate the contributions of development. An intrinsic bias to generate new taxa could be based either on a conserved trait (like non-planktotrophic development or small body size —Jablonski 1986a; Brown and Maurer 1989; Brown 1995), that happens to incidentally enhance the chances o f speciation, or on developmental flexibility which, for structural reasons, allow for greater production of variant phenotypes (Wake and Larson 1987; Buss 1988; Lauder and Liem 1989; Liem 1990). W hether a conserved trait or a developmental bias appeared by previous selection is irrelevant in this context. Once they are fixed (with heritable variation


10

absent), their maintenance is not the product of stabilizing selection, but o f developmental dynamics (Maynard Smith et al. 1985). Stasis at deeper levels in the hierarchy is almost synonymous with fixation: development must then operate via mechanisms o f self organization and self-regulation, generation after generation, to produce stable morphologies. Hence, even adaptive radiations are prone to be influenced by the limitations o f development, to the extent that key innovations become entrenched in the developmental system. Thus, whenever stasis holds and can be understood as the partial result o f developmental constraints, development is liable to have a direct impact on processes o f clade sorting. W hether development can be a clade-level property (McKinney 1988a; see Rieppel 1986,1991; L0vtrup 1987; Nelson 1989), justifying its involvement in clade selection sensu stricto, or must rather be viewed as an organismal property conducive to incidental effects via emergent fitnesses (see Vrba and Gould 1986), the macroevolutionary implications o f clade sorting, such as the generation of trends, can be tied in directly with development. Therefore, the study of the role of development in evolutionary radiations would benefit from the identification and description o f radiations that are exaptive and can thus better be understood by reference to morphogenetic, rather than adaptive, opportunities; it would also benefit from the placement o f key innovation hypotheses in an explicit developmental context that helps understand not only origin (the materials available) but also maintenance (fixation and developmental entrenchment). This means that in order to dissect radiations empirically, as much attention should be paid to the dynamics of diversification in developmental spaces as in ecospaces. Developmental spaces (the among-individual analogue of epigenetic landscapes; see Ellers 1993) imply stability and preferred pathways o f form generation and transformation; adaptedness arises from the interplay of structural possibilities and the set o f realizable functions (ecospace). There is


11 no necessary one-to-one mapping o f an adaptedness function onto a stability or complexity function, although some attempts in this direction have been made (Kauffman 1993,1995). The challenge is to investigate developmental spaces in a paleobiological context, where most radiations are ultimately described, while acknowledging that paleontological data are often consistent with both ecological and developmental arguments. As all organisms develop and have an ecology at the same time, answers are not likely to come in simple form. The framing of appropriate questions with diverse kinds of data should be the target of continual evaluation. In what follows I try to provide an outline of research options relevant to the evolution of biodiversity.

Data

Development has many aspects, but it is through its morphological results that most paleobiological and neontological interpretations at various hierarchical levels and time scales are made. Morphological data are usually represented as characters or character combinations, and from them three partially overlapping kinds of data are assembled: cladistic or phylogenetic, taxonomic and phenetic. Character combinations may involve synapomorphies only, as in cladistics, revealing primarily a branching hierarchy o f nested sets to be explained by history. Or they may involve a mixture of synapomorphies, symplesiomorphies and autapomorphies, as in conventional taxonomy and phenetics, revealing relatively discrete groupings in morphospace, although not necessarily hierarchically nested. On genealogical grounds, a cladistic hierarchy clearly assumes precedence. Ultimately, though, cladistics, taxonomy and phenetics would reflect a (non branching) hierarchy of levels of inclusiveness or distinctness (disparity) that is partially independent from the cladistic hierarchy of relationships. The role o f development, if any, depends on the level of morphological distinctness considered. Therefore, all three kinds


12

o f data can provide meaningful information about the way a hierarchy of disparity is organized. The deeper the hierarchical level, the more likely that similarities among taxa involve constraints imposed by development, due to invariance in the face of different environments (Levinton 1988; McKinney and McNamara 1991). Morphological distinctness is mainly a phenetic issue, but we are faced with the paradox that our ability to use phenetic data is inversely related to the level of analysis in the genealogical hierarchy: as dissimilarity increases, homologous features disappear. Although attempts have been made to circumvent the problem by sidestepping the issue o f homology (e.g., the skeleton space o f Thomas and Reif 1993), such approaches may diminish the potential developmental meaning of observed patterns o f morphological distribution. In the absence of phenetic measures, taxonomic and cladistic ranks have been accepted as proxies for morphological distinctness (e.g., Bambach 1985; Bambach and Sepkoski 1992; Erwin 1994; Erwin et al. 1987; Jablonski and Bottjer 1990a,b,c, 1991; Jacobs 1990; Lpvtrup 1988; Patzkowsky 1995; Raup 1983, 1985; Raup and Boyajian 1988; Valentine 1990a; Valentine and Erwin 1987; Valentine et al. 1991). This approach avoids the problem of dwindling homologies. However, taxonomic and cladistic information should not be viewed as surrogates for phenetic data, but as complementary data that can stand as a signal on their own (Foote 1996). Taxonomic and cladistic data can be useful in describing evolutionary radiations and in suggesting (1) whether coherent patterns are present; (2) whether phenetic data might be fruitfully added to the analysis; and (3) whether development or any other explanatory process should be considered. A null hypothesis of exponential diversification, for example, provides a baseline against which deviations can be measured and interpreted. On its own, an exponential pattern may imply no absolute constraint to diversification, either ecological or developmental (but see Miller and Sepkoski 1988). However, different exponential patterns suggest differences in rates


13 of speciation and extinction as well as taxonomic structure, the ratio o f lineages to paraclades (see Patzkowsky 1995), and accordingly possible underlying ecological and morphological correlates. A logistic pattern, in turn, almost automatically suggests the existence o f constraints. Although the expected increase in extinction rate and decrease in origination rate with time presupposes ecological mediation of diversity-dependence, the relative constancy o f extinction rates (Walker and Valentine 1984; Van Valen 1985) makes "density-dependence" of origination rates as reasonably compatible with dwindling o f possibilities in developmental space as with crowding in ecospace (Gilinsky and Bambach 1987). This implies the existence of developmental "carrying capacities" (finite sets of feasible morphologies), as opposed to ecological ones. This is a possibility rarely explored, but maybe appropriate over macroevolutionary time scales. Nevertheless, taxon-free studies of overall morphological variation (see Foote 1991, 1993a,b) are important, by potentially permitting analyses that directly link developmental constraints or allowances to morphological consequences without the confounding effects of the genealogical hierarchy. It is usually a very tricky exercise to use characters at the organismal level to explain proliferation o f taxa (Lauder and Liem 1989; Cracraft 1990; but see Carlson 1992; Baumiller 1993). All in all, in studying the role o f development in evolution it seems best to experiment with as many different kinds o f data as possible, while keeping in mind the methodological limitations inherent in the very production o f empirical knowledge.

The Nature of the Questions

Consistency arguments are an important part o f historical explanations, and consilience o f inductions has been an acceptable strategy o f investigation at least since Darwin (Ruse 1986). Different kinds of data sometimes inductively suggest or prohibit


14 certain questions, but we may also be interested in the inverse, deductive problem: posing a priori hypotheses that stand on their own and await testing in a variety o f ways. There are general and specific approaches, depending on whether developmental information or its proxies are used. General (indirect) approaches use adult phenotypic data, and implicitly assume that constraints on origination and extinction of adult form that are not consistent with ecological arguments must be due to development. Specific approaches attempt to test for the role of development explicitly, by direct inclusion o f developmentally relevant information (growth series, larval characters, etc.) or its proxies in evolutionary studies. Current controversies can serve as a guide for causal investigation o f radiation and diversification in general, and I will draw on them as appropriate. In particular, the Cambrian explosion, however unique, has become a focus for discussions pertaining to ecological and developmental causality behind radiations (Valentine and Erwin 1987; Gould 1991, 1993; McShea 1993; Ridley 1993; Erwin 1994; Valentine 1995; Valentine et al. 1996).

Below, and without claiming exhaustiveness, I review some previous approaches

(see also McShea 1993; Erwin 1994) while suggesting a few additional possibilities that have been generally neglected.

General Approaches

Study o f origination and extinction.

Data on origination are naturally tied in

with developmental arguments. By looking at origination in different ways, insight can be gained into the nature of innovation across levels, taxa and time (Eble 1995a). To a first approximation, the role o f development can be studied by interpreting variation in rates visa-vis ecological hypotheses. Arguments for occupation o f empty ecospace invariably rely on an extrinsic change in the rate of success o f novelties in time or space (see Table 1.1), and imply that the rate of production should vary with environmental changes. Rebounds from mass extinctions provide an ideal situation for testing such covariation; bursts after


15

Table 1.1. M ajor hypotheses to account for evolutionary radiation and accompanying temporal asymmetry, expressed as later quiescence. Processes involved in the generation o f asymmetry are briefly described. Mutation-driven and genome hypotheses (Erwin 1994) are subsumed under the development hypothesis.



Hypothesis

Radiation

Quiescence

Process

Empty Ecospace

Ecology, Development

Ecology

extrinsic change in rate of success

Development

Rugged Fitness Landscape

Development

Development

(h ig h e r flex ibility)

( l o w e r flexibility)

Development + Ecology

Development + Ecology

intrinsic change in rate of production

intrinsic change in rate of success in rugged fitness landscapes

ON

17

mass extinctions have been shown for genera (Sepkoski 1995), families (Sepkoski 1984; Erwin et al. 1987) and, to a lesser extent, orders (Sepkoski 1978; Erwin et al. 1987). Although this general association between disturbance and production o f novelties is amenable to ecological interpretation, it is worth emphasizing that many rebounds would be an expected consequence of diversification even if it was stochastic or if it was primarily driven by occupation of developmental space. Analysis o f origination through cumulative origination functions (Eble 1995a) might be useful here, for its ability to test for long-term regularities in change of rates that might indicate particular constraints on novelty introduction. Rosenzweig (1995, pp. 63-69) discussed cumulative origination functions in time at the species level, looking primarily for interpretability in terms of species-area relationships. Later in this chapter I address cumulative origination functions at higher taxonomic levels. More studies of rate of origination in the context o f functional/ecological data are needed (e.g., trophic categories -- Erwin et al. 1987; degrees of environmental disturbance —Jablonski and Bottjer I990a,b,c,1991; body size -- Brown and Maurer 1989; Brown 1995), as well as in conjunction with developmental information (e.g., frequency of heterochrony - McNamara 1986, 1988; timing o f germ-line determination -- Buss 1988; developmental architecture - Jacobs 1990; developmental control mechanisms - Valentine et al. 1996), while refining the level of analysis by detailed study o f individual clades. Jacobs (1990), for example, analyzed the pattern o f ordinal origination in the Phanerozoic for two different developmental Bauplane (serial and non-serial), and argued that the more dramatic decline in origination of serial forms is not predicted by ecological considerations, being instead associated with the underlying regulatory hierarchy in development. Intrinsic factors are also important in extinction, however, (a) Variation in degree of developmental constraint (and thus degree o f plasticity) across groups might play a role


18 analogous to genetic variation in extinction. In the face of environmental change, highly constrained species may be unable to adapt, and therefore go extinct. Studies of morphological variability in pre-extinction intervals should be promising in testing for extinction selectivity, and in assessing to what extent such selectivity may be blind to fitness in normal environments and a contingent result o f fixation o f developmental pathways -- Raup's (1991) "wanton extinction", (b) Also, immediately after an extinction event, higher developmental or growth rate may favor rebounding capabilities within a group (see Lidgard 1986 on bryozoans). (c) Further, a certain pattern o f development leading to a particular adult morphology might become frozen simply by reason of extinction: Paleozoic echinoids showed much variation in numbers o f plate columns in ambulacra and interambulacra, but the survival o f only two lineages at the end-Permian (with 2 plate columns per ambulacrum/interambulacrum) historically constrained the whole radiation of post-Paleozoic echinoids (all with the same arrangement). Extinction here led to the fixation o f a previously variable morphogenetic sequence, (d) Finally, in coevolutionary avalanches of extinction, small perturbations in internally cohesive ecosystems can cascade into varying degrees of breakdown (Kauffman 1993, 1995; Plotnick and McKinney, 1993). To the extent that perturbations can be purely endogenous, arising for example from speciation events affecting the connectedness o f ecosystems, the nature and timing of avalanches has to depend on the ultimate raw material for origination provided by development. Support for such coevolutionary, biotically driven avalanches is at present based on modelling and on claims for ecosystem cohesiveness (Plotnick and McKinney 1993; Kauffman and Levin 1987; Kauffman 1993; Jablonski and Sepkoski 1996). It must be balanced against the extensively documented extrinsic forcing factors that underlie extinction in the history of life. Nevertheless, susceptibility to disturbance, biotic or abiotic, must partially depend on taxon-specific traits that translate into differential metapopulation dynamics, speciation and extinction


19 (McKinney and Allmon, 1995; Brown 1995). Taxon-specific traits are a function of the developmental context of their generation and maintenance. Origination and extinction propensities can be concurrently interpreted in such context, and might be even predicted when specific developmental information is available. How might development be involved in producing the observed temporal asymmetry in turnover of taxa at higher levels o f the taxonomic hierarchy (phyla, classes, orders)? The Vendian-Cambrian history of diversity at such taxonomic levels is marked by great increases, with origination outweighing extinction (Valentine et al. 1991). Turnover at lower levels (families, genera) follows the same pattern (Sepkoski 1992a), but the ratio o f higher taxa to families and genera is at a Phanerozoic peak during the Cambrian (Valentine et al. 1991). Regardless of the relative importance o f ecological opportunity in triggering the Cambrian explosion, the uniqueness of taxonomic structure in such period is in retrospect an expression o f the way constructional themes (Bauplane) became available at least at the same rate as variations within themes. Such constructional themes, being very conservative, retrospectively imply a high degree of developmental o r generative entrenchment (Wimsatt 1986). But when such themes were being established, characters likely to become more entrenched (with more "entrenchability", or capacity for integration) were probably as readily produced as any other character (provided they were ecologically relevant or at least neutral). The contrast between relatively high diversity o f higher taxa and relatively low diversity o f lower taxa in the early Paleozoic may then be a simple result o f the inherently smaller number of all possible more entrenchable characters (corresponding to higher taxa) in comparison with that o f less entrenchable ones (corresponding to lower taxa). Thus, the burst in originations at the highest taxonomic levels in the early Paleozoic might reflect the exploration of a given range o f possibilities for entrenchment. Concomitantly, other characters would build up through time and move integrated phenotypes further and further away from the opportunity o f incorporating


20

highly entrenchable characters (see also Erwin et al. 1987; Levinton 1988; Stearns 1992). This would reduce the chances o f higher-level origination, but at the same time reduce the chances of extinction (which was high in the Cambrian) by accumulation o f subtaxa within clades. Such decline in turnover ultimately equilibrates into a relative constancy in number o f phyla, classes and orders observed after the Cambrian and Ordovician (Sepkoski 1978; Erwin et al. 1987; Valentine 1990b), although a small burst in origination o f orders after the end-Permian mass extinction may demand special explanation (e.g., availability of developmental or ecological space). In any case, studies of rate of origination and extinction of characters with different degrees of entrenchment should help test such a developmental hypothesis. Note that this treatment of turnover decline in terms of entrenchment is distinct from debates concerning the decline in origination and extinction at the family and lower levels (see Gilinsky 1994), although it does rely on a temporal change in taxonomic structure (see Flessa and Jablonski 1985) and it does attempt to address origination and extinction in coordinated fashion (Gilinsky 1994). The potential role of developmental factors in the decline in turnover at lower levels deserves further examination, however.

Analysis of morphological disparity. The quantitative study o f morphological variation through time is becoming a major research agenda in paleobiology (Foote 1991,1993a,b, 1995; Gould 1991; Jablonski 1995; Wagner 1995). Consistent regularities in phenotypic covariance through time and across taxa, despite variation in environment, are strongly suggestive o f ontogenetic constraints. Crinoids expanded their occupation of morphospace only slightly after their initial radiation, despite 250 m.y. of evolution in the Paleozoic (Foote 1995). Burgess Shale arthropods have at least the same (Briggs et al. 1992; Wills et al. 1994), and arguably more (Foote and Gould 1992), morphological disparity than Recent ones, despite more than 500 m.y. separating the two samples. Natural selection is opportunistic, with diffusion in morphospace being a null


21

expectation (Fisher 1986; Foote, 1993a). However, quiescence over such long spans of time strongly suggests a decrease in the generation of opportunities themselves. Reduction in variance through time, despite mass extinctions clearing out ecospace, also points to morphogenetic entrenchment (Gould et al. 1987; Valentine et al. 1996). W e are not dealing with definitive tests here, because "environment" is a complex parameter (Lewontin 1983; Brandon 1992) and variability in the intensity of natural selection is almost an irrefutable alternative (see McShea 1993). Although a particular pattern o f character covariance may arise by selection or chance, the continual persistence of a pattern of character covariance across taxa can be viewed as a potential result o f developmental constraint. This approach is implicit in the standard definition of developmental constraint (Maynard Smith et al. 1985), and in genetic terms implies pleiotropy and gene regulatory interactions. A role for selection is by no means discarded, but operationally, intrinsic correlations of growth are bound to be expressed as regularities in the covariance matrix (see Lande 1986; Arnold 1992). An unconstrained phenotype will show equal variance for all characters, and no significant covariance (Wagner 1988). More entrenched characters should have higher covariances with a wider range of other characters (McShea 1993). Developmentally linked or serially homologous characters, with no obvious functional coupling, should display significantly higher correlation than non-linked or non-homologous characters (Bader and Hall 1960; Lande 1986). Thus, Gould (1984) argues that covariance sets in Cerion reflect many automatic correlations in growth o f a constrained structure, and stand as evidence for nonadaptation. In the same vein, Hughes (1990) suggests that isometric growth is evidence o f a more canalized development, and conversely that allometry implies less canalization and more mosaicism; degree of isometry and allometry could thus be compared with phylogeny and stratigraphy for tests of hypotheses o f developmental flexibility.


Test of model predictions. Explicit, a priori model predictions can facilitate testing by avoiding the retrospective fallacy, the danger to impose a particular explanation on a historical pattern . Lauder and Liem (1989) cogently discussed a phylogenetic procedure to allow testing of historical factors such as developmental constraints and allowances, which I summarize here. Given a certain morphological novelty (e.g., metamery), one might explicitly predict certain consequences (e.g., a bias toward greater morphological variation as a result of semi-independent units, the segments), and test such consequences via multivariate analysis of taxa in a phylogenetically defined ingroup (metameric clade) and outgroup (non-metameric clade). The prediction can be further tested in other ingroup/outgroup comparisons, assuming that the feature in question is homoplastic. By analogy with the subclade test (McShea 1994), one might also wish to analyze morphological variation in component subdades (e.g., annelids, arthropods) to check whether an increase in morphological variation over time is indeed a necessary correlate o f the structural novelty. Table 1.1 presents more general models of evolutionary radiation. These have been very refined from a theoretical point of view, generating a plethora o f predictions in different empirical circumstances. They provide a platform for testing o f specific hypothesis. Wagner (1995), for example, tested the empty ecospace and the development models against data on the diversification o f early gastropods. He partitioned his data into "internal" characters (more prone to developmental constraints) and "external" characters (reflecting general trophic strategies), and asked whether the associated composite variables changed in magnitude through time in agreement with the observed decrease in disparity. A marked decrease was observed in the magnitude o f composite variables associated with internal characters, which could be interpreted as support for the development model and the associated prediction of increasing developmental constraint.


23 Kauffman’s model o f rugged fitness landscapes predicts early radiation and later quiescence of higher taxa (Kauffman 1989, 1993, 1995). The model relies on long-jump adaptation achieved by early ontogenetic mutants under a tight interplay o f developmental constraints and selection. Although some workers have commented on the in-principle untestability of the model (e.g., Charlesworth 1995; Levinton 1995), certain predictions are testable (Eble 1995a,b). Further below I present a case study in the context of the CambroOrdovician radiation o f phyla, classes and orders.

Specific Approaches

Description and study o f heterochrony. In many ways, discussions about the importance of heterochrony are reminiscent of the debate about the importance of mutation in evolution. Mutation is ubiquitous, varies in effect, and interacts with selection; so does heterochrony (McKinney and McNamara 1991). But heterochrony is clearly nonrandom (in a statistical sense), and this is what makes it so interesting and relevant from the perspective of development and evolution. Not all evolutionary changes are reducible to heterochrony (David 1989; Muller 1990; McKinney and McNamara 1991), but many are (see contributions in McKinney 1988b). Much attention has been paid to inference of heterochrony from phylogeny and characterization of heterochronoclines. From the point o f view of evolutionary radiations, data on heterochrony could be further explored to allow testing of hypotheses of epigenetic entrenchment through time (Valentine and Campbell 1975; Levinton 1988; Wimsatt and Schank 1988): the frequency o f heterochrony, regardless of mode, can be an index o f developmental lability (e.g., McNamara 1988). Such an index could also be used in testing hypotheses about the dynamics o f novelty in space; for example, the post-Paleozoic onshore ordinal origination bias (Jablonski and Bottjer 1990a,b,c, 1991) may or may not be a direct result of involvement of heterochrony


24 in the origin o f orders (for contrasting views, see Jablonski and Bottjer 1990a,b,c; McKinney and M cNamara 1991).

Study o f fluctuating asymmetry. As suggested by Jablonski and Bottjer (1990a), and more recently by Ridley (1993) and M cShea (1993), decrease in levels of fluctuating asymmetry through time could indicate more labile development early on, to the extent that fluctuating asymmetry is a proxy for developmental buffering (see Palmer 1986; Palmer and Strobeck 1986). Studies in a paleontological context include L. Smith (1994), on trilobites. Although sample size was small, he found that high fluctuating asymmetry is not characteristic for every Cambrian trilobite species.

Inferences from mode o f larval development. Mode of larval developm ent in marine invertebrates can affect both origination and extinction: planktotrophic development is conducive to wider geographic ranges and higher resistance to extinction; non-planktotrophic development leads to the reverse, with associated higher probabilities of origination (Jablonski 1986a,b). Although such correlations may break down during mass extinctions (Jablonski 1986b), they provide important opportunities to investigate the direct impact of a developmental feature on rates of diversification, as shown for mollusks (e.g., Jablonski 1986b), cheilostome bryozoans (Taylor 1988) and echinoids (Emlet 1989, 1995).

Mapping of developmental information onto cladograms. There is no necessary correspondence between biological homologies (shared developmental constraints --W agner 1989, 1994; see also Roth 1984, 1994) and historical homologies (synapomorphies —see Rieppel 1992), but the finding o f matches between these different kinds o f homology points to development as a control on innovation or extinction over evolutionary time. One can map developmental characters on an independently derived cladogram and test whether phylogenetically defined clades can also be defined by developmental features (Wray 1992; Erwin 1993, 1994; Meyer et al. 1995), whether


25 particular adult features are consistently associated with developmental ones (Wray and Bely 1994), whether rampant homoplasy has a developmental basis (Wake and Larson 1987; Wake 1991), or whether extinct clades or living fossils (clades experiencing little speciation) might owe a turnover bias to go extinct or originate to developmental apomorphies. Wray (1992) found that most orders and families o f echinoids (but not less inclusive taxa) have synapomorphies in larval morphology, suggesting a possible control of development on origination at least at those levels. Valentine et al. (1996) assembled molecular, developmental and fossil evidence to show that each distinct metazoan phylum or class for which the Hox homeobox cluster has been described is associated with a distinct pattern of gene duplication and loss. This is an interesting result that must be interpreted in terms of different models for the timing of homeobox gene diversification relative to the timing o f splitting of lineages in the context o f the Cambrian explosion (see Valentine et al. 1996). It does indicate that developmental flexibility was a major determinant in the radiation of higher taxa, at least as far as the homeobox evidence is concerned. Rather than mapping the distribution of developmental characters or character-states on independently constructed trees, one may choose to optimize such data on trees constructed using a total-evidence approach. Smith et al. (1995) argued that there is no consistent congruence among molecular, adult morphological and developmental (larval) data in echinoids. This does not rule out the potential importance of development in echinoid evolution; but it does suggest that changes in larval morphology are partly decoupled from changes in adult morphology, and that partly decoupled radiations within the same genealogical nexus must have taken place. Alternatively, one can hypothesize that the evolution o f development must be isomorphic with the evolution of, say, bodyplans (Davidson et al. 1995) and then


26

concentrate on detailed interpretation o f developmental characters in total-evidence trees and their implications for the timing and structure o f radiations (for an example involving the metazoan radiation, see Peterson and Marshall 1995). At the limit, one can concentrate exclusively on developmental characters and the reconstruction o f developmental trees either through parsimony or through direct study of bifurcations in the developmental process (Ho 1990, 1992). In this case, the goal is to understand the diversity o f forms purely in terms of the diversity of developmental transitions (see Goodwin 1990, 1994).

Morphospace comparisons. While the significance o f developm ent to systematics is still hotly debated, among-group developmental transformations and hierarchies are not always recoverable through parsimony (Alberch 1985; Ho 1990, 1992; Marshall et al. 1994). One may expect the orderliness o f ontogeny to underlie the order recovered by phylogeny, but the developmental and the genealogical hierarchies are logically separate -- developmental constraints can affect both homology and homoplasy. The developmental hierarchy is one o f levels of distinctness, and overall morphological relatedness. This should naturally invite the use of phenetic methods in investigation o f the role of development in evolution (Gould 1991, 1993). Comparison of morphospaces constructed with and without developmental information provides a way of consistently studying the impact o f development in constraining or facilitating changes in diversity [in a manner analogous to Hickman’s (1993) design spaces, where the focus is on functional causation]. Congruence in range and/or location in "developmental" and "non-developmental" morphospaces, when properly interpreted to account for possible stochastic effects, is powerful evidence for a controlling influence o f development. To allow comparison, homologous features must be used. This restricts our choices (with qualifications) to three kinds o f contrasts: theoretical/empirical, abnormal/normal and juvenile/adult morphospaces.


27 Theoretical morphospaces (Raup 1966, 1968; see McGhee 1991) rely on models that generate form on the basis of a few fundamental parameters. Models vary in how explicitly they incorporate actual processes o f morphogenesis (e.g., McGhee 1991; Savazzi 1995). Naturally, the more elaborate the model, the more likely it is to accurately reflect the morphogenetic processes involved. But simplicity has its appeal. If a certain range o f variation can be successfully accounted for by a simple model, this means that the actual biological system and pathways o f control need not be more complicated than the model itself (Raup 1968). From a morphogenetic point o f view, we should be interested in constraints of both physical and biological nature: they are equally expressed in development (Gould 1989) and as such have equivalent causal status relative to functionally based extrinsic controls. Seilacher's (1994) candle experiments, where a simple process o f wax accretion at the interface of two media effectively simulates certain shell forms, is an example of such reasoning; the exercise is purely physical, but it underscores the potential simplicity of morphogenetic processes. The classical example of comparison of theoretical and empirical distributions is the mapping o f ammonoid cephalopods in simple coiling space (Raup 1967). That the actual range and locations are mostly consistent with theoretical possibilities o f the model used is telling in itself: whatever the causes of particular phylogenetic transitions among ammonoids, their radiation was fundamentally limited by the developmental constraint o f coiling (Maynard Smith et al. 1985). The density o f occupation o f particular regions is a separate problem (Ward 1980; Hickman 1993). More attention should thus be paid to realistic modelling of domains in morphospace, where different parameter values result in the same phenotype, the stability of phenotypes being proportional to the area of a domain. (Alberch 1989). Discontinuities in morphospace are usually reflected in the existence of teratologies, which brings us to the potential usefulness of comparisons of abnormalities with normal


28

forms. Teratologies, however defined, are highly unfit. Their appearance cannot be the result of natural selection, and must reflect internal constraints and opportunities (Alberch 1989). Under proper sampling, the set o f teratologies and the directions o f transformations should represent a space o f possible forms that is purely developmental, and empirically derived. Trends in the generation of teratologies can then be scrutinized for similarities with trends in actual origination. If abnormal variation mimics normal variation, a case is made for a direct impact o f the logic of development (and its preferred channels of transformation) in the dynamics of origination. Examples of such approaches include the match of patterns of limb variation between intraspecific teratologies and ev o lu tio n ary established genera (Alberch 1989), in conformity with Shubin and Alberch's (1986) morphogenetic model for tetrapod limb diversity; the extensive similarity between Drosophila melanogaster mutants and variation in the Drosophilidae as a whole (DeSalle and Carew 1992); and the remarkable resemblance between pollutional teratologies of regular sea-urchins and a variety of irregular forms typical of other orders (Dafni 1986, 1988). These studies are all neontological and amenable to experimental manipulation. Teratologies in the fossil record are very rare, but abundant taxa should be tractable. In addition, teratologies could be more operationally defined in terms of morphological outliers; if properly characterized, outliers become data to allow comparison o f normal and abnormal forms much in the same way as above, and with the advantage of allowing paleobiological insights. Finally, juvenile/adult morphospace comparisons can be especially relevant in assessing the role o f development in radiations. Similarities in morphological distributions of pooled juveniles and pooled adults can indicate if changes in ontogenetic trajectories are likely to be involved in the shaping of general trends. Alternatively, differences can indicate whether more or less intergroup variation is present earlier in development, thus testing von Baerian recapitulation (which states that more specific characters are developed


29 from more general ones). David (1989) does exactly that in an analysis o f diversity in deep-sea echinoids of the family Pourtalesiidae, showing morphoclines from generalized juveniles to divergent adults. The inclusion o f fossil data in such studies would be most interesting, so that the diversity o f extinct taxa is accounted for and the whole history o f a group is considered.

A Case Study: Testing Higher-Taxon Innovation in Rugged Fitness Landscapes

I here provide an empirical study that relies on two previously outlined general, and admittedly indirect, approaches to the investigation of the role of development in evolutionary radiations: study of origination and test of model predictions. This study tests a particular model o f higher-taxon innovation relevant to the debate over developmental versus ecological flexibility in accounting for evolutionary radiations.

Explanations for Temporal Asymmetry Two main explanations for evolutionary radiations pervade the debate o f the past 10 years, at least as far as the problem of the "Cambrian explosion" and later quiescence is concerned. According to the empty ecospace hypothesis, there is an extrinsic, ecologically driven change in the rate of success o f evolutionary innovations. According to the development/genomic hypothesis, there is an intrinsic change in the rate of production of novelties (Table 1.1). This dichotomy has dominated discussion, but recently a third alternative was proposed: Kauffman's rugged fitness landscapes model (Kauffman 1989, 1993, 1995), where an extrinsic change in the rate of success in an intrinsically constrained fitness landscape occurs during diversification (see Table 1.1). Kauffman's model is


30 accompanied by explicit analytical predictions, which are not necessarily reducible to ecology per se (see Kauffman 1989).

Data The work here relies on compilations from compendia o f global stratigraphic ranges including Sepkoski's familial (1992b) and generic (unpublished) database for marine invertebrates and Benton's familial compilation (1993). The data were culled to minimize taxonomic and sampling biases by removing monotypic, single-stage, incertae sedis, problematic or poorly preserved taxa from the analysis. The observed patterns thus represent a conservative case, with stronger biological signal.

Evaluating the Pattern o f Temporal Asymmetry The temporally asymmetric pattern of origination for phyla, classes, and orders, if not for lower levels, is almost universally agreed upon (Valentine 1969; Erwin et al. 1987). Taken at face value, plots o f total origination through time indicate that a major burst indeed happened by the Cambrian and Ordovician, with later reduction in the intensity of originations (Fig. 1.1). I performed bootstrap analyses to test the statistical significance of the Cambro-Ordovician burst, against the null hypothesis that the burst could have happened even if different magnitudes of origination were equally likely through time. Origination increments were sampled at random with replacement from the whole of the Phanerozoic plus upper Vendian, and a bootstrap distribution (based on a thousand replications) for number o f originations up to the end of the Ordovician constructed and compared with the actual num ber (Fig. 1.2). For phyla (p=0.006), classes (p* j= CL 1>)

Marine Invertebrates n = 13 phyla

3 £ 3 u 0

1000

3000

2000

4000

5000

Cumulative Genera

141

Marine Invertebrates

c _o 3 "5b *w C

10-

o

E

•S CL £> >

n = 13 phyla

s 3

c

3

u

4

4.5

5

5.5

6

6.5

7

7.5

8

8.5

9

In (Cumulative Genera)

Figure 1.4. Plot o f cumulative number o f phylum originations versus cumulative number of genera through time for marine invertebrates. Top: untransformed plot. Bottom: semi-logarithmic plot. Note clear fit to linear prediction.


40 551 e

.2 .2

'Sb •e O


V3

X. 5 3

n = 53 classes

>

3 3 u

10-

0

5000

10000

15000

20000

Cumulative Genera

55 “|

M arine Invertebrates

C

po 3 £ ’5) *u

40

o

V 3I 0) > JS

n = 53 classes

3

C 3 u 4

5

6

7

8

9

10

11

In (Cumulative Genera)

F igure 1.5. Plot o f cumulative number of class originations versus cumulative number o f genera through time for marine invertebrates. Top: untransformed plot. Bottom: semi log plot. Note roughly sigmoidal shape of the underlying function. Next page: residual semi-logarithmic plot of cumulative class origination versus cumulative generic origination, Results of Durbin-Watson test are shown.


41

Classes Durbin-Watson v=0.659, p=0.01

o

C /5

C

o H c '5b

o

°0

5

0

>

o°

H 3 S 33 -o 'yj U

£ 0D O

(S> o

O

o o o

-6 -8

4

5

6

7

8

9

10

11

In (C um ulative G enera)

Figure 1.5. (continued)


42 225200175150125-


loo-

n = 219 orders

7550250

5000

10000 15000 20000 25000 30000 35000 C u m u la tiv e G enera

225 n


200 c3 ’3) u O c

175150125-

'■B O u>

100 -

3

50-

u

25-

s3

75-

4

n = 219 orders

5

6

7

8

9

10

11

In (C u m u la tiv e G enera)

F ig u re 1.6. Plot of cumulative num ber o f ordinal originations versus cumulative number of genera through time for marine invertebrates. Top: untransformed plot. Bottom: semi-log plot. Note roughly quadratic shape of the underlying function. Next page: residual semi-logarithmic plot o f cumulative ordinal origination versus cumulative generic origination. Results of Durbin-Watson test are shown.


43

Orders 100c

_o ac 'So

Durbin-Watson v=0.159, p=0.01

8060 40-

u> 3

20

-

3 r*

c

u 33 t: X u

&

0

3

-

20

-

%

-40-

OS

-60

t

7

* ■■

■■ - r

8

9

10

11

In (C um ulative G enera)

Figure 1.6. (co n tin u ed )

at the phylum level were discarded due to small numbers and if one would still not be confident in assignment of taxa to class level, the cumulative pattern of appearance o f marine invertebrate orders by itself disproves the prediction of the rugged fitness landscape model. Orders have been used successfully in the past as proxies for morphological distinctness (see Jablonski and Bottjer 1990a; Valentine et al. 1991), and there is clearly a break in patterns o f diversity at the ordinal level: lower levels show self-similar patterns that differ from ordinal ones (Jablonski and Bottjer 1990a,b,c; 1991; Bambach and Sepkoski 1992). If anything, the larger sample size for orders (n= 2l9 in the present study, a very conservative estimate that tends to diminish biases arising from the quality o f the fossil record and taxonomic practice) greatly reduces the chance that a certain proportion of spurious orders would significantly affect the results.


44 Discussion It must be noted that testing of the logarithmic prediction (yielding a linear prediction in a semi-log plot) is plagued by the cumulative nature o f the variables. In a different context, one faces the same issues surrounding demonstration of linearity for survivorship curves (Van Valen 1973; Raup 1975; Pearson 1995). Cumulative curves can look deceptively regular, and in borderline cases statistical testing (however difficult) should be attempted. In the preceding analyses I concentrated on the pattern of the residuals for its visual appeal (with apparent lack o f homogeneity for classes and orders) and used the Durbin-Watson test statistic (with expected value o f 2) to rigorously check positive serial correlation of the residuals: they should be independent and with constant variance for the linear regression model to be correct (this again was apparently not the case for classes and orders). The rugged fitness landscapes model incorporates serial correlation of the predicted values due to its reliance on cumulative variables, but implies no expectation of serial correlation of the residuals, which should behave randomly around the regression line (see Kauffman and Levin 1987). Apart from residual analysis, the strong curvilinearity of the semi-log pattern for orders, and to a lesser extent for classes, is sufficient to "test", albeit qualitatively, the fit o f the model prediction. More rigorously, one can show that a parabola fits the ordinal pattern better than a straight line, the same being true of a sigmoidal curve for the class pattern. The meaning o f such cumulative origination functions is not of concern here, but one might suspect that an unconsidered variable was involved, with historical effects being a likely factor. I do not claim taxonomic data to be perfect. Measures of morphological disparity could, in principle, be used in the same way, with character combinations being traced back to their first appearances in the fossil record, or inferred with the help of morphological and molecular phylogenies. In view of documented discordances between taxonomic and morphological diversity (Foote 1993a) there is nothing that should preclude morphological


45 data, more tractable at lower levels, from being subjected to the same kind of analyses and perhaps even support the rugged fitness landscapes model. A difficulty with the rugged fitness landscapes model is that one does not know a priori at what level the long-jump adaptation end o f the spectrum ceases to be meaningful for analyses. If the same cumulative plots are constructed for families and genera, deviation from the model prediction again occurs. Rosenzweig (1995, P.68) analysis o f a local assemblage of Ordovician invertebrate species over 5 myr suggests again deviation from linearity in a semi-log plot (although he was testing species-area models). It remains to be seen whether species-level cumulative curves could be extended over longer time scales, thus allowing greater comparability with the present analyses. But Rosenzweig does provide a context in which Kauffman’s model might tie in with more refined ecological arguments. Rosenzweig's call for renewed attention to cumulative species-time curves should be expanded to all scales. Another way o f testing Kauffman's hypothesis w ould be to concentrate on more circumscribed clades or contrasting environmental contexts. Although additional analyses are beyond the purposes of this paper, it is worth mentioning that for orders o f mollusks, echinoderms, mammals and insects (with cumulative number of genera being used as a proxy for cumulative number of tries for mollusks and echinoderms, and cumulative number o f families as such a proxy for mammals and insects), the rugged fitness landscapes model also does not seem to hold, at least as far as the assumption of a single landscape is concerned (Eble 1995a). Deformation of landscapes with environmental change is a relevant issue here. It has been said that the ad hoc invocation of landscape deformation through time would render Kauffman's arguments essentially untestable (Levinton 1995). However, Kauffman (1993) implicitly assumes that deformation, even in the face of pronounced environmental perturbation, would be minor and not affect the general structure of the


46 landscape at the higher taxonomic levels considered here. That the generation o f morphological innovations in long-jump adaptation would be achieved by early ontogeny mutants lends support to claims for a relative stability o f landscapes at higher levels, since the developmentally constrained set of possible changes at deeper levels of the hierarchy would be quite independent o f environmental changes. Here is the irony o f the rugged fitness landscapes model: by relying on ahistorical principles to explain history, historical effects m ust a priori be assumed to be largely ineffective; if history could change the nature of those ahistorical principles, then there would be nothing left to test. Landscapes in the present context would only change if fundamentally different genetic and morphogenetic controls were repeatedly appearing through time. This may well be the case in view o f the present results, suggesting that history would need to be taken into account from the outset. The whole approach to the rugged fitness landscapes model assumes that adaptation during radiation will occur by successively shorter long-jumps, because of the correlational structure o f the landscape —phyla are founded, followed by classes and then by orders (Simpson 1953). The argument breaks down if there is no obvious top-down pattern. Sepkoski’s (1992a) newest analysis of the Vendian-Cambrian diversification shows that the top-down scenario may not apply anymore to the metazoan radiation, because the patterns o f diversification for genera, families, orders and classes are all convergent, showing approximately the same pattern o f diversification and the same timing o f diversity increases. Although in other instances (e.g., the Ordovician radiations) a case could be made for a top down pulse of filling (see Erwin et al. 1987), the generality of the pattern is greatly reduced by its absence early on. In the face of coeval diversification, the basic problem is reduced to explaining why that pattern o f diversification (and accompanying disparity) has not recurred for higher taxa. While a hypothesis of differential developmental flexibility through time remains unscathed, ecology-dependent alternatives, including not only the rugged fitness landscapes model but also, for that matter, the empty


47 ecospace model, may need to be framed in different terms —without reliance on lower rates of diversification at lower taxonomic levels —if earlier higher disparity is a problem to be explained.

An Alternative Procedure to Test the Model of Rugged Fitness Landscapes Although the discrepancy in cumulative origination patterns across higher levels, and in particular the visibly curvilinear (as opposed to log-linear) pattern for orders, is sufficient to challenge the generality o f the rugged fitness landscapes model, it is worth investigating alternative ways of testing the model that treat data non-cumulatively. This is desirable to give some statistical confidence beyond the indirect approach based on the Durbin-Watson test, the results of which indicated not only non-homogeneity o f residuals, the issue of interest, but also some (inevitable) degree of nonindependence o f points arising from the cumulative nature of the data. The logarithmic prediction o f the rugged fitness landscapes model expresses an expectation regarding the waiting-time for successive improvements. This is equivalent to an exponential decrease in the number o f improvement opportunities (see Kauffman 1989, 1993), which immediately suggests direct testing of an exponential decline in the intensity of origination against the empirical data. This effectively allows treatment o f the data in non-cumulative form. Analytically, the expectation becomes S = e('k*G), with S now as the number of originations, G as before the cumulative number o f genera, and k representing a decay constant. In a semi-log plot, and allowing "measurement" error, as discussed before, one has InS = a - kG as the prediction to be tested. Figs. 1.7, 1.8 and 1.9 test this prediction against the observed pattern of phylum, class and ordinal origination relative to the accumulation of genera through time. Once again, of interest is both the degree o f linearity


48 and the assumptions o f the linear model, of which the most critical is the homogeneity of the variance o f the residuals around the regression line (Manly 1992). For phyla (Fig. 1.7), beyond being consistent with the normality assumption (about 68% of standardized residuals within the range from -1 to +1 and about 95% of them within the range from -2 to +2) and not being serially correlated (Durbin-Watson test), the residuals are also homoscedastic around the regression line. Homoscedasticity is present if there is no significant rank correlation (here Spearman's p was used) between the absolute values of the residuals and the independent variable (Rock 1988). Otherwise, the residuals are heteroscedastic and the regression is invalid. As shown in Fig. 1.7, there is no significant rank correlation between the residuals of the logarithm o f phylum originations and cumulative genera through time. As discussed earlier, it is difficult to assess the robustness of the phylum pattern due to the small sample size involved, but taken at face value the observations are consistent with the model prediction. More interesting is the class pattern (Fig. 1.8), which in non-cumulative form support an exponential decrease in the intensity of origination. The residuals are normally distributed and show no significant serial correlation, and Spearman's p is again not significant. The inclusion o f a removed outlier does not change the results. In contrast with the apparent nonlinearity of the data in cumulative form (see Fig. 1.6), a fit to the rugged fitness landscapes model is granted here, and is statistically more robust. The ordinal pattern, however, once again does not fit the model prediction (Fig. 1.9). Although the residuals are normally distributed and serially uncorrelated, there is significant rank correlation between the residuals of the regression and cumulative genera, indicating violation of the assumption of homogeneity o f the variance o f the residuals and thus invalidation of the model.


49

Marine Invertebrate Phyla R 2 = .189 n = I3

1000

2000

3000

4000

5000

Cumulative Genera through Time

Residuals are homoscedastic and regression is valid

•J S

C

.o

_e

'Sb

D u r b in -W a ts o n v = 1 .8 2 4 ; n o t s ig n if ic a n t ( 5 % le v e l)

.6

.4

S p e a rm a n 's p = - 0 .4 8 6 p < 0 .2 7 7

.2

0 -C

a.

-.2 -.4

a3 T3 *V 3

0c

-.6 -.8

-1 1000

' - i —■ —r '—i ■ i 2000 3000

— i—

■— i—

4000

'—

i

5000


Fig. 1.7. Alternative test of the rugged fitness landscapes model -- phyla. Semilogarithmic axis is the y-axis. Plotted is the number of phylum originations versus the cumulative number o f genera through time for marine invertebrates. Top: semi-logarithmic plot. Bottom: residual semi-logarithmic plot. The Durbin-Watson statistic and the rank correlation coefficient between the absolute values o f the residuals and the cumulative number o f genera are shown.


50

Marine Invertebrate Classes R 2 = .3 2 6 n=53

2.5C 3 C SO 0 •3 c

.5-

1 O 'W ' c

oo

oo O u t l i e r re m o v e d

0

5000

10000

15000

20000

25000


Residuals are homoscedastic and regression is valid

1.251c _o 3c .75 .5 's o

8 Ji

D u r b i n - W a i s o n v = 2 .0 3 2 ; n o t s i g n i f ic a n t (59c level) S p e a r m a n ’s p = - 0 .2 9 8 p < 0 .2 1 9

.25 i 0

J u c -.25■33 -.5•a -.754> a: -1 -1.25

cP — i— i— i— i— i— i— i— i— i— i— i— i— i— i— i— i— i— i— i— r — i— i— i

0

2000

4000

6000

8000

10000

12000


Fig. 1.8. Alternative test o f the rugged fitness landscapes model — classes. Semilogarithmic axis is the y-axis. Plotted is the number of class originations versus the cumulative number o f genera through time for marine invertebrates. Top: semi-logarithmic plot. Bottom: residual semi-logarithmic plot. The Durbin-Watson statistic and the rank correlation coefficient between the absolute values o f the residuals and the cumulative number of genera are shown.


51

Marine Invertebrate Orders 3.5-

R 2 = .4 5 3 n=219

Oq

C

o

C5

C . 2? *c

O 13 .E •5

O CD OO

OO OO

0 0 (0)

c

OODO O

0

5000

OO

10000 15000 20000 25000 30000 35000 Cumulative Genera through Time

Residuals are heteroscedastic and regression is invalid v:

D u r b in - W a ts o n v = 1 .5 7 3 ; n o t s ig n if ic a n t (59c le v e l)

2-i

C

_o

1.5rs c ‘ob 1 o

oo

S p e a r m a n 's p = - 0 .3 9 1

o

p < 0 .0 0 5

o

.5 ■H O S3 =3

0 -.5-

T3 ' 7. ] o

Collyrites

,,

'1 " » 1 » ~1 ~ i 1 i -2

-1.5

-1

-.5

0

.5

■ 1

1.5

2

2.5

PCI

Aalenian-Bajociar (Ji)

P ygom ^us

Ol U a.

\ l

-1.5

»

T

-1

Orbigny ma

r --------i - ■

-.5

i

0

i

.5

«

i

n - r

1.5

i

i

2

i

i

2.5

PC I


77

Albian (K3)

(b) o Disasteroida * Holasteroida A Spatangoida

[^] [*]

X

E*1 -4^4-

n u o

[O]

[O]

/

-i

-2

L abrotaxis

-3

"I

-2

*

-1.5

I

-1

I

I-----

-.5

0

T" i .5 PC I

i I

• 1 1.5

" I 2

1 | 2.5

Barremian-Aptian (K2) Hemiaster

A [*}

[*

Neocomian (K l) 3

Toxaster africanus

[O] [O]

.1

Hola: ter cordatus ■3 1.5

.5

0

.5

1.5

2.5

PC I


78

3

Paleocene leftovers (Pal)

(c)

2

* Holasteroida A Spatangoida

0 [41*1 •2 3 1.5

0

.5

.5

9

1.5

2.5

PC i

Campanian-Maestrichtian (K5)

1

Cenomanian-Santonian (K4) M icraster------

"A [A] A

[A ] [AI

u o

X Infulaster A

. ' IL , i * A [* ] [*] * [*]

A

/

/

X — Ha :enowia Ovulaster -I— 1— i 1 ' i— ■ i ■■ r ' i -2 -1.5 -1 -.5 0 .5 PC I

i

•

i 1.5

•

i

*" ■i

2

2.5


79 Holasteroids and spatangoids expand in morphospace in very similar ways, increasing both range and pairwise distance through the Cretaceous, and generating distinctive morphologies in the process (e.g., the spatangoid Hemiaster and the holasteroid Labrotaxis). Interestingly, primitive disasteroid locations seem to be less densely occupied. By the Cenomanian-Santonian (Fig. 2.3 -- c), "typical" spatangoid forms like Micraster appear at the edges of morphospace, and so do holasteroids like Infulaster and the bizarre bottle-shaped Hagenowia. Overall, the amount o f morphospace occupied is comparable but certain regions are preferentially occupied (holasteroids in the third clockwise quadrant). Morphospace occupation and disparity eventually achieve a m axim um , with very similar total spread in morphology, although the effect o f outliers inflates holasteroid disparity (see Fig. 2.4). The picture for the Paleocene survivors of the K-T extinction is quite different, though: spatangoids maintain essentially the same spread in morphospace, whereas holasteroids are eliminated preferentially at the edges o f their Late Cretaceous distribution, shrinking in morphospace occupation. These results suggest nonselective (or less selective) thinning in morphospace for spatangoids, and morphological selectivity o f extinction for holasteroid morphologies. This contrast magnifies the conclusions reached on the basis of taxonomic diversity, and invites consideration o f clade-specific traits affecting susceptibility to extinction, as discussed above. O f course, traits within holasteroids explaining the survivorship o f forms like Echinocorys, H olaster and Offaster are worth investigating. Abundance per se does not seem to be a factor (Smith, pers. comm.; but see Ernst 1972). An ordination space, however intuitive, offers only a hint at the actual changes in total disparity. Changes in total disparity, measured as the total variance in the original variables at each interval, are shown in Fig. 2.4. Changes in generic diversity are also shown for comparison, since the contrast between disparity and diversity can be very


80

Fig. 2.4. Contrasting patterns of morphological disparity (thick lines) and taxonomic diversity (thin lines) for the Atelostomata as a whole (2.4 - a), disasteroids (2.4 - b), holasteroids (2.4 —c), and spatangoids (2.4 —d). Taxonomic (generic) diversity on the right o f each graph; morphological disparity (measured as the sum o f the variances of original variates) on the left. Disparity error bars were calculated by boostrapping (200 replicates). Diversity error bars as in Fig.2.2. Stratigraphic intervals are the same as in Fig.2.3 and are coded as follows: J l - Aalenian-Bajocian (Middle Jurassic); J2 - BathonianCallovian (Middle Jurassic); J3 - Late Jurassic; K1 - Neocomian; K2 - Barremian-Aptian; K3 - Albian; K4 - Cenomanian-Santonian; K5 - Campanian-Maastrichtian ; Pal. Paleocene survivors of the end-Cretaceous extinction. Values are plotted at the beginning o f each interval.


81

A r8 0

50 ] Atelostomata T

-7 0

-40 -3 0

-20 -10

J1

J2 J3

K1

K2

K3

K4

K5

Pal

G e o lo g ic T im e


Generic D iv ersity

45-

82

B r 18

50• 14 4030-

20

-

J2

J3

K1

K2

K3

G eo lo g ic T im e


Generic D iversity

Disparity (Total V arian ce)

601 Disasteroida

83

-4 0

Holasteroida

40;

-3 5

35-

-3 0

30;

-2 5

25-

20

-20

-

-15 15-

-10

10-

- 5

5 o

■ i r | t i rj t

K1

K2

i it

r l T"r n

K3

i "| r i i | r - 1 i i r r t i T"T'T

K4

K5

- 0

Pal

G e o lo g ic T im e


Generic Diversity

Disparity (Total V arian ce)

45'

84

D

Generic Diversity

Spatangoida

K1

K2

K3

K4

K5

Pal

G eo lo g ic T im e


85 informative about underlying modes of diversification and extinction (see Foote 1993a). For the Atelostomata as a whole (Fig. 2.4 -- a), there is a general concordance between morphological disparity and taxonomic diversity through time, and both have high levels in the Campanian-Maastrichtian (Late Cretaceous - K5). Several discordances are apparent, however. First, disasteroid disparity initially increased at a comparatively higher rate than number o f taxa. Second, the remarkable burst in disparity in the Neocomian (K1) is not matched by taxonomic diversity. Spatangoids and holasteroids appear for the first time then, which suggests that their appearance is not just an automatic consequence o f the accumulation of disasteroid taxa, but instead represents a period o f truly enhanced morphological experimentation. Third, the Late Cretaceous increase in disparity does not occur in proportion with the massive increase in diversity. Finally, the marked drop in taxonomic diversity into the Paleocene is not matched by a proportionate drop in morphological disparity. The drops do point, however, to some selective taxonomic attrition with respect to morphology (Foote 1993a), clearly the contribution o f holasteroids. At a lower taxonomic level, disasteroids, holasteroids and spatangoids are illustrated (Figs. 2.4 -- b, c and d, respectively). The Jurassic disasteroid pattern of diversity and disparity is obviously identical to that shown for the Atelostomata, albeit at a different scale (the disasteroid initial increase in disparity occurs here at a comparatively lower rate than the increase in number of taxa). What is remarkable is the absolute nonselectivity of their decline (Fig. 2.4 - b). The major loss of taxa into the Cretaceous is simply not matched by disparity, which remains the same partly as a result of Neocomian originations and partly by reason of thinning in morphospace, the latter prevailing as the decline proceeds (see Fig. 2.3). Given the small size and paraphyly o f the group, one must also add the very origin of holasteroids and spatangoids from disasteroids as another reason contributing to thinning o f disasteroid morphospace.


86

As for holasteroids (Fig. 2.4 -- c) and spatangoids (Fig. 2.4 —d), notice that their levels of disparity do not add up to those o f the Atelostomata, as here the interest lies more in an absolute measure of internal disparity (within-group dispersion) than in an additive partitioning of overall disparity into partial disparities (see Foote 1993b). Holasteroids show a rough concordance in morphological disparity and taxonomic diversity (2.4 —b), but an early high in morphological distinction is not mirrored by a high number of taxa. A burst in diversity and disparity is very proportionate between the Albian and the Cenomanian-Santonian, but is followed by another burst in diversity not matched by disparity, suggesting a change in morphological diversification mode from diffusive to constrained evolution (see Foote 1993a). The substantial drop in both disparity and diversity with the K-T is consistent with negative sorting o f holasteroid morphologies. Throughout the Cretaceous history of spatangoids, there is a rather good concordance between the sign of the increments in both morphological and taxonomic evolutionary histories, although not in proportion: the tracking o f diversity by disparity is very dampened. In the Late Cretaceous, disparity fails to increase substantially despite two near doublings of standing diversity. Much in the same way as in holasteroids, diffusion followed by constraint (deceleration of diffusion) is the hallmark o f spatangoid morphological diversification in the Cretaceous. Contrary to holasteroids, however, disparity remains the same through the K-T despite considerable extinction of genera. Spatangoid extinction is clearly nonselective, supporting the discussion on taxonomic patterns.


87

C onclusions

The main outlines o f diversification presented here constitute a necessary step in approaching causation in the evolution o f disasteroids, holasteroids and spatangoids, complementing previous work. More detailed work of ecological, developmental and phylogenetic nature (some o f it in progress) should help in identifying correlates o f some of the patterns here documented. However, some interesting generalizations emerge: a welldefined exponential phase in the expansion of holasteroids and spatangoids; possible absence of interclade interaction, at least with likely candidates, during most o f the Mesozoic history o f holasteroids and spatangoids; the origin o f holasteroids and spatangoids from disasteroids representing a long jum p in terms o f disparity o f the Atelostomata; only rough concordance and considerable discordance between morphological and taxonomic diversification in all groups; diffusion with later deceleration as the main feature of both holasteroid and spatangoid changes in disparity during the Cretaceous; and preferential survival o f spatangoids over holasteroids with the endCretaceous extinction, both in taxonomic and morphological terms. Morphospace occupation and morphological disparity were roughly comparable during the Cretaceous history o f holasteroids and spatangoids (Fig. 2.5), especially in terms of the common underlying dynamics of diffusion and later deceleration in the pace of morphological diversification. This suggests that similar controls (intrinsic and/or extrinsic) may have persisted despite the passage o f time. Relative separability o f habitat would point to the involvement of intrinsic factors. More studies o f disparity in other echinoid groups can put the patterns here documented in a broader perspective. Packing in morphospace was clearly occurring at the end o f the Cretaceous for both holasteroids and spatangoids. At face value holasteroids displayed somewhat higher standing disparity in the Late Cretaceous, but the error bars on their M esozoic disparity


88

Fig. 2.5. Holasteroid and spatangoid disparity contrasted. Same disparity curves as in Fig. 2.4, but with error bars removed for clarity. Error bars overlap throughout, but there is marginal statistical significance for the differences in disparity in K4 and K5. Intervals and abbreviations as in Fig. 2.4.


89

* Holasteroida Spatangoida

a

40 -i

Disparity (Total Variance)

35 30 25 -

20

-

15 10

-

K1

K2

K3

K4

K5

Pal

Geologic Time


90

histories (not shown in Fig. 2.5) overlap throughout, and there is no statistical significance for the differences. However, marginal statistical significance is present in the Cenomanian-Santonian (p < 0.1, z-test based on bootstrapped standard errors -- see discussion of assumptions of the test in Ch. EH; Efron and Tibshirani 1993) and Campanian-Maastrichtian (p < 0.14), clearly as a result o f a more profuse generation o f peripheral forms (Fig. 2.3). To some extent, holasteroids were showing a tendency to break away in morphospace, in an apparent trend that was drastically interrupted by the end-Cretaceous extinction. Holasteroids had the potential to produce some unusual morphologies early, but such potential was more fully explored only much later, in the deep-sea of the near Recent. Conversely, much morphological turnover within the Spatangoida had yet to occur as the clade rebounded from the K-T and continued its exponential diversification. Despite the regularities in the Mesozoic diversification of heart urchins, their full history could not avoid being completely changed by a major extinction crisis.


C H A PT E R IH

CONTRASTING MORPHOLOGICAL DIVERSIFICATION IN SISTER CLADES: HOLASTEROID AND SPATANGOID ECHINOIDS IN THE MESOZOIC

Introduction

How regular is evolution in related clades? What explains regularities, when present? These are long-standing issues in evolutionary paleobiology, and a variety o f approaches have been developed to tackle them, most notably theoretical morphology (e.g., Raup 1966, 1968; McGhee 1991; Niklas 1994), constructional morphology and morphodynamics (Seilacher 1970, 1991; Reif et al. 1985; Gould 1989), and the quantitative study o f biodiversity through geological time (e.g., Valentine 1969; Raup 1972; Sepkoski 1978, 1979, 1981, 1984, 1993; Sepkoski et al. 1981; Bambach 1985; Erwin et al. 1987; Gilinsky and Bambach 1987; G ould et al. 1987; Jablonski and Bottjer 1991; Valentine et al. 1991; Jablonski 1993). More recently, the quantitative study of morphological disparity (the spread o r spacing of forms in morphological space) in the fossil record has contributed to a renewed awareness of large-scale heterogeneities to diversification and has put the scrutiny of temporal asymmetries under new light. After several important studies of Paleozoic groups (e.g., Foote 1990a, 1992a, 1995; Wagner 1995, 1997), attention now begins to shift to other timeframes, like the Mesozoic (Roy 1994; Dommergues et al. 1996; Foote 1996a; Eble in press) and Cenozoic (Van 91


92

Valkenburgh 1991; Jem vall et al. 1996; Roy 1996). Regardless o f timeframe, however, there is still a need for a more comprehensive theoretical framework for disparity (e.g., Foote 1993a, 1997), with enrichment of the theme o f temporal asymmetry and constraint with a host of other aspects of disparity, including its relationship with origination and extinction, its partitioning into character complexes (Foote 1995; W agner 1995), its connection with sorting and selection processes, and its expression in the biological hierarchy (Foote 1993b). Such aspects are explored in this paper, as an empirical contribution to the general theoretical goal. Is phylogeny needed to attain the general goal? Phylogenetic information is clearly important in paleobiologic studies (Smith 1994), although the level o f detail required in the context of the analysis o f disparity has been a matter o f heated debate, which may well turn out to recede if phylogenetic and phenetic approaches are viewed as complementary rather than orthogonal (Foote 1996b). Some knowledge o f phylogeny is implicit in the choice o f units of study and in the framing o f questions. Phylogenetic data on every measured specimen, however, are neither necessary nor sufficient for the study o f disparity through time. They can be very informative, but are best viewed as a complement, adding important historical information. This information can be of at least three kinds: - metrics of disparity that take phylogeny into account, like those based on sisterspecies contrasts (Wagner 1995) or branch length (Smith 1994; Wills et al. 1994); - assessment of total change as opposed to net change (Foote 1995, 1996b); - provision of a framework for mapping and interpretation o f temporal patterns of disparity, and investigation of underlying evolutionary processes. It is this third approach, not yet extensively applied, that I wish to explore, by studying changes in disparity in a major component of post-Paleozoic benthic marine assemblages, the echinoid superordinal clade Atelostomata (Smith 1981, 1984). The Atelostomata comprises all heart urchins sensu lato.


93 Heart urchins are an important constituent of the post-Paleozoic radiation of echinoids. After almost complete annihilation by the end-Permian extinction, with only two lineages surviving into the Triassic (Smith and Hollingsworth 1990), echinoids radiated spectacularly in the aftermath. Many orders were produced (substantially contributing to the onshore ordinal origination bias —see Jablonski and Bottjer 1990, 1991), suggesting that much morphological disparity was being generated, despite the historical constraints imposed by the surviving Paleozoic bodyplan. The generality of patterns and causes of evolutionary radiations has not been extensively studied beyond the early Paleozoic (but see Erwin et al. 1987; Foote 1996a). As a sample o f the echinoid radiation, heart urchins provide an opportunity to assess the amount and the kinds of evolutionary change that might be characteristic of post-Paleozoic echinoids. In addition, echinoid turnover rates are sufficiently different from other major post-Paleozoic groups like bivalves and gastropods (Stanley 1979; Sepkoski 1981) to justify an echinoid-based study as an independent, complementary account on the nature o f the post-Paleozoic diversification of marine invertebrates. Finally, although preservability in echinoids varies spatio-temporally and with taxonomic membership (Smith 1984; Kidwell and Baumiller 1990; Greenstein 1993), it is generally regarded as quite high (Raup 1979), especially in irregular echinoids, thus allowing fairly detailed analyses. Descriptively, the aim is to compare morphological diversification and changes in disparity in sister clades, taken to be alternate natural experiments in evolution. As a compromise between available phylogenetic information and temporal coverage, I will concentrate attention on a well-corroborated (Smith 1981, pers.comm.) three-taxon statement for the three orders comprising the heart urchin clade: "Disasteroida", Holasteroida, and Spatangoida (Fig. 3.1). As irregular echinoids, heart urchins display bilateral symmetry superimposed over the primitive pentamerous radial symmetry. In addition, heart urchins have evolved a number of elaborations of the test (Fig. 3.2), albeit


94

Fig. 3.1. Stratigraphic and phylogenetic framework for the irregular echinoid orders Holasteroida, Spatangoida and Disasteroida. Disasteroids appear in the Middle Jurassic (possibly in the Early Jurassic -- Mintz 1968) and range to the mid-Cretaceous. Spatangoids and holasteroids appear coevally in the Early Cretaceous (Berriasian), and range to the Recent. Phylogenetically, the three orders nest into the three-taxon statement shown in the left, which is robust relative to other echinoids at least at the ordinal level (Smith 1981, 1984). Together they form the superorder Atelostomata, with synapomorphies that include elongate/disjunct apical system, plastron differentiation (minor in most disasteroids), and peristome in anterior position (Smith 1981). Disasteroids are the paraphyletic stem-group for the monophyla Holasteroida and Spatangoida, which share a non-disjunct apical system, fascioles, differentiated ambulacrum in, well-developed phyllodes, branched respiratory tube-feet, differentiated spines and tubercles, and a differentiated plastron (ibid.). Holasteroids uniformly have an elongate apical system and a primitively meridostemous plastron; spatangoids have a compact apical system and an amphistemous plastron.


Spatangoida

Holasteroida

Recent Mi OI

Eo Pal

65

u-K

Disasteroida

97 I-K

u-J

145

m-J I I I

I-J

208 Disasteroida Holasteroida

Spatangoida


96

Fig. 3.2. A representative disasteroid (top), holasteroid (center) and spatangoid (bottom). Top: Disaster granulosus, aboral fa.b). oral (c) and side (dt view. Center: Holaster nodulosus. aboral (a), oral (b), and side (c) view. Bottom: Micraster coranguinum. aboral (a), oral (b) and side (c) view. All illustrations o f natural size, from Mortensen (1950).


Disaster granulosus

Holaster nodulosus

a.

b.

c.

Micraster coranguinum

a.

H.

c.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission

98 not all synapomorphic: they include elongation and disjunction o f the apical system, enlargement and complexification o f plates, and formation o f a plastron in the oral side; peristome and periproct migration; differentiation of ambulacrum HI into a frontal furrow and of the other ambulacra into petals; and functional differentiation o f spines and tubercles, facilitating a spectacular expansion into a burrowing, infaunal mode o f life (Kier 1974, Smith 1984). In many ways (but not all, given the uniqueness o f sand-dollars; Seilacher 1979, 1990), heart urchins are representative of the observed limits to experimentation in the Echinoidea: the diverse morphologies cover a wide portion o f the perceived range of variation in the class (e.g., globose forms, flattened forms), not to mention the unique cordiform, wedge-shaped and bottle-shaped architectures that appear in spatangoids and holasteroids. Spatangoids and holasteroids are sister clades that appeared coevally in the Early Cretaceous (Berriasian —Jablonski and Bottjer 1990a), 145-140 m.y. ago. Both groups became highly effective at living at or below the sediment-water interface (Ernst 1972), and several burrowing and feeding mechanisms have evolved (Smith 1984; Kanazawa 1992). At the same time, the two groups show considerable differences in taxonomic diversity histories. Spatangoids became taxonomically much more diverse than holasteroids, with higher generic and species diversity. Although they diversified exponentially at roughly the same rate during the Cretaceous (Eble 1994; in press), differential response to the endCretaceous mass extinction was critical in the groups' history: whereas spatangoids were less affected and rebounded to attain their highest diversities during the Tertiary, holasteroids were hard-hit by the event and quickly declined, remaining mostly restricted to the deep-sea afterwards. This pattern, robust in the face o f taxonomic and stratigraphic updates (Stokes 1979; David 1988; Eble in press; Jeffery and Smith in press), suggests extinction selectivity. Still, almost nothing is known about morphological correlates (but see Markov and Solovjev 1995). Beyond the usual Gestalt perception o f holasteroids


99

including some o f the most unusual echinoids (e.g., Hagenowia - Gale and Sm ith 1982, and the Recent pourtalesiids and urechinids - David 1990; Mooi and David 1996), there is still much to be learned about the quantitative history o f morphological evolution in heart urchins as a whole. In this paper, I analyze the morphological and taxonomic diversification patterns among spatangoids and holasteroids during the period in which both groups were comparably diverse —the Cretaceous; I also include in the study Paleocene survivors o f the K-T extinction, to assess the possibility o f morphological selectivity. Disasteroids appeared in the Jurassic and range to the mid-Cretaceous; they form the paraphyletic stem group containing putative ancestors to crown group Holasteroida plus Spatangoida (Fig.3.1). Thus, disasteroids are also included as a plesiomorphic baseline for analysis of the effects of phylogenetic affinity on morphological change. Accordingly, I trace their disparity throughout the Jurassic and Cretaceous, comprising the radiation and decline of the group until its ultimate extinction. In addition, the more inclusive patterns o f the Atelostomata as a whole are studied and compared with those of the individual orders, thus providing an account o f concordances and discordances at different levels in the hierarchy .

Materials and Methods

Morphological Disparity I studied morphological disparity o f heart urchins through time by means o f a threedimensional landmark-based morphometric quantification o f test architecture (Fig. 3.3). Such a morphometric description permits an assessment o f general morphological patterns through time in a relatively taxon-free manner (Foote 1990b). Heart urchin test architecture is comprehensively captured by the morphometric scheme illustrated in Fig. 3.3. This scheme, with 18 repeatable and meaningful landmarks (Roth 1993), codifies many


100

Fig. 3.3. Morphometric schem e used in the study. Points represent landmarks from which x,y,z coordinates were collected for all of the specimens considered in this study.



anterior notch peripetalous fasciole

ambulacrum III

2a peristome

labrum anterior petal —,

phyllode

pore pairs

4b plastron apical system lateroanal fasciole posterior petal

5 b ^ ^ ^ 5 a >vsubanal fasciole periproct

o

102 standard classical morphometric descriptions (height, length, width, etc.) and accounts for important test features such as general profile, relative positions and relative size of ambulacra and its modifications (petals), extent o f frontal furrow, geometry and relative size of plastron plates, and relative positions of the apical system, peristome and periproct. Because plastron landmarks vary little in the z-direction, and because such variation might be confounded by measurement and orientation error, they were assumed to lie in the same plane. Also, redundant landmark information was averaged and non-informative coordinates were removed. Thus, from 18 landmarks, 38 variables resulted. Landmark data collection consisted of measurement of three-dimensional Cartesian coordinates and involved image acquisition, digitalization, and analysis using a video camera and BioScan OPTIMAS® image analysis software. Although the morphometric scheme utilized incorporates aspects o f all parts of the test, it is not exhaustive. Features like the structure o f spine tubercles, apical system plating or ambulacral pore structure were not included, because they are difficult to quantify morphometrically and are not always preserved. Still, as a subset of possible characters, the 38 variables measured correspond to a large portion o f observed foci o f variation in heart urchins, and were deemed appropriate to estimate relative changes in disparity within the clade. Complete landmark configurations were consistently obtained, which was facilitated by the replication of several landmarks on oral, aboral and side views, and by the serial homology of other landmarks as a result of bilateral symmetry. This permitted reduction in measurement error on the one hand (by averaging replicate/serially homologous landmarks), and accounting for preservation problems (e.g., in specimens with eroded landmarks) on the other. Note that although a certain degree o f serial homology extends to the plastron plates (see Fig. 3.3), they are usually not symmetrical along the longitudinal body axis; in fact, most Mesozoic heart urchins have plastrons with


103 varying degrees of asymmetry, so that "corresponding" landmarks on both sides o f the plastron are not really homologous, being at best fractional o r partial (Shubin and Wake 1996) in both a structural and phylogenetic sense. Landmark coordinates for all specimens were scaled to a common size using a reference baseline (the y-coordinate for the tip o f the periproct, with the origin of the coordinate system centered between the frontal genital pores o f the apical disc), and row normalization was performed for each specimen, rendering the sum of squares for each object equal to 1. Row normalization retains the proportionality o f variables within objects, and destroys differences in magnitude between objects (Reyment and Joreskog 1993). Further, standardization was carried out to guarantee equal weighting o f the variables; although they are all on the same scale, the range of plastron coordinates tended to be an order of magnitude smaller than that of other coordinates, thus making standardization necessary. The final data matrix used in the analyses consisted o f standardized variates. For the production o f an ordination space, the resulting correlation matrix was used as input in a principal components analysis. The first eight ordered PC's were retained for further analysis, based on the combined criteria of inspection o f natural breaks in the decay o f the variance of ranked eigenvalues that would be indicative o f nonrandom decay up to some point, and analysis of maximal loadings in the eigenvector matrix, leading to selection o f n PC's that comprise the maximal loadings for all variables. The first break in the decay o f the variance occurs between PC IV and PC V, and most (though not all) variables load maximally in the first four PC's. Thus, scores on the first four rotated principal components, summarizing roughly 70% of the total variance, were used for graphic portrayal of dispersion and location of forms in morphospace. Morphological disparity per se at each stratigraphic interval was measured as the total variance in the original, size normalized and standardized morphospace. The total variance is the sum of the univariate


104 variances and is equivalent to the mean squared Euclidean distance between each point and the centroid; it is also equal to the sum of the eigenvalues (Van Valen 1974, 1978). The amount or range of morphospace occupied was also measured, in this case as the sum of the univariate ranges. Although empirically the total variance and the total range tend to correlate (Foote 1992b), the relationship is not always proportional or monotonic, since the total range is very sensitive to sample size. Although sample size is roughly equivalent for spatangoids and holasteroids for most o f the Cretaceous, estimation of changes in total range was deemed useful to ascertain how morphologically innovative the coeval origin of holasteroids and spatangoids (with intrinsically small sample sizes) was relative to disasteroids. Additionally, total range provides an alternative metric of morphological distinction for the supposedly "bizarre" experiments in morphology thought to occur in the Late Cretaceous. Error bars on total variance and total range were calculated by bootstrapping of the data within each interval (Efron and Tibshirani 1993); the taxa in each interval were resampled randomly, with replacement, 200 times, except when noted otherwise. The relative overlap of error bars did not change appreciably with more than 2 00 replications. Because of phylogenetic autocorrelation, data points are to some extent not strictly independent, so that error bars cannot be taken as absolute and must be interpreted for comparative purposes only. The issue of nonindependence is a contentious one, but tends to disappear if sampling is viewed as being performed not with reference to an underlying evolutionary process, but to a finite population instead (Sanderson 1995). Statistically speaking, the central assertion of bootstrapping is that the relative frequency distribution of the resampled statistic is an estimate of the original, unknown sampling distribution of the statistic. Violation of the independence assumption is thus just a special case of having a statistic without a known sampling distribution (Mooney and Duval 1993). The minimal assumption of bootstrap, that the empirical distribution function represented by the sample


105 is a good estimator of the population distribution function (Mooney and Duval 1993), remains intact. Nonindependence is thus part o f the unknown sampling distribution, and random subsets of the data are biased in the same way. This situation is in fact implicit in the application of bootstrap to covariance matrices and phylogenetic data, and it is in fact desirable in any neutral model of morphological evolution (Raup 1987). In m ost biological and paleobiological situations, bootstrap procedures are thus best seen as producing estimates of analytical error given some degree of nonindependence. This is the approach adopted here. For several bootstrap samples, parametric statistical tests are possible because bootstrap sample statistics are normally distributed, even if the original distribution is non normal. Thus, in several situations o f interest, I tested for differences between two samples with z-tests based on the normal distribution and using bootstrapped standard errors. Another approach, demanding many replications, is to compare a given statistic with a tail of the resampled distribution. It is used in Section IV.

Sampling and Stratigraphic Resolution Specimens from 59 genera were used in this study: 13 of the 19 described genera o f disasteroids (the remaining genera were erected on the basis of poorly preserved or incompletely known specimens); 21 o f the 34 described genera of holasteroids appearing in the Cretaceous (at least 5 of the remaining genera are known only from poorly or incompletely preserved specimens); and 25 of the 34 described genera of spatangoids appearing in the Cretaceous (see Appendices). Genera are appropriate units o f sampling in this study because they circumvent the relative fluidity o f echinoid species-level taxonomy and increase the reliability of stratigraphic ranges (cf. Sepkoski and Kendrick 1993). One species per genus was sampled for most o f the genera; exceptions included a few longranging and species-rich genera. Turnover of genera can be high, which might render use


106 of a range-through approach unnecessary. However, the size o f the groups is such that, to maximize inclusiveness, the measured form of each genus is used throughout its range. At least in terms o f shape, there is strong conservatism in generic morphologies —genera have often been used as operational taxonomic units in phylogenetic analyses (e.g., David 1988; Jensen 1988), and stasis has been documented or implied in a number o f studies (see Smith 1984; Donovan and Veale 1996). Nevertheless, long-ranging, species-rich genera in the analyses (Holaster. Toxaster. Heteraster. Hemiaster) were not dealt with by averaging multiple species, so as to preserve phylogenetic (and phenetic) differentiation. Thus, several subgenera of Hemiaster (some actually having been assigned genus-level status recently —Smith and Bengtson 1991; Neraudeau 1994) were sampled and treated separately, along with their ranges. Species of Holaster and Heteraster were also treated separately, but here the early range o f the genus was sampled by the earliest species and the remainder by later species. For Toxaster. both of the sampled species were early representatives of the genus and were allowed to range throughout the range of the genus. Temporal resolution is at the epoch to sub-epoch level, and follows the partitioning of stratigraphic intervals and the time scale of Harland et al. (1990). This resolution is a necessary compromise between time subdivision and sample size per interval. Nine intervals were used: (1) Jurassic 1 (Aalenian-Bajocian); (2) Jurassic 2 (BathonianCallovian); (3) Jurassic 3 (Late Jurassic); (4) Cretaceous 1 (Neocomian); (5) Cretaceous 2 (Barremian-Aptian); (6 ) Cretaceous 3 (Albian); (7) Cretaceous 4 (Cenomanian-Santonian); ( 8 ) Cretaceous 5 (Campanian-Maastrichtian); and (9) Paleocene. In this paper, only the Paleocene survivors of the K-T event are included, regardless o f whether they are recorded from the Danian, Thanetian, or even later, i.e., Lazarus taxa are included. The time scale used, generic diversity per interval, and sample size per interval are listed in Table 3.1.


107

Table 3.1. Time scale used, generic diversity per interval, and sample size per interval for heart urchins in the Mesozoic. The Paleocene sample contains only the survivors o f the Cretaceous-Tertiary extinction event. Stratigraphic intervals are coded as follows: J lAalenian-Bajocian, Middle Jurassic; J2 - Bathonian-Callovian, Middle Jurassic; J3 - Late Jurassic; Kl - Neocomian, Early Cretaceous; K2 - Barremian-Aptian, Early Cretaceous; K3 - Albian, Early Cretaceous; K4 - Cenomanian-Santonian, Late Cretaceous; K5 Campanian-Maastrichtian, Late Cretaceous; Pal. - Paleocene. Intervals and codes are the same throughout the text and figures.

Time Scale (Intervals)

Generic Diversity

Sampled Genera

J1

5

4

J2

10

8

J3

13

10

Kl

12

11

K2

13

11

K3

16

12

K4

41

24

K5

66

37

Pal

24

17


108 Taxonomic Diversity To allow comparisons between disparity and diversity, taxonomic diversity is here studied at the genus level. Global stratigraphic ranges (see Appendix I) were derived mainly from J.J. Sepkoski's unpublished generic database, with modifications and additions where appropriate. From tabulations o f stratigraphic ranges, time series of standing diversity were produced. Error bars on standing diversity estimates were calculated as the square root o f of the number o f taxa (Sepkoski and Raup 1986). This procedure provides an estimate of counting error associated with discrete events, but must be used with caution since it was initially suggested for counts o f extinction only, and because diversity at adjacent intervals is not truly independent. It does not allow absolute statistical confidence, but only relative statements if it is assumed that the sampling error is constant throughout the time-series.

Comments on the Use o f Phvlogenv and the Notion o f Constraint A phylogenetic framework, with explicit sister-group statements, provides one way o f dividing a larger group into subgroups that can be viewed as multiple natural experiments. One could study variation at the level of lineages (say, comparing intraspecific variability between sister-species), but lack of divergence at this level may often be an artifact of limited temporal extension . With short timeframes, the finding o f equivalent levels of disparity is hard to interpret as more than the simple result of phylogenetic inertia (a time-dependent property o f any phylogeny, producing autocorrelation) —lack of divergence is to be expected by chance alone. As one expands the timeframe, the danger o f mistaking phylogenetic inertia for a true nonrandom pattern is diminished, and shared levels o f disparity can then be more confidently ascribed to historical/phylogenetic constraints, which are less time-dependent. The distinction between different degrees of time-dependence is not the only criterion to distinguish between inertia


109 and historical/phylogenetic constraint (see Derrickson and Ricklefs 1988), but it is more readily applicable to temporal patterns in the fossil record. Time-dependent properties are interesting in their own right as they can be informative of rate phenomena and the pace of diffusion in morphospace. More often, however, one is interested in the persistence and breaking o f more stable, less time-dependent patterns of historical/phylogenetic constraint. Time-dependence is relative and has to be viewed operationally, because the nature of constraints is probabilistic (see Oster et al. 1988; Schwenk 1995) Care must be exercised in using the notion o f "constraint", a useful but often misused term. As most words in the lexicon, constraint has multiple meanings. Although in evolutionary biology it is sometimes used as a purely descriptive term, akin to "pattern", it is best used in connection with statements about process, thus becoming imbued with explanatory value. To constrain is "to force or produce in an unnatural or strained manner" (Webster's Dictionary). This usually implies a local context of interest that can be seen as natural or anticipated, to which constraints would be external and "unexpected" (Steams 1986). For some, the local context is a received theory of process, say, Darwinism and the expectation o f adaptation by directional natural selection (e.g.,Wagner 1986; Gould 1989; Steams 1992). Constraints then become equivalent to developmental constraints, i.e., biases to variation arising from morphogenetic design limits (Maynard Smith et al. 1985; Wagner 1986), or to stabilizing selection, a bias due to functional necessity. At least one author considered both directional and stabilizing selection under the received theory, and synonymized constraint and developmental constraint (Gould 1989). In paleobiology, it is more common for the local context to be established on the basis of theories of pattern (usually o f uniformitarian nature), assumed to hold as null hypotheses for empirical inferences from the fossil record. When expectations arising from a certain standard of pattern are not met (e.g., rate constancy, random diffusion in morphological space, etc.), a hypothesis o f process in the form of some kind o f constraint


110 is then posed. Thus, for example, heterogeneities like deceleration o f morphological diversification, limits or boundaries, saturation o f morphospace, change o f evolutionary step size, differential flexibility through time and related patterns are all deviations from an expectation of uniformity in the unfolding o f macroevolutionary patterns. Explanations for such heterogeneities may then involve constraints qualified as developmental o r selective. The distinction of different kinds of constraint by reference to (deviations from) a favoured theory of pattern instead of process reflects the eclecticism of approaches like theoretical (see McGhee 1991) and constructional (see Seilacher 1970) morphology, as well as the indeterminacy of current debates over the relative importance of genomic and ecological factors in evolutionary radiations (Valentine 1995; Eble in press). In many ways, the field of paleobiology has matured to the point o f not having a received theory of process. Theories or expectations of pattern (implying no constraint beyond specified boundary conditions) are usually the local context o f interest, and deviations are subject to different hypotheses of constraint. For the present purposes, the kinds of constraints, w ith "no constraint" as a distinct category, are defined in Table 3.2. Phylogenetic constraint is distinguished from historical constraint for conceptual clarity, and phylogenetic inertia presented as a confounding factor. Notice that constraint here is supposed to encapsulate both a negative (limitations) and positive (opportunities) meaning (Alberch 1982, Gould 1989). In Table 3.3, at the end of the chapter, the empirical results o f this paper are listed with implications and processes deduced with emphasis on the presence or absence of different kinds of constraints in so far as they are testable with the data at hand. As defined in Table 3.2, historical/phylogenetic constraint should be used as no more than a descriptive term, when emphasis is directed towards placement o f a selective or developmental constraint in a phylogenetic hierarchy. To the extent that this is the goal, the distinction between historical and phylogenetic constraint proposed here can make hypotheses about the origin of a constraint more precise, by specifying the level of


Ill

Table 3.2.

Categories o f constraint

No constraint:

given boundary conditions, conformance to an expected disparity or diversity path, morphological distribution, or any other null hypothesis implied by a theoretical expectation (e.g., diffusion, rate constancy) or by the data at hand.

Developm ental constraint:

a bias in the production or maintenance o f phenotypic variation caused by inherent properties o f the developmental system, in relative independence from the environment (see Maynard Smith et al. 1985).

Selective constraint:

a bias in the production or maintenance o f variation caused by a specific organism-environment interaction. Can be synonymized with adaptive or functional constraints (see Reif et al. 1985; Seilacher 1991).

Phylogenetic constraint:

a bias in the production or maintenance o f variation which can be ascribed to nodes in a cladogram. It corresponds methodologically with the definition of monophyletic groups by means o f historical homologies (Brooks and McLennan 1991; Harvey and Pagel 1991; McKitrick 1993). It is no more than a descriptive term, and may correspond to either developmental or selective constraints.

Historical constraint:

a bias in the production o f variation which is historically contingent to a phylogenetically defined group. It embraces phylogenetic constraints (historical homologies), plesiomorphies and homoplasies Wake 1991; Schwenk 1995). It is no more than a descriptive term, and may correspond to either developmental or selective constraints.

Phylogenetic inertia:

a pseudoconstraint. A time-dependent property of any phylogeny that creates a spurious pattern o f autocorrelation. The deeper the branches in a phylogeny, the easier its distinction from historical/phylogenetic constraint.


112 inclusiveness in which meaningful explanations (causal linkages) are to be sought. Thus, if changes in disparity are limited across entities within a group, they might be ascribed to a phylogenetic constraint if particular synapomorphies defining that group are causally linked to the characteristic pattern of variation. For example, echinoderms have recently been homologized in terms o f axial (linear, non-isotropic radii) and extraxial elements, with different echinoderm clades being defined by the relative proportion of axial and extraxial skeletons (M ooi et al. 1994; David and Mooi 1996). Echinoids are uniquely defined by extraxial skeleton reduced to the apical system, with axial skeleton exclusively composing the test. Accordingly, the regularity o f echinoid variation that arises from plate addition always along the radii can be viewed as a phylogenetic constraint across echinoids. In contrast, as defined in Table 3.2, a historical constraint is a more general concept, embracing not only phylogenetic constraints (historical homologies) but also homoplasies and plesiomorphies. In other words, a historical constraint implies a more inclusive level o f analysis, such that a characteristic pattern o f change in a group might be causally linked to the homoplastic evolution o f a certain character (e.g., as a result o f heterochrony) or to a plesiomorphic character defining a more inclusive group than the one under consideration (e.g., the existence o f axial skeleton in the echinoid example above). It is worth emphasizing that phylogenetic analysis per se does not guarantee the identification of phylogenetic/historical constraints; apomorphies, plesiomorphies and homoplasies can only be hypothesized as such after a causal linkage between a feature and a certain nonrandom pattern in taxonomic diversity or morphological disparity is established (see also Lauder and Liem 1989; Sanderson and Donoghue 1996). Key innovations (or positive constraints) —historical factors that permit or enhance diversification (Lauder and Liem 1989; Jablonski and Bottjer 1990b) have been increasingly studied in this manner.


113

Morphological Disparity and Taxonomic Diversity through Time

Forms are ordinated in a principal component space (Fig. 3.4) to provide a baseline for visualization o f temporal changes in both range and pairwise distance at the same time. PC I is plotted against PC H, and PC III against PC IV. Based on inspection o f the magnitude o f the loadings, PC I (27% of the variance) is mostly contributed by variation in petaloid development o f ambulacra; PC H (20% o f the variance) has landmarks associated with longitudinal translation of the upper plastron loading highly, as well as those related to lateral extension of the body; PC HI (15% of the variance) mostly reflects the depth of the ambitus; PC IV ( 8 % o f the variance) is related to the asymmetry of the plastron. The ordinations are best viewed as a guide to general change in morphospace occupation. The discussion below is based both on the ordination and on the full morphospace.

Atelostomata Fig. 3.5 shows the pattern o f disparity and diversity through time for the three orders combined into the superorder Atelostomata. There is an overall increase in disparity from the Jurassic to the Cretaceous, with a mid-Cretaceous drop which is quickly erased in the Late Cretaceous. A statistically significant maximum (p < 0.0001, z-test based on bootstrapped standard errors) is achieved relatively early, in the Neocomian (K l), when there was still low generic richness. The maximum coincides with the origin of the orders Holasteroida and Spatangoida (see below). When compared with the Jurassic and given the wide overlap of error bars, Cretaceous levels o f disparity remain relatively constant throughout the period -- but notice the appreciable Albian drop (K3) relative to the Neocomian —with no surpassing o f the Neocomian maximum despite a five-fold increase in taxonomic diversity.


114

Fig. 3.4. Pattern of morphospace occupation in disasteroids, holasteroids and spatangoids, from the Middle Jurassic to the Late Cretaceous (as a sample o f standing diversity) and for the Paleocene (survivors of the end-Cretaceous extinction only). Ordination is based on a principal components analysis, of which the first four axes (70% o f total variance) are used to illustrate changes in morphological distributions. PC I is plotted against PC II (top) and PC HI is plotted against PC IV (bottom). See text for explanation of the meaning o f axes. The ordinations are meant to codify only the general pattern o f change in morphospace. Disparity patterns are derived from all original variables. Shown is a sample of standing diversity. Range-through taxa at each interval are indicated by brackets.


115

Aalenian-Bajocian (Jl) 3 n 2 -

CN

if

O

-Q

0

-1

-

-2

•

-i—

2

Disasteroida

— ^O -

|--------1------- 1--------1------- [—

-1.5

-1

-.5

-> T" 0 .5 PC 1

t

—

1

>— i—

■— i— ■— I

1.5

2

2.5

Aalenian-Bajocian (Jl) 2.5

2 1.5

O

Disasteroida

PC 4

1

5 0

.5 OO

1

1.5 •2 3

■2

1

0

1

2

3

PC 3


116

Bathonian-Callovian (J2)

Fig. 3.4 (continued)

3 2

O

Disasteroida

0

•2 •3 •2

25

-1.5

1

-.5

5

0

2

1.5

2.5

Bathonian-Callovian (J2)

Disasteroida

PC 4

O

to] to] 3

-2

-1

0

2

3

PC 3


117

Late Jurassic (J3)


O

o

Disasteroida

[Ol

CN „ Cl

r\

[Ol [% 1

0

[Ol

O O O

u

-1

-2 ---- 1— 1— 1— 1— 1— 1— 1— I---- 1— 1 0 .5 I 1.5 2 2.5 PC 1

-3 -2

-1.5

-1

-.5

Late Jurassic (J3) 2.5 -i

O

Disasteroida

[0 ] [Ol [O] [O]

3

2

1

0

9

3

PC 3


118

Neocomian (Kl)


3

2 1

EO]

Disasteroida X Holasteroida A Spatangoida O

0

1

•2

■3 2

-1.5

-.5

0

.5

1

2

1.5

2.5

PC 1

Neocomian (Kl) 2.5 -i

O Disasteroida X Holasteroida A Spatangoida [0 ] [O]

[O]

[Ol

3

-2

0

1

2

3

PC 3


119

Barremian-Aptian (K2)


3

2

0

Disasteroida Holasteroida Spatangoida

[Ol

1

-2 3 -2

-1.5

1

-.5

.5

0

2

1.5

2.5

PC 1

Barremian-Aptian (K2) 2.5 -]

[X]

PC 4

O X A


[Ol

3

■2

1

0

2

3

PC 3


120

Albian (K3) Fig. 3.4 (continued)

3

2 I O Disasteroida 7—7 X Holasteroida [O] A Spatangoida

0

[Ol 1

■2 •3 2

-1.5

1

5

0

.5

1.5

2

2.5

PC 1

Albian (K3) 2.5

2

PC 4

1.5 O X A

.5 0


.5 [Ol

-1

1.5

2 3

D

1

0

2

3

PC 3


121

Cenomanian-Santonian (K4)


3

2 1

X Holasteroida A Spatangoida

0

[XI [X]

1

2 •3 2

-1.5

1

5

0

5

1.5

2

2.5

PC 1

Cenomanian-Santonian (K4) 2.5 i

[X] PC 4

X Holasteroida A Spatangoida -.5

3

2

1

0

1

2

3

PC 3


122

Campanian-Maestrichtian (K5) Fig. 3.4 (continued)

X Holasteroida Spatangoida

Campanian-Maestrichtian (K5)

PC 4


[A] a [A 1


123

Paleocene leftovers (Pal)


3

2 [A]

PC 2

[A] X Holasteroida A Spatangoida

0

[XI

2 3 2

-1.5

1

-.5

0

1

1.5

2

2.5

PC I

Paleocene leftovers (Pal) 2.5

[MI

9

1.5

PC 4

1

,5


0

-.5 1

1.5 •2

3

•2

0

1

2

3

PC 3


124

Fig. 3.5. Patterns o f morphological disparity (top) and taxonomic diversity (bottom) for the Atelostomata (three orders combined). Morphological disparity is measured by the total variance, the sum of the variances o f all original variates. Taxonomic diversity is the standing diversity of genera at each interval. Error bars on disparity in this and subsequent figures were generated by bootstrapping (200 replicates). Error bands on diversity in this and subsequent figures correspond to ±Vd, an estimate o f counting error, where D is the standing generic diversity.


Atelostomata

c«

30 25 '

Kl

K2

K3

K4

K5

Pal

K4

K5

Pal

Geologic Time

70 60 “ ‘33 §

50 40 30 -

20

'

10

‘

Jl

J2 J3 Kl

K2

K3

Geologic Time


126 Inspection o f the distribution o f forms in principal component space (Fig. 3.4) suggests that morphological diversification until the Early Cretaceous proceeded mainly by the generation of outliers and a monotonic increase in disparity; once the outliers become the norm, morphospace is filled and spacing diminished, with the pattern o f disparity then becoming basically one o f fluctuation around a plateau. In other words, the pattern is roughly one of initial diffusion, abrupt change in step size with the origin o f holasteroids and spatangoids (see below), and later stability. The end-Cretaceous mass extinction does little to lower the plateau, with a marked drop in taxonomic diversity not matched by a proportionate drop in disparity (no statistical significance for disparity difference, z-test). As far as the Atelostomata are concerned, this extinction is thus nonselective with respect to morphology. However, some selectivity might be present, since there is a drop o f disparity after all. The drop is not statistically significant, but o f interest here is a relative issue, involving the degree of reduction in disparity relative to diversity. Thus, if only the survivors of the pre-extinction interval are taken into account (thereby excluding biotic recovery), the disparity drop for the Atelostomata was about 1/5 o f the drop in diversity. This generates an expectation to be contrasted with the pattern of extinction in the holasteroid and spatangoid subclades (see below).

Disasteroids The pattern o f disparity and diversity in the Disasteroida is qualitatively concordant in the Jurassic, when both keep increasing (Fig. 3.6). A doubling o f diversity from the Aalenian-Bajocian (Jl) to Bathonian-Callovian (J2) is accompanied by a near doubling of disparity, but a further increase in diversity into the Late Jurassic is not followed by disparity, which reaches a plateau that persists through the Jurassic-Cretaceous transition. This plateau is determined mostly by the appearance and persistence o f the genus Collvrites. which displays subpetaloid development of the ambulacra. The Late Jurassic


127

Disasteroida ?

c

50 -

.2 >

40 '

l

30-

Q 10

-

Jl

J2

J3

Kl

K2

K3

Geologic Time 18 16 14

< u >

12

5

10

•u < u c u a

8

6 4 2

0 Jl

J2

J3

Kl

K2

K3

Geologic Time

Fig. 3.6. Patterns o f morphological disparity (top) and taxonomic diversity (bottom) for the Disasteroida.


128 genus Proholaster. another subpetaloid form, bridges the gap between the outlying Collyrites and the primitive region o f morphospace. The Tithonian marks the beginning o f the end of disasteroids, for diversity falls precipitously into the Cretaceous. However, the decline o f disasteroids is apparently nonselective with respect to morphology: morphospace thinning seems to occur, with persistence o f both primitive and outlying forms (Fig. 3.4), and disparity increases accordingly. The fact that genera are still being produced in the Neocomian despite the overall loss in diversity -- while at the same time spatangoids and holasteroids appear (from disasteroids) -- suggests the possibility that disasteroids may have declined by actively evolving into something else, the establishment of their paraphyletic status then becoming the very reason for their extinction. However, this can only explain why certain taxa possessing new suites o f morphological attributes are pruned away from their paraphyletic stem group. Disasteroids persist for some time after the origin of spatangoids and holasteroids, but typically disasteroid morphologies, with a disjunct apical system and a protostemous plastron, become less common in absolute terms, with disasteroid originations ceasing after the Neocomian. Biogeographic factors might have been involved. The decline o f disasteroids seems to correlate with the origin and initial diversification of spatangoids. In the Jurassic, disasteroids were an important component o f marine assemblages (Thierry 1984) and more geographically widespread. In contrast, by the Early Cretaceous they are known only from Europe and immediate surroundings (Solovjev 1971). On the other hand, the earliest representatives of the Spatangoida (e.g., Toxaster) were widespread very early in the Cretaceous (Devries 1960), a pattern mirrored by other genera later on. This suggests an hypothesis that at least part of the disasteroid decline could be explained by the appearance and rapid geographical expansion o f spatangoids. The implication is that population sizes in early spatangoids should have been sufficiently high so as to lead to preemption o f environments. David (1979) provides some evidence for this, by documenting the changes


129 in relative abundance of the spatangoid Toxaster relative to the disasteroids Disaster and Collvropsis in the Valanginian and Hauterivian o f subalpine France. From an initial spatangoid-disasteroid alternation of dominance mediated by environmental change (with spatangoids thriving in less disturbed sediments), spatangoids eventually become dominant. Whether competition was actually involved is unclear, given the role played by changes in environmental regime. Additional detailed paleoecological studies could provide more data.

Holasteroids The history of disparity of Mesozoic holasteroids (Fig. 3.7) seems to have two distinct phases. After an early high in disparity that persists to the mid-Cretaceous, disparity rises to a new level that is maintained from the Cenomanian-Santonian (K4) to the Campanian-Maastrichtian (K5). The first phase might be interpreted as the establishment of the genus Holaster and closely related derivatives like Cardiaster and Labrotaxis (David 1988). The second phase represents a massive exploration o f new regions of morphospace, in conjunction with a general increase in generic richness and with the evolution of derived characters uniting distinct subclades within the Holasteroida (David 1988). One such subclade is the Infulaster-Hagenowia lineage (Gale and Smith 1982), where the elongation of the test and the formation o f a rostrum gives the test a bottle-like shape. The two genera are present in both K4 and K5. Other subclades are characterized by novel plate configurations in the plastron, which reaches a maximum in variation in the Late Cretaceous. As seen in Fig. 3.4, much o f the holasteroid action in the second phase is related to colonization of morphospace away from the centroid, with no clear preferred direction of change. The overall impression is one o f diffusive evolution and continual morphological experimentation. A ceiling to such experimentation may have been reached by the Campanian-


130

Holasteroida 40 -

C3

p b

20 15 -

Q

10

‘

Kl

K2

K3

K4

K5

Pal

K5

Pal

Geologic Time

35 -

>

(5

O u c

20 3

00 10

0

J1

J2

J3

K1

G eologic Time

Fig. 3.15. Oral and aboral disparity for the Disasteroida.


151

Oral

25 1 2 2 .5

-

8 c

20

5

1 7 .5 '

■

7 .5

2 .5

K1

K2

K3

K4

K5

Pal

K5

Pal

G eologic Tim e

35 I

-y.

Aboral

30 -

V

u §

•e CS

>

^

o E £

25 -

20

-

15

'

10

‘

K1

K2

K3

K4

Geologic Tim e

Fig. 3.16. Oral and aboral disparity for the Holasteroida.


152

22.5 1

20

uu■■r. §

’C > < O e3

-

1 7 .5 “ 15 1 2 .5 “

4 -4

10

-

in

K1

K2

K3

K4

K5

Pal

K5

Pal

Geologic Time

Aboral

X

14 -

u S

12 -

>

10 -

•I—

o

s 3

CO

K1

K2

K3

K4

Geologic Time

Fig. 3.17. Oral and aboral disparity for the Spatangoida.


153 thereafter, while less functionally relevant features were more continuously modified, in accretionary fashion. As to the Cretaceous-Tertiary drop in disparity, its somewhat nonselective character is equally partitioned between oral and aboral disparity. Disasteroid total disparity (Fig. 3.6) is almost completely the contribution of variation in aboral landmarks (Fig. 3.15). The protostemous plastron is very stereotypical o f disasteroids (with Proholaster as an exception), and the very small size o f the plastron plates must have prevented extensive allometric change relative to other regions of the body. Therefore, the small magnitude of oral disparity and the overall tracking o f the rise and fall o f generic richness is somewhat predictable. In contrast, so many of the basic themes later developed by spatangoids and holasteroids (frontal furrow, petaloid ambulacra, oral periproctal position, gibbose tests) are represented among disasteroids, however sporadically, that it becomes tempting to describe the disasteroid saga as one o f aboral differentiation and change that adumbrates typically descendant holasteroid and spatangoid themes. The contrast o f oral and aboral disparity in holasteroids is impressive (Fig. 3.16). Initially, most of the disparity is the result of aboral variation in Holaster. The appearance o f new species of Holaster and new genera by the Albian essentially marks a huge drop in aboral disparity and a huge rise in oral disparity. The drop in aboral disparity must result from a particularly homogeneous filling in of morphospace in many dimensions; the later recovery indicates a more inhomogeneous continuation o f filling initially powered by a marked rise in taxonomic diversity (Fig. 3.7). As for oral disparity, the early low is best seen as the result of a low diversity lag in what later becomes a pattern o f rapid diffusion and later stability. The respective contributions of oral and aboral disparity to the total become more commensurate by the Late Cretaceous, and the attainment of a plateau in both cases suggests a uniform exhaustion of morphological themes in the group, regardless of functionality or developmental accessibility. The presence o f a drop in disparity into the


154 Paleocene in both time-series, despite different degrees o f functionality, leads to the possibility that sorting (sensu Vrba and Gould 1986) o f holasteroids may have had as much to do with clade membership as with the morphologies exhibited per se. It is thus conceivable that the drop in holasteroid disparity is more the by-product of a massive reduction in taxonomic richness than the result of within-group morphological selectivity. As discussed above and elsewhere (Eble 1997; Chapter H), holasteroid extinction may hinge on broad ecological correlates (e.g., depth of burrowing, abundance), regardless of the particulars of morphology. If so, the reduction in disparity is better described as "morphological sorting", and the causes o f taxonomic attrition need to be sought in the realm of spatial ecology and biogeography (see Jablonski 1989). Spatangoids are as impressive as holasteroids for the discrepancy exhibited between oral and aboral disparity (Fig. 3.17). Instead of symmetrically opposite patterns, however, the differences are o f a wholly different nature. Oral disparity changes follow a very monotonic increase throughout the Cretaceous, mirroring and accentuating the overall diffusive pattern (Fig. 3.8). Much in the same way as with total disparity, a slowing down o f diffusion is apparent in the latter part o f the Cretaceous, despite a marked increase in taxonomic diversity, thus implying a temporal decline in step size. Aborally, disparity changes are not directionally monotonic, and seem to have gone on average nowhere. Disparity is high in the beginning and remains high, with no clear response to increases in taxonomic diversity (Fig. 3.8). Definite boundaries seem to be reached very early on — aboral spatangoid morphological evolution is clearly constrained. This means that the steady increase in number of genera and families during the Cretaceous implied mostly packing in aboral morphospace and accretion in oral morphospace. Although the relative contribution of oral disparity to the total is initially small, it later drives the continual increase in total disparity (thus accretionarily) until the Campanian-Maastrichtian, when both oral and aboral disparity first differences have the same sign. The sign changes again


155 through the Cretaceous-Tertiary, as oral disparity of Paleocene leftovers rises, driving the overall nonselectivity o f spatangoid extinction, and aboral disparity falls somewhat. The drop in aboral disparity is not negligible relative to taxonomic diversity; it is in fact quite commensurate with the holasteroid one, and by itself it suggests some morphological selectivity in extinction within spatangoids. Such selectivity would seem to be o f a different nature than that affecting holasteroids, however (see above), so that the possible ecological correlates discussed in connection with the holasteroid-spatangoid contrast may or may not apply. In the context o f the previous section, where it was argued that the origin of holasteroids and spatangoids from disasteroids represented a large jump in terms o f the disparity of the Atelostomata, the partitioning of disparity into oral and aboral components implicitly allows an additional test. It could be argued that the observed burst in atelostomatan disparity (in fact the most pronounced increment observed throughout the Mesozoic) is a possible artifact o f the way in which petals were scored. In non-petaloid forms, petal landmarks assumed the coordinates of the position of the apical system (0 , 0 , 0). This is biologically justified since petals are peramorphic derived features produced from plate addition from the apical system. There is a whole continuum of petal lengths in ontogeny and phylogeny, from very small to very large, and variation in rates o f plate addition is also viewed as being continuous (Raup 1968). Still, although the origin of novelties cannot simply be dismissed as artifactual (it either happens or it does not), one might be interested in the effect of the acquisition of petals at the origin of holasteroids and spatangoids, and in particular how robust the disparity burst is to the removal o f these characters. The pattern o f oral disparity (Fig. 3.14), which does not include petal landmarks, provides a test of the robustness of the overall pattern. There is a pronounced Neocomian burst in both oral and aboral disparity, in absolute terms and relative to taxonomic diversity


156 —in fact the oral disparity burst is the more significant one. The substantial increase in disparity at the origin o f holasteroids and spatangoids is not simply driven by the petals, but is instead morphologically pervasive. One could go even further and argue that two subpetaloid disasteroid forms, Collvrites and Proholaster. arose before the origin o f holasteroids and spatangoids (see Fig. 3.4) but did not automatically entail a rise in disparity as substantial as the one in question. Therefore, the conclusions of the previous section still hold. The very morphological pervasiveness of the Neocomian Atelostomatan burst gives a hint at the kind of release of constraint involved: if ecospace occupation was exclusively driving the pattern, only the clearly functional aboral landmarks should contribute to a burst. The oral disparity burst suggests that developmental flexibility itself was at issue, regardless of functionality or degree of developmental entrenchment. If so, the origin o f holasteroids and spatangoids in onshore environments might be a result of an ability o f some aspect of such environments (characteristic population structures, UV radiation) to trigger novelties by affecting development at the level of, say, pleiotropic interactions or mutation rates. The consideration o f groups in isolation may hide regularities that extend from group to group, and which are not necessarily apparent when a more inclusive group is considered. If one concentrates not on the oral and aboral disparity histories of each group and the mode of diversification implied by each taxonomic diversity history, but instead traces the levels of disparity from disasteroids to holasteroids and spatangoids, some interesting generalizations concerning process emerge. Oral disparity is remarkably low in disasteroids, although it does increase through the Jurassic. This low level persists in both early holasteroids and early spatangoids, implying the persistence of a pattern of developmental entrenchment and therefore developmental constraint. High developmental entrenchment does not mean impossibility of change, and indeed both holasteroids and


157 spatangoids ultimately break away from the disasteroid historical constraint (in this case a developmental constraint) and achieve much higher levels of oral disparity in the Late Cretaceous. The high levels come late in the time series in comparison with aboral disparity, which is expected since the waiting-time for change in more developmentally entrenched systems should be longer. Although there are idiosyncrasies in the Mesozoic histories o f holasteroids and spatangoids, in both cases the higher levels o f oral disparity are achieved with concomitant deceleration (see K4 and K5 in Figs. 3.16 and 3.17), suggesting that a new regime of historical (developmental) constraint was established. From the Middle Jurassic to the Late Cretaceous, the change in degree of oral developmental constraint is consistent with an increase in modularity by parcellation (Wagner 1996; Wagner and Altenberg 1996) —in other words, higher character dissociability. This is consistent not only with the observed change in level o f disparity, but also with the diversity of discrete morphological classes to which plastronal plate arrangements can be assigned in Late Cretaceous holasteroids and spatangoids (see Smith 1984; David 1988). An increase in oral developmental modularity might be understood in terms of the larger size of plastron plates and an ability to behave somewhat more independently as units o f growth in the context o f tension gradients in the morphogenetic whole (see Raup 1968; McNamara 1987; Seilacher 1991). This scenario for change in modularity in a region o f the heart urchin phenotype which has, in terms o f the geometry of measured landmarks, very little functionality, challenges the notion that modular units of the phenotype have to fulfill the criterion o f serving a primary functional role (Wagner 1996). Phenotypic modules can arise by reasons o f structure that have little to do with direct function. When aboral disparity is traced from disasteroids to holasteroids and spatangoids, a different pattern is suggested, with maintenance o f a disasteroid level of aboral disparity throughout the histories of spatangoids and holasteroids. Although there is a possible


158 excursion in holasteroids, both spatangoids and holasteroids display in the Late Cretaceous essentially the same level of aboral disparity observed in disasteroids in the Late Jurassic (Figs. 3.15, 3.16 and 3.17), despite an independent history o f 60 million years and the progressive occupation o f quite distinct tiers in the sediment and the evolution o f different habits. Clearly, there are aspects o f heart urchin aboral morphology that were not measured and must have evolved in conjunction with habitat/habit divergence —different kinds o f spines and ambulacral pores are obvious suspects . However, the opportunities and limits to variation in aboral test architecture appear to have remained quite stable regardless of phylogenetic and ecological divergence. Even though the aboral landmarks in the present analysis are less developmentally entrenched and more functional, the persistence of a pattern of variation through time, groups and habitats is strong circumstantial evidence for the persistence of a historical constraint in the form of a developmental constraint. This by no means implies that selection was not involved in the evolution of the obvious functionality of aboral features; indeed it may have been involved in each and every evolutionary transition. At the level o f all Mesozoic heart urchins, however, limits to variation consistent with design limits imposed by development would seem to have regulated the extent to which selection could lead the massive input of taxonomic diversity into newer reaches o f aboral morphospace. Although the relative contribution of aboral disparity to the total was consistently large in all three groups through time, it was oral disparity, with less obvious functionality and more developmental entrenchment, which most clearly expressed phylogenetic differentiation within the Atelostomata, at least at the level o f orders.


159

Disparity of Originations: Disparity, Versatility, and Evolutionary Potential

Wagner and Altenberg (1996) emphasize the distinction between variation, the actually present intraspecific or interspecific differences in a group, and variability, the potential or propensity to vary. Analogously, in macroevolution disparity is the appropriate quantity to describe large-scale variation in an evolving group. Concomitantly, what is the macroevolutionary analogue of variability? Genetic variability is measured by the mutation rate and mutational variance, for single traits, or by the mutational covariance matrix, for multivariate cases; biological versatility (Vermeij 1973), the number or range of independently varying morphogenetic parameters, has been suggested as a proxy for variability (Wagner and Altenberg 1996). Biological versatility is best studied when a few parameters can be described and categorized a priori through developmental and theoretical morphology (e.g., Vermeij 1971, 1973, 1974). This requirement at the same time makes it difficult to expect more than a tenuous relationship between patterns o f temporal change inferred from a small set o f parameters and changes in total disparity o f a comprehensive set of characters. In disparity terms, a description o f the potential to vary is perhaps best afforded by the disparity o f originations in each interval. Although radiating clades under relatively constant extinction rates will usually display a correlation between the first differences of standing disparity and the disparity o f originations, the fact that standing disparity can increase when extinction events are nonselective suggests that origination data would more closely capture changes in the propensity to vary (variability). In addition, standing disparity may change even when the disparity o f originations is nil (as in some declining clades). Much in the same way that there is no necessary one-to-one mapping between variation and variability and their relationship is informative about underlying generative biases, the distinction between standing disparity and disparity of originations may add insight to discussions of macroevolutionary constraints -- significant changes in


160

the rate of production o f novelties must be tied in with changes in the range of opportunities (ecological or developmental) available for origination. A discussion o f disparity o f originations was undertaken earlier in the context o f the origin of holasteroids and spatangoids from disasteroids, with the implication that a change in the propensity to vary accompanied the increase in atelostomatan disparity observed in the Neocomian (K l). The disparity o f originations throughout the study period in the Atelostomata, Disasteroida, Holasteroida and Spatangoida is presented in Figs. 3.18 to 3.21. For the Atelostomata as a whole (Fig. 3.18), two main differences are clear relative to the pattern of total disparity (Fig. 3.5). First, a burst in standing disparity is not matched by a burst in the disparity o f originations in the Neocomian. Although counterintuitive, such mismatch is understandable if the distinction between within-group and among-group disparity is kept: relative to Late Jurassic disasteroid disparity, the Neocomian pool of originants (including a couple of disasteroids and the newly appeared holasteroids and spatangoids) significantly added disparity (p < 0.008, z-test), signalling an among-sample jum p in morphospace). Such addition is mostly the contribution o f holasteroids and spatangoids, since total Neocomian disasteroid disparity is essentially the same as in the Late Jurassic (Fig. 3.6; see also Fig. 3.4). Within the originant pool, disparity also increased but not significantly so (z-test) in comparison with Late Jurassic (solely disasteroid) disparity of originations. There is a trend towards increase in disparity of originations (disparity in K l differs with marginal statistical significance from J2 - p < 0.067, z-test), but such increase is somewhat gradual. This contrast o f long-jump "fugitive" morphospace and within-sample cohesion is an interesting aspect of diversification in morphospace, and highlights the perception that controls on origination may at times be different than those operating on a balance between origination and extinction (e.g., Eble 1998). Comparisons between standing disparity and disparity of


161

Atelostomata 1

50 ' t:

a o a c '5h O o *c u* V! Q

40 35 30 25 ' 20 •

15 * 10 '

5’ o -I

I"

Jl

i

i

t

|

i

J2 J3

i" i

Kl

|

i

r~

t

|

K2

i

r

K3

n

|

i

K4

i

i

|

i

i

KS

Geologic Time

Fig. 3.18. Disparity of originations for the Atelostomata. Originations may more closely reflect the potential or propensity to vary (versatility of Vermeij 1973 and variability of W agner and Altenberg 1996), and thus more closely capture changes in underlying generative constraints, or lack thereof, on disparity.


162 originations in other groups, at times in which new subgroups appear, could determine the generality of the present results. Second, the "plateau" observed for total disparity (Fig. 3.5) is simply not supported by the disparity of originations. The significantly reduced (p < 0.022, z-test) disparity of originations (contributed by spatangoids only) in the Barremian-Aptian (K2) is not matched by a proportionate decrease in total disparity, indicating inhomogeneous filling o f morphospace -- the originants fail to occupy newly extinct (disasteroid) portions o f morphospace (see Fig. 3.4). Conversely, the ensuing rise in disparity o f origination in the Albian does not agree with a slight decrease in total disparity. Late Cretaceous increases in the disparity of originations are expressed in the total disparity pattern, but have different magnitude, significantly exceeding the Barremian-Aptian level (p < 0.022, z-test for increase from K2 to K4). Overall, accretion to the edges o f morphospace would seem to be happening (see Fig. 3.4). Thus, what appears to be a plateau in standing disparity for the Atelostomata as a whole is more the result of changing origination and extinction rates, which balance to produce a plateau, than the expression o f a stable propensity to vary (here measured as disparity of originations). Disasteroids (Fig. 3.19) are interesting in that the change in disparity o f originations closely follows changes in taxonomic diversity. This is not surprising in view o f the high turnover from interval to interval: most o f the disparity at each interval is contributed by newly appeared genera. In contrast, holasteroids (Fig. 3.20) have disparity o f originations closely matching the total disparity pattern. There is virtually no extinction at the generic level until the K-T, and thus the total disparity pattern up to the Campanian-Maastrichtian is an almost perfect expression of the accumulation o f originations through time. Therefore, the two phases in holasteroid morphological diversification, with an early plateau followed by a late plateau that signals a decrease in morphological experimentation, are for the most part a result of increasing constraint on the underlying generative process -- on the potential


163

Disasteroida 45 1 40 ' C /5

|

35 -

a .5 oa Q

30 25 -

°

20

a a.

15 -

5

10"

-

J1

J2

J3

K l

K2

K3

G eologic Time

Fig. 3.19. Disparity of originations for the Disasteroida.


164

Holasteroida V5 •1 C3 .£

40 “ 35 30 '

10

-

K l

K2

K3

K4

K5

G eologic Time

Fig. 3.20. Disparity o f originations for the Holasteroida.


165 to produce novel morphologies. Given this deceleration in origination, the uniqueness of Recent deep-sea holasteroids (treated in Eble, Ch. IV) is all the more puzzling. The spatangoid pattern o f change in disparity o f originations (Fig. 3.21) is also reminiscent of the pattern of total disparity. Little extinction happens until the CenomanianSantonian, so that total disparity is mostly contributed by cumulative originations. However, the contribution o f new originations to total disparity in the CampanianMaastrichtian is very much damped, as a result o f accretion to the edges o f remaining morphospace from the previous period (see Fig. 3.4). Origination in the CampanianMaastrichtian was substantially higher than before (p < 0.064, z-test for increase from K4 to K5), suggesting that the potential to generate novel morphologies was far from reduced. That this is so is suggested by the history of spatangoids in the Cenozoic (Eble, in prep.).

A Comment on Disparity and Sorting

What is the relationship between disparity and sorting? Morphology is quantified at the individual level, but disparity is inferred from multiple samples across a clade. While the total variance (and total range) in morphospace, just like the univariate variance, is ultimately an aggregate character, disparity viewed as a clade potential might be conceptualized in ways similar to geographic range, a putative emergent trait (Jablonski et ai. 1985; Jablonski 1987). To the extent that a hierarchy o f levels of inclusiveness or distinctness has ontological reality (Valentine and May 1996; Eble 1997), the relationship between disparity and hierarchy theory clearly deserves more attention. The debate over emergent versus aggregate characters is unresolved (e.g., Vrba and G ould 1986; Lloyd and Gould 1993; Grantham 1995) and should not concern us here, for the only requirement for clade sorting is that o f emergent fitnesses. That disparity and sorting are connected is true almost by definition; of interest is whether such connection is direct or indirect. A direct


166

Spatangoida 45

K o J3 c '3) O r— o

40

s

’C C3 o. v .

35 30 25

20 15

10 5

0 K l

K2

K3

K4

Geologic Time

Fig. 3.21. Disparity o f originations for the Spatangoida.


167 connection is implied by discussions o f evolvability (e.g., Dawkins 1989; W agner and Altenberg 1996), where variation is assumed to be the fodder o f adaptive change, and by discussions o f variability (Lloyd and Gould 1993), where clade selection is thought to directly favor more variable clades. Lloyd and Gould argued that "clades with greater interspecific variability due to more copious speciation, may gain a macroevolutionary edge" and that in mass extinctions "survival... must often depend on fortunate success o f a few variants ... and limited variability must often lead to elimination" (1993, p.598). While variability can be the cause of increased survivorship o f clades, it is by no means a guarantee for survival over the long run, as exemplified by the holasteroidspatangoid contrast. Holasteroids and spatangoids have similar levels o f disparity (and of disparity o f originations) in the Late Cretaceous. Both groups displayed a wide array of morphologies and were thriving then. Yet the rules governing survival through the K-T would seem to have had nothing to do with whatever advantages disparity might have imparted, for holasteroids suffered a disproportionately larger drop in diversity (and disparity). It is possible that pre-extinction holasteroid disparity guaranteed the persistence of a few forms into the Danian. In light o f the lack o f a similar drop in spatangoids, however, it is possible that it was an ecological correlate o f clade membership that might have favoured survival in the Late Cretaceous crisis, in ways clearly unrelated to overall disparity before then. Disparity changes may often hitchhike on clade sorting, their relationship thus being indirect, and not direct as a result o f clade selection on disparity. The generality o f this statement and the relative importance o f direct and indirect links between disparity and sorting remain to be determined.


168

C onclusions

The present study, while documenting morphological diversification in holasteroids and spatangoids, with disasteroids as a baseline for comparisons, also served as an opportunity to apply a number o f approaches to the understanding o f changes in disparity through time as well as their relationship with underlying taxonomic diversity patterns (see Table 3.3 for a summary o f empirical results). Cross-hierarchical relationships were explored, with the concomitant analysis o f disparity in the superorder Atelostomata. Within-group disparity is not additive (Foote 1993b), so that distinct patterns at different levels may often accrue, as was the case in this study. The higher-level phylogenetic framework available, however limited, allowed the long-term pattern o f disparity in individual groups to be contrasted in light o f various morphological diversification modes. Disasteroids showed a pattern consistent with diffusion and later deceleration in disparity; such deceleration was interrupted with a change in step-size and the origin o f holasteroids and spatangoids, which marked a period o f true morphological innovation. Once established, each of these clades then proceeded, in slightly idiosyncratic ways, through diffusion and later deceleration under very similar taxonomic diversification rates. The common underlying dynamics o f disparity change suggests that morphological evolution may have proceeded under very similar controls, and that a historical constraint was expressed in terms of the dynamics o f diversification. Whether such an historical constraint is a unique phylogenetic constraint at the level o f (Disasteroida (Spatangoida+Holasteroida))) can only be determined after similar studies are carried out in outgroups to the Atelostomata. More unique patterns were found when disparity was partitioned into components with different degrees of developmental entrenchment and functionality, and different inferences on process thus achieved. There seems to be an increase in developmental


169 modularity and decrease in developmental constraint on oral landmarks from disasteroids to holasteroids and spatangoids. For aboral landmarks, it is apparent that a level of disasteroid aboral disparity is not effectively surpassed by spatangoids and holasteroids despite their independent evolutionary histories over 60 million years and progressive differences in habit and habitat —a historical constraint expressed as a developmental constraint would seem underlie this regularity and the maintenance o f a pattern of aboral architectural modularity. Consideration of disparity of originations alone not only improved understanding of the dynamics of overall patterns but also allowed, by analogy with the distinction between variation and variability (W agner and Altenberg 1996), reconsideration o f the theme of versatility (Vermeij 1973) in disparity terms. Much in the same way that insight in diversity studies is possible by partitioning diversity into origination and extinction components, further disparity studies might benefit from consideration o f time-series of disparity o f originations and of disparity of extinctions. Sample sizes are inherently smaller in disparity studies, but such approach should prove to be often tractable. It is the challenge o f evolutionary theory to bring the burgeoning field of disparity studies in line with the recent advances made in our conceptualization o f macroevolution, including hierarchy theory. Most of previous discussions concerned taxonomic diversity; they should be complemented and expanded with disparity studies and ultimately interpreted with the help o f phytogenies. Eventually, the trinity disparity-diversityphylogeny, in integrative fashion, should lead to maximal information in the study o f the great themes of regularity and contingency, origination and extinction, and a host of other subjects revolving around the evolution o f biological entities through geological time.

Reproduced with permission of the copyright owner. Further reproduction prohibited w ithout permission.

170 TABLE 3.3 SUMMARY OF EMPIRICAL RESULTS, WITH EMPHASIS ON THE PRESENCE OR ABSENCE OF DIFFERENT KINDS OF CONSTRAINTS PATTERNS

IMPLICATIONS AND PROCESSES

Exponential taxonomic diversification in the No constraint given boundary conditions Mesozoic histories o f holasteroids and (e.g., intrinsic diversification rate). spatangoids. Holasteroids show much higher K-T drop in taxonomic diversity than spatangoids, despite nearly equal diversification rates.

Mediation of extinction by ecological habit and/or habitat —selective constraint.

Peak disparity o f Atelostomata early.

Early higher flexibility as a result of less constrained developmental or selective controls.

Early burst in the disparity of Atelostomata despite low taxonomic diversity.

Early higher flexibility as a result o f less constrained developmental or selective controls.

Deceleration of morphological diversification relative to taxonomic diversification in the Atelostomata.

Later lower flexibility as a result of more constrained developmental or selective controls.

Failure of spatangoid and holasteroid morphological diversification to generate as much disparity as the Atelostomata.

Morphological evolution proceeds differently at different hierarchical levels — within-group variation is distinct from among-group variation, implying discreteness in morphospace - spatangoids and holasteroids were more developmentally or selectively constrained than the more inclusive Atelostomata.

Disasteroid Cretaceous decline not matched by disparity.

No constraint. Extinction (or pseudoextinction) is morphologically nonselective.

Proportionate burst in holasteroid diversity and disparity between Albian and Cenomanian-Santonian.

No constraint. Diffusion in morphospace.

Burst in holasteroid diversity not matched by disparity between the CenomanianSantonian and the CampanianMaastrichtian.

Some morphological boundaries may have been approached —such boundaries might be rooted on developmental or selective constraints.


TABLE 3.3 (continued) PATTERNS


Substantial drop in holasteroid diversity and Morphological selectivity o f extinction — disparity with the K-T. selective constraint on certain aspects of morphology. Failure of spatangoid disparity to increase substantially in the Late Cretaceous despite two near doublings o f diversity.

Some morphological boundaries may have been approached —such boundaries might be rooted on developmental or selective constraints.

Little change in spatangoid disparity No constraint. Extinction is through the K-T but considerable extinction morphologically nonselective. o f genera. First holasteroid and spatangoid appearances significantly add to the disparity of Jurassic disasteroids.

Origin of orders Holasteroida and Spatangoida represent a long jump in terms of the disparity o f Atelostomata. This signals enhanced morphological experimentation and flexibility. Less constrained developmental or selective controls.

Neocomian burst in Atelostomatan disparity A change in developmental flexibility is morphologically pervasive, affecting both should affect both categories of partial oral and aboral disparity. disparity, in contrast with ecological controls, which should not drive changes in oral disparity. Peak atelostomatan oral disparity in the Late Progressive increase in developmental Cretaceous. modularity, with reduction o f developmental constraints. Peak atelostomatan aboral disparity in the Early Cretaceous.

Functional themes established early and mostly elaborated thereafter. By itself this pattern is consistent with either early developmental or ecological flexibility (but see above).

Disparity of oral landmarks initially remains Increase in developmental modularity and the same from disasteroids to holasteroids decrease in developmental constraint. and spatangoids, but the latter two groups ultimately exceed the disasteroid level.


172 TABLE 3.3 (continued) PATTERNS


Essentially the same level of aboral disparity persists from disasteroids to holasteroids and spatangoids.

Because this happens despite the passage of time and holasteroid-spatangoid differences in habit and habitat, a historical constraint expressed as a developmental constraint persists.

Disparity of atelostomatan originations monotonically increases despite the plateau in standing disparity in the Cretaceous.

No constraint in disparity o f originations. Plateau in standing disparity is a dynamic balance between disparity o f originations and disparity of extinctions.

Disasteroid changes in disparity o f No constraint. Increase in disparity originations closely tracks diversity changes afforded byu originations is diffusive. Two-phase holasteroid standing disparity pattern, with apparent deceleration, is matched by disparity of originations.

Increasing constraint (developmental or selective) on the underlying generative process.

Spatangoid disparity of originations in the Campanian-Maastrichtian is not damped as standing disparity is.

No constraint. No reduction in the potential to generate novel morphologies.

Both holasteroids and spatangoids display high disparity in the Late Cretaceous but only holasteroids are hard-hit at the K-T.

No necessary relationship between disparity and sorting. Ecological correlates o f clade membership may be more important than disparity per se in this instance.



THE UNIVERSITY OF CHICAGO

THE MACROEVOLUTIONARY HISTORY OF DIVERSITY AND DISPARITY IN DISASTEROID, HOLASTEROID AND SPATANGOID HEART URCHINS VOLUME TWO

A DISSERTATION SUBMITTED TO THE FACULTY OF THE DIVISION OF THE BIOLOGICAL SCIENCES AND THE PRITZKER SCHOOL OF MEDICINE IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY COMMITTEE ON EVOLUTIONARY BIOLOGY

BY GUNTHER JENSEN EBLE

CHICAGO, ILLINOIS AUGUST 1997


C H A P T E R IV

THE RECOVERY OF DISPARITY IN SISTER CLADES: SPATANGOIDS AND HOLASTEROIDS AFTER THE K-T

Introduction

That clades recover after mass extinction events is a canonical statement in paleobiological studies (Erwin 1996; Harries et al. 1996). Recovery may not happen if decline is associated with or followed by persistence in refugia (Vermeij 1986), but it is the norm, however long it may take for ecosystems to reassemble or for low diversity lags to be erased. In fact, it is usually the expectation if vacation o f ecological space, at least at the lower taxonomic levels, is assumed. A burgeoning of interest in the dynamics o f recovery from mass extinction is in part a natural consequence of the now widely recognized fact that mass extinctions are distinct phenomena in the history o f life (Raup and Sepkoski 1982; Raup 1986), need have no relation with background times (Jablonski 1986a), and may contingently change evolutionary history in unanticipated directions (Jablonski 1986b; Gould 1989). In addition, biotic recoveries are usually rapid and may be accompanied by an intensification of diversification rates (see Sepkoski 1984, 1992; Jablonski 1994; Harries et al. 1996; MacLeod et al. 1997), with clear implications for the understanding of general controls on evolutionary radiations. While killing is the obvious immediate effect o f mass extinctions, their impact on survivors can only be understood by careful analysis o f their dynamics in the aftermath. A number of studies have now begun to tackle in more 173


174

detail the initial recovery following extinction events (e.g., Hansen et al. 1993; Erwin 1996; see Hart 1996), in a trend that promises to shed much light over the nature o f extinctions themselves, as well as have implications for the current biodiversity crisis (see Eldredge 1992). From the standpoint o f long-term consequences of extinction events and their relationship to large-scale evolutionary patterns, however, the full course of rediversification has always remained a fascinating issue. Most prominent are cases of clade incumbency and replacement —"double wedge" reciprocal waxing and waning of clades as well as more complicated patterns that may not always find a direct explanation in terms o f competition (Jablonski 1986b; Sepkoski 1996). Long-term, sustained or non reversed individualistic waning of clades following an extinction episode has received much less attention (with brachiopods as a prime example in individualistic accounts of the group's decline —e.g., Gould and Calloway 1980; but see Sepkoski 1996). Similarly, more symmetric patterns in pre- and post-extinction evolutionary dynamics, while implicit in modelling studies o f diversification (Stanley 1979; Sepkoski, 1984), have not been extensively documented or tested (but see Erwin et al. 1987, for marine animal families through the P-T; Hansen 1988, for U.S. G ulf Coast mollusks through the K-T; M iller and Sepkoski 1988, for bivalves; Sepkoski 1992, for marine animal genera at the OrdovicianSilurian; and Foote 1996a, for Paleozoic and post-Paleozoic crinoids). More empirical studies are needed before a truly general theory o f biotic recovery can be advanced (for initial attempts, see Harries and Kauffman 1990; Harries et al. 1996; Erwin 1996). An assessment o f changes in morphological disparity (the spread or spacing o f forms in morphospace) is clearly relevant to issues o f extinction, survival, and recovery as discussed above. Morphological disparity provides an independent metric of evolutionary flexibility and underlying dynamics in morphological space, complementing and refining evolutionary patterns recovered from study o f taxonomic diversity. Most disparity studies


175

have concerned large-scale patterns and the testing o f hypotheses about the unfolding of secular morphological trends. Explicit accounts of short- and long-term recovery from mass extinction in terms of disparity have now begun to be provided (Foote 1996a), complementing previous work in systematic and stratigraphic paleontology. A host of more specific questions arises once the comparison between taxonomic diversity and morphological disparity is recast around mass extinctions. Does disparity peak early on in the recovery process? Is there a disparity counterpart to macroevolutionary recovery lags? Do comparable pre-extincion and post-rebound high levels o f taxonomic diversity display the same degree of disparity? Are certain aspects of morphology more prone to increase or decrease in disparity after an extinction event? These questions provide an agenda that complements recent efforts to understand biotic recovery in terms of diversification dynamics. In this paper, I will address some o f these questions by tracking diversity and disparity in heart urchins (superorder Atelostomata) after the Cretaceous-Tertiary mass extinction. The Atelostomata comprises three echinoid orders, the Disasteroida (paraphyletic stem group), Holasteroida and Spatangoida. In a previous paper (Eble, ch. Ill), I discussed their morphological features and their history o f diversity and disparity during the Mesozoic, while including in the analysis the survivors o f the K-T extinction. Disasteroids appeared in the mid-Jurassic, gave rise to holasteroids and spatangoids in the Early Cretaceous, and disappeared in the mid-Cretaceous. Holasteroids and spatangoids exhibit comparable histories throughout the Cretaceous, but the K-T marks a severe decline in holasteroid diversity, which never recovered in taxonomic diversity (Fig. 4.1). Spatangoids have milder losses, recover to achieve their ultimate peak diversity in the Eocene, and slowly decline afterwards, partly as a result o f Eocene and Miocene extinction events (Fig. 4.1). An interesting sister-clade contrast thus can be placed in the framework o f disparity, and can be contrasted with the more commensurate holasteroid and spatangoid


176

#

Spatangoida

O Holasteroida 70 60 ’35

o 40 'C >

‘m 50 ■ U«