Global Biodiversity Patterns During the Cambrian Radiation Globale Biodiversitätsentwicklung Während Der Kambrischen Radiation

Home , Cambrian Stage 2, Cambrian Stage 3, Cambrian Stage 4, Drumian, Early Jurassic, Fortunian

Global Biodiversity Patterns during the Cambrian Radiation Globale Biodiversitätsentwicklung während der kambrischen Radiation

Der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg Zur

Erlangung des Doktorgrades Dr. rer. nat vorgelegt von Lin Na aus Heilongjiang, V. R. China

Erlangen 2016

Als Dissertation genehmigt von der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-Universität Erlangen-Nürnberg

Tag der mündlichen Prüfung: 2.11.2016. Vorsitzender des Promotionsorgans: Prof. Dr. Georg Kreimer Gutachter: Prof. Dr. Wolfgang Kießling Prof. Dr. Axel Munnecke Acknowledgments Foremost, I thank my supervisors Prof. Dr. Wolfgang Kießling. Only with his guidance, I could now fulfill my PhD study. From the day I decided to study abroad in Germany, Prof. Kießling gave me the greatest possible help with all aspects, including the design of research, analyses of data, publishing paper, and dissertation corrections. Along with his professional assistance and guidance, I would like to thank him for his help in daily life as well.

My most heartfelt gratitude goes to my fiancé Qijian Li (Jason) for his care, his help, his understanding, his patience, and keeping me company all along in Germany. I thank my parents for not holding me back and kindly tolerated my absence from their life. I thank my parents for their understandings and cares and also thank them for giving me financial support whenever I needed.

Without many other prerequisites and supports, this PhD study cannot be accomplished and presented here. I thank the following people: - Ex-supervisor in Nanjing Institute of Geology and Palaeontology: Guoxiang Li who gave me professional advices on taxonomy of SSF; - Leader of pre-project in Nanjing Institute of Geology and Palaeontology: Maoyan Zhu who kindly provided financial support for the field trips in South China; - Other colleagues in Nanjing Institute of Geology and Palaeontology for their help in organizing the field trip in China; - Christopher Heubeck and Michael Steiner from Freie Universität for their professional instructions in field trips of Kazakhstan and South China; - Colleagues from Museum für Naturkunde Berlin: Uta Merkel for her fantastic input of data from my study period, Heike Mewis, Melanie Hopkins, Melanie Tietje, and colleagues for all sorts of supports; - Colleagues in my Department from Friedrich-Alexander-Universität Erlangen- Nürnberg: Axel Munnecke, Birgit Leipner-Mata, Mihaela-Cristina Krause, Emilia Jarochowska, Manja Hethke, Patric Chellouche, Adam Kocsis, Christian Schulbert for all sorts of supports. Also, my work was supported and funded by Deutsche Forschungsgemeinschaft KI 806/9-1, embedded in the Research Unit “The Precambrian-Cambrian Biosphere (R)evolution: Insights from Chinese Microcontinents” (FOR 736). Summary The fossil record offers unique insights into the environmental and geographic partitioning of biodiversity during global diversifications. Global (gamma) diversity can be decomposed into local (alpha) and turnover (beta) components. Relationship of alpha and beta diversity in deep time during major evolutionary radiations has seldom been studied in paleobiology. Pathways of alpha- and beta- diversity in diversifying ecosystems notably differ depending on the relative role of various ecological interactions and environmental parameters. By combining the analysis of macroecological patterns of taxonomic composition and spatial distribution, we can assess the question which factors primarily controlled the diversification of life over geological times. Here I explore biodiversity patterns during the Cambrian radiation, the most dramatic radiation in Earth’s history. That the main pulse of the Cambrian radiation was in the early Cambrian has long been known, but defining the time of peak diversity has been hampered by problems with stratigraphic correlations. Many mechanisms of the Cambrian Explosion have been proposed, which can be subsumed as abiotic change, biotic interactions and genetic. This thesis aims at elucidating the role of biotic interactions using rigorous quantitative analyses of Ediacaran-Cambrian biodiversity dynamics. I use data from the Paleobiology Database (http://paleobiodb.org), which is a core infrastructure for modern paleobiological analyses. Based on well-established biostratigraphic correlation charts, I revised the stratigraphic assignments of all collections from the Ediacaran into the earliest Ordovician. I use sampling-standardized analyses of fossil occurrence data to derive accurate time series of alpha, beta, and gamma diversity. The vetted data demonstrate a prominent diversity peak in Stage 3, which is roughly equivalent to the Siberian Atdabanian and most of the Siberian Botomian. As diversity stabilized or declined subsequently, I conclude that Cambrian radiation only spanned the first three Cambrian stages in terms of biodiversity at the genus level. I assess how the overall increase in global diversity was partitioned between within-community (alpha) and between-community (beta) components and how beta

i diversity was partitioned among environments and geographic regions. Alpha and beta diversity increased from the Fortunian to Stage 3, and fluctuated erratically through the following stages. Changes of Cambrian gamma diversity were chiefly driven by changes in beta diversity. The pattern of diversity partitioning along onshore-offshore gradients demonstrate that there is no clear trend. Major ecological innovations appeared randomly with respect to bathymetric gradients and the variation of genus turnover between communities was not governed by environmental heterogeneity. My study elucidates that global biodiversity during the Cambrian radiation was driven by niche contraction at local scales and vicariance at continental scales. At local scales, the combined trajectories of alpha and beta diversity during the initial diversification suggest low competition and high predation within communities. At continental scales, the increase of beta diversity was controlled by the high rate of community turnover among adjacent continents. Beta diversity has similar trajectories both among environments and geographic regions, but turnover between adjacent paleocontinents was probably the main driver of diversification. This finding supports the general importance of plate tectonics in large-scale diversifications.

Zusammenfassung

Der Fossilbericht bietet einzigartige Einblicke in die ökologische und geographische Partitionierung der biologischen Diversität bei globalen Diversifikationen. Globale (Gamma) Diversität kann in unterteilt werden in die Diversität lokaler Vergesellschaftungen (Alpha) und Unterschiede zwischen Vergesellschaften (Beta). Das Verhältnis von Alpha- und Beta-Diversität über geologische Zeitskalen während großer evolutionärer Radiationen ist bislang selten paläobiologisch untersucht worden. Die zeitlichen Muster von Alpha- und Beta-Diversität in diversifizierenden Ökosystemen unterscheiden sich vor allem in der relativen Rolle von verschiedenen ökologischen Interaktionen und Umweltparametern. Durch die Kombination der Analyse von makroökologischen Mustern, taxonomischer Zusammensetzung und deren räumlicher Verteilung können wir die Frage nach den Faktoren beantworten, die maßgeblich die Diversifizierung des Lebens über die geologischen Zeiten kontrollierten. Hier untersuche ich Muster der Biodiversität während der kambrischen Radiation, der dramatischsten Radiation der Erdgeschichte. Dass der Hauptimpuls der kambrischen Radiation im frühen Kambrium war, ist seit langem bekannt, aber die eine genauere Festlegung wurde durch Probleme mit globalen stratigraphischen Korrelationen behindert. Viele Mechanismen der kambrischen Explosion sind vorgeschlagen worden, die als abiotische Änderung, biotische Interaktionen und genetische Faktoren subsummiert werden können. Diese Arbeit zielt darauf ab, die Rolle von biotischen Interaktionen mit Hilfe von probenstandardisierten Diversitätsdynamiken im Ediacarium und Kambrium aufzuklären. Ich verwende Daten aus der Paleobiology Database (http://paleobiodb.org), die eine zentrale Infrastruktur für moderne paläobiologische Analysen ist. Auf Basis etablierter biostratigraphischer Korrelationsdiagramme, wurde die stratigraphische Einstufung aller Sammlungen vom Ediacaran bis in das früheste Ordovizium überarbeitet. Ich benutze probenstandardisierte Analysen von Fossilvorkommen um daraus genaue Zeitreihen von Alpha-, Beta- und Gamma-Diversität abzuleiten. Die bereinigten Daten zeigen eine herausragende Diversitätsspitze in Stufe 3, die in etwa sibirischen Atdabanium und Botomium entspricht. Weil sich die Diversität anschließend stabilisierte

iii oder sogar verringerte, schließe ich, dass die kambrische Radiation sich nur über die ersten drei Stufen des Kambriums vollzog, zumindest in Bezug auf die Biodiversität auf der Gattungsebene. Ich beurteile, wie sich die Zunahme der globalen Diversität auf die Diversität innerhalb (Alpha) und zwischen (Beta) Lebensgemeinschaften aufteilte und wie Beta- Diversität zwischen Ablagerungsräumen und geografischen Regionen verteilt war. Alpha und Beta-Diversität erhöhten sich vom Fortunium zu Stufe 3 und schwankten ohne Trend in den folgenden Stufen. Änderungen der kambrischen Gamma-Diversität wurden vor allem durch Veränderungen in der Beta-Diversität beeinflusst. Die Muster der Diversitätsverteilung entlang von Onshore-Offshore-Gradienten, zeigen keinen klaren Trend. Wesentliche ökologische Innovationen erschienen zufällig in Bezug auf bathymetrische Gradienten und die Variation der Beta-Diversität war nicht durch Umweltheterogenität gesteuert. Meine Arbeit verdeutlicht, dass die globale Biodiversität während der kambrischen Radiation durch Nischenkontraktion auf lokaler Ebene und Vikarianz auf kontinentaler Ebene getrieben wurde. Auf lokaler Ebene lassen die kombinierten Muster von Alpha und Beta-Diversität während der Hauptradiation ein System mit wenig Konkurrenz und starkem Räuberdruck innerhalb der Lebensgemeinschaften vermuten. Auf kontinentaler Ebene wurde der massive Anstieg der Beta-Diversität durch Vikarianz zwischen Gemeinschaften benachbarter Kontinente gesteuert. Beta-Diversität war zwischen Lebensräumen und zwischen geographischen Regionen gleichermaßen bedeutsam, aber zunehmender Provinzialismus zwischen benachbarten Paläo- Kontinenten war wahrscheinlich die Hauptursache der Diversifizierung. Dieser Befund unterstützt die allgemeine Bedeutung der Plattentektonik in großen Diversifikationen der Erdgeschichte.

Table of contents

1 Chapter 1 Introduction ...... 1

1.1 Basic features of the Cambrian “Explosion” ...... 1

1.1.1 Uniqueness and timing ...... 1

1.1.2 Origin of modern marine ecosystems ...... 3

1.2 Paleoenviromental conditions ...... 6

1.2.1 Oxidation event ...... 6

1.2.2 Ocean chemistry and nutrient input ...... 8

1.2.3 Plate tectonics ...... 10

1.3 Biodiversity in the Cambrian ...... 12

1.4 Diversity partitioning: Alpha, beta and gamma diversity ...... 13

1.5 The worldwide database of Cambrian genus occurrences ...... 15

1.6 Purposes of this thesis ...... 16

2 Chapter 2: Stratigraphic resolution ...... 19

2.1 Introduction ...... 19

2.2 History of the Ediacaran and Cambrian time scales ...... 20

2.3 Biostratigraphic correlations in this study ...... 21

2.3.1 Ediacaran ...... 23

2.3.2 Cambrian ...... 24

2.4 Comparison between diversity trajectories by old and new time scales ...... 28

3 Chapter 3: Effects of sampling standardization ...... 33

3.1 Raw data ...... 33

3.2 Sampling bias ...... 34

3.3 Counting of taxa ...... 37

3.4 Subsampling ...... 42

3.4.1 Occurrence-based subsampling (Classical Rarefaction or CR) ...... 43

3.4.2 Sample-based subsampling (by-list subsampling) ...... 47

3.4.3 Shareholder quorum sampling (SQ) ...... 51

3.5 Correlations among CR, UW, OW, and SQ diversities ...... 52

3.6 Discussion ...... 53

4 Chapter 4: Diversity partitioning during the Cambrian radiation ...... 59

4.1 Introduction ...... 59

4.2 Methods ...... 60

4.2.1 Measure of alpha and beta ...... 60

4.2.2 Measure of origination and extinction rates ...... 60

4.3 Results ...... 61

4.3.1 Trajectories of alpha, beta, and gamma diversity ...... 61

4.3.2 The main driver of gamma diversity ...... 64

4.3.3 Origination and extinction rates ...... 67

4.4 Discussion ...... 67

5 Chapter 5: Onshore-offshore patterns in the evolution of Cambrian shelf communities: insight from alpha, beta, and gamma diversity...... 71

5.1 Introduction ...... 71

5.2 Methods ...... 71

5.3 Results ...... 74

5.3.1 Gamma diversity ...... 74

5.3.2 Alpha diversity ...... 74

5.3.3 Beta diversity ...... 76

5.3.4 Origination and extinction rates along onshore-offshore gradient ...... 80

5.4 Discussion ...... 81

6 Chapter 6: Escalatory ecological trends in Cambrian invertebrates and key role of predators in driving the Cambrian radiation ...... 87

6.1 Introduction ...... 87

6.2 Methods ...... 88

6.3 Results ...... 89

6.3.1 The role of predators for the Cambrian radiation ...... 89

6.3.2 Increase of proportional diversity of ecological groups in the Cambrian ...... 90

6.3.3 Low-competition mode ...... 97

6.4 Discussion ...... 100

7 Chapter 7: Did plate tectonic drive the Cambrian radiation? ...... 103

7.1 Introduction ...... 103

7.2 Methods ...... 104

7.2.1 Measure of geodisparity ...... 104

7.2.2 Measure of ecodisparity ...... 105

7.2.3 Tectonic reconstruction ...... 106

7.3 Results ...... 106

7.3.1 Geodisparity and ecodisparity during the Cambrian radiation ...... 106

7.3.2 Distance-decay pattern during the Cambrian ...... 107

7.3.3 Geographic biases ...... 112

7.3.4 Geodisparity among different environmental affinities ...... 113

7.4 Discussion ...... 118

7.5 Future outlook ...... 120

REFERENCES ...... 123

Appendix: [R] CODE ...... 139

vii

Get and clean data ...... 139

Diversity estimates ...... 140

Origination/extinction rates among onshore-offshore gradient ...... 143

Geodisparity ...... 146

Ecodisparity ...... 149

Escalation ...... 151

Model selection ...... 154

Paleomaps ...... 155

viii

1 Chapter 1 Introduction

1.1 Basic features of the Cambrian “Explosion”

1.1.1 Uniqueness and timing

The geologically sudden appearance of most modern animal phyla in the fossil record is perhaps the most significant evolutionary event in Earth history (Gould 1989, Valentine 1994, 2002, Valentine 2004) (see Figure 1.1). The Cambrian explosion of metazoans represents the greatest and most rapid expansion in higher-order animal disparity, with crown members of nearly every animal phylum originating within 10–20 million years (Knoll and Carroll, 1999; Erwin et al., 2011). However, stem group representatives of animal phyla and even some crown groups were already present in the Ediacaran (Knoll and Carroll 1999, Valentine et al. 1999, Valentine 2002, Marshall 2006, Budd 2008, Yin et al. 2013, Yin et al. 2015). The earliest known metazoan fossils are found in the Doushantou Formation, Guizhou, China during the interval between 590 and 550 mya (Knoll & Xiao 1999). A recent study has revealed that the sponge-like metazoan animals may have existed even 10 million years earlier (Yin et al. 2015). Molecular clock estimates confirm that the earliest members of many animal groups existed prior to the Cambrian. The combined evidences from molecular clocks and the fossil record indicate that the origin of high level animal groups occurred earlier than the increase in diversity and disparity in the Cambrian. The timing of the diversification of animal groups allows a clear distinction to be made between the Ediacaran Biota and the following main diversification of animal groups at the beginning of the Cambrian (Figure 1.1). The Ediacaran Biota is a suite of large complex multicellular organisms of considerable diversity and morphological disparity. In spite of a rich Ediacaran fossil record and a surprisingly complex community structure (Clapham and Narbonne 2002), global species richness was probably moderate in the range of a few hundred to a thousand (Valentine et al. 1978, Raup 1979, Valentine and Erwin 1983, Clapham et al. 2003, Droser et al. 2006, Laflamme et al. 2013, Darroch et al. 2015). In comparison, from the beginning of the Cambrian, animal life tended to

leave pervasive evidence of its existence, both as body fossils and trace fossils. From the first appearance of animals with durable skeletons to the first body fossils of trilobites, the radiation lasted about 20 million years. The high rate of diversification (Sepkoski 1992) is represented by the rapid increase in the number of species in the early Cambrian (Zhuravlev 2001, Marshall 2006, Maloof et al. 2010).

Figure 1.1 Animal diversity and disparity across the Proterozoic-Cambrian transition (after Marshall, 2006).

1.1.2 Origin of modern marine ecosystems

Modern marine ecosystems are characterized by a huge variety of organisms that interact and transfer mass and energy via a trophic cascade (Hairston et al. 1960, Carpenter et al. 1985, Terborgh and Estes 2013). To understand marine ecosystems during the Ediacaran- Cambrian period, Bambach et al. (2007) and Bush et al. (2011) introduced the concept of “ecospace” as representing the number of ecological and functional guilds within ecosystems. An ecospace cube is graphically represented by tiering, motility level, and preferred feeding mechanism of an organism. Due to small realized ecospace (Bambach et al. 2007, Xiao and Laflamme 2009, Bush et al. 2011), the ecosystems in the Ediacaran are considered underdeveloped. Since the beginning of the Cambrian, organisms have realized progressively more modes of life (Figure 1.2). The increase in ecological modes is suggested to be fundamental in permitting increases in diversity (Bambach 1983, 1985, 1993, Bambach et al. 2002), and thus in the development of level-bottom ecosystems during the Cambrian radiation.

Figure 1.2 Ecospace Occupation Cubes representing behavioural ecological guilds based on ecospace conception of Bush et al., 2011. Vertical axis represents vertical tiering in the water column and sediment, horizontal axis representing major feeding strategies, and the z axis represents the level of mobility. Dark blue squares represent occupied niches, with the diversity indicated within. Light blue squares indicate niches only represented by trace fossils. Pink cubes represent novel Cambrian innovations (after Laflamme et al., 2013).

1.1.2.1 The increase of predation

The Ediacaran-Cambrian transition also marks the emergence and formation of marine ecosystems with “modern” trophic structures (Butterfield 2007, Butterfield 2009, Butterfield 2011) (see Figure 1.3). Ediacaran organisms were mainly non-motile osmotrophs inhabiting the sediment surface (Laflamme et al. 2013) (also see Figure 1.2). Trophic capacities in the Ediacaran are simple, with no signs of predation, except for bore holes in some of the earliest skeletonized fossils, Cloudina, from the latest Ediacaran (Hua et al. 2003). In contrast, predation was probably widespread at the very beginning of Cambrian (Vermeij 1989). The expansion of mineralized exoskeletons in earliest Cambrian animals is thought to be one of the evolutionary expressions of escalation between prey and predator (Ginsburg and Jablonka 2010). Skeletal structures (e.g. spines) that served a defensive function can be observed in numerous groups such as halkieriids, Wiwaxia and lobopodians (Dzik and Krumbiegel 1989, Skovsted 2003, Conway Morris and Caron 2007, Liu et al. 2011). The drill holes found in mineralized skeletons (Conway Morris and Bengtson 1994) support predator–prey relationships in the Cambrian. Direct evidence of predators in the fossil record stem from later Cambrian strata and include fragmentary eodiscoid trilobite remains found in the gut of a fossil arthropod from the lower Middle Cambrian (Zhu et al. 2004) and bite marks and healed injuries in the exoskeletons of trilobites and trilobite-like animals (Babcock 2003, Skovsted et al. 2007).

Figure 1.3 The critical change of ecosystem across the Proterozoic-Cambrian transition from the pre- metazoan and Phanerozoic marine biospheres (after Butterfield, 2011).

1.1.2.2 The increase of tiering

At the same time tiering increased the vertical arrangement of organisms (Ausich and Bottjer 1982). The increase of trophic levels in marine ecosystem may have facilitated the increase in tiering within Cambrian benthic communities (Vannier et al. 2007). Only a small fraction of Ediacaran assemblages were deposit feeders or grazers (Xiao and Laflamme 2009). Cambrian level-bottom fauna have a wider range in tiering, mostly represented by sponge communities (Wu et al. 2014). There was little bioturbation in the Ediacaran. The earliest trace fossils stem from near the top of the Ediacaran showcasing burrows close to the sediment-water interface (Droser et al. 1999, Gehling et al. 2001, Droser et al. 2002, Jensen 2003). The increase in trace fossil diversity across the Ediacaran-Cambrian boundary implies that animals started to explore niches of greater depth rather than just remaining horizontally on the surface of the sediment. This increase of tiering is a prelude to the rapid increase in diversity during the Cambrian radiation (Mangano and Buatois 2014).

1.1.2.3 The rise of metazoan reef ecosystems

The history of Ediacaran-Cambrian reef building documents the replacement of predominantly microbial communities by those in which skeletal animals participated in reef construction, heralding a new reef ecosystem with elaborate trophic webs, complex organism interactions, increased niche partitioning, and high taxonomic diversity. Lower Cambrian archaeocyath-cyanobacterial reefs were taxonomically diverse and ecologically complex, and they were differentiated into distinct open surface and cryptic communities (Pratt et al. 2001). There are reports of Ediacaran metazoan reefs from the Nama Group, Namibia (Grotzinger et al. 2000, Wood 2011, Wood and Curtis 2015). These Cloudina- Namacalathus reefs formed open frameworks without a microbial component but with mutual attachment and cementation between individuals (Penny et al. 2014). In contrast to the succeeding archaeocyathan-dominated reefs, the diversity and complexity of communities in the Cloudina reefs is low. Reef-building metazoans represent an important ecological innovation whereby individuals collectively enhance feeding efficiency and gain protection from competitors and predation. Ecosystem engineering by new metazoan reef-builders played an important role in the further diversification during the rest of Paleozoic (Marenco et al. 2007).

1.2 Paleoenviromental conditions

Although the increase of both biodiversity and disparity during the Cambrian Explosion has long been recognized, the discussion of causes is ongoing. We can broadly categorize current explanations into: 1, changes in the abiotic environment, such as oxygen and nutrients, 2, changes in the biotic environment, such as novel biotic interactions, 3, factors in genetic or developmental capacity of the taxa involved (Erwin et al. 2011). It seems likely that no single factor can be the ultimate trigger or driver of the Cambrian radiation (Smith and Harper 2013). However, some of the abiotic changes were unique, such they may well explain the timing if not the mechanism of the Cambrian Explosion.

1.2.1 Oxidation event

Rising atmospheric oxygen levels were a crucial prerequisite for the emergence of animals (Knoll 2003, Canfield et al. 2007, Campbell and Squire 2010) and oxygen is also

considered as a potential “trigger” of the Cambrian explosion (Canfield and Teske 1996, Holland 2006, Liang et al. 2006, Acquisti et al. 2007, Canfield et al. 2007). There were two major increases of atmospheric oxygen levels over the course of pre-Cambrian era. The first one, termed the Great Oxidation Event, occurred about 2.3 billion years ago (Holland 1994, Farquhar et al. 2000, Bekker et al. 2004), after which oxygen levels were high enough to oxidize continental iron (“red beds”) and to preclude mass-independent fractionation (MIF) of sulfur isotopes (Bekker et al. 2004). Nevertheless, atmospheric oxygen levels were probably not higher than 1%-10% present atmospheric level (PAL) (Canfield 2005, Holland 2006, Kump 2008) over the entire Proterozoic until a second increase during the late Neoproterozoic period (Levin 2003, Fike et al. 2006, Campbell and Squire 2010), which is widely considered to have stimulated the evolution of macroscopic multicellular animals (the Ediacaran biota) and the subsequent radiation of calcified invertebrate (the Cambrian fauna) (Des Marais et al. 1992, Knoll and Carroll 1999, Catling et al. 2005, Canfield et al. 2007). Despite additional studies showing regional anoxic conditions in coeval deep- water sediments during the late Proterozoic period (Canfield et al. 2008, Johnston et al. 2013, Sperling et al. 2015), the evidence for widespread deep water oxygenation is compelling after the Gaskiers glaciation (580 Ma) (Canfield et al. 2007). The first known members of the Ediacara biota arose shortly after the Gaskiers glaciation, suggesting a causal link between their evolution and this oxygenation event. However, this oxygenation event only caused a limited increase of the ocean oxygen level (around 10% PAL oxygen level), which was sufficient for the evolution of Ediacaran biota in large and thin forms (maximized external surfaces) but restricts bilaterians to small, thin body plans (Runnegar 1982a, Runnegar 1982b). This might be a potential reason of the bilaterian paucity in Neoproterozoic fossil record. Once rising oxygen levels removed physical barriers to the evolution of large body size, the traces of earliest bilaterian animals are more likely to be preserved in the fossil record. Atmospheric oxygen level has been suggested to have exceeded 10% PAL in the beginning of the Cambrian (Levin 2003, Fischer 2016), permitting widespread oxidative metabolism required by Cambrian biota (Berkner and Marshall 1965, Rhoads and Morse 1971, Zhang and Cui 2016). A widespread oxygen deficiency in shallow marine environments at the

Ediacaran-Cambrian boundary is supported by a negative δ13C anomaly that occurred worldwide at the Ediacaran-Cambrian boundary (Kimura and Watanabe 2001). This oxygen deficiency in shallow water is considered to be responsible for the disappearance of Ediacaran-type fossils in Cambrian strata (Wille et al. 2008). However, after a brief duration of this anoxia event, increased proportions of carnivores in a community and more complex food web confirmed a pivotal role of increasing level of oxygen during the Cambrian radiation (Sperling et al. 2013). That the level of oxygen was still not near the average Phanerozoic oxygen levels might be the reason of low level of diversity in the Cambrian compared with the rest of the Phanerozoic (Saltzman et al. 2011a).

1.2.2 Ocean chemistry and nutrient input

The sudden onset of widespread biomineralization in the Early Cambrian is still an ongoing debate. The perturbation of seawater chemistry has been proposed to explain this event. Brennan et al. (2004) calculated the major ion composition of fluid inclusions from terminal Proterozoic (ca. 544 Ma) and Early Cambrian (ca. 515 Ma) marine halite and found that the concentrations of Ca in seawater increased approximately threefold 2- during the Early Cambrian. In contrast, over the same period, [SO4 ] decreased more than twofold, and [Mg2+] decreased by 20%. This increase in the seawater [Ca2+] may have created a chemical environment favorable for the initial biotic secretion of calcium carbonate and calcium phosphate hard parts, which have dominated marine biota ever since and marked the onset of biomineralization. Potential links between secular changes in the Mg/Ca-ratio of seawater and evolution have been explored at the Phanerozoic scale. The results suggest that the link is much less pronounced than usually admitted (Kiessling 2008) and it is impossible to predict the evolutionary success (in terms of diversity or relative abundance) of groups with a particular skeletal mineralogy, given the chemical state of the oceans (e.g., calcite or aragonite seas). However, the match between inferred ocean chemistry and biominerals in the Early Cambrian (Porter 2007) suggests that Cambrian seawater chemistry set the stage for the appropriate skeletal mineralogy. The addition of Ca and bicarbonate may have increased CaCO3 supersaturation levels facilitating the secretion of carbonate skeletons (Squire et al. 2006). The Early Cambrian surge in oceanic [Ca2+] was

likely the first increase following the rise of metazoans and may have spurred evolutionary changes in marine biota. Increasing Ca concentration in seawater (Brennan et al. 2004), along with increasing Sr, Fe and P delivery to the oceans from land, witnesses extensive mountain building, rapid erosion at the Ediacaran-Cambrian boundary (Squire et al. 2006, Campbell and Squire 2010), and enhanced continental crustal weathering (Walker et al. 2002). Associated with the transgression and expansion of shallow marine habitats, nutrients from large-scale continents weathering that resulted from long history of continental denudation in Proterozoic were surged in the Cambrian oceans (Peters and Gaines 2012). Enhanced nutrient concentrations could have facilitated algal growth leading to marked increase of oxygen production and increasing sediment fluxes; carbon burial would have prevented oxidation of organic matter. The Early Cambrian was particularly favorable for the formation of phosphorites. Phosphorite deposits occur in many parts of the world (Cook and Shergold 1984). This phosphorite period may have been the result of changes in sea level, oceanic upwelling, rifting and particular phases of seafloor spreading, transgression, elevated sea level, climate, plate tectonics, and continental drift. Moreover, the increase in phosphorus content of the marine could also support the process of phosphogenesis, which was consistent with the biomineralization event at the onset of the Cambrian radiation. Some authors have suggested that elevated nutrient concentrations were also directly responsible for the diversification of animals (Brasier 1992), but this is unlikely. High nutrient levels may support high population sizes and perhaps diversity, but originations are more likely in nutrient-depleted settings with small population sizes (Palumbi 1994, Kiessling and Aberhan 2007). However, unusually high concentrations of phosphorous and silica may explain the unusual taphonomy in the earliest Cambrian, which needs to be considered in large-scale diversity analyses (Brasier and Lindsay 2001, Porter 2004). Signor and Vermeij (1994) noted that suspension feeders were present in benthic communities from the beginning of the Cambrian. Freeman and Lundelius (2005, 2007) interpreted Cambrian brachiopods and molluscs as having planktotrophic larval development and suggested that planktotrophy is original in Metazoa. The evolution of

planktotrophy and the expansion of suspension feeders might probably be triggered by the same underlying reasons: enhanced nutrient supply and paleo-productivity (Nützel et al. 2006).

1.2.3 Plate tectonics

Plate tectonic movements can cause geochemical changes in oceanic and atmospheric reservoirs (Dalziel 1997, Brasier and Lindsay 2001, Kirschvink and Raub 2003, Dalziel 2014), and thus can be another potential trigger of the Cambrian radiation. Paleogeographic reconstructions at the Neoproterozoic-Phanerozoic transition highlight the breakup of the late Mesoproterozoic supercontinent Rodinia. Rodinia was formed 1.3- 0.9 billion years ago (Li et al. 2008). Its subsequent breakup resulted in the separation of Laurentia, Siberia, Baltica and Gondwana. Following Rodinia’s initial breakup, a younger supercontinent Pannotia has been suggested to have shortly existed near Ediacaran-Cambrian transition (Figure 1.4). The formation of Pannotia is characterized by the assembly of Laurentia and newly completed Gondwanaland (Dalziel 1997). The initiation of the breakup of Pannotia occurred at the same time as the radiation of metazoan life (Erwin and Valentine 2012). The breakup of Pannotia initiated the separation of Laurentia and amalgamated Gondwana along with the opening of the Iapetus Ocean, resulting in long-term sea level rise (Sauk transgression). During the Sauk transgression, the precipitation of carbonate sediments reached a Phanerozoic peak (Ronov et al. 1980, Walker et al. 2002). The disassembly of the supercontinent Pannotia has been suggested to be a potential trigger of the Cambrian radiation (Dalziel 2014). The separation of Laurentia and Gondwana initiated a major deep oceanic connection between the opening Iapetus Ocean basin and previously established paleo-Pacific Ocean basin (Dalziel 1997). The process involved increasing continental runoff and nutrient delivery.

Basal Ediacaran

Early Cambrian

Late Cambrian

Figure 1.4 Paleogeographic reconstruction maps from the basal Ediacaran to the late Cambrian (from http://jan.ucc.nau.edu/~rcb7/Cambrian.html).

1.3 Biodiversity in the Cambrian

The dramatic diversification of metazoans during the Ediacaran-Cambrian period has attracted wide attention (Stanley 1976, Bengtson 1977, Sepkoski 1978, 1992, Marshall 2006, Erwin et al. 2011, Erwin and Valentine 2012, Erwin 2015). There are, however, a number of problems in unravelling the original course of events that dictated the style of these evolutionary radiations. The most popular analysis of animal diversity during the Cambrian stem from Sepkoski’s compilations (Sepkoski 1979, Sepkoski 1981). When Sepkoski investigated taxonomic diversity in deep time, one of his observations defined the Cambrian radiation as the first of two intervals of logistic diversification within the Paleozoic (Figure 1.5). This interval marked an exponential increase of biodiversity at the family level across the Precambrian-Cambrian Boundary and a “pseudo-equilibrium” through the Middle and Late Cambrian, caused by diversity-dependent decrease in origination rate and increase in extinction rate. To explain the Cambrian radiation, understanding what controls the patterns of diversity is critically important. Most previous studies of disparity targeted Lagerstätten with exceptional preservation (see above). Temporal patterns of diversity, however, are usually based on less well-preserved sites and, in excluding stratigraphic singletons, they omit most of the information in Lagerstätten. I included all taxa in my study because I am not interested in true total diversity but a robust trajectory. This is especially true for Sepkoski’s compendia on the stratigraphic ranges of fossil marine families and genera (Sepkoski 1993b, Sepkoski 2002). According to Sepkoski’s genus compendium, diversity during the Cambrian rose in a stepwise fashion from the Nemakit-Daldynian (now Fortunian) to the Botomian (now Stage 4) “stages”, whereas disparity (measured by the number of classes) reached its maximum already by the Atdabanian (Marshall 2006). Thus the oft-observed pattern of disparity increase predating diversity rise (Foote 1997) seems to be confirmed in the Cambrian radiation. The Cambrian fossil record comprises different taphonomic pathways (Butterfield 2003). All current analyses on Ediacaran- Cambrian biodiversity suffer from being based on raw data without sampling standardization. Because diversity is strongly dependent of sampling intensity and preservation quality, these raw patterns might not trace relevant biologic information and tell true patterns of diversity.

Figure 1.5 Global taxonomic richness of marine animals through the Paleozoic Era based on the compilation of genera described in Sepkoski (1986). Diversity was calculated for intervals averaging 5.5 Ma in duration (Sepkoski, 1988). The Cambrian radiation witnessed the first of two intervals of logistic diversification within the Paleozoic.

1.4 Diversity partitioning: Alpha, beta and gamma diversity

A central question in biodiversity research concerns how species richness is distributed in space and where it varies and changes the most. By studying aggregate species distributions, macroecological studies can access hidden patterns and help reveal the main factors explaining these patterns. Whittaker (1960) decomposed regional (gamma) diversity into local (alpha) and turnover (beta) components. Alpha diversity measures species packing within a community. It is usually assessed by mean number of taxa in a communities or a particular habitat, and thus reflects how finely species divide ecological resources. Alpha diversity was originally considered point diversity and can be also measured based on some biodiversity indices such as Shannon-Wiener and Simpson. Beta diversity is used to assess the amount of turnover in species composition among

communities. There are different ways of assessing beta diversity, for example, by the simple equation or by directly measuring ecological distance. Beta diversity patterns along environmental gradients may indicate the degree of habitat specialization. Gamma diversity is the overall number of taxa in a region and was originally defined as combination of alpha and beta in Whittaker’s diversity classification (Whittaker 1960, Whittaker 1972). Figure 1.6 illustrates the relationship between alpha (within- community), beta (between-community or between-region), and gamma (within-region or within-global) diversity.

Site 1 α = 4 Site 2 1 α2= 3

α = 2 β Site 3 α3= 3 Site 4 4 β

Region A Region B

Figure 1.6 The illustration of the concept of alpha, beta, and gamma diversity.

Both alpha diversity and beta diversity are important to understand how diversity is distributed across space and reflect the group’s biogeographic history as well as the ecological opportunities and challenges it has encountered over the course of its diversification. Sepkoski (1988) was the first who applied Whittaker’s concept to global

diversity patterns in the Paleozoic. He concluded that beta diversity rose from the Cambrian to the Ordovician and remained roughly constant through the later periods. A more recent study focused on reefal communities in the Cambrian and concluded that beta diversity was a major driver of the global increase of reef-builder diversity (Zhuravlev and Naimark 2005). Both alpha and gamma diversity have long been discussed regarding to evolution of global marine diversity (Bambach 1977). The regularly preserved skeletal faunas, suggest that community structure was predominantly simple in the Cambrian, that is, rank-abundance distributions followed a geometric series or zero-sum multinomial distribution (Wagner et al. 2006). Although the latter study did not extend into the Ediacaran, community structure in this period appears to be comparable, even though a surprisingly high between-community variation has been observed (Clapham et al. 2003). In this work I focus on the interplay of alpha and beta diversity in the Cambrian Radiation at multiple scales.

1.5 The worldwide database of Cambrian genus occurrences

Diversity curves can vary significantly when calibrated on the basis of different literature compilations and methods of counting and sampling-standardization. Previous studies were either based on Sepkoski’s (2002) global compilation of geologic ranges of marine genera or personal databases (e.g. Sepkoski’s and Zhuravlev’s compilations). Although many synoptic studies and large data compilations exist on the Cambrian “explosion”, the primary paleontological data are scattered in the literature and in small databases of individual researchers. In order to reach a better understanding of global biodiversity patterns and dynamics, this information must be compiled, cleaned and analyzed using modern methods of analytical paleobiology. In contrast to synoptic, range-through based compilations (Newell 1959, Valentine 1969, Sepkoski 1978, 1979, 1984, Zhuravlev and Riding 2001), the Paleobiology Database (PaleobioDB, https://paleobiodb.org) documents individual fossil collections, containing lists of genera, subgenera, and species, and, where available, abundance data. Any taxon may have multiple recorded occurrences in the database. Some collections pertain to entire outcrops or even basin-wide stratigraphic units, but most correspond to

single localities and many to bed-scale inventories (Table 1.1). The Paleobiology Database is currently the largest, publicly available dataset on Ediacaran-Cambrian fossil occurrence data. The data can be downloaded with the Paleobiology Database application programming interface (Peters and McClennen 2015).

Table 1.1 Counts of fossils in different collecting scales per stage from the Paleobiology Database.

single local local Multiple recorded Collecti regional regional Intervals locality section bed occurrence no. on no. section no. bed no. no. no. no. Avalon 104 85 53 16 11 3 2 assemblage White Sea 113 67 46 10 6 2 3 assembalge Nama 65 45 37 3 3 1 1 assemblage Fortunian 727 294 218 29 44 2 1 Cambrian 1449 282 87 64 118 6 7 Stage 2 Cambrian 5876 1015 597 127 277 11 3 Stage 3 Cambrian 1819 612 297 75 225 6 9 Stage 4 Cambrian 3831 863 409 86 333 14 21 Stage 5 Drumian 1583 455 237 37 169 7 5 Guzhangian 2992 553 189 62 292 4 6 Paibian 1635 477 189 39 235 11 3 Jiangshanian 2447 670 274 64 322 4 6 Cambrian 1819 469 229 53 178 4 5 Stage 10 Tremadocian 3651 1091 625 98 313 3 52

1.6 Purposes of this thesis

The potential of Whittaker’s concept to provide insights into evolutionary processes has been demonstrated in a few paleobiologic studies (Miller 1997, Harper 2006). Although Whittaker’s concept is widely used and the principal levels of biodiversity are well

explored in time and space, there were few analyses explicitly exploring the diversity partitioning during the Cambrian Explosion. The question as to how the Cambrian radiation was partitioned within and between communities is old (Sepkoski 1988) but still unresolved. Particularly important is how the evolutionary radiation in the Cambrian is manifested in different environmental settings and the global diversification was driven by a higher species richness in local communities (alpha diversity) or rather by an increase of between communities (beta diversity)? Here I reveal the patterns and causal processes of animal community assembly and disassembly over Ediacaran-Cambrian times. Another basic question is when and where did the Cambrian radiation occur? In terms of global diversity earlier analyses suggested that the main increase was in the Atdabanian with no significant increase thereafter (until the Ordovician) (Zhuravlev and Riding 2001), which is quite similar to the pattern of disparity. Regional patterns may provide interesting departures from this general trend. Finding systematic relationships of diversity pattern with latitude and sedimentary environments is thus a focus of this thesis. For example, the oldest Ediacaran macrofossils are known from Newfoundland in high paleolatitude, deep-water siliciclastic environments, whereas the skeletal fauna marking the onset of the Cambrian radiation appears to have evolved in low-latitude, shallow- water, and calcium carbonate settings. These raw data, of course, cannot be taken at face value without considering taphonomic and sampling biases. I begin this dissertation with a revision of Ediacaran-Cambrian stratigraphy and discuss the effects a volatile time scale and stratigraphic resolution on the estimates of the Cambrian diversification. I will then discuss the effects of sampling standardization on the estimation of Cambrian marine diversification. Next, I look into the diversity partitioning along onshore-offshore gradients and reveal global diversity patterns in the Cambrian. Finally, I focus on the triggers of the Cambrian radiation by investigating how the Cambrian diversification is manifested ecologically and geographically.

2 Chapter 2: Stratigraphic resolution

2.1 Introduction

It has been long acknowledged that changes in the precision and accuracy of stratigraphic correlation have a great impact on the estimation of global diversity patterns (Sepkoski 1997). Our understanding of the history of global diversity changes with the refinement of global stratigraphy. For instance, the series-level data available several decades ago suggested a mass extinction at the end of the Cambrian but did not distinguish the explosive diversification in the Cambrian from the Ordovician radiation (Sepkoski 1979). It was until later when finer resolution at the stage level was developed with the aid of the International Commission on Stratigraphy. The stage-level data allowed the recognition of successive extinction events throughout the Late Cambrian and a distinct quiescent interval of approximately steady diversity between the Cambrian Explosion and the Ordovician radiation (Sepkoski 1979, 1995). Earlier time series studies on this subject were famously conducted by Sepkoski in his papers on Phanerozoic biodiversity patterns (Sepkoski 1978, 1979, Sepkoski et al. 1981, Sepkoski 1992). Sepkoski used the Siberian stages for the Lower Cambrian, an informal subdivision of Middle Cambrian and the North American stages for the Upper Cambrian. Notably, in one of his following papers of which he focused on the partitioning of alpha, beta, and gamma diversity through the whole Paleozoic (Sepkoski 1988), he subdivided the Cambrian into eight intervals based on North American biozones and “stages” to attain a high stratigraphic resolution. This biostratigraphic framework worked well because most of assemblages in his data set were derived from the Laurentian portion of North America. Early Cambrian diversity analyses in some later publications are at a stunningly high stratigraphic resolution (e.g. Li et al. 2007, Maloof et al. 2010, Zhuravlev 2001). The stratigraphic resolutions used in these papers tend to separate the Early Cambrian in terms of regional bio- or chrono-stratigraphic frameworks. Although regional subdivisions can be very fine, the biozones can rarely be correlated globally such that diversity curves may differ markedly. For example, Li et al (2007) analyzed a data set of 876 genera reported from the lower Cambrian of south China for biodiversity changes 19

and concluded that the early Cambrian metazoan diversity had two pulses in the regional Meishucuan and Qiongzhusian stages, respectively. Another analysis based on the data calibrated by Russian stages and zonal scales for the Early and early Middle Cambrian shows two peaks during the Tommotian and Botomian stages (Zhuravlev and Riding 2001), which were slightly earlier than the ones found in South China, suggesting that global stratigraphic correlations are an issue. To revisit the diversity patterns of the Cambrian radiation using a large fossil data set, the priority is to attain a solid stratigraphic framework for binning. The standard subdivision of the Phanerozoic in the Paleobiology Database separates 49 intervals of roughly equal duration by grouping short geographic stages when necessary. This stratigraphic resolution was used to derive the basic patterns of the Phanerozoic biodiversity (e.g. Aberhan and Kiessling 2012, Alroy et al. 2008, Kiessling 2008). Most recent analyses of long-term patterns have chosen to ignore the whole Cambrian period (e.g. Alroy et al. 2008, Hopkins et al. 2014, Kiessling et al. 2008, Nürnberg and Aberhan 2015), particularly the traditional Early Cambrian, because of the difficulty to include a large number of collections from this interval relative to the later intervals in the Phanerozoic, as well as the fact that the dynamics in the Cambrian are strange as long intervals lead to high turnover. The stratigraphic resolution for the Cambrian period in the Paleobiology Database is variable, ranging from single trilobite zones to epochs. The information is often imprecise (e.g. epoch rather than stage level) and inaccurate. The formal subdivision of the Cambrian in the PaleobioDB is still based on the traditional Siberian stages for the former Early Cambrian and the Laurentian subdivision of the Upper Cambrian. The Ediacaran period is not subdivided at all. Therefore, the first task of this study is to define objective criteria for the assignment of fossil collections in the database according to the updated international chronostratigraphic chart.

2.2 History of the Ediacaran and Cambrian time scales

Global correlation of Cambrian biozones has always been hindered by the strong provincialism (Landing 1992, Lindsay et al. 1996). The traditional “lower”, “middle”, and “upper” Cambrian was formalized by Charles Walcott based on Trilobite faunas

(Walcott 1890). The stratigraphic nomenclature of the Cambrian has gone through considerable revisions since then (Babcock and Peng 2007). The Lower, Middle, and Upper Cambrian have been provisionally replaced by the Terreneuvian Series, Series 2, Series 3, and the Furongian Series. The Terreneuvian is roughly equivalent to the pre- trilobite portion of the Early Cambrian, whereas Series 2 consists of the remainder of the former Early Cambrian, after the first appearance of trilobites. The base of Series 3 is close to the Early-Middle Cambrian boundary in many parts of the world, and Series 3 is an expanded version of the Middle Cambrian, incorporating the lower part of the traditional Upper Cambrian as well. The Furongian Series is a reduced version of the Upper Cambrian. The 2009 ICS time scale had no formal subdivision of the Ediacaran period. Only two of four Cambrian series and only four of ten stages are formally named. The updated Cambrian chronostratigraphic time scale given by 2016 International Commission on Stratigraphy (ICS) contains four series and ten stages, but only two series and five stages have formal names, with the others still undefined (Ogg et al. 2016). That is, the lower boundaries of five stages have been successfully nailed by Global Boundary Stratotype Section and Points (GSSP), and thus have provide standard criteria for international correlation of the Drumian, Guzhangian, Paibian, and Jiangshanian stages. The pre- trilobitic part of the Cambrian is traditionally subdivided based on small shelly fossils (SSF). SSF are especially common in shallow water settings of the Yangtze Platform and are used for the regional biostratigraphic division and correlation of south China. For example, the mollusk Watsonella crosbyi has achieved a worldwide distribution in the sub-trilobitic strata of the lower Cambrian, including North America (southeastern Newfoundland), Siberia, south China, Mongolia, and Kyrgyzstan. Watsonella crosbyi has thus been suggested as a potential candidate of GSSP (Global Boundary Stratotype Section and Points) index fossil for the base of the Cambrian Stage 2 (Li et al. 2011).

2.3 Biostratigraphic correlations in this study

Figure 2.1 Biostratigraphic correlation chart used to assign collections in the Paleobiology Database to 14 international stage-level intervals. The biostratigraphic separation for the late Ediacaran period is based on ref. (Narbonne et al. 2012). The correlation part for the Cambrian is based on body fossil zonal schemes of principal regions. Siberia: refs. (Geyer and Shergold 2000, Peng et al. 2012, Landing et al. 2013). South China: refs. (Zhu et al. 2001, Steiner et al. 2007) for SSF zones; ref. (Peng et al. 2012) for trilobite zones. Laurentia: ref. (Sepkoski 1997, Babcock et al. 2011). Australia: refs. (Young and Laurie 1996, Sepkoski 1997, Kruse et al. 2009). Baltica: refs. (Sepkoski 1997, Geyer and Shergold 2000).

I use the correlation chart showed in Figure 2.1 to assign collection to stages. The criteria I use to establish the correlation chart are based on the informal subdivision of the late Ediacaran ecosystem (three distinguished faunal assemblages) and well-established subdivision of the Cambrian period (four Epochs and ten stages).

2.3.1 Ediacaran

The Ediacaran spans the time from 575 to 542 Myr and marks the first appearance of complex, macroscopic organisms in earth history (Narbonne 2005) and the first metazoan radiation prior to the “Cambrian Explosion”. Apart from acritarchs, algae and related microfossils, which are outside the scope of this study, three assemblages of Ediacara- type fossil impression, have been recognized: Avalon assemblage, White Sea assemblage and Nama assemblage (Waggoner 2003, Narbonne 2005, Xiao and Laflamme 2009). Each assemblage exhibits a major evolutionary innovation and represents a significant development in the evolution of life (Xiao and Laflamme 2009). Because of their distinctive character, they can be readily applied to the subdivision and biostratigraphic correlation of the late Ediacaran time. The Avalon assemblage (575-560 Ma) is characterized by rangeomorphs such as Charnia, Bradgatia, Fractofusus, Arboreomorpha, and Triradialomorpha. The Avalon organisms are known only from deep-water settings in Newfoundland (Narbonne and Gehling 2003, Hofmann et al. 2008), England (Brasier and Antcliffe 2009), and the Mackenzie Mountains of northwestern Canada (Narbonne and Aitken 1990). The appearance of these fossils is perhaps related to a noticeable rise in oxygen after the Gaskiers glaciation and nutrient-rich waters supported by upwelling (Laflamme et al. 2013). The White Sea assemblage (560-550 Ma) is represented by fossils from the White Sea and Urals of Russia (Grazhdankin 2004), the Flinders Ranges of south Australia, the Wernecke and Mackenzie Mountains of northwestern Canada (Narbonne and Aitken 1990) and the Olenek Uplift of Siberia (Grazhdankin et al. 2008). Most of the White Sea organisms are restricted to shallow water environments. Apart from few rangeomorphs, the White Sea assemblage contains new developmental plans including Erniettomorpha, Dickinsoniomorpha, Kimberellomorpha, Bilaterialomorphs, Tetraradialomorphs, and

Pentaradialomorphs. Almost all higher groups in the White Sea assemblage have achieved a global distribution (Laflamme et al. 2013). The Nama assemblage (550-542 Ma) is the youngest one that is known from the Kuibis and Schwarzrand subgroups of Namibia (Grotzinger et al. 1995, Narbonne et al. 1997, Grotzinger et al. 2005), the deep-water siliciclastics and carbonates of the Khatyspyt Formation in central Siberia (Grazhdankin et al. 2008), the Mojave Desert of western USA (Hagadorn et al. 2000, Hagadorn and Waggoner 2000), shallow-water carbonates in Oman (Amthor et al. 2003), the Dengying Formation of China (Hua et al. 2005). The fossils in the Nama assemblage are mainly rangeomorphs and erniettomorphs. Of most biostratigraphic significance is the worldwide appearance of calcified macrofossils, Cloudina and Namacalathus in the Nama assemblage.

2.3.2 Cambrian

2.3.2.1 Terreneuvian (Fortunian-Stage 2)

The Terreneuvian series comprises two stages, the Fortunian and an overlying stage provisionally called Cambrian Stage 2. The Fortunian is the lowest stage of the Terreneuvian Series and the Cambrian System. The GSSP is at Fortune Head in eastern Newfoundland and defined by the first appearance datum of the trace fossil Treptichnus pedum. The first appearance datum (FAD) of T. pedum defines the base of the T. pedum ichnozone, representing the appearance of complex substrate-disturbing behaviors by epifaunal and infaunal animals. However, the ichnofossil is not generally accepted as a global time marker. The base of Stage 2 has no GSSP and is provisionally defined based on the FAD of a small shelly fossil (SSF) or an archaeocyath (Babcock 2005). The traditional Lower Cambrian on the Siberian platform is subdivided into regional stages namely the Nemakit-Daldynian, Tommotian, Atdabanian, Botomian and Toyonian based on SSF and archaeocyathan assemblages; whereas, the lower Cambrian on the Yangtze platform is subdivided into the Meishucunian, Qiongzhusian, Qianglangpuan, and Longwangmiaoan stages based on SSF assemblages. The Nemakit- Daldynian is contemporaneous to the global Fortunian stage and zoned into the Anabarites trisulcatus and Purella antiqua biozones, which can be correlated with Anabarites trisulcatus-Protohertzina anabarica and Paragloborilus subglobosus-Purella

squanulosa assemblage zone of coeval facies of lower part of traditional Meishucunian stage in south China respectively (Khomentovsky and Karlova 1993). Khomentovsky and Karlova (1993) also proposed that Nochoroicyathus (formerly referred to Aldanocyathus) sunnaginicus Zone marks the start of Tommotian stage, which is generally equivalent with the Meishucunian stage in south China and global Cambrian stage 2. The traditional Meishucunian stage in south China covers an interval occupied by abundant and diverse small shelly fossils and consists of mainly three SSF assemblage biozones in most local regions. Watsonella crosbyi has been suggested as a potential candidate of global correlation for the beginning of the Stage 2 (Li et al. 2011).

2.3.2.2 Series 2 (Stage 3-Stage 4)

Series 2 is characterized by the appearance and rapid rise of trilobites. In Series 2, the Cambrian fauna is suggested to have experienced an explosive diversification of metazoans, particularly polymerid trilobites and archaeocyaths. The bases of both Stage 3 and Stage 4 have not been defined. The boundary between stage 2 and stage 3 is roughly positioned at the FAD of trilobites, and thus divides the lower Cambrian into a sub- trilobitic series and a trilobite-dominated series. This criterion of subdivision is applicable in Laurentia (Palmer 1998a), in south China (Peng 2003), in western Gondwana (Geyer and Landing 2004), and in Avalonia (Landing 1991). Cambrian Series 2 is also marked by a biogeographic distinction between two main faunal provinces (Peng et al. 2012). One is the Redlichiid province of Gondwana, characterized by endemic redlichiids, pandemic ellipsocephaloids, and ecodiscids; the other is Olenellid province of mostly Siberia, Laurentia, and Baltica, characterized by endemic olenellids, pandemic ellipsocephaloids, and ecodiscids. On account of this strong provincialism, trilobites tend to be a poor correlation tool in Series 2 relative to trilobites from younger epochs. The first appearance of trilobites also appears to be diachronous in different regions and hence it is difficult to use the FAD of trilobites as time marker. Even the few genera and species that are thought to be globally distributed are of limited use for global correlation due to limited local stratigraphic range (Palmer 1998b). Other fossils associated with endemic trilobite assemblages, such as archaeocyaths, hyoliths, molluscs, and other small shelly fossils, either have limited occurrences or strong affinities to particular deposition

environments (Geyer and Shergold 2000). The earliest occurrence of trilobites recognized in Siberia and western Gondwana (Morocco) includes Profallotaspis jakutensis, which occurs shortly above the base of Siberian Atdabanian stage (Astashkin and Rozanov 1991), whereas fallotaspidids trilobites appear later in tropical western Laurentia. The oldest identifiable trilobites in Baltica belong to the Schimidtiellus mickwitzi biozone. The Parabadiella assemblage zone contains the oldest trilobites in eastern Yunnan of south China and marks the base of the Qiongzhusian stage. The boundary of stage 3 and stage 4 is hard to define. The base of Stage 4 has been suggested at a level roughly corresponding to the base of the Dyeran stage of Laurentia (Palmer 1998a), and the Canglangpuan stage of south China. The Botomian correlates approximately to upper Stage 3 and the Toyonian correlates approximately to lower-middle Stage 4 (Peng et al. 2012). The Stage 4 spans almost the entire range of redlichiids and olenellids, both of which became extinct at the termination of Cambrian Series 2 (Zhao et al. 2008).

2.3.2.3 Series 3 (Stage 5-Guzhangian)

The base of the provisional Series 3 and Stage 5 has been traditionally referred to as Lower-Middle Cambrian boundary. This concept of Lower-Middle Cambrian is still popularly used in Siberia, Laurentia, and south China. Most recent investigations suggest that the FAD of Oryctocephalus indicus is a suitable candidate for defining the base of Series 3 and stage 5 (Peng et al. 2012) and has good potential for international correlation (Geyer and Peel 2011, Naimark et al. 2011). Oryctocephalus indicus biozone correspond closely to the Lower-Middle Cambrian boundary of Laurentia (Palmer 1998a), northern Siberia (Korovnikov 2006), and eastern Gondwana (Peng et al. 2012). The base of the Drumian is defined by a GSSP coinciding with the FAD of an agnostoid trilobite Ptychagnostus atavus, which has achieved a worldwide distribution. Together with the first occurrence of the preceding Ptychagnostus gibbus and the first occurrence of the succeeding Ptychagnostus punctuosus, Ptychagnostus avatus has been used as a zonal index fossil in deposits of Laurentia, Baltica and Gondwana (Geyer and Shergold 2000, Peng and Robison 2000).

The base of the Guzhangian is defined by a GSSP coinciding with the FAD of a cosmopolitan agnostoid trilobite Lejopyge laevigata that occurs in Siberia, Laurentia, Baltica, and Gondwana (i.e., south China, Australia, and Avalonia). Notably, the traditional index fossil Agnostus pisiformis for the base of upper Cambrian in Baltica has been reported from this biozone in Siberia (Geyer and Shergold 2000), suggesting that the base of Agnostus pisiformis zone in Olenus stage of Baltica should be restricted in Guzhangian, specifically, the base of Linguagnostis reconditus biozone recognized in south China (Ahlberg and Ahlgren 1996, Ahlberg 2003).

2.3.2.4 Furongian (Paibian-Stage 10)

The Furongian is the uppermost series of the Cambrian system. It is traditionally referred to as the upper Cambrian. The GSSP for the base of the Furongian and Paibian coincides with the FAD of agnostoid trilobite Glyptagnostus reticulatus. The Furongian epoch has experienced a series of extinction events among polymerid trilobites in Laurentia, which define the classical biomeres, namely the Marjumiid biomere, the Pterocephaliid biomere, and the Ptychaspid biomere. The Glyptagnostus reticulatus zone marks the base of the Paibian. The index fossil Glyptagnostus reticulatus has been recovered from Russia, China, Canada, Kazakhstan, Australia, Norway, Sweden, Denmark, USA, UK, and Argentina (Geyer and Shergold 2000). It has been used as zonal index fossil in Siberia, south China, Laurentia and Australia. The interval represented by this biozone marks the base of the Franconian stage in Laurentia (Palmer 1984), the base of the Idamean Stage in Australia, the base of Waergangian in south China, the base of the Maduan stage in Siberia, and the base of Homagnostus obesus zone in the Olenus stage of Baltica. The level also corresponds to a strong, globally recognized carbon isotope (δ13C) shift, namely the Steptoean Positive Carbon Isotope Excursion (SPICE) (Saltzman et al. 2000). The base of the Jiangshanian is defined by a GSSP marked by the first occurrence of the cosmopolitan agnostoid trilobite Agnostotes orientalis, which usually co-occurs with the polymerid trilobite Irvingella in many regions. Their concurrent range has been used as stratigraphic marker in South China, Siberia, and Laurentia. Particularly, Irvingella has achieved extensive paleogeographic distribution (Geyer and Shergold

2000), which permits correlation among regions. Occurrences of Irvingella have been recovered in Russia, China, Sweden, UK, Australia, Kazakhstan, South Korea, Vietnam, and USA. This horizon ranges across the Steptoan-Sunwaptan stage boundary in Laurentia, and closely corresponds to the base of the Iverian stage in Australia, the base of the Taoyuanian stage of south China, the base of the Parabolina spinulosa zone of the Olenus stage in Baltica. The base of the uppermost stage (Stage 10) is suggested to be close to the FAD of an agnostoid, Lotagnostus americamis (Babcock 2005). The species recovered from this level appears to be widely identifiable and recognizable in all major paleocontinents of Cambrian. This horizon is important for correlations between Siberia, south China, Laurentia, and Baltica. Figure 2.1 shows the current stratigraphic correlation between Siberia, south China, Australia, Laurentia, and Baltica. The stratigraphic criteria used here led to a global recognition of 41 biozones in 14 stage-level time intervals from the late Ediacaran to the earliest Ordovician (parts highlighted in blue and green in Figure 2.1). The stage- level bins correspond to the chronostratigraphic time scale given by 2016 ICS (Ogg et al. 2016). The standard of biozonal bins were defined by the combination of biostratigraphic schemes for the Fortunian and stage 2 of the Cambrian in South China, biostratigraphic scheme for stage 3 and stage 4 of the Cambrian in Russia, and biostratigraphic scheme for the rest of the Cambrian in South China. These regional biozones are most often referenced to discuss stratigraphic correlation in published literatures.

2.4 Comparison between diversity trajectories by old and new time scales

The raw, genus-level diversity trend is presented in Figure 2.2, counted on the basis of both traditional and present timescales. The curves in the plot show only the amount of raw data available for analysis and have no necessary biological meaning. Log scale is used because the variables are differed in orders of magnitude. I tabulated the number of genera along with occurrences and fossil collections through the Ediacaran-Cambrian periods. The diversity curves based on the present timescale shows several different features of the ones of traditional timescale, as shown in Figure 2.2: (1) the relatively

coarse, regional stage-level stratigraphic treatment appears to inflate the level of standing diversity during the lengthy Ediacaran period, relative to its depiction at the finer, stage- level. Diversity curves of the new timescale shows a more dramatic increase in number of genera through the Ediacaran-Cambrian boundary than that of the old ones; (2) the number of genera in the old timescale declines prominently in the Toyonian stage, while the contemporaneous drop (Stage 4 to Stage 5) is less severe in the new timescale; and (3), more details about the post-Stage 4 dynamics are evident in the new time scale. Also there is only one peak in the new one but a plateau in the old one. These findings suggest that temporal resolution has a significant effect on the pattern of diversity during the studied interval. Throughout this thesis, I will attempt to use the modern terminology, but in a few cases, for example when referring to older citations, I will use the traditional stratigraphy when it was not immediately evident in which series the occurrences of interest belong.

occurrences collections genera Counts per interval per Counts

N-D Tomm At Bo T M.Camb U.Camb Ediacaran Cambrian 5 10 50 200 1000 5000

560 540 520 500

Age (Ma)

occurrences collections genera Counts per interval

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian 5814 5 10 50 100 500 5000

560 540 520 500 480

Age (Ma)

Figure 2.2 Data inventory in the PaleobioDB through Ediacaran-Cambrian times (37994 occurrences). Raw counting is show in Error! Reference source not found.. Note log scale of y-axis. Ma, million years ago. Abbreviation of Stages: (A), N-D, Nemakit-Daldynian; Tomm, Tommotian; At, Atdabanian; Bo, Botomian; T, Toyonian; M. Camb, Middle Cambrian; U. Camb, Upper Cambrian; (B), Ava, Avalon assemblage; Whs, White Sea assemblage; Nam, Nama assemblage; For, Fortunian; St2, Stage 2; St3, Stage 3; St4, Stage 4; St5, Stage 5; Dru, Drumian; Guz, Guzhangian; Pai, Paibian; Jia, Jiangshanian; St10, Stage 10; Tre, Tremadocian (Ordovician).

Table 2.1Data inventory by stratigraphic interval. Coll. represent counts of collections; Occu. represent counts of occurrences; Gen. represent number of genus.

Age Age Stage Coll. Occu. Gen. Stage Bottom Middle Abbr. (Ma) (Ma) Avalon Assemblage Ava 575 567.5 76 151 62 White Sea Assemblage Whs 560 554.5 67 179 66 Nama Assemblage Nam 549 545 45 126 61 Fortunian For 541 535 195 932 205 Stage 2 St2 529 525 284 1785 335 Stage 3 St3 521 517.5 1013 7021 1153 Stage 4 St4 514 511.5 612 2393 574 Stage 5 St5 509 506.75 863 4472 641 Drumian Dru 504.5 502.5 455 2063 480 Guzhangian Guz 500.5 498.75 553 3451 459 Paibian Pai 497 495.5 477 2008 373 Jiangshanian Jia 494 491.75 670 2802 355 Stage 10 St10 489.5 487.4 469 2237 418 Tremadocian Tre 485.4 482 1090 4461 811

3 Chapter 3: Effects of sampling standardization

3.1 Raw data

Fossil occurrence data of Ediacaran to Tremadocian age have been entered into the Paleobiology Database (PaleobioDB, http://paleobiodb.org) by PBDB contributors worldwide and were downloaded alongside all other occurrences on 23.05.2014. This dataset comprises 7,117 collections and 39,737 taxonomic occurrences with information of taxonomy, paleogeography, geology and ecology. The download used all occurrence data from the marine invertebrate working group except for vertebrates and genera listed in quotation marks, or qualified as “?”, “cf.”, “aff.” or “ex gr.”. Data were not filtered to include only marine invertebrate genera classified to a higher taxa (family, order and/or class), because the result would be impractical considering the uncertain affinities of many Cambrian biota. Because I only focus on marine invertebrates, data exclude occurrences of algae, acritarchs and ichnofossil. This left a dataset of 37,994 occurrences in 6,718 collections comprising a total of 3,858 genera. Multiple occurrences of a genus within a collection were treated as the same occurrence. According to the correlation chart in Figure 2.1, I vetted the stratigraphy of all collections in the dataset. Within this framework, the Ediacaran includes three assemblage zones, and the Cambrian contains ten stages. I also include the data from the Tremadocian stage of the Ordovician in the dataset. Then I reassigned collections, and thus the occurrences among each of them, in the dataset to both biozones and stages. Figure 3.1 displays the frequencies of occurrences in both stage-level and biozonal bins, as well as the percentages of precise bin assignments. Collections that could not be assigned to a biozone or a stage were excluded from further analyses. As shown in figure 3.1, only 31.4% generic occurrences can be 100% precisely assigned to certain biozone and 94% generic occurrences assigned to certain stage-leveled bins. Although it has been suggested that it would be advantageous for diversity studies to use the highest possible resolution (Foote and Raup 1996), precise bin assignments are too rare for any meaningful analysis. It is evident that biozonal assignments show a lesser bin-to-bin variation compared with stage-leveled assignments (SD(biozone-to-biozone variations) = 0.094,

SD(stage-to-stage variations) = 0.244). The same pattern is also shown in overall variability of 33

occurrences between both approaches (SD(biozones) = 788.15, SD(stages) = 2305.33). The estimate of diversity can be highly variable when intervals are short relative to lumped time intervals (Figure 3.1B). The number of genera counted with biozonal assignments shows a rather depressed pattern during the period from the Stage 3 to the Stage 4 relative to that with stage-leveled assignments in the sense of both diversities with and without singleton. This result is mainly due to the fact that stage-level assignments can lump more occurrences into relatively long intervals as opposed to short intervals. This is also the reason why the diversity with singletons peaked in Stage 3 (Figure 3.1A) when the interval is long enough to lump the generic occurrences with short durations. In my thesis, my analyses on diversity base on stage-level assignments rather than biozone-level assignments.

3.2 Sampling bias

Global diversity curves depict the changes in the number of taxa that have existed through time. Diversity curves are potentially of high significance for interpreting macroevolutionary patterns and processes but they also mirror variation in the nature of the fossil record and the way the record is reported. Thus, our conception on paleodiversity through time is exclusively dependent upon the fidelity of the fossil record. The incompleteness of preservation biases the biological pattern. It is clear that the fossil is no more than a small representation of the life of the past, that is, only a small portion of the members of life assemblages can be preserved as fossils eventually. Ecologically speaking, the estimated number of species is the downward-biased estimator for the total species richness of a local assemblage. Therefore, fossil-based diversity curves can never record the true, absolute biodiversity. If all potential biases are considered, the trajectories of biodiversity might, however, reliably record the relative diversity change over time. As long as secular factors that influence the fossil record remain constant through time or vary in concert (Raup 1991), paleodiversity trajectories would be able to approach to the true diversity variation through time.

Diversity without singleton Diversity with singleton

0.98 %

0.99 % 1 % Frequency

1 % genera of Number

0.96 % 0.99 % 1 % 0.98 % 0.99 % 0.94 %

0.98 % Frequency Frequency

0.38 % 0.97 % 1 % 0 2000 4000 6000 8000 10000 0 200 400 600 800 1000 1200

0 2 4 6 8 10 12 14 Oldest level B

Diversity without singleton Diversity with singleton

0.96 Frequency Number of genera of Number

0.04 0.06 0.32 0.62 0.39 0.320.42 0.55 0.38 0.56 0 0.08 0.080.13 0.16 0 0.15 0.690.96 0.78 0.58 0.580.84 0.180.350.11 0.09 0.23 0.32 0.01 0.97 0.33 0.25 0.38 0 0.150.14 0 0.06 0 0 0.28 0 1000 2000 3000 4000 5000 6000 0 200 400 600 800

0 10203040

Oldest level Oldest level

Figure 3.1 Occurrences in stage-level (A) and biozonal (B) with precise bin assignments shown on top of each bar. The curves show the number of counted from raw data genera with and without singletons for each of bins.

The quality of the fossil record depends strongly upon the amount of exposed sedimentary rocks, taphonomic heterogeneity, and sampling efforts through time (Raup 1972, Smith 2001, Kiessling 2005, Peters 2005, Smith and McGowan 2005, Smith 2007a, b). Among these sources of bias, paleobiologists pay much attention to sampling bias, that is, the non-random variation of the number of fossils recovered over time and space. Obviously, looking at more fossils generally means finding more kinds of fossils and new fossil species can add to the total number of known species. Therefore, we can expect a curve depicting the relationship between the amounts of fossil data and perceived fossil diversity to increase rapidly at the beginning and then become flatter as the amounts of fossil data increases. As shown in Figure 2.2B, differential sampling efforts do exist in the database, and there is a strong correlation between the number of collections and the number of generic occurrences and genus richness (Table 3.1). It is thus intuitive that estimates of diversity are biased by collecting efforts. Therefore, to understand the diversity patterns during the Cambrian radiation, the very purpose of estimation is to get rid of sampling biases.

Table 3.1 Cross correlations among numbers of collection, occurrence, and genus through the Ediacarna-Cambrian transition based on data in table 2.1 with first differencing order.

Independent Dependent N Spearman’s ρ collection no. occurrence no. 11 0.93*** collection no. genus no. 11 0.88*** occurrence no. genus no. 11 0.92*** *** P < 0.001

A rarefaction curve is a plot of the number of species as a function of the number of samples, showing how increasing but randomly drawn amounts of data correspond with increasing sampled diversity. Figure 3.2 shows rarefaction curves relating diversity to the total number of occurrences described through the Ediacaran and the Cambrian respectively. In either time interval, the total number of genera revealed in the dataset

increases with the process of data collection. The problem raised by this curve is that collecting effort varies through time. Therefore, the shape of a raw diversity curve must be biased and may even include little or no biological information, especially when comparing diversities in different geological time intervals.

Ediacaran Cambrian Genera sampled Genera 0 500 1000 1500 2000 2500 3000 3500

0 1000 2000 3000 4000 5000 6000 Occurrences drawn

Figure 3.2 Rarefaction curves relating diversity to the number of randomly drawn fossil collections through two time intervals of Ediacaran and Cambrian. Coloured bars depict 95% confidence bands around the curves for Ediacaran and Cambrian.

3.3 Counting of taxa

Counting methods are crucial for diversity assessments, and there is still no consensus which method is to be preferred. The time-series of diversity seek an estimate standing diversity of the total number of taxa that exist during any part of an interval rather than at a point in time. Thereby, there are two kinds of data from which the counts can be drawn.

The traditional approach is to include all taxa ranging anywhere into a time interval. This method was widely used by Sepkoski in analyzing his unpublished global compilation of geologic ranges of marine genera (Sepkoski 1996, 1997). This compilation attempts to capture all published information on first and last appearances through time but lacks information on actual occurrences within ranges. Without this information, diversity estimate cannot be sampling standardized and is thus prone to biases. Therefore, most modern literature depends on taxonomic occurrence data, which consist of records showing which taxa are found in which individual fossil sample. Modern analyses of the diversity applied to paleontological data relies on fundamental classification of taxa that is present during a stratigraphic interval (Foote 2000). All of the common counting methods involve four fundamental categories of taxa present during a stratigraphic interval: (1) Taxa that cross both of top and bottom boundaries (bt), (2) Crossing the bottom boundary and going extinct in an interval (bL), (3), Originating within an interval and crossing its top boundary (Ft), (4), Originating and going extinct within an interval (FL), or singletons. Several counting methods could be inferred from these four counting categories. Traditional curves (Sepkoski 1996, 1997) employ all four counts (Range-through, RT),

Nbt + NbL + NFt + NFL, which follows from the inference that each genus existed through all of the intervals between its first and last appearance. The second counting method is boundary crosser (BC): Nbt + NbL. This method is counting the number of taxa restrictedly present before and after the boundary of a time interval. Therefore BC taxa have coexisted in time, whereas RT has not necessarily coexisted, e.g., when a taxon is going extinct early in a time interval and another taxon originates late in the same time interval. The third counting method is sampled in bin (SIB) (Miller and Foote 1996) which only count taxa actually sampled in an interval without extrapolating ranges between their first and last appearance in the fossil record. That is, SIB does not count taxa that are found before and after a temporal bin and sampled within it. This has the advantage that edge effects at the beginning and end of time series are avoided (Alroy 2008). Differences between BC and RT have been illustrated by applying them to identical data sets (Bambach 1999, Alroy 2000b). Likewise, BC and SIB were both applied to a Paleobiology Database Phanerozoic invertebrate data set by Alroy et al.

(2001), showing that a wide variety of diversity curves can be generated from the same data set just by varying sampling and analytical protocols. Therefore, choosing standardizing sampling levels and defining counts of taxa are crucial to diversity analyses. Figure 3.3 shows the genus-level diversity curves for the Ediacaran-Cambrian time interval based on those three counting methods. Both sampled counts of taxa with SIB and RT increases from Ediacaran to the Cambrian Stage 3 and then declines through the rest of the Cambrian, whereas the peaks of BC counts appear later during the period from the Cambrian Stage 4 to the Drumian stage. The BC curve shows less variation and depressed in general than the SIB and RT curves. Relatively long temporal bins (e.g. Figure 3.1A) could probably increase SIB counts because this effect inflates the available pool of genera for sampling. On the contrary, The long temporal bins depress BC counts in way of decreasing the chances of taxa being repeatedly sampled, which is intuitive because taxa cannot cross boundaries unless they are drawn repeatedly (Alroy et al. 2001). But I find no significant correlations between interval duration and either proportion or number of singletons (Diff(interval duration)/ NFL: Spearman’s rho = 0.2, p-value = 0.53;

Diff(interval duration)/ PFL: Spearman’s rho = -0.08, p-value = 0.81), suggesting that interval duration does not cause a substantial sampling bias. Additionally, the Cambrian radiation was characterized by the appearance and a rapid proliferation of higher taxa (Figure 3.4) that would come to dominate marine settings through the balance of early Paleozoic (Sepkoski 1981). It is recognized in published time-series of global marine biodiversity as one of the most extensive intervals of genus and family diversification in the history of life (Sepkoski 1993a, Benton 1995). This relatively fine, stage-level stratigraphic treatment appears not to influence the basic diversity trend as the level of standing diversity is not inflated during the lengthy Ediacaran and the first two Cambrian intervals (Table 3.3).

BC RT SIB Counts of genera

Ava Whs Nam For St2 St3 St4 St5 Dru GuzPai Jia St10 Tre

-200 0 200 400 600Ediacaran 800 1200 Cambrian 560 540 520 500 480 Age (Ma)

Figure 3.3 Raw genus-level diversity curves for the Ediacaran-Cambrian time interval based on all marine taxa of animal from the Paleobiology Database. Note log scale of y axis. The calculations are based on Table 3.2. Brown line depicts totals of genera crossing the boundaries between each bin and preceding bin (Boundary-crossers, BC); Black line depicts totals of genera ranging into or across each bin (Range-through with singletons, RT); Orange line depicts totals of genera sampled in bin (Sampled-in-Bin, SIB). Ma, million years ago. Abbreviation of Stages: Ava, Avalon assemblage; Whs, White Sea assemblage; Nam, Nama assemblage; For, Fortunian; St2, Stage 2; St3, Stage 3; St4, Stage 4; St5, Stage 5; Dru, Drumian; Guz, Guzhangian; Pai, Paibian; Jia, Jiangshanian; St10, Stage 10; Tre, Tremadocian (Ordovician).

Table 3.2 Counts of four fundamental categories of taxa present during a stratigraphic interval from the Ediacaran to the earliest Ordovician. The column of %singleton presents the proportions of singleton in total range-through counts.

Intervals Nbt NbL NFt NFL/singleton %singleton Avalon assemblage 0 0 22 42 0.66 White Sea assembalge 14 12 13 42 0.58 Nama assemblage 8 19 10 30 0.51 Fortunian 9 10 114 79 0.39 Cambrian Stage 2 77 99 153 120 0.36 Cambrian Stage 3 90 158 290 744 0.66 Cambrian Stage 4 146 243 199 189 0.31 Cambrian Stage 5 167 193 237 279 0.41 Drumian 154 245 176 122 0.24 Guzhangian 122 186 185 144 0.29 Paibian 134 168 174 84 0.20 Jiangshanian 156 162 171 89 0.21 Cambrian Stage 10 122 202 139 129 0.29 Tremadocian 0 186 0 592 0.76

Classes Orders Families l erva t n i s per t oun C 1 0 50 100 150 200

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian 560 540 520 500 480

Age (Ma)

Figure 3.4 Time series plot of higher taxa diversity (families, orders, classes) from the Ediacaran to the earliest Ordovician. Note log scale of y axis.

Table 3.3 Correlation chart between interval duration and taxonomic diversities. Correlations are based on data after first differencing order. Div-species, -genus, -family, -order, and –class represent diversities at five taxonomic levels.

Independent Dependent N Spearman’s ρ p-value Diff(interval duration) Diff(div-species) 10 0.25 0.44 Diff(interval duration) Diff(div-genus) 10 0.28 0.38 Diff(interval duration) Diff(div-family) 10 0.4 0.2 Diff(interval duration) Diff(div-order) 10 0.45 0.14 Diff(interval duration) Diff(div-class) 10 0.51 0.09

3.4 Subsampling

An effective solution to this sampling problem is sampling standardization or subsampling. Sampling standardization is derived from the concept of rarefaction, which is most often used by ecologists as to compare species richness data among sets with different sample sizes (Sanders 1968). The assumption of sampling standardization is to get a uniform sampling effort through time in order to make assessable of diversity comparisons among time intervals. The basis of subsampling method is a process in which one starts with a set of “items” that include taxa; sets a uniform sampling “quota” to be used in all the time intervals; draws one item randomly, counts the taxa, and grabs another item without replacement, counts the taxa, collecting items until the quota is reached. This procedure is repeated several times (subsampling trials) and the results are averaged (normal means). Based on this concept, Miller and Foote (Miller and Foote 1996) directly rarefied the occurrence data, drawing them randomly from the pooled stage-level data, to calibrate the Ordovician Radiation of marine life. Paleobiologists call this method classical rarefaction (CR) even though the items drawn are occurrences instead of specimens in original rarefaction practice in ecology. Special protocols and methods have been developed later (Alroy 1996, Alroy et al. 1998, Alroy et al. 2001, Kiessling 2005, Smith 2007a, Chiarucci et al. 2008, Alroy 2010a, Alroy 2010b) including the by-list unweighted (UW) subsampling, by-list occurrence-weighted (OW) subsampling, and shareholder

quorum (SQ) subsampling. These randomized subsampling protocols seek to hold each time interval to a uniform quota of data items. The quota may be expressed in terms of the number of fossil collections (UW), occurrences represented by the collections (OW), or a fixed proportion of the overall species pool (SQ). Alroy (2010b) and Jost (2010) independently proposed that samples be standardized to a common level of sample completeness (as measured by sample coverage, SQ), and developed algorithmic approaches for comparing rarefied samples. Because uniformly fair sampling is not the same as uniformly accurate sampling (Alroy 2010b), SQ relaxes such assumption by drawing a fixed proportion of the overall species pool. Here I explore ways to address sampling issues at the Ediacaran-Cambrian transition. The aim is to investigate the nature of diversity data of the Cambrian and to test four distinct analytical methods. Basically, the approach is to standardize the amount of sampling in each time unit to approximate the best estimation of the true biological pattern. I employ CR, UW, OW, and SQ methods on the data separately and compare the trajectories among different subsampling methods. The ultimate objective of sampling standardization is to find the best, that is, the most biologically meaningful, method for assessing diversity during the Cambrian radiation.

3.4.1 Occurrence-based subsampling (Classical Rarefaction or CR)

On the coarsest temporal scale, the rarefied diversities of the Ediacaran and Cambrian periods differ markedly (Figure 3.2). Cambrian diversity was much higher than that in the Ediacaran even when sampling is taken into account. However, sampling in the Ediacaran so much lower than in the Cambrian that diversities can only be compared at low levels of sampling. Applying the same methods over time and focusing on the better sampled Cambrian, diversity can be compared at the reasonable subsampling quota of 1161 occurrences (Figure 3.5), that is, the level of sampling of the Fortunian stage. A direct comparison of the raw diversity pattern (Figure 2.2B and Figure 3.3 SIB counts) with the rarefied one (Figure 3.6) suggests that the diversity increase over the first three Cambrian stages is not an artifact of sampling. The comparison of the Stage 4 and the Stage 5 curves, however, suggests that the Stage 4-Stage 5 diversity increase in the raw data is,

indeed, an artifact of sampling.

St3

St5

St4

Dru Guz St10

Genus occurrences Genus Pai Jia St2

For 0 200 400 600 800 1000

0 2000 4000 6000 8000

Genera

Figure 3.5 Comparative rarefaction curves for ten Cambrian stages, based on the analysis of all genus occurrences in the Paleobiology Database after stratigraphic vetting (see Chapter 2). The vertical line represents the smallest sample size of 1161 genus occurrences with which all stages can be compared.

Rarefied No. of Genera of No. Rarefied

For St2 St3 St4 St5 Dru Guz Pai Jia St10 Cambrian 0 100 200 300 400 500 600

540 530 520 510 500 490

Age (Ma)

Figure 3.6 The SIB diversity trajectory based on rarefaction to 1161 genus occurrences. Error bars represent standard deviations of 100 subsampling trials.

To take the whole studied interval into account, that is, from the Ediacaran to the earliest Ordovician, the sampling level has to be much lower (136 occurrences) and therefore the rarefied trajectories appear smoother (Figure 3.7). However, the basic patterns within the Cambrian remain stable even at this low sampling level: There was a strong diversity rise in the first three Cambrian stages, a drop in Stage 4, and diversity declined erratically through the rest of the Cambrian (Figure 3.8).

St3

Tre

St5

St4 Genera Dru Guz St10 Pai Jia St2

For 200 400 600 800 1000

NamWhsAv a

2000 4000 6000 8000

Genus occurrences

Figure 3.7 Comparative rarefaction curves for 14 stage-level intervals over the studied time, based on the analysis of global database of genus occurrences. Vertical line represents the smallest sample size of 136 genus occurrences with which all stages can be compared; horizontal lines represent genus richness for each Cambrian stage at a sample size of 136 genus occurrences.

Rarefied No. of Genera of No. Rarefied

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian 0 20 40 60 80 100 120 140

560 540 520 500 480

Age (Ma)

Figure 3.8 The SIB diversity trajectory based on rarefaction to 136 genus occurrences. Error bars represent standard deviations of 100 subsampling trials.

3.4.2 Sample-based subsampling (by-list subsampling)

3.4.2.1 Occurrences-weighted subsampling (OW)

In contrast to classical rarefaction, in which occurrences are independently drawn from pooled data, occurrences-weighted subsampling (OW) draws entire collections (faunal lists) until a quota of occurrences is reached. The method is similar to rarefaction, but occurrences are kept in their assemblage context rather than treated independently. Several studies (Alroy et al. 2001, Crampton et al. 2009) have shown that the results are similar for CR and OW and this study is no exception (Figure 3.9). Diversity exhibits a strong increase in the first three Cambrian stages in standardized curve employing sampled-within-bin counts, which is also seen in the trend of diversity based on the

rarefaction estimates in Figure 3.8. Like the CR diversity trajectory, the OW diversity did not drop dramatically, but exhibits ebb and flow through the rest of the Cambrian, seemingly reaching a plateau.

SIB BC Number of genera 0 50 100

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian

560 540 520 500 480 Age (Ma)

Figure 3.9 The SIB and BC diversity trajectories estimated by the occurrence-weighted subsampling from the Ediacaran to the Tremadocian (the earliest Ordovician). Error bars are SDs of 100 subsampling trials.

3.4.2.2 Unweighted subsampling (UW)

The unweighted subsampling method or UW bases the quota on the number of lists (collections) independent of the number of occurrences they contain. In our case, the minimum number of collections in a time bin was 45, which was accordingly defined as the subsampling quota. UW can differ markedly from CR or OW (Alroy et al. 2001, Alroy et al. 2008) and is rarely used for reconstructions of global diversity. UW is problematic if the modal number of occurrences per collection varies systematically through time (Alroy et al. 2001). But this is unlikely in my data because there is no modal

occurrences per collection through the Ediacaran-Cambrian time except for the Avalon assemblage interval and Cambrian Stage 3, in which the proportions of modal occurrences are 17% (14/82) and 14% (138/1003) respectively (also see Figure 3.10).

Modal number per number collection Modal 0 5 10 15 20 25 30

1234567891011121314

Figure 3.10 Boxplot showing the variation in modal numbers of occurrences per collection through Ediacaran-Cambrian time. 14 time intervals from the Ediacaran to the earliest Ordovician are represented as 1 to 14 numbers in axis.

Nevertheless, in the Ediacaran-Cambrian case the results are similar to the previous ones (Figure 3.11). Diversity exhibits a strong increase from the Ediacaran to Cambrian Stage 3 and then drops in Stage 4. The drop of UW diversity in Stage 4 is more pronounced than that of either CR or OW. These differences are due to variable average list lengths, which reflect alpha diversity (see Chapter 5). Cambrian Stage 2 marks the

first appearance of archaeocyath sponges, which achieved their maximum diversity in Stage 3. Because of the monographic treatment and potential oversplitting of archaeocyathans (Sepkoski 1979, Wood et al. 1992), their collections are usually long lists of genus occurrences. This explains the diversity peak in Stage 3 to a large degree.

SIB BC Number of genera

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian -50 0 50 100 150 200 250

560 540 520 500 480 Age (Ma)

Figure 3.11 The SIB and BC diversity trajectories from the Ediacaran to the Tremadocian (the earliest Ordovician) based on unweighted subsampling of 43 collections per stage. Error bars are SDs of 100 subsampling trials.

The three methods applied until now, yield the same basic patterns in spite of some differences. For example, all methods decrease the artificial Cambrian Stage 3 peak of five-fold increase in the number of genera since the Fortunian. Furthermore, the two methods that weight on the basis of occurrences (CR and OW) unsurprisingly yield almost indistinguishable trajectories. However, important differences among the three curves are evident.

First, all two by-list methods generate broader standard deviations than does classical rarefaction. The reason is that rarefaction alone assumes complete independence among occurrences, which effectively inflates the number of independently drawn items. Because the fossil record presents itself in the form of fossil collections and not isolated occurrences, broader standard deviations are arguably more realistic. Second, all sampling-standardized curves fail to show a significant decline after Cambrian Stage 3. I argue that subsampling has efficiently removed a sampling bias in this case, even though several authors argue for a substantial diversity drop after the Sinsk Event (Zhuravlev and Wood 1996, Kiessling et al. 2002, Luchinina et al. 2013). Third, the boundary-crosser curves generally suggest slightly higher Cambrian Stage 2 diversity. But in general, the boundary-crosser curves show less variation and smaller confidence intervals than the sampled-within bin curves. This difference perhaps reflects real differences in diversity trajectories between common, long-ranging taxa and rare, short-lived taxa, because the former should feature even more heavily in boundary- crosser counts. This explains inflated sampled-in-bin counts in Stage 3, when rare, short- lived taxa (singletons) are abundant (see Table 3.2).

3.4.3 Shareholder quorum sampling (SQ)

Shareholder quorum sampling was introduced by Alroy (2010b) in order to solve the problem that the previous subsampling methods are sensitive to evenness (and in fact measure evenness more than richness) and that rare taxa are underrepresented in diversity estimates (Alroy 2010b). This method does not employ a uniform quota of items, but involves random sampling of individuals to equal frequency coverage, the shareholder quorum. The shareholder quorum is superior to traditional subsampling methods (e.g., rarefaction) in that variations in evenness are explicitly taken into account. Therefore, it is less biased by different abundance distributions and represents a nearly unbiased estimate of richness (Aberhan and Kiessling 2014). Figure 3.12 depicts SQ-standardized genus-level diversity based on 70% frequency coverage per stage for marine animals from the Ediacaran to the earliest Ordovician. SQ diversity reached its climax in Cambrian Stage 3, dropped in Stage 4, and declined through the rest of the Cambrian. The pattern of standardized diversity is consistent with that of raw data and is insensitive

to including or excluding the archaeocyath sponges, which are potentially oversplit (Sepkoski 1979, Wood et al. 1992). The trajectory of SQ diversity is supposed to most closely approach to the true diversity variation through time. Counts of genera (SQ) 0 100 200 300 Ava Whs Nam For St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Ediacaran Cambrian

560 540 520 500 480 Age (Ma)

Figure 3.12 Global genus-level diversity of marine animals from the Ediacaran to the earliest Ordovician. Sampling-standardized genus-level diversity based on shareholder quorum subsampling with 70% frequency coverage per stage. The brown line refers to all marine genera, and the blue line excludes archaeocyaths. Error bars are SDs of 100 subsampling trials for all marine genera.

3.5 Correlations among CR, UW, OW, and SQ diversities

I test the relationships among diversities estimated by four subsampling methods (employing SIB counts) in order to address which one represents the most biologically meaningful analysis for the Cambrian diversification. Figure 3.13 shows plots for pairwise relationships among diversities standardized by four methods, CR, UW, OW, and SQ. All plots show significant correlations, suggesting that using any standardization

makes no big difference in estimating diversification during the Ediacaran-Cambrian time. However, stronger correlations are indeed shown in CR-OW, CR-SQ, and OW-SQ plots, as in the scatters in these three plots are distributed more closely to corresponding regression lines. This finding further implies that CR, OW, and SQ estimate the most corresponding values in diversity. It is also obvious that the variance in sampled diversity is much higher when using UW subsampling method. The reason is that by drawing whole collections one could fill the quota either with many small collections or a few large, diverse collections, and both will happen commonly (Alroy 2010a). To test if the estimate of genus richness is relatively independent of additional collecting effort, I tested how the changes in number of collection data downloaded from the Paleobiology Database before 2011 (when this project started) have affected the basic pattern present here. Figure 3.14 shows the data inventory of 20,000+ occurrences in the Paleobiology Database through the Ediacaran-Cambrian times counted by the old binning system. The pattern does show some differences such as a rapid increase in the number of genera after the dramatic decline in the Toyonian, but the basic patterns are robust, suggesting that the future compilation in the database is unlikely to introduce a substantial change to the basic pattern of generic diversity.

3.6 Discussion

Initially, my results confirm that sampling bias is perceivable in the fossil record through the Ediacaran-Cambrian time. Counting methods (RT, SIB, and BC) are supposed to respond differently to sampling bias (Alroy et al. 2001), which could influence the outcomes of diversity estimates based on standardized data. The mirrored patterns between SIB and RT imply that Lazarus taxa (taxa that range completely through an interval but not sampled) were not dominant either among Ediacaran fauna or Cambrian fauna, and thus would not cause a substantial sampling bias. The discrepancy between SIB and BC curves is mainly due to a large number of Cambrian taxa with short duration so that they were only found within sampled interval (singletons), such as those found in exceptional Lagerstätten (e.g. Chengjiang biota, Burgess Shale fauna etc.). I prefer using SIB to assess diversity because it does not interpolate taxon occurrences between their first and last appearance in the fossil record. This counting method has the advantage that

edge effects at the beginning and end of time series are avoided (Alroy 2010a).

) )

diversity W diversity (SQ (O log log log log 3.5 4.0 4.5 5.0 5.5 3.5 4.0 4.5 5.0 5.5 3.8 4.0 4.2 4.4 4.6 4.8 3.8 4.0 4.2 4.4 4.6 4.8

log (CR diversity) log (CR diversity)

) )

diversity UW diversity (SQ ( log log log log 3.5 4.0 4.5 5.0 5.5 3.5 4.0 4.5 5.0 5.5 6.0 3.8 4.0 4.2 4.4 4.6 4.8 4.0 4.2 4.4 4.6

log (CR diversity) log (OW diversity)

) )

diversity W diversity (SQ (O log log log 4.0 4.5 5.0 5.5 6.0 3.8 4.0 4.2 4.4 4.6 4.8 4.0 4.5 5.0 5.5 4.0 4.5 5.0 5.5

log (UW diversity) log (UW diversity)

Figure 3.13 Cross-correlations among CR, UW, OW, and SQ diversities. The test is based on Spearman’s rho. All tests show significant correlations.

occurrences collections genera Counts per interval per Counts

N-D Tomm At Bo T M.Camb U.Camb Ediac. Cambrian 2 5 20 50 200 1000 5000

560 550 540 530 520 510 500 490

Age (Ma)

Figure 3.14 Data inventory in the Paleobiology Database through Ediacaran-Cambrian times as of December 2011 (20,000+ occurrences). Note log scale of y-axis.

The broad Cambrian trend is in agreement with other published compilations at the genus and family levels and indicates an order of magnitude diversity increase (SIB counts) (Sepkoski 1979). All subsampling methods agree that the Cambrian radiation took place over a short time span from the Fortunian to Stage 3, which is in concert with diversity trajectory based on raw data (Figure 2.2). Basically, all standardized curves exhibit similar patterns: diversity increased dramatically from the Ediacaran to the Cambrian Stage 3 and then declined through the rest of the Cambrian; despite continuous decrease after the Cambrian Stage 4, diversity at its lowest level in the Paibian stage was markedly higher than during almost all parts of the preceding period. The pattern of Archaeocyathan sponges represented here is consistent with earlier investigations stating that archaeocyaths appeared in the Tommotian, progressively colonizing Atdabanian carbonate platforms, reaching their acme of development in the Botomian, and then declining in the Toyonian (Pratt et al. 2001). Randomized subsampling protocols such as CR, UW, and OW seek to hold each

time interval to a uniform quota of data items (Alroy 1996, Miller and Foote 1996, Alroy 2000a, Alroy et al. 2001, Alroy et al. 2008, Alroy 2010a). The quota may be expressed in terms of the number of taxonomic occurrences (CR), fossil collections (UW), or occurrences represented by the collections (OW). These three methods make important assumptions about uniformity through time in the evenness of abundance distributions and the average number of specimens found in collections (Alroy 2000a, Alroy et al. 2008). That is, these methods work the best when the dataset contains a large amount of high-quality specimen count data for each time interval (Alroy et al. 2008), otherwise, the outcome would be problematic (Alroy 2010b). The Cambrian period is characterized by the most short-lived genera among time intervals in the Phanerozoic represented by stratigraphic singletons in the dataset (Table 3.2, on average 42% singletons out of total range-throughers per time interval). Likewise, these three methods will give similar outcomes if generic occurrences are evenly distributed in collections in the dataset. However, the distribution of genus occurrences in collections for each time interval in the Cambrian is uneven (Figure 3.1). Applying any of such method to the diversity curve is supposed to flatten it. Therefore, the best way to calculate global diversity is shareholder quorum sampling, which does not draw a fixed number or fraction of data items or taxa but randomly sample individuals to equal frequency coverage (Alroy 2010b). In addition, a comparison made between databases of fossil marine animal genera downloaded in 2011 and 2014 suggests that diversity curves compiled from the two databases are very similar. Despite that genera have been added and deleted in three years, apparent macroevolutionary patterns for the entire marine fauna have remained constant. All major events of radiation and extinction are identical. Therefore, errors in large paleontological data bases and arbitrariness of included taxa are not necessarily impediments to the analysis of pattern in the fossil record, so long as the data are sufficiently numerous. In summary, that the main pulse of the Cambrian radiation was in the early Cambrian has long been known (Zhuravlev and Naimark 2005, Li et al. 2007), but defining the time of peak diversity has been hampered by problems with stratigraphic correlations. The improved data in my study and updated stratigraphic assignments demonstrate a prominent peak in Stage 3, which is roughly equivalent to the Siberian

Atdabanian and most of the Siberian Botomian (Peng et al. 2012). The subsequent decline in global diversity after the diversification pulse observed both in raw data and standardized data may have been partly due to the Sinsk extinction event in Stage 4 (Zhuravlev and Wood 1996, Kiessling et al. 2002, Luchinina et al. 2013) but environmental factors are more plausible to explain the lack of further diversification in the later Cambrian. For example, prolonged tropical warming has been suggested to explain the scarcity of metazoan reefs in the later Cambrian (Rowland 2002) and would also fit the gamma diversity trajectories given that cooling is thought to have facilitated the subsequent Ordovician radiation (Trotter et al. 2008). Alternatively, a late Cambrian rise in oxygen levels has been proposed to explain the onset of the Ordovician radiation (Saltzman et al. 2011b). By inference, oxygen-limitation may have hindered further diversification after the early Cambrian. Furthermore, the analysis indicates that diversity reached a plateau after Cambrian Stage 4, perhaps reflecting a stable and complex ecosystem in the late Cambrian.

4 Chapter 4: Diversity partitioning during the Cambrian radiation

4.1 Introduction

Regional (gamma) diversity can be partitioned into local (alpha) and turnover (beta) components (Whittaker 1960, Whittaker 1972). Alpha diversity measures species packing within a community, that is, richness of taxa at a single locality or in a particular community. It is usually assessed by mean number of taxa in a communities or a particular habitat, and thus reflects how finely species divide ecological resources. Beta diversity is used to assess the amount of turnover in species composition among communities, that is, the taxonomic differentiation of fauna between sites or communities. Gamma diversity is understood as global genus richness, whereas beta diversity is used to characterize turnover between individual assemblages as well as depositional environments or geographic areas. Sepkoski (1988) has proposed that increasing differentiation in local communities could have also played an important role. Regional studies in South China (Li et al. 2007) and Siberia (Zhuravlev and Naimark 2005) on skeletal metazoans and archaeocyath sponge assemblages respectively suggest that the main pulse of the Cambrian radiation was in the early Cambrian (Qiongzhusian stage in China and Atdabanian stage in Siberia to be exact). They also argued that the degree of provinciality at the level of genus was the principal factor of taxonomic diversity in the earliest benthic paleocommunities. However, the role that beta diversity played in governing the Cambrian radiation is still underexplored. I use sampling-standardized analyses of fossil occurrence data from the Paleobiology Database to derive accurate time series of alpha, beta, and gamma diversity from the Ediacaran into the earliest Ordovician. The purpose of this chapter is to investigate beta diversity and its contribution to global diversification during the Cambrian radiation. The basic question is: was global diversification during the Cambrian radiation driven by a higher species richness in local communities (alpha) or rather by an increase of between communities (beta)? The measure of alpha and beta diversity is based on the conception that most fossil collections recorded in the Paleobiology Database pertain to single localities or bed-scale inventories. Each 59

collection in the Paleobiology Database contains the information referring to its sedimentary environment and lithology. Therefore, each collection can be treated as a fossil community or fossil assemblage in a certain environment. I prefer using by-list unweighted subsampling on fossil occurrence data to derive accurate time series of alpha and beta diversity as UW weights all collections equally. I then apply non-parametric statistical tests on the relationships between gamma and alpha and between gamma and beta to assess if global diversification during the Cambrian Explosion was driven by higher genera richness level in local communities (alpha diversity) or rather by an increase of differences between communities (beta diversity).

4.2 Methods

4.2.1 Measure of alpha and beta

Alpha diversity is calculated as mean number of taxa sampled within community. When abundance data is available, alpha diversity can be also measured based on some biodiversity indices such as Shannon-Wiener and Simpson. There are several ways of measuring beta diversity. I herein use two commonly used measures of beta diversity: (1) =⁄ , where is total diversity at the global scale, and is the mean genus richness within assemblages. This is a global measure of beta diversity based on presence-absence data (Whittaker 1960, Whittaker 1972); and, (2)

= 1 − 2 ∑ , ∑ + ∑ , where is the number of occurrences of th th the k genus in one of the sites, is the number of occurrences of the k genus in the other site, and , selects the lesser of the two values. This measure is based on the Bray–Curtis index (Bray and Curtis 1957), which assesses the pairwise dissimilarity among sampling units with respect to environmental or geographic variables. This dissimilarity approach is related to Czekanowski’s similarity coefficient (Sepkoski 1974) between the same sites, both of which evaluate the percentage of sum of the lesser values for only those species in common between both sites.

4.2.2 Measure of origination and extinction rates

I use two rate measures: per-taxon rates and per-capita rates (Table 4.1). The per-taxon rates are originations or extinctions per genus per interval. They have been used as a way

to normalize the number of origination or extinction events by total diversity and interval length. As interval length increases, a progressively larger proportion of total diversity consists of singletons. Because these taxa first appear and last appear within the same interval, proportional origination and extinction asymptotically approach unity as ∆t increases, and the per-taxon rates consequently decline as interval length increases (Foote 1994). This method includes single interval taxa (singletons). Singletons are considered to be especially sensitive to variation in preservation and interval length, and thus produce many undesirable distortions of the fossil record (Foote 2009). The estimates of per-capita rates used here, pˆ and qˆ, are unaffected by interval length, and thus does not need to be standardized by stage durations (Jablonski and

Bottjer 1990, Foote 2005). Nbt / (Nbt + NFt) gives the proportion of lineages extant at the start of the interval that survive to the end, and Nbt / (NbL + Nbt) gives the proportion of lineages extant at the end that were already extant at the start. These ratios decay exponentially with time if rates are constant within the interval; thus the logarithm of each ratio declines linearly with time (Foote 2000).

Table 4.1 Definitions of taxonomic rate metrics for intervals of length . Measures are expressed in terms of numbers belonging to the four fundamental classes of taxa, , , , and (see Chapter 3: Counting of taxa).

Measure Definition Origination: +⁄ + + +⁄∆ Per-taxon rate Extinction: +⁄ + + +⁄∆ Net diversification rate: origination - extinction ̂ = −⁄ + ⁄∆ Foote’s per-capita rate = −⁄ + ⁄∆ Net diversification rate: ̂ -

4.3 Results

4.3.1 Trajectories of alpha, beta, and gamma diversity

Raw gamma diversity estimate (global diversity in previous chapter) based on shareholder quorum subsampling method (see Chapter 3 Figure 3.12) exhibits a strong

increase in the first three Cambrian stages. In order to clearly exhibit diversity trajectories in terms of alpha, beta, and gamma diversity in this chapter, I revisit sampling standardized gamma diversity in Figure 4.1. Gamma diversity dropped in Stage 4 and declined further through the rest of the Cambrian. Alpha and beta diversity increased from the Fortunian to Stage 3, and fluctuated erratically through the following stages (Figure 4.2). My estimate of alpha (and indirectly beta) diversity is based on the number of genera in published fossil collections and, thus, may be affected by monographic biases. However, the same basic pattern is seen in diversity estimates of paleocommunities with abundance data (Figure 4.3).

1200 A

1000

800

600 (raw data)(raw

Counts ofgenera 400

200

0 B 300

200

100 Counts of (SQ) genera

0 Ava Whs Nam For St2 St3 St4 St5DruGuzPaiJiaSt10 Tre Ediacaran Cambrian 560 540 520 500 480

Age (Ma)

Figure 4.1 Revisit of global genus-level diversity of marine animals from the Ediacaran to the earliest Ordovician. (A) Raw counts of the number of genera (sampled-in-bin) (see Figure 3.3). (B) Sampling- standardized genus-level diversity (sampled-in-bin) based on shareholder quorum subsampling with 70% frequency coverage per stage (see Figure 3.12).

Figure 4.2 Alpha diversity and beta diversity from the Ediacaran to the earliest Ordovician based on unweighted by-list subsampling of 45 collections per stage. Error bars are standard deviations of 100 subsampling trials.

Figure 4.3 Alpha diversity measured by the Shannon index (H') of quantitative collections comprising at least 80 specimens per collection (293 collections in total). Individual estimates are represented by symbols separated by substrate lithology. These exclude values of less than 0.2 to reduce the bias from extreme habitats. The blue line connects median values.

4.3.2 The main driver of gamma diversity

To test whether the global diversification was driven by a higher species richness in local communities (alpha diversity) or rather by an increase of between communities (beta diversity), I employed non-parametric correlations applying Spearman’s rho (ρ) on overall-alpha/gamma and overall-beta/gamma diversities. Because there are no significant autocorrelations in any of the relevant variables (overall alpha, overall beta, gamma) (Figure 4.4), correlations are based on non-differenced data.

Gamma diversity (SQ) ACF -0.5 0.0 0.5 1.0 0246810 Lag

Alpha diversity (UW) ACF -0.5 0.0 0.5 1.0 0246810 Lag

Beta diversity (UW) ACF -0.5 0.0 0.5 1.0 0246810 Lag

Figure 4.4 Autocorrelations of gamma (Chapter 3), alpha, and beta diversity.

I find a strong correlation between global genus richness and beta diversity (Spearman’s ρ = 0.93, p-value < 0.001), whereas there is no significant correlation

between global genus richness and alpha diversity (Spearman’s ρ = 0.42, p-value = 0.137). The same relationship is evident with a moving-window approach of five successive stages (Figure 4.5). Here the beta-gamma link is significant over the interval from Stage 2 to the Drumian. The strong correlation between beta diversity and gamma diversity could be biased by the multiplicative approach I am using to derive beta diversity from both alpha and gamma diversity, which are independently assessed. Although this method probably has a better biological underpinning than the additive approach (Veech et al. 2002, Holland 2010, Hautmann 2014), a correlation between gamma and beta diversity is still evident when using an additive method (Spearman’s ρ = 0.84, p-value < 0.001). Therefore, gamma diversity in the Cambrian was largely governed by differentiation among communities or assemblages rather than by genus packing within assemblages.

Figure 4.5 Spearman rank correlations between gamma-alpha diversity and gamma-beta diversity measured within a moving window of five successive bins. Values are plotted in the middle of window. The correlation of gamma and beta diversity is high throughout the Cambrian and often significant but there is no significant correlation between gamma and alpha diversity. Empty blue squares denote correlations with a p-value < 0.1 and solid blue squares p-values of < 0.05.

4.3.3 Origination and extinction rates

Mass extinctions may affect diversity at multiple levels. Several episodes of profound turnover and mass extinction have been noted at the Ediacaran-Cambrian boundary and throughout the Cambrian (Bambach et al. 2004, Zhu et al. 2006). However, the sampling- standardized trajectory of extinction rates is high but fairly smooth without prominent peaks (Figure 4.6). There is no significant correlation between extinction rates and either alpha or beta diversity (Table 4.2), indicating that the observed extinctions are unlikely to introduce a substantial bias on alpha-beta-gamma diversity patterns.

Table 4.2 Cross-correlations among rate measures and alpha, beta diversity.

Dependent Independent alpha beta Spearman’s ρ p-value Spearman’s ρ p-value per-taxon ext. -0.19 0.52 0.47 0.09 per-taxon ori. 0.02 0.94 -0.38 0.19 per-capita ext. -0.23 0.45 0.44 0.14 per-capita ori. 0.27 0.37 -0.26 0.38

4.4 Discussion

Although a surprising diversity of benthic organisms is now known from the Ediacaran (Erwin and Valentine 2012), diversity was substantially lower at the gamma and alpha level than in the Cambrian (Figure 4.1B and Figure 4.2). The differences between the Ediacaran and Cambrian faunas are profound and must in part reflect the rapid rates of evolution that included the appearance of various groups. There is, however, some evidence for the survival of Ediacaran taxa into the Lower Cambrian (Glaessner 1985, Conway Morris 1993, Shu et al. 2006), and several authors have emphasized that taphonomic factors may have accentuated the differences between Ediacaran and Cambrian faunas (Seilacher 1989, Hagadorn et al. 2002, Laflamme et al. 2011).

per taxon rate per capita rate A Extinction rate Extinction Av a Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediac ar an Cambr ian -0.5 0.0 0.5 1.0 1.5 2.0 560 540 520 500 480

per taxon rate per capita rate B

Origination rate Av a Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediac ar an Cambr ian -1 0 1 2 3 560 540 520 500 480

per taxon rate per capita rate C

Av a Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre

Diversification rate Ediac ar an Cambr ian -2 -1 0 1 2 3

560 540 520 500 480

Age (Ma)

Figure 4.6 Sampling-standardized extinction, origination, and diversification rates through the Ediacaran and Cambrian. Solid lines are per taxon rates. Dashed lines are per-capita rates of Foote (Jablonski and Bottjer 1990, 2000) but not standardized by stage durations (Jablonski and Bottjer 1990, Foote 2005). Error bars are standard deviations of 100 subsampling trials.

The diversity during the Cambrian radiation could have been caused by simply random processes of speciation and extinction. Rates of diversification were low in the late Ediacaran and abruptly much higher in the Early Cambrian. This pronounced origination and diversification pulse may be exaggerated by the closure of the Ediacaran taphonomic window with widespread soft-body preservation and the opening of a novel taphonomic window by the widespread acquisition of skeletons (Maloof et al. 2010). However, the prominent peak of diversification rate in the Fortunian (Figure 4.6C) marks a dramatic change physically or ecologically at the beginning of the Cambrian. Such hypotheses have included changes in the chemistry of the ocean (Lane 1917; Vinogradov 1959; Chilingar and Bissell 1963; and others), increase in concentration of atmospheric oxygen (Nurshall 1959; Berkner and Marshall 1964, 1965; Towe 1970; Rhoads and Morse 1971; and others), changes in the stability of the global climate (Rudwick 1964; Fischer 1965; Valentine and Moores 1972; and others), and the initial appearance of predators (Schuchert and Dunbar 1933; Raymond 1935; Hutchinson 1961; Bengtson 1977; and others). Furthermore, the high taxonomic turnover that marks the disappearance of Ediacaran soft-bodied organisms may be tempered by secular variation in global taphonomic regimes (Narbonne 2005), but can also be attributed to ecosystem engineering and predation (Laflamme et al. 2013). Environmental perturbations were invoked to explain the extinction of the few skeletal organisms (Amthor et al. 2003). In summary, by assessing how the overall increase in global diversity was partitioned between within-community (alpha) and between-community (beta) components, I find that changes in gamma diversity in the Cambrian were chiefly driven by changes in beta diversity.

5 Chapter 5: Onshore-offshore patterns in the evolution of Cambrian shelf communities: insight from alpha, beta, and gamma diversity.

5.1 Introduction

The analysis of global diversity in the previous chapter addressed the question when the Cambrian radiation occurred. Here I focus on geographic and environmental patterns to address the question if diversification occurred earlier in some habitats or regions than others. Finding systematic relationships of diversity trends with latitude and sedimentary environments will thus be a major focus. In this chapter, I will look into the diversity patterns along onshore-offshore environments. Onshore-offshore evolutionary gradients are among the most important discoveries in evolutionary paleobiology. The onshore-offshore pattern posits that major evolutionary innovations first appear onshore and later manifest offshore, often with the result that older faunas are replaced (Jablonski et al. 1983, Bottjer and Jablonski 1988, Jablonski and Bottjer 1990, Jablonski 2005, Smith and Stockley 2005, Kiessling and Kocsis 2015). Still, case studies also show exceptions: paleogeographic dispersal of benthic fauna took place in both onshore and offshore directions (Zhan and Jin 2014). I dissect the alpha, beta, and gamma diversity patterns along an onshore-offshore environment. I test the distribution of diversity among environmental zones along an onshore-offshore gradient. This may help deciphering the environmental partitioning of diversity trajectories from the Ediacaran to the Early Cambrian.

5.2 Methods

The Paleobiology Database records fossil collections along with their depositional environments, identified mostly by sedimentary structures and trace fossils (https://www.paleobiodb.org/public/tips/environtips.html). Fossil assemblages were grouped into six environmental zones representing a broad spectrum of carbonate and siliciclastic shelf deposits. The inner shelf zone is defined by near coast tide-dominated areas and shoreface including peritidal, subtidal environments, shoreline deposits and sand shoal with varying and restricted salinity. This zone is above the fair-weather wave

base, the water depth above which the sea bed is affected by the motion of waves that occur during calm weather (typically 5-15m). Within inner shelf environments, there are three zones that can be separated in terms of physical conditions: (1) high stress environment, including peritidal and shoreline (Zone 1), (2) variable nearshore and protected subtidal environment (Zone 2), such as lagoons and delta platforms, and (3) wave-agitated shallow subtidal environment (Zone 3). The mid-shelf (Zone 4) is an extended subtidal zone and transition zone between fair-weather wave zone and normal storm wave base, the latter is the water depth (typically about 15-40m) to which average storm waves can affect the sea floor (Peters and Loss 2012). The outer shelf (Zone 5) is characterized by mud-dominated, offshore environments below normal storm wave base. The deep water (Zone 6) is a carbonate slope or siliciclastic submarine fan environments characterized by slumps, slides, debrites, and turbidites, as well as basinal oozes. This environment framework of six zones is used to assign fossil assemblages along an onshore-offshore gradient. The correspondence between environments in the PaleoDB and the environmental zones described above are summarized in Table 5.1. Everything else (e.g., unknown or imprecise calls) was excluded from environmental assignment. For example, the occurrences with an environmental call of "marine indet." in the Paleobiology Database cannot be classified into any finer determination than a siliciclastic coastal or marine system. With this filter, a total of 20692 occurrences could be assigned to one marine environmental zone. The approaches of estimating alpha, beta and gamma diversity and the rate measures have been introduced in previous chapter (see Chapter 4: Methods). The multiplicative measure of beta diversity is applied to analyze within-zone beta diversity; whereas the pairwise dissimilarity measure of beta is applied for analysis of between- zone beta diversity. Origination and extinction rates are based on Foote’s per-capita rates (Foote 2005). Within 14 stratigraphic intervals from the late Ediacaran to the earliest Ordovician and six environmental zones, the data could be converted into a 14-by-6 matrix for between-zone beta analysis. Assemblages or collections within each cell of the matrix were pooled. Only zones that contained at least 5 fossil collections were used in

computing pairwise dissimilarity. Percent dissimilarity was used to estimate beta diversity between pairs of zones within the same stratigraphic interval.

Table 5.1 Description of generalized environmental zones used to array fossil assemblages into an onshore- offshore gradient.

Onshore- Environmental Number of offshore Environmental calls in the PaleobioDB zones collections gradient “peritidal”, “estuary/bay”, “lagoonal”, “paralic Zone 1 110 indet.”, “delta plain”. “open shallow subtidal”, “lagoonal/restricted Inner Zone 2 731 shallow subtidal”, “shallow subtidal indet.”, shelf “foreshore”. “sand shoal”, “intrashelf/intraplatform reef”, Zone 3 252 “shoreface”, “delta front”, “perireef or subreef”, “reef, buildup or bioherm”. "deep subtidal ramp", "deep subtidal shelf", "deep Mid-shelf Zone 4 598 subtidal indet.", "platform/shelf-margin reef", "transition zone/lower shoreface", "prodelta". Upper "offshore ramp", "offshore shelf", "offshore indet.", Zone 5 665 shelf "slope/ramp reef", "offshore". "slope", "basinal (carbonate)", "basinal (siliceous)", Deep Zone 6 1034 "basinal (siliciclastic)", "submarine fan", "deep- water water indet.".

The interval estimates of taxonomic rates tend to be affected by interval length (Foote 2009). Here I use the per-capita rate (Foote 2000) to accurately assess variation of origination/extinction rates along onshore-offshore gradient. Because this approach is estimated from numbers of boundary crossers, and thus, is independent of interval length. This is also the only metric for which origination rate does not affect the estimation of

extinction rate and vice versa.

5.3 Results

5.3.1 Gamma diversity

Gamma diversity in the environmental zones shows different patterns through the whole Cambrian (Figure 5.1), with deep subtidal settings harboring most of the genera in the Fortunian, and the diversity increasing sharply in Stage 3 in both nearshore and offshore settings. Peritidal environments consistently yield the lowest diversity in the Cambrian. Diversity patterns in environmental Zone 2, 3, and 5 increased from the Fortunian to Cambrian Stage 3 and then dropped in Cambrian Stage 4. The generic richness in shallow subtidal environments shows peaks in Stage 3 and in the Drumian. The number of genera in wave-agitated environments (Zone 3) is depressed through the Cambrian after achieving its maximum in the Stage 3. The same pattern of diversity can also be seen in offshore settings (Zones 4, 5, and 6), where generic richness peaked in the Stage 3 (Zone 5) and in the Stage 4 (Zones 4 and 6), and then dropped erratically through the rest of the Cambrian. Truly deep water genera first appeared in Stage 4, quickly achieved their diversity maximum in the following stage (Stage 5), and then fluctuated through the rest of the Cambrian. The generic richness in deep subtidal environments remained low compared to other environmental zones during the first four Cambrian stages. Post-stage 4 pattern in deep subtidal environments corresponds well to the one in deep water settings if not considering the values in Jiangshanian.

5.3.2 Alpha diversity

Figure 5.2 illustrates time series of mean alpha diversities computed for each environmental zone from the late Ediacaran to the earliest Ordovician. Diversity is lowest in environmental Zone 1, where sampling standardization could not be performed for the entire time series. This zone is largely inhabited by trilobites, brachiopods, and mollusks. Curves of alpha diversity in environmental Zone 2 and Zone 3 tend to increase dramatically from the Fortunian to the Stage 3. Beyond environmental Zone 3, mean alpha diversity shows less variation. Across environmental zones mean alpha diversity increased only slightly from the Fortunian to Stage 2 and showed no persistent increase

or decrease thereafter (see Figure 5.3 and Table 5.2). This is qualitatively different from the pattern of alpha diversity in previous chapter (Figure 4.2).

Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Number of genera (SQ) genera of Number

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian -40 -20 0 20 40 60 80 100 560 540 520 500 480

Age (Ma)

Figure 5.1 Gamma diversity of marine animals for each environmental zone and each geologic stage from the Ediacaran to the earliest Ordovician. Sampling-standardized genus-level diversity (sampled-in-bin) based on shareholder quorum subsampling with 60% frequency coverage per stage. Error bars are SDs of 100 subsampling trials for all marine genera.

Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 Alpha diversity 0 5 10 15 20 Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian 560 540 520 500 480 Age (Ma)

Figure 5.2 Mean alpha diversity of marine animals for each environmental zone and each geologic stage from the Ediacaran to the earliest Ordovician. Sampling-standardized genus-level diversity (sampled-in-bin) based on by-list unweighted subsampling of 6 collections per stage. Error bars are SDs of 100 subsampling trials for all marine genera.

5.3.3 Beta diversity

5.3.3.1 Within-zone beta diversity

Like within-zone alpha diversity, I calculated generic turnover among collections based on multiplicative approach (see Chapter 4) to assess the variation of beta diversity for each environmental zone from the Ediacaran to the earliest Ordovician. Compare to alpha diversity, beta diversity shows less variation among environmental zones (Figure 5.4 and Table 5.3). In order to test whether within-zone beta diversity might be changing through time during the whole Cambrian, beta diversity was firstly computed for each environmental zone in each stratigraphic interval. One-way analysis of variance (one-way

ANOVA) detected no significant differences among mean values for ten stage-level intervals of the Cambrian (F = 0.94, df = 11.17, P > 0.5), indicating that variation in faunal composition within certain environments is unlikely to cause a substantial bias on patterns of between-zone beta diversity.

Table 5.2 Average generic alpha diversities for each environmental zone and each geologic period of the Cambrian.

Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 mean For 2.27 5.66 3.97 St 2 4.30 2.87 8.50 3.18 4.89 4.75 St 3 2.93 14.19 1.88 3.18 5.54 St 4 3.12 7.92 2.44 3.83 6.05 4.67 St 5 4.11 4.20 4.43 3.38 13.40 5.90 Dru 6.83 3.50 4.46 3.25 6.18 4.84 Guz 4.34 5.86 4.64 3.49 9.45 5.56 Pai 2.44 4.85 8.61 4.98 5.22 Jia 3.44 3.98 4.17 6.11 4.42 St 10 4.42 7.79 4.93 4.24 6.62 5.60 Mean 0.43 3.68 7.42 4.04 4.34 7.54

5.3.3.2 Between-zone beta diversity

Both mean and median values are plotted in Figure 5.5 to show how genus turnover varied with increasing environmental separation in the Cambrian. Both patterns are basically identical. There are increases in pair-wise dissimilarity for environmental separations of one to two and three to five zones in the Cambrian. This finding is similar to the basic trend documented in Sepkoski’s analysis of beta diversity in the Cambrian (Sepkoski 1988). Beyond the adjacent zones, there is no obvious change in pair-wise dissimilarity among two to three environmental separations in the Cambrian, but a clear increase among three to five environmental separations. The basic pattern is also seen in mean between-zone beta diversity (Figure 5.6).

Mean within-zone alpha diversity within-zone Mean

For St2 St3 St4 St5 Dru Guz Pai Jia St10

Cambrian 02468

540 530 520 510 500 490

Age (Ma)

Figure 5.3 Average generic alpha diversities for each environmental zone through the whole Cambrian period. Error bars are SDs of mean estimates.

Table 5.3 Average generic beta diversities for each environmental zone and each geologic period of the Cambrian.

Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 6 mean For 4.83 1.80 1.11 St 2 3.67 3.90 3.77 4.49 4.25 3.35 St 3 3.71 3.73 4.71 5.10 5.47 3.79 St 4 4.77 4.27 5.31 3.82 3.03 St 5 4.08 5.62 5.22 4.42 3.22 Dru 4.55 5.06 4.68 4.53 3.14 Guz 3.78 5.20 4.73 4.39 3.02 Pai 3.91 5.16 4.17 4.35 2.93 Jia 4.67 4.67 3.19 4.41 4.01 4.57 4.25 St 10 4.29 5.11 3.53 4.09 2.84 Mean 1.21 1.23 3.71 4.93 4.32 3.02

Beta diversity

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre 0246810 Ediacaran Cambrian

560 540 520 500 480 Age (Ma)

Figure 5.4 Beta diversity of marine animals for each environmental zone and each geologic stage from the Ediacaran to the earliest Ordovician. Sampling-standardized genus-level diversity (sampled-in-bin) based on by-list unweighted subsampling of 6 collections per stage. Error bars are SDs of 100 subsampling trials for all marine genera.

Figure 5.6 illustrates the variation of mean beta diversity measured by pairwise dissimilarity between adjacent environmental zones for each stage-level interval through the Cambrian (also see Table 5.4). Beta diversity is lowest in the Fortunian and higher, but not continually increasing, from Stage 3 onwards, with a dramatic drop in the Guzhangian stage. This pattern reflects that Fortunian, Stage 2 and Guzhangian have high degrees of taxonomic homogeneity in onshore-offshore settings. This observation could be due to environmentally unconstrained distribution of the earliest Cambrian skeletal fauna, ranging from nearshore to offshore in open marine environment. The early increase in beta diversity within adjacent environmental zone may be a result of transgression from the late Ediacaran to the early Cambrian (Brasier 1982), which largely produced a bias towards to preservation of suitable habitats (Mount and Lignor 1992). Figure 5.7 shows the variation of mean beta diversity measured by pairwise dissimilarity

with increasing environmental separation through Ediacaran-Cambrian times. There is no pronounced difference among the curves, which is consistent with what I observed in Figure 5.5.

Percent dissimilarity Percent dissimilarity Percent dissimilarity Percent

Mean Median Ediacaran (Median) 0.70 0.75 0.80 0.85 0.90 0.95 1.00

12345

EnvironmentalEnvironmental separation separation

Figure 5.5 Beta diversity (percent dissimilarity) as a function of environmental separation. Red and black lines are for the Cambrian. Mean values of the percent dissimilarity are plotted on ordinate against separation of the environmental zones on the abscissa (1 = adjacent zones, 2 = one zone in between, etc.).

5.3.4 Origination and extinction rates along onshore-offshore gradient

The variation of origination and extinction rates in each environmental zone along an onshore-offshore gradient may affect beta diversity during the Cambrian. Origination rates tend to be higher in shallow water environments than deep water environments (Kiessling et al. 2010). High origination rates may eventually increase beta diversity in shallow water settings more than in deep water settings. Figure 5.8 and Figure 5.9 display plots of origination/extinction rates in adjacent environmental zones from Cambrian Stage 3 to Cambrian Stage 10. There is no strong difference between

origination/extinction rates in adjacent environmental zones. This finding indicates that the variation of origination/extinction rates in different environmental settings is unlikely to introduce a strong bias on the pattern of beta diversity during the Cambrian.

Mean between-zone beta diversity diversity Mean beta between-zone Mean between-zone beta diversity beta between-zone Mean

For St2 St3 St4 St5 Dru Guz Pai Jia St10

Cambrian 0.5 0.6 0.7 0.8 0.9 1.0 540 530 520 510 500 490

Age (Ma) Age (Ma)

Figure 5.6 Mean beta diversities for each environmental zone through the whole Cambrian period. Error bars are SDs of mean estimates

5.4 Discussion

Mean alpha diversity shows that ecological innovation appeared first in open marine environments and then expanded into offshore settings. Beyond environmental Zone 1, alpha diversity tends to increase dramatically from the Fortunian to the Stage 3. The pronounced alpha diversity in Cambrian Stage 3 seems to result from the inclusion of large composite faunal lists of Archaeocyaths from Russia. This finding differs from Sepkoski’s result (Sepkoski 1988) in which alpha diversity changed little among environmental zones, because he excluded reef environments from his data. The low level of alpha diversity in the other environmental zones spanned the first five Cambrian

stages. Environmental Zone 3 is defined as wave-agitated settings that include bioherm- rich areas. This zone is dominated by archaeocyath sponges which are quite diverse (Sepkoski 1979). That means, alpha diversity in Zone 3 achieved its maximum in Cambrian Stage 3 may indicate a high degree of niche packing in reefal communities (Zhuravlev et al. 2015).

Table 5.4 Percent dissimilarity between adjacent environmental zones.

zone1-2 zone2-3 zone3-4 zone4-5 zone5-6 For NA NA NA 0.76 0.71 St 2 0.81 0.72 0.72 0.84 - St 3 0.93 0.82 0.92 0.86 - St 4 0.76 0.98 0.95 0.91 0.86 St 5 0.96 0.92 0.87 0.72 0.84 Dru 1.00 0.95 0.92 0.77 0.69 Guz 0.79 0.52 0.61 0.74 0.88 Pai 0.93 1.00 1.00 0.71 0.71 Jia 0.66 0.95 0.97 0.81 0.89 St 10 0.78 0.84 0.73 0.91 0.83

Mean alpha diversity shows less variation from mid-shelf to deep water environments (Figure 5.2). The differences in diversity between Zones 4 and 5 are minor. The diversity in Zone 6 is pronounced in Cambrian Stage 5 and Guzhangian. The evolution of deep marine type communities starts in Cambrian Stage 4, when global diversity declined markedly by an extinction event (Zhuravlev and Wood 1996, Kiessling et al. 2002, Luchinina et al. 2013). Compared to diversities in shallower environmental zones, higher diversity in deep water settings might be due to tropical warming in later Cambrian. During tropical warming, the temperature in shallow marine areas tends be higher than that in deep marine areas. Consequently warming-resistant taxa in deep water tend to diversify while severe physical conditions subdued the increase in number of taxa in shallow water to inhabit.

Percent dissimilarity dissimilarity Percent Percent dissimilarity (mean) dissimilarity Percent

1 sep. 2 seps. 3 seps. 4 seps. 5 seps.

Ava Whs Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Ediacaran Cambrian 0.0 0.2 0.4 0.6 0.8 1.0 560 540 520 500 480

AgeAge (Ma) (Ma)

Figure 5.7 Mean beta diversity (percent dissimilarity) of marine animals with increasing environmental separations from the Ediacaran to the earliest Ordovician.

The Ediacaran period displays virtually no clear trend of beta diversity beyond the adjacent zones (Figure 5.5), which may explains the absence of typical Ediacaran fossils from carbonates (Laflamme et al. 2013) and near-global distribution of many genera (Waggoner 2003). Likewise, the Cambrian period shows no clear trend in beta diversity in terms of pair-wise dissimilarity in relation to increasing environmental heterogeneity. This pattern indicates that there is a substantial component of the Cambrian fauna that was without habitat specialization and living in shallow-water environmental zones across the shelf. Inspection of the raw data indicates that this such substantial component includes many of the trilobite genera that lend their names to platform-facies biostratigraphic zones (e.g., Blountia, Olenellus, Elrathina, Cedarina, and others) as well as some inarticulate brachiopods (e.g., Acrotreta, Lingulella, and Micromitra) and other taxa (e.g., Hyolithes). However, in deep-water environments, the number of Cambrian

fauna with specialized habitat increased regardless of that there were numerous taxa achieving a wide range of distribution in deep water settings, such as Agnostus, Anabarites, and Beltella etc.

Guz Jia Dru St 2 Pai St 4 St 10 St 2 St 5 St 5 St 3 St St10 4 St 3 DruGuz Whs Origination rateZone in 3 Origination rateZone in 4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Origination rate in Zone 2 Origination rate in Zone 3

Guz

Jia St 10 Guz St 4Dru St 5 Pai Pai St 2 StDru 4 Jia St St3 5

Tre Origination rate in Zone 5 Origination rate in Zone 6 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Origination rate in Zone 4 Origination rate in Zone 5

Figure 5.8 Plots of origination rates between adjacent environmental zones along onshore-offshore gradient during the Cambrian. The rates are standardized based on UW subsampling. Rates are per- capita rates of Foote (2005). Error bars are standard deviations of 100 subsampling trials.

St 4 St 2

Dru Pai Jia Guz St 4 St 5 St 10 Dru St 5 St 3 Guz St 3 St 10 St 2 Whs Extinction rate in Zone 3 Extinction rate in Zone 4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Extinction rate in Zone 2 Extinction rate in Zone 3

Jia Dru Guz Pai Dru Jia Guz Pai St 5 St 10 St 3

Tre St 4 St 5

St 2 St 4 Extinction rate in Zone5 Extinction rate in Zone6 -0.20.00.20.40.60.81.0 -0.20.00.20.40.60.81.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 Extinction rate in Zone 4 Extinction rate in Zone 5

Figure 5.9 Plots of extinction rates between adjacent environmental zones along onshore-offshore gradient during the Cambrian. The rates are standardized based on UW subsampling. Rates are per- capita rates of Foote (2005). Error bars are standard deviations of 100 subsampling trials.

Earlier analysis of onshore-offshore patterns for the replacement of evolutionary faunas through the Phanerozoic has suggested that major ecological innovations first appeared in nearshore environments and then expanded outward across the shelf (Jablonski et al. 1983). Later analyses of paleoecology of the Early Cambrian fauna suggested that major shelly taxa and trace fossils appeared offshore from shallow-water, physically stressed environments and later spread to nearshore habitats (Mount and Signor 1985). My study shows that gamma diversity in deep subtidal settings harbors

most of the genera in the Fortunian (Figure 5.1). This finding suggests that major ecological innovations in the Cambrian did not appear firstly in nearshore environments, but rather in the deep open marine environments. Moreover, there is no pronounced difference of beta diversity among the different environmental zones (Figure 5.5 and Figure 5.7). This finding supports that variation of genus turnover between communities was not governed by environmental heterogeneity. It has been long been recognized that some major quantitative discrepancies exist between alpha diversity and global diversity has long been known. For example, the increase in alpha diversity across the Ordovician radiation is only about 50 percent (Bambach 1977), whereas the concurrent increase in global diversity is on the order of 300 percent at both familial (Sepkoski and Sheehan 1983) and generic (Sepkoski 1988) taxonomic levels. This missing diversity has been argued, to a great extent, to have been contributed by provinciality (Valentine 1970, 1971, Valentine and Moores 1972, Valentine et al. 1978). This finding of alpha diversity confirms a problem with “missing diversity”. The increase in alpha diversity summarized in Table 5.2 is insufficient to explain the dramatic increase in global generic diversity illustrated in Figure 3.12 (see Chapter 3). In summary, the pattern of diversity partitioning along onshore-offshore gradients demonstrate that there is no clear trend. Major ecological innovations appeared randomly with respect to bathymetric gradients and the variation of genus turnover between communities was not governed by environmental heterogeneity. Furthermore, there seems no such trend as neither the origination rate nor extinction rate displayed a clear environmental bias (see Figure 5.8 and Figure 5.9). This implies that the Cambrian diversification is unlikely to be manifested in an onshore-offshore framework.

6 Chapter 6: Escalatory ecological trends in Cambrian invertebrates and key role of predators in driving the Cambrian radiation

6.1 Introduction

Ecological interactions, such as predation and bioturbation, are thought to fundamental determinants of macroevolutionary trends (Vermeij 1987). The role of predators in controlling taxonomic diversity may be substantial (Paine 1966, Murdoch 1969, Paine 1971, Porter 1972) and the appearance of predators may also partly explain the massive skeletonization event that characterizes the fossil record of the explosion (Vermeij 1989). Although both Bengtson (2002) and Marshall (2006) considered predation as a shaper rather than the initial trigger of the Cambrian Explosion, the importance of predation on the evolution, diversification, biomineralization, and ecology of metazoans has been the focus of intensive research interest over the last 20 years (Vermeij 1987, Conway Morris and Bengtson 1994, Kowalewski 2002). Although there is strong coupling of predation intensity and diversity in the Phanerozoic fossil record (Huntley and Kowalewski 2007), statistical evidence is still lacking to support predation as major evolutionary force of the diversification during the Cambrian radiation. The escalation hypothesis is the most commonly cited hypothesis explaining trends in the fossil record, which invokes negative interactions among ecological groups. The escalation hypothesis predicts that long-term ecological shifts are caused by the evolutionary response of prey to predation pressure (Vermeij 1987). This provokes three potential faunal responses: infaunality, mobility, and skeletonization. Infaunality focuses on prey moving into the substrate for protection from predators; mobility focuses on prey actively and facultatively invading into areas on or above the substrate to evade predators; skeletonization focuses on prey utilizing the ambient environments to form shells in order to defend themselves from predators. Alternatively, the bulldozer hypothesis (Thayer 1983), a subsidiary of the escalation hypothesis, predicts that epifaunal immobile taxa will decline when there is increased biological disturbance of the sediment by infaunal, actively mobile taxa (bioturbators). Trends involving particular taxa (Vermeij 1994) or over short time scales (Aberhan et al. 2006) confirm with these predictions. However, the

long term patterns of ecological proportions did not support escalation as an important shaper in the history of Phanerozoic marine life (Madin et al. 2006). Still, the escalation hypotheses need to be tested by examining trends in relative diversity among marine invertebrates during the Cambrian radiation.

6.2 Methods

I use subsampling diversity of carnivores and non-carnivorous skeletal fauna in addressing the question to which extent the diversity of predators influenced the biomineralization of non-carnivorous taxa. The correlation between diversity of carnivores and global diversity is to assess if the predators played an important role in the diversification during the Cambrian radiation. My approach to testing escalatory ecological trends is similar to Madin et al. (Madin et al. 2006) but focuses on a narrower time interval. The escalation hypothesis is to be tested by examining trends in relative diversity (proportions of genera) among marine metazoans as a whole. Cross-correlation is to be assessed between the patterns in the proportion of carnivorous marine invertebrates and the proportions of infaunal or mobile prey or skeletal prey and between the proportion of bioturbators and the proportion of immobile epifauna. Using proportions within each ecological category instead of raw counts is to avoid problems associated with variation in sampling intensity that plague estimates of global diversity (Alroy et al. 2001). I classified marine invertebrate taxa into three ecological categories (diet, life habit, and locomotion) and one preservational category (mineralogical composition). As the analyses concern interactions among benthic invertebrates as well as pelagic organisms, counts of nektonic taxa and occurrences were also included. However, planktonic taxa and occurrences were excluded because planktonic taxa are not strong indicators for predator-prey interactions. Carnivores are assigned for taxa characterized as “carnivore” in the Paleobiology Database, and non-carnivores are the rest of the taxa in the dataset. Mobile fauna were defined by occurrences characterized as locomotion calls in the PaleoDB of “actively mobile”, “facultatively mobile”, “fast-moving”, and “slow-moving”. Then infauna was defined as all infaunal taxa in non-carnivores. Skeletal prey is defined by occurrences characterized as compositional calls in the PaleoDB of

“calcite”, “aragonite”, “high Mg calcite”, “hydroxyapatite”, “low Mg calcite”, “silica”, and “phosphatic”. Accordingly, infaunality is evaluated by proportions of genera for taxa classified as infaunal non-carnivores. Mobility is evaluated by the proportions of genera classified as mobile non-carnivorous including actively mobile, facultatively mobile, fast-moving, and slow-moving. Bioturbation is assessed by proportions of genera for taxa classified as infaunal, actively mobile. For statistical independence of groups, proportions of carnivores are given relative to all taxa, but proportions of non-carnivorous groups are given relative to all non- carnivorous taxa. Proportions of skeletal prey are given relative to all non-carnivorous taxa. Bioturbating and immobile epifaunal genera do not exhibit both modes of life habit or diet; therefore, proportions of bioturbators are given relative to all taxa, whereas proportions of immobile epifauna are given relative to all non-bioturbating taxa. Cross- correlations between logit-transformed proportions of occurrences of carnivorous and non-carnivorous infaunal, carnivores and mobile, carnivores and skeletal prey, and bioturbating and immobile epifaunal marine invertebrates are conducted to assess if escalation drives macroecological trends at global and million-year time scales. Model selection was performed using Akaike information criterion (AIC) in a Stepwise Algorithm (MASS package in R). AIC is a measure of goodness-of-fit that is penalized according to the number of parameters (Akaike 1998, Aho et al. 2014): the model with the lowest AIC is the estimated best model. It is a convenient metric for the evidence favouring each model and can be used to indicate which model or models are most consistent with the changes of global diversity.

6.3 Results

6.3.1 The role of predators for the Cambrian radiation

Figure 6.1 shows that the standardized diversity of carnivores increased remarkably from the Fortunian to Stage 3, which is consistent with the pattern of global diversity estimated in Chapter 3. There is a significant correlation between the diversity of predators and global diversity during the Cambrian (Figure 6.2 A, Spearman’s rho = 0.84, p-value = 0.003). This correspondence is not an independent auto-correlated trends (Figure 6.2 B,

first-difference ordering: Spearman’s rho = 0.81, p-value = 0.008), suggesting that the predation might be one of the major drivers of the “Cambrian Explosion”.

Diversity of carnivores of Diversity 0 20406080 Nam For St2 St3 St4 St5 Dru Guz Pai Jia St10 Tre Edi Є O

540 530 520 510 500 490 480

Age (Ma)

Figure 6.1 Counts of genera for taxa classified as carnivores in the Cambrian based on shareholder quorum subsampling with 60% frequency coverage per stage. Error bars are SDs of 100 subsampling trials for all genera.

6.3.2 Increase of proportional diversity of ecological groups in the Cambrian

Figure 6.3 depicts the variation of proportional diversity of ecological groups during the Cambrian, showing that proportional diversity of carnivores increased strongly from the Fortunian to Stage 5 and fluctuated through the rest of the Cambrian period (Figure 6.3 A). The raw data in Figure 6.3 shows concerted changes in the dominance of ecological strategies. Increase in non-carnivorous mobility and skeletal prey may imply an evolutionary response to an increase in carnivores. Raw cross-correlations between the ecological categories superficially suggest that only correlation between proportions of carnivores and non-carnivorous mobility is significant (Figure 6.4B, Spearman’s rho = 0.79, p-value = 0.006), suggesting that two groups were caught in an escalatory struggle.

However, analytical test shows that there are independent autocorrelated trends proportional diversity of ecological groups of carnivores, non-carnivorous mobile, skeletal prey and immobile epifaunal (Figure 6.5). The cross-correlation weaken considerably after differencing (Figure 6.6B and Figure 6.7B, Spearman’s rho = 0.45, p- value = 0.19). The same pattern is also shown in cross-correlations among proportions of occurrences (Table 6.1).

A B Log (Global diversity) (Global Log Log (Global diversity) (Global Log  234567 -1 0 1 2 3 012345 -10123 Log (Diversity of carnivores) Log (Diversity of carnivores)

Figure 6.2 Correlations between log-transformed diversities of global and carnivorous marine invertebrates in the Cambrian. A is based on raw data; B is based on data after first-order differencing.

A B Carnivores Non-carnivorous infaunal

Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre Nam For St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Edi Є O Edi Є O -0.1 0.0 0.1 0.2 0.3 0.4 0.5 540 530 520 510 500 490 480 -0.02 0.02540 0.06 530 0.10 520 510 500 490 480

C D

Non-carnivorous mobile Skeletal prey Proportion Proportion

0.0 0.2Nam 0.4 0.6 For 0.8 St2 St3 St4 St5 DruGuzPai Jia St10 Tre 0.0Nam 0.2 0.4 For 0.6 St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Edi Є O Edi Є O

540 530 520 510 500 490 480 540 530 520 510 500 490 480

E F

Bioturbators Immobile epifaunal

Nam For St2 St3 St4 St5 DruGuzPai Jia St10 Tre 0.0Nam 0.2 0.4 For 0.6 St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Edi Є O Edi Є O

-0.02 0.02540 0.06 530 0.10 520 510 500 490 480 540 530 520 510 500 490 480

Time (Ma)

Figure 6.3 Proportions of genera (solid line) and occurrences (dashed line) for taxa classified as carnivorous (A), non-carnivorous infaunal (B), non-carnivorous mobile (C), skeletal prey (D), bioturbating (E), immobile epifaunal (F). For statistical independence of groups, proportions of carnivores are given relative to all taxa, but proportions of non-carnivorous groups [(B), (C) and (D)] are given relative to all non-carnivorous taxa. Bioturbating and immobile epifaunal genera do not exhibit both modes of life habit or diet; therefore, proportions of bioturbators are given relative to all taxa, whereas proportions of immobile epifauna are given relative to all non-bioturbating taxa. Letters at the bottom of each panel are abbreviations for geological periods and stages. Ma, million years ago. 92

A B Logit (proportion) (proportion) Logit Non-carnivorous mobile Non-carnivorous infauna -6 -5 -4 -3 -2 -2 -1 0 1 2 -4 -3 -2 -1 0 -4 -3 -2 -1 0 Carnivores Carnivores

C D Skeletal prey Skeletal Immobile epifauna Immobile -3 -2 -1 0 1 -3.0 -2.0 -1.0 0.0 -4 -3 -2 -1 0 -6.0 -5.0 -4.0 -3.0 Carnivores Bioturbators

Logit (proportion)

Figure 6.4 Cross-correlations between logit-transformed proportions of genera of carnivorous and non-carnivorous infaunal (A), carnivorous and mobile (B), carnivorous and skeletal (C), and bioturbating and immobile epifaunal (D) marine invertebrates based on raw data. Correlations on the left involve raw data. Solid lines represent least-square lines of best fit.

Carnivores Infaunal Mobile ACF ACF ACF -0.5 0.0 0.5 1.0 -0.5 0.0 0.5 1.0 -0.5 0.0 0.5 1.0

0246810 0246810 0246810 Lag Lag Lag

Skeletal Bioturbating Immobile epifaual ACF ACF ACF -0.5 0.0 0.5 1.0 -0.5 0.0 0.5 1.0 -0.5 0.0 0.5 1.0

0246810 02468 0246810 Lag Lag Lag

Figure 6.5 Autocorrelation test for proportional diversity of ecological groups (genera).

A B Δ logit (proportion) (proportion) logit Non-carnivorous mobile Non-carnivorous infauna -1 0 1 2 3 -2 -1 0 1 2 -10123 -10123 Carnivores Carnivores

C D Skeletal prey Skeletal Immobile epifauna Immobile -2 -1 0 1 2 -2 -1 0 1 2 -10123 -10123 Carnivores Bioturbators

ΔLogit (proportion) Figure 6.6 Cross-correlations between logit-transformed proportions of genera of carnivorous and non-carnivorous infaunal (A), carnivorous and mobile (B), carnivorous and skeletal (C), and bioturbating and immobile epifaunal (D) marine invertebrates based on data after first-order differencing. Solid lines represent least-square lines of best fit. ΔLogit, change in the logit- transformed proportion.

Table 6.1 Summary of test statistics of cross-correlations for the ecological groups. Independent = independent variable; Dependent = dependent variable; N = sample size; Occurrences = correlations involving counts of occurrences; Genera = correlations involving counts of genera.

Occurrences Genera Independent Dependent N (Spearman’s ρ) (Spearman’s ρ) A. Undifferenced Carnivores Infauna 11 0.6 0.39 Carnivores Mobile 11 0.65* 0.79*** Carnivores Skeletal 11 ‐0.56 ‐0.57 Bioturbators Immobile epifauna 10 0.28 0.04 B. First order differencing Carnivores Infauna 10 0.3 ‐0.38 Carnivores Mobile 10 0.006 0.45 Carnivores Skeletal 10 ‐0.13 ‐0.47 Bioturbators Immobile epifauna 9 0.32 0.12 C. Generalized differencing Carnivores Infauna 0.24 ‐0.14 Carnivores Mobile 0.08 0.45 Carnivores Skeletal ‐0.2 ‐0.49 Bioturbators Immobile epifauna 0.42 0.2 * P < 0.05 ** P < 0.01 *** P < 0.001

A B Δ logit (proportion) (proportion) logit Non-carnivorous mobile Non-carnivorous infauna -2 -1 0 1 2 -10123 -10123 -10123 Carnivores Carnivores

C D Skeletal prey Skeletal Immobile epifauna Immobile -2 -1 0 1 2 -2 -1 0 1 2 -10123 -10123 Carnivores Bioturbators

ΔLogit (proportion) Figure 6.7 Cross-correlations between logit-transformed proportions of genera of carnivorous and non-carnivorous infaunal (A), carnivorous and mobile (B), carnivorous and skeletal (C), and bioturbating and immobile epifaunal (D) marine invertebrates based on data after generalized differencing. Solid lines represent least-square lines of best fit. ΔLogit, change in the logit- transformed proportion.

6.3.3 Low-competition mode

Model calculations predict that pathways of alpha- and beta- diversity in diversifying ecosystems notably differ depending on the relative role of various ecological interactions and environmental parameters (Hautmann 2014). The relationship between alpha and beta diversity during the time of unbounded diversification (Figure 6.8), that is, from the Fortunian to Cambrian Stage 3, suggests that a low-competition model best

explains the Cambrian radiation. When competition is low, the addition of new species to a community happens either by exploitation of previously unutilized resources or by packing more species in marginal ecospace (Whittaker 1972). This process will initially not shrink niches of pre-existing species such that alpha diversity will increase steeply, whereas beta diversity will initially increase only moderately. As diversification continues, competition will increase, niches will contract and consequently beta diversity will increase profoundly, whereas alpha diversity levels off. Niche contraction facilitates niche partitioning, and thus changes of taxonomic composition among habitats. Low competition can have several different causes. One of the potential causes is intensive predation which can keep abundances of organisms well below the carrying capacities (Stanley 2007, 2008). Global biodiversity patterns can be explained by three key ecological variables: within-community diversity (alpha), between-communities turnover (beta), and the extent of predation (proportional diversity of carnivores). I found that the changes of gamma diversity during the Cambrian radiation were mainly governed by taxonomic differentiation between communities, that is, beta diversity (Chapter 4). I also found a strong correlation between diversity of carnivores and gamma diversity (see above). The question is to which degree predations drove gamma diversity by influencing generic richness and turnover in local communities? This question can be best answered by testing to include or exclude explanatory variables in models. Thereby, five independent parameters are considered to explain gamma diversity: alpha, beta, diversity of predators, proportional diversity of predators based on both occurrences and genera. I calculate Akaike's information criterion (AIC) for one or several fitted model objects for which a log-likelihood value can be obtained When the variables are combined in spatial multiple regression analysis, model selection by AIC indicates that beta diversity + alpha diversity + proportions of genera of carnivores is the best model to explain the changes in global diversity (gamma) during the Cambrian radiation (Table 6.2), in terms of that AIC value is the lowest. Akaike weights () provide measure of the strength of evidence for each model. The highest value of Akaike weight indicate the probability that the model is the best among the whole set of candidate models.

Figure 6.8 Alpha-beta-gamma plot for the first three Cambrian stages, the time of continuous diversification. Note log-scale of all axes. Error bars are standard deviations of 100 subsampling trials.

Table 6.2 Model selection for the analyses of global biodiversity. Dependent variable is gamma diversity. Four independent variables are alpha diversity, beta diversity, proportional occurrences of carnivores, and proportional genera of carnivores respectively. The models include all possible combinations of the explanatory variables. Carni represents diversity of carnivores; Pgens is the proportion of carnivore genera; Poccus is the proportions carnivore occurrences. Lower AIC and AICc values mean a better fit.

Number of Akaike Models AIC AICc parameters weight () gamma ~ beta + alpha + carnivore + Pgens + Poccus 5 0.09 103.18 140.51 gamma ~ beta + alpha + Pgens + Poccus 4 0.26 101.19 122.19 gamma ~ beta + alpha + Pgens 3 0.62 99.44 111.44 gamma ~ beta + alpha 2 0.02 106.21 112.88 gamma ~ beta 1 0.00 117.09 120.52 gamma ~ alpha 1 0.00 132.54 135.97

6.4 Discussion

There is no clear evidence that escalation drove ecological trends during the Cambrian radiation. The principal role of predation in the Cambrian diversification was thus to keep the system in a low-competition mode (Araujo et al. 2011, Na and Kiessling 2015). Experimental studies indicate that removal of a top predator in marine ecosystem leads to a reduction in the number of species by making space available for competitively subordinate species (Paine 1966, Hautmann 2015). Such experiments have been performed in the environment of hydrothermal vents at a depth of 2500 m in the eastern tropical Pacific Ocean (Micheli et al. 2002). Top predators have an indirect beneficial effect on inferior competitors by depressing abundance of a superior competitor. Predation universally defeats competition in benthic marine systems (Stanley 2008) suggesting that the Cambrian radiation was not exceptional in this aspect. “The Cambrian Explosion” involves radiations of many groups of taxa, including molluscs, brachiopods, arthropods, poriferans. The theory of adaptive radiation proposes that competition for resources drives ecological divergence, the diversification of forms, and, ultimately (with reproductive isolation), the formation of new species (Schluter 2000, Losos 2010). Broad-niched, generalized species are more ecologically abundant and trophically dominant, and they also tend to evolutionarily expand into more diverse clades because their adaptations allow them to preemptively evolve into niche space in a finite world. Moreover, it is hypothesized that available niches decline and the rate of diversification slows down as the diversity of competing species increases (Schluter 2000). This competition-related selection is an efficient way in which resources can be better retained, appropriated, and regulated (Vermeij 1994). The extent to which adaptation in any one direction can occur is limited by conflicting functional demands, that is, by trade-offs among incompatible requirements. Thereby, which functional trait or sets of traits operates in the light of fitness determines what kind of selective pressure that species are under (Reich et al. 2003). According to my results, there seems no evolutionary force such as escalating between predation and prey driving the Cambrian biodiversity. This finding supports the argument at Phanerozoic time scales that escalation did not drive macroecological trends (Madin et al. 2006). However, my findings support that the role of predators in

100

controlling taxonomic diversity may be very substantial (Paine 1966, Murdoch 1969, Paine 1971, Porter 1972). However, whether there is predation in an ecosystem or not, niches will always contract because of inter- or intra-specific competition. But predation will determine if niche contraction happens earlier or later. Assume that there is no predation in the earliest Cambrian at all, species will quickly achieve their maximum population and mutual competition will be very strong due to limited resources. In this case, the new addition of species will increase niche filling (alpha) and niche contraction (beta) at the same time. However, this is not case in my study. My finding of alpha-beta- gamma relationships during the first Cambrian stages suggests that alpha diversity initially increased faster than beta diversity. Intense predation keeps prey species at low population size, which means species slightly detract from each other's resources and exhibit a large degree of niche overlap. This will result first in an increase of alpha diversity as diversification proceeds. Increase in competition level will later force species narrow their habitats to achieve an ecological optimum, which is the case in the earliest Cambrian (Hautmann 2014). The analysis of model selection suggests that the best model of explaining the increase of gamma diversity during the Cambrian radiation is the one include alpha, beta, and proportional genera of carnivores. This result supports my speculation that the increase in proportional diversity of predators initiated local richness and enhanced niche contraction and thus increased the beta diversity eventually. Competitive interaction over geological times is often assessed by waxing and waning of clade diversity (Gould and Calloway 1980). The species that first captures a resource or habitat can rule the niches until something unexpected removes it (Rosenzweig and Mccord 1991). After local or global extinction, a competitor can move in and take over (Pimm 1991). Extinctions will tend to eliminate preferentially organisms with high metabolic demands as well as those with few energy reserves. Consequently, those species that are most functionally specialized, most highly escalated, and most energy-demanding are especially prone to extinction. In summary, through the Cambrian time additional species can fit into the community by occupying new niches and the niche space can become increasingly complex. This complexity is expressed by the increase in alpha diversity, the number of

101

species coexisting in communities. Alternatively, species can be added along environmental gradient such that beta diversity increased later. The total or gamma diversity is a product of the mean alpha diversity of its communities and the degree of beta differentiation among them (Whittaker 1972). To conclude, I found that gamma diversity or global diversity is mainly governed by between-community diversity, that is, beta diversity. Here I reconcile the finding of predation as a potential driver of the Cambrian radiation and the dominant role of beta diversity. The drivers of beta diversity varied with spatial scale. At local scale, an increase of beta diversity was due to niche contraction, which in turn may have been fueled by predation. At a global scale, the trajectory of diversity might be more dependent on abiotic changes of environments.

102

7 Chapter 7: Did plate tectonic drive the Cambrian radiation?

7.1 Introduction

Large-scale patterns of biodiversity depend on changes in geographical distributions of species over time (Servais et al. 2009). Understanding the spatial distribution of species offers clues to the drivers of diversity. It has been proposed by Li et al. (2007) that substantial geographic variation might exist in the patterns of global diversity in the Cambrian. I demonstrated that an increase of beta diversity was the principal driver of increasing gamma diversity during the Cambrian radiation (see Chapter 4). The pivotal role of beta diversity in structuring global diversity patterns has also been suggested for early Cambrian reef communities (Zhuravlev and Naimark 2005). Beta diversity strongly depends on both the sizes of sampling area and sampling units (Hortal et al. 2010). Biotic interactions facilitated the increase of beta diversity in terms of niche contraction in a low competition mode (see Chapter 6). However, at regional or continental scales, beta diversity or community differentiation between regions does not necessarily increase when niches contract, because environmental gradients can be strongly interrupted by dispersal limitation and topographic isolation. Ecological factors that regulated genus diversity in a local community might not be a major driver of diversity at a regional or continental scale. Therefore, there must be other mechanisms that drive beta diversity at continental scales in the Cambrian. Tectonically speaking, profound continental reconfigurations occurred during the Ediacaran-Cambrian periods. These are often referred to as the breakup of the supercontinent Pannotia (Powell and Young 1995, Dalziel 1997, Scotese 2009, Dalziel 2014), which occurred at the very end of the Precambrian and at the beginning of the Cambrian. Pannotia was a short-lived supercontinent. It assembled in the interval from 650 to 550 Ma (Scotese 2009) and comprised Laurentia and the newly completed Gondwanaland. Pannotia breakup was characterized by the opening of the Iapetus and Ægir oceans (Cawood et al. 2001, Torsvik and Rehnstrom 2001, Cawood et al. 2007), resulting in the separation of Laurentia, Baltica, Siberia, and Gondwana, but also orogenies leading to the amalgamation of distinct cratonic blocks within Gondwana, which formed the core of Pannotia. The disassembly of Pannotia (particularly the 103

separation between Laurentia and Gondwana) is suggested to have started close to the onset of the Cambrian (Fortunian; 541 Ma) and ended before Cambrian Stage 3 (521 Ma) (Dalziel 2014). Consequently, I address two questions in this final chapter: (1) Does beta-diversity vary over different spatial scales? (2) If the patterns can shed light on the underlying processes on a continental scale (the breakup of a supercontinent, environmental variation or dispersal limitation)? To answer these two questions, I first assess geodisparity (see Methods below) of genus occurrences for Cambrian stages. This analysis aims to investigate whether there was a secular Cambrian increase in geodisparity. I then look into how the rates of geographic disparity of community assemblages changed over three geographic scales. The purpose of doing this is to find the underlying causes of the variation of geodisparity at continental scales. I discuss the important role that tectonic fragmentation has played in triggering the Cambrian radiation. Finally, I assess the interplay of geodisparity and environmental heterogeneity.

7.2 Methods

7.2.1 Measure of geodisparity

In order to assess the global geographic disparity of marine biotas, Miller et al. (2009) introduced the concept of geodisparity defined as “the degree of global compositional disparity among coeval biotas as a function of geographic distance” (Miller et al. 2009). Geodisparity is related to beta diversity (percent dissimilarity) in that it is a measure of faunal differences, but along geographic rather than environmental gradients. Geodisparity is an approach to estimate beta diversity at pre-defined geographic scales. The analysis of geodisparity through the Phanerozoic has shown that no substantial trends are evident at long-term timescales and thus an increase in provincialism is unlikely over the course of the Phanerozoic (Miller et al. 2009). Additionally, the decrease in community similarity with geographic distance, also called “distance-decay” (Tobler 1970), is a universal biogeographic pattern observed in communities from all domains of life (Nekola and White 1999, Green et al. 2004, Horner-Devine et al. 2004, Martiny et al. 2011). A distance-decay curve is observed because environmental variables tend to be spatially auto-correlated, and organisms with

104

differing niche preferences are selected from the available pool of taxa as the environment changes with distance. Species turnover along spatial environmental gradients (beta diversity or between-communities diversity) produces a decrease of similarity with distance (e.g. Whittaker, 1975; Cody, 1985). Therefore, characterizing the “distance-decay” pattern during the Cambrian radiation may yield insights into the underlying causes of the rise of global beta diversity. Geodisparity was estimated in the context of paleogeographic distance, by using the geodisparity method introduced by Miller et al. (Miller et al. 2009). In order to create a paleogeographic distance matrix, occurrence data for a given stratigraphic interval were pooled in 5 × 5° paleogeographic grids. A dissimilarity matrix of these pooled data was computed, and the data were then parsed into 2,000-km intervals of great circle distances between grid centers. I focused on three geographic scales: within a continent (0–2,000 km), between adjacent continents or within a larger continent (2,000-4,000 km), and between remote continents (greater than 4,000 km). The turnover, or the rate of change in community composition, is represented by the slope of the relationship between dissimilarity and geographic distance. This distance–decay relationship was estimated by linear regression.

7.2.2 Measure of ecodisparity

Ecological disparity (or ecodisparity) was measured by pooling collections in the same basic environmental setting and comparing the faunas between environments that differed by one, two or three environmental categories. Environmental beta is thus estimated by the mean dissimilarity in the context of habitat disparity. Environmental categories were tropical/non-tropical (separated at 30° absolute paleolatitude), carbonate/siliciclastic (distinguished by dominant lithology, marls were ignored), and shallow-water/deep-water (separated by storm wave base). This method differs from the ecodisparity approach in Chapter 6. Ecodisparity in Chapter 6 is measured by taxonomic dissimilarity among environmental zones along an onshore-offshore gradient. It more focuses on the bathymetric category of genus occurrences regardless of information on taxa’s lithology and paleogeographic distribution.

105

7.2.3 Tectonic reconstruction

Plate-tectonic reconstructions for the Cambrian are still poorly constrained and controversial. Even the concept of the late Ediacaran supercontinent Pannotia (Dalziel 1997, Scotese 2009) as implemented in Scotese’s (Scotese 2001) maps may not be universally accepted (Hoffman 1991, Hoffman 1999, Li et al. 2008). However, the most recent revision of Ediacaran-Cambrian plate tectonics agrees with my basic assumption that a transient Pannotia supercontinent did exist and that the Ediacaran (in the north) to Early Cambrian (in the south) saw its breakup (Dalziel 2014). Pannotia was probably short-lived and thus difficult to establish. The largest part of Pannotia was Gondwana, which assembled following the closure of three ancient paleo-oceans during the Ediacaran (Meert 2003, Collins 2006, Mikhalsky et al. 2006). The core of Gondwana consisted mainly of Australia, India, Africa, Antarctica, and South America, with numerous microcontinents at its margins (i.e. South China, Avalonia, Cadomina, Iran, Turkey, etc.) (Golonka 2002). Although the precise position of individual tectonic plates or cratonic blocks within Pannotia is debated, it is known that Laurentia was located between East and West Gondwana at the end of the Ediacaran (Hoffman 1991, Dalziel 1997). The breakup of Pannotia is constrained by the opening of the Iapetus Ocean (Condie 2003, 2013, Dalziel 2014) that also marked the separation of Laurentia and Baltica (Torsvik et al. 1996), and the separation of Laurentia and Siberia close to Precambrian-Cambrian boundary (Pelechaty 1996). Plate tectonic configurations and paleopositions of fossil occurrences are based on Scotese’s paleomap software (Scotese 2001). Paleopositions were automatically provided upon download of the occurrences from the PaleobioDB based on a routine implemented by John Alroy (Macquarie University, Sydney). The concept of Pannotia is implemented in Scotese’s reconstructions and although newer reconstructions (Dalziel 2014) differ in some aspects such as the degree of continental dispersal in the earliest Cambrian, they agree that the Early Cambrian was a time of continental disassembly.

7.3 Results

7.3.1 Geodisparity and ecodisparity during the Cambrian radiation

The geodisparity represented by percent dissimilarity between communities in the three

106

spatial intervals is shown in a time series plot (Figure 7.1A). The dissimilarity between assemblage compositions increased with geographic distance among the grids that the assemblages were sampled from. Figure 7.1B depicts environmental beta in way of percent dissimilarity in assemblage composition among environmental categories in tropical areas. It is evident that turnover in community composition increased with greater environmental separation. These results indicate that geodisparity and ecological disparity appear to be equally important to global diversity as they both increase in the first three stages and then stagnate. It is also obvious that turnover in community composition increases with more heterogeneized environments and increasing geographic distances in the Cambrian.

7.3.2 Distance-decay pattern during the Cambrian

Turnover in assemblage composition over all Cambrian intervals increased gradually with paleogeographic distance in the intervals of less than 2,000 km and greater than 4,000 km (Figure 7.2), likely reflecting normal Cambrian distance-decay patterns. In contrast, turnover in the 2,000- to 4,000-km interval during the first three Cambrian stages increased dramatically with geographic distance (Figure 7.2A), whereas turnover in assemblage composition during the rest of the Cambrian stages show less steep distance-decay curves (Figure 7.2B and C). This difference can be seen by the linear regressions between assemblage dissimilarity and paleogeographic distance conducted on all ten Cambrian stages (Table 7.1). Endemism tended to be much greater in the late Ediacaran compared with the Cambrian (Figure 7.3). Most early Cambrian genera exhibit a high degree of endemism, indicating the substantial differences in community composition observed over large spatial scales. Increasing biogeographic provincialism has been associated with the breakup of supercontinents. I plotted on Scotese’s Paleogeographic maps the faunal dissimilarity among geographic grids within distance from 2000 km to 4000 km for the Fortunian, Cambrian Stage 2, and Cambrian Stage 3 (Figure 7.4). It can be confirmed that the unusually high rate of turnover between 2000 and 4000km geographic distance mainly occurred between adjacent continents or distant cratonic blocks within Gondwana.

107

0-2000km 2000km-4000km > 4000km Percent dissimilarity Percent

For St2 St3 St4 St5 Dru Guz Pai Jia St10 Tre Cambrian 0.6 0.7 0.8 0.9 1.0 540 530 520 510 500 490 480

Age (Ma)

Percent dissimilarity dissimilarity Percent

For St2 St3 St4 St5 Dru Guz Pai Jia St10 Tre 0.40.50.60.70.80.91.0 Cambrian

540 530 520 510 500 490 480

Age (Ma)

Figure 7.1 A: Bray-Curtis dissimilarity in assemblage composition among geographic grids in three intervals of paleogeographic distance. Only grids/categories with at least 10 occurrences were considered. B: Bray-Curtis dissimilarity in assemblage composition among environmental categories in tropical areas. Brown line indicates one change and blue line two changes in environmental categories, which are carbonate/siliciclastic lithology and shallow-water/deep-water bathymetry. Vertical bars indicate standard errors.

108

Figure 7.2 Distance-decay curves during the Cambrian stages plotted as taxonomic dissimilarity among 5 × 5° paleogeographic grids. A: blue circles, Stage 3; green circles, Stage 2; red circles, Fortunian. B: grey circles, Stage 4; orange circles, Stage 5; purple circles, Drumian. C: light blue circles, Guzhangian; light purple circles, Paibian; light green circles, Jiangshanian. The black lines denote separate regressions for three distance intervals.

109

Table 7.1 Linear regressions between Cambrian geodisparity and paleogeographic distance in three intervals of great circle distance. Int. = Intercept, Sl. = Slope, ρ = Spearman’s rho. Significant code: *** P < 0.001, ** P < 0.01, * P < 0.05. Significant slopes of greater than 0.15 are in bold.

< 2000 km 2000-4000 km > 4000 km Int. Sl. ρ Int. Sl. ρ Int. Sl. ρ Fortunian -0.67 0.05 0.03 -5.72 0.66 0.55 * -0.9 0.06 0.27 Stage 2 -1.19 0.12 0.13 -4.67 0.56 0.62 * -0.5 0.04 0.13 * Stage 3 -0.76 0.09 0.18 * -1.57 0.19 0.28 *** -0.1 0 -0.01 Stage 4 -0.75 0.08 0.1 -1.78 0.21 0.09 -0.3 0.02 0.09 * Stage 5 -0.47 0.04 0.11 -0.23 0.01 0.07 -0.2 0.02 -0.03 Drumian -1.08 0.13 0.22 -0.5 0.05 0.12 -0.1 0 0 Guzhangian -1.03 0.12 0.26 * -0.63 0.07 0.09 -0.2 0.02 0.04 Paibian -1.78 0.23 0.43 * 0 -0.01 -0.05 -0.1 0 -0.03 Jiangshanian -1.17 0.14 0.14 -1.38 0.16 0.23 -0.7 0.06 0.12 Stage 10 -0.85 0.1 0.14 -0.34 0.04 0.16 -0.2 0.02 0.02 Degree of endemism Degree of

0.55 0.60Cambrian 0.65 0.70 Ediacaran

Figure 7.3 Boxplot shows the variation of degree of endemism in the Ediacaran and Cambrian. The degree of endemism is measured by the proportion of genera unique to a geographic grid over time.

110

Figure 7.4 Geographic representation of dissimilarity among geographic grids within distance from 2000 km to 4000 km plotted on paleogeographic maps for Fortunian (A), Stage 2 (B) and Stage 3 (C) in the Cambrian. Lines are only shown between grids with at least 10 occurrences. The thickness of connecting lines is proportional to the Bray-Curtis dissimilarity between grids. Maps are based on Scotese (ref. 10) and show his 540 Ma (A and B) and 520 Ma (C) reconstructions. Abbreviations: EGOND‒East Gondwana; WGOND‒West Gondwana; LAU‒Laurentia; SIB‒Siberia; BAL‒Baltica. 111

7.3.3 Geographic biases

Beta diversity can be biased by the variation of geographic clustering among sites over time. Therefore, I test the correlation between the median paleogeographic distance of collections and beta diversity, as well as between the median distance of grid centroids and beta diversity. There is no significant correlation in both cases (median distance between collections/beta: ρ = 0.1, P = 0.727; median distance between grid centroids/beta: ρ = -0.24, P = 0.418), implying that geographic clustering does not cause a significant bias. In order to test whether ecological differentiation is more important than continental separation, I analyzed beta diversity within the collections from a single continent such as Gondwana. Beta diversity within Gondwana shows more muted fluctuations than global beta diversity. Subtracting the global beta from Gondwana beta reveals a major increase in Stage 3 (Figure 7.5). Because this increase is concurrent with the main increase of gamma diversity, my conclusion of geodisparity being a main component in the Cambrian radiation is supported. Meanwhile, I assessed separately for Gondwana alpha and beta diversity, geodisparity, and ecodisparity (Figure 7.6 and Figure 7.7). The clear trend of beta diversity increases with increasing geographic distance and environmental heterogeneity is not seen in patterns of geodisparity and ecodisparity within Gondwana. Compare to curves showed in Figure 7.1A, three curves of geographic beta in terms of three geographic intervals (Figure 7.7A) show less variation through the whole Cambrian time and tend to be intertwined with each other. The same pattern is also shown in environmental beta within Gondwana (Figure 7.7B). Different is percent dissimilarity among environmental categories only separated by one shows a remarkable increase from Fortunian to Stage 2 This finding simply suggest that Therefore, geodisparity appears to be the main driver of the Cambrian diversity rise. Figure 7.1B depicts environmental beta in way of percent dissimilarity in assemblage composition among environmental categories in tropical areas. It is evident that turnover in community composition increased with greater environmental separation. These results indicate that geodisparity and ecological disparity within a single continent, Gondwana, appear not to be able to affect the geodisparity and ecodisparity at global scale. These results also

112

support my conclusion that the increase of beta diversity was mainly governed by between-continents taxonomic turnover rather than within-continent.

Figure 7.5 Trajectories of global beta (see Figure 4.2) minus within-Gondwana beta. Beta diversity within Gondwana from the Ediacaran to the earliest Ordovician based on unweighted by-list subsampling of 13 collections per stage.

7.3.4 Geodisparity among different environmental affinities

The high rate of turnover in community composition between adjacent continents (2000-4000km) can be the result of environmental heterogeneity (such as the increase in geodisparity corresponding with the increase of ecodisparity, Figure 7.1). In order to investigate if the increase in geodisparity beta was influenced by the heterogeneity of environments, I calculated geodisparity separately within uniform substrates and bathymetric settings, separating carbonates and siliciclastics on the one hand and inshore (shallow water) and offshore (deeper water) settings on the other hand, following the protocol of Kiessling et al. (2007). Occurrence data for a given stratigraphic interval and environmental category were again pooled in 5 × 5° paleogeographic grids. For each

113

dataset, a dissimilarity matrix of these pooled data was computed for each Cambrian stage, and so was a great circle distance matrix of between grid centers. The rates of change in community composition represented by the linear regression between dissimilarity and geographic distance are listed in Table 7.2. Similar to the pooled data, there are significant correlations between assemblage dissimilarity and paleogeographic distance in the 2,000- to 4,000-km interval during the first four Cambrian stages in carbonate settings (Figure 7.8). However, the patterns in the other three environmental categories are different, probably due to the scarcity of data points. For carbonates at least the role of continental disassembly is thus confirmed. Facies heterogeneity among adjacent continents (or within a larger continent) cannot explain the particularly high turnover in community composition in the 2000-4000 km distance interval.

Figure 7.6 Alpha diversity and beta diversity within Gondwana from the Ediacaran to the earliest Ordovician based on unweighted by-list subsampling of 13 collections per stage. Error bars are standard deviations of 100 subsampling trials.

114

Geodisparity (Gondwana) A

0-2000 km Percent dissimilarity 2000-4000 km > 4000 km

For St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Cambrian 0.2 0.4 0.6 0.8 1.0 1.2 540 530 520 510 500 490 480

Age (Ma)

Ecodisparity (Gondwana) B Percent dissimilarity one change two changes For St2 St3 St4 St5 Dru GuzPai Jia St10 Tre Cambrian 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 540 530 520 510 500 490 480

Age (Ma)

Figure 7.7 A: Bray-Curtis dissimilarity in assemblage composition among geographic grids in three intervals of paleogeographic distance within Gondwana. B: Bray-Curtis dissimilarity in assemblage composition among environmental categories in Gondwana. Only grids/categories with at least 10 occurrences were considered. Error bars indicate standard errors.

115

Table 7.2 Linear regressions between Cambrian geodisparity and paleogeographic distance in three intervals of great circle distance among different environmental affinities. Int. = Intercept, Sl. = Slope, ρ = Spearman’s rho. Significant code: *** P < 0.001, ** P < 0.01, * P < 0.05. Double minus represents invalid value. Significant slopes of greater than 0.15 are in bold.

Carbonates < 2000 km 2000-4000 km > 4000 km Int. Sl. ρ Int. Sl. ρ Int. Sl. ρ Fortunian -1.84 0.22 0.23 -7.44 0.86 0.94 * 0.7 -0.09 -0.39* Stage 2 -2.01 0.24 0.39 -4.88 0.59 0.76 * -1.09 0.11 0.36*** Stage 3 -0.81 0.09 0.4** -2.3 0.27 0.44* -0.12 0 0.02 Stage 4 -1.77 0.23 0.39 -2 0.24 0.55* -0.1 0 0.07 Stage 5 -0.16 0 0 0.28 -0.05 0.02 -0.01 0 -0.33** Drumian -0.97 0.12 0.24 0.15 -0.03 0.09 -0.23 0.02 0.07 Guzhangian -0.53 0.06 0.28 -1.74 0.2 0.41 -0.15 0.01 -0.04 Paibian -3.44 0.45 0.54 0.91 -0.38 -0.53 -0.76 0.08 0.28* Jiangshanian -1.31 1.14 -0.09 1.2 -0.17 -0.25 -1.03 0.1 -0.01 Stage 10 -0.57 0.05 0.6 -77.8 9.5 1 -0.85 0.09 0.48**

Siliciclastics < 2000 km 2000-4000 km > 4000 km Int. Sl. ρ Int. Sl. ρ Int. Sl. ρ Fortunian -0.15 -- -- -0.33 -- -- 0.31 -0.05 -- Stage 2 0 ------Stage 3 -1.44 0.2 0.33 935.8 -113 -1 -0.37 0.04 0.34** Stage 4 -2.34 0.31 0.4 -10.6 1.28 0.26 -0.2 0.01 0.3 Stage 5 0.04 -0.03 -0.15 -1.1 0.13 0.49* -0.3 0.03 0.04 Drumian 0.13 -0.06 -0.5 -0.61 0.06 0.26 -0.86 0.09 0.32* Guzhangian 0.87 -0.14 -0.6 1.24 -0.16 -0.2 -0.71 0.07 0.5** Paibian ------1.73 0.18 0.5 Jiangshanian 0.26 -0.05 0.11 -8.9 1.14 1 ------Stage 10 -4.15 0.57 0.52 -0.51 0.06 0.4 -0.02 0 0.06

Shallow water

116

< 2000 km 2000-4000 km > 4000 km Int. Sl. ρ Int. Sl. ρ Int. Sl. ρ Fortunian ------0.26 -- -- Stage 2 -0.5 -- -- 0 -- -- -1.47 0.14 0.4 Stage 3 -0.88 0.1 0.02 -3 0.36 0.82* 0.04 0 -0.09 Stage 4 -1.8 ------0.1 -0.01 -0.2 Stage 5 -1.26 0.16 0.5 ------0.22 0.02 0.05 Drumian -0.3 -- -- 0.48 -0.06 -1 0.08 -0.08 -0.13 Guzhangian 2.9 -0.5 -1 -4.74 0.58 0.75 0 0 -- Paibian ------0.15 -0.02 -0.44 Jiangshanian -0.71 0.07 0.15 -0.08 ------Stage 10 -1.38 0.17 0.44 -0.63 0.07 0.26 -0.45 0.05 0.38*

Deep water < 2000 km 2000-4000 km > 4000 km Int. Sl. ρ Int. Sl. ρ Int. Sl. ρ Fortunian -0.34 ------Stage 2 0 0 ------Stage 3 ------0 0 -- Stage 4 -0.29 -- -- -0.03 -- -- 0.19 -0.02 -0.28 Stage 5 -1.14 0.15 0.3 -2.5 0.29 0.52 -0.7 0.07 0.2 Drumian 21.7 -3.5 -0.87 1.1 -0.15 0.03 -0.1 0 -0.24 Guzhangian 0 -- -- 2.02 -0.26 -1 -0.31 0.03 0.11 Paibian -0.71 -- -- 0 0 -- -0.49 0.05 0.37 Jiangshanian -0.06 0 -0.32 -4.22 0.53 1 -5.15 0.53 0.17 Stage 10 -0.27 -- -- -0.03 -- -- -0.46 0.05 0.74*

117

Fortunian Stage 2

-1.2 -1.0 -0.8Stage -0.6 -0.4 3 -0.2 0.0 log (Assemblage dissimilarity in carbonates) 345678910

log (Geographic distance between grids)

Figure 7.8 Distance-decay curves during the first three Cambrian stages plotted as taxonomic dissimilarity among 5 × 5° paleogeographic grids for taxa with an affinity for carbonate substrates affinity. Top plot: blue circles, Stage 3; green circles, Stage 2; red circles, Fortunian. The black lines denote separate regressions for three distance intervals.

7.4 Discussion

The steep distance-decay in the 2000-4000 km distance interval is the distance typically measured between adjacent paleo-continents in the early Cambrian, suggesting that the continents became biogeographically distinct in the first two Cambrian stages and that deep oceans between continental shelf areas caused effective migration barriers, which likely enhanced allopatric speciation (McCartney et al. 2000). The opening and widening of paleo-oceans among paleocontinents following the breakup of Pannotia was causing effective migration barriers at that time. Rapid allopatric speciation and rapid turnover in community composition has occurred among paleocontinents, which in turn facilitated further adaptive evolution. The commencement of the Cambrian radiation occurred immediately after the major end-Proterozoic glaciations, when global sea-level rose with

118

low-magnitude fluctuations and huge well-connected oceans provided ample habitat space (Ruban 2010). The appearance of new groups of diverse fauna in Stage 2, for example archaeocyathan sponges, was probably initiated by the start of transgression (Luchinina et al. 2013). At continental scale, beta diversity was governed by the strong increase of provincialism among paleo-continents. Although the disassembly of Pannotia as a trigger of the Cambrian diversification is not widely accepted, my results support the contention that the disassembly of Pannotia increased geodisparity and thereby gamma diversity in the Cambrian. Disassembly of supercontinents is rare and always seems to have dramatic evolutionary consequences (Valentine and Moores 1970). Examples are (1) the breakup of the supercontinent Rodinia some 750 Myr ago has been linked with global glaciations, which are often held responsible for the emergence of metazoans (Shields-Zhou and Och 2011); (2) further continental dispersal in the Ordovician is thought to have facilitated the Ordovician diversification (Servais et al. 2009); (3) the link between the breakup of Pangea in the early Jurassic and the Mesozoic-Cenozoic radiation has long been suggested (Valentine and Moores 1972) and is supported by sampling-standardized diversity curves (Alroy et al. 2008); and, (4) a recent study has suggested that the Cretaceous breakup of Gondwana was the main trigger of tropical reef biodiversity (Leprieur et al. 2016). The capacity of biological dispersal could have changed substantially in the Early Cambrian. Our knowledge of dispersal in Ediacaran biota is limited (Figure 6.3) but at least some Ediacaran taxa have been reported to be able to cross oceanic basins and achieved a cosmopolitan distribution (Darroch et al. 2013, Laflamme et al. 2013). Cambrian animals probably dispersed with non-planktotrophic larval stages (Peterson 2005), which would limit their geographic distribution relative to animals with planktotrophic larvae (Jablonski and Lutz 1983). But animals with direct development (no larval stages for most of the Ediacaran fauna) should have a lower geographic range than those with larvae (Figure 7.3). Nevertheless, the increase of geodisparity I observe during the Early Cambrian could be due to the widespread appearance of animals with larval stages combined with continental disassembly. In other words biological innovation – the evolution of larval stages (Peterson 2005) – in tandem with

119

supercontinent break-up best explains the increase of geographic beta diversity. The most common mode of speciation is allopatric (Begon et al. 2006), characterized by the emergence of geographic barriers. However, little consideration is usually given how speciation contributes to diversity at different spatial scales. A strict allopatric speciation without further migration tends to contribute little to the increase in the diversity of the local community. The generation of both local and regional diversity depends on range expansion and secondary sympatry (Mittelbach and Schemske 2015, Tomašových et al. 2016). The larger geographic ranges of Cambrian taxa relative to Ediacaran fauna imply range expansions and a second contact between endemic species following allopatric speciation that occurred during the first three stages of the Cambrian. It is also evident that geographic turnover rate between adjacent continental regions remained at low levels over the entire post-stage 3 interval.

7.5 Future outlook

My study provides evidence for niche partitioning, plate tectonics, and key innovations as strong and persistent evolutionary forces for the diversification event occurred during the Ediacaran-Cambrian transition. To the end of this thesis, I have to point out that all the results presented here are actually a beginning of the detailed quantitative analysis of Cambrian radiation. Given that the radiation has been characterized as the rapid increase of both biodiversity and morphological disparity. One of the future directions would be exploring the links between the ecological dynamics described above and the astonishing increase of morphological disparity in the early Cambrian. Equally interesting is how biogeographic distributions of taxa affect global biodiversity patterns during the Cambrian radiation. Following questions would be: How did tectonic events affect the morphological disparity? What is the trigger of the burst of disparity? Could it be any ecological causes? Besides, the Ordovician Period witnessed another important diversification event of marine life in Earth history. In contrast to the preceding “Cambrian Explosion”, few higher taxa originated during the Ordovician Radiation but a massive turnover in the overall composition of the marine biota took place. It would be also worth to explore the relationship between the Cambrian radiation and the Ordovician radiation by combining

120

quantitative analysis with the theoretical models. This will provide invaluable clues to the big picture of the earliest evolution in metazoan during the early Paleozoic.

121

122

REFERENCES

Aberhan, M., and W. Kiessling. 2012. Phanerozoic marine biodiversity: a fresh look at data, methods, patterns and processes. Pp. 3‐22. Earth and Life. Springer. Aberhan, M., and W. Kiessling. 2014. Rebuilding biodiversity of patagonian marine molluscs after the end‐Cretaceous mass extinction. PLoS ONE 9(7):e102629. Aberhan, M., W. Kiessling, and F. T. Fursich. 2006. Testing the role of biological interactions in the evolution of mid‐Mesozoic marine benthic ecosystems. Paleobiology 32(2):259‐277. Acquisti, C., J. Kleffe, and S. Collins. 2007. Oxygen content of transmembrane proteins over macroevolutionary time scales. Nature 445(7123):47‐52. Ahlberg, P. 2003. Trilobites and intercontinental tie points in the Upper Cambrian of Scandinavia. Pp. 0127‐134. Geologica Acta. Ahlberg, P., and J. Ahlgren. 1996. Agnostids from the upper Cambrian of Västergötland, Sweden. GFF 118(3):129‐140. Aho, K., D. Derryberry, and T. Peterson. 2014. Model selection for ecologists: the worldviews of AIC and BIC. Ecology 95(3):631‐6. Akaike, H. 1998. Information theory and an extension of the maximum likelihood principle. Pp. 199‐213. In K. T. Emanuel Parzen, Genshiro Kitagawa, ed. Selected Papers of Hirotugu Akaike. Springer, New York. Alroy, J. 1996. Constant extinction, constrained diversification, and uncoordinated stasis in North American mammals. Palaeogeography Palaeoclimatology Palaeoecology 127(1‐ 4):285‐311. Alroy, J. 2000a. New methods for quantifying macroevolutionary patterns and processes. Paleobiology 26(4):707‐733. Alroy, J. 2000b. Successive approximations of diversity curves: Ten more years in the library. Geology 28(11):1023‐1026. Alroy, J. 2008. Dynamics of Origination and Extinction in the Marine Fossil Record. Proceedings of the National Academy of Sciences of the United States of America 105:11536‐11542. Alroy, J. 2010a. Fair sampling of taxonomic richness and unbiased estimation of origination and extinction rates. Pp. 55‐80. Quantitative methods in paleobiology. Paleontological Society Papers. Alroy, J. 2010b. The shifting balance of diversity among major marine animal groups. Science 329(5996):1191‐1194. Alroy, J., M. Aberhan, D. J. Bottjer, M. Foote, F. T. Fürsich, P. J. Harries, A. J. W. Hendy, S. M. Holland, L. C. Ivany, W. Kiessling, M. A. Kosnik, C. R. Marshall, A. J. McGowan, A. I. Miller, T. D. Olszewski, M. E. Patzkowsky, S. E. Peters, L. Villier, P. J. Wagner, N. Bonuso, P. S. Borkow, B. Brenneis, M. E. Clapham, L. M. Fall, C. A. Ferguson, V. L. Hanson, A. Z. Krug, K. M. Layou, E. H. Leckey, S. Nürnberg, C. M. Powers, J. A. Sessa, C. Simpson, A. Tomašových, and C. C. Visaggi. 2008. Phanerozoic Trends in the Global Diversity of Marine Invertebrates. Science 321(5885):97‐100. Alroy, J., C. R. Marshall, R. K. Bambach, K. Bezusko, M. Foote, F. T. Fursich, T. A. Hansen, S. M. Holland, L. C. Ivany, D. Jablonski, D. K. Jacobs, D. C. Jones, M. A. Kosnik, S. Lidgard, S. Low, A. I. Miller, P. M. Novack‐Gottshall, T. D. Olszewski, M. E. Patzkowsky, D. M. Raup, K. Roy, J. J. Sepkoski, Jr., M. G. Sommers, P. J. Wagner, and A. Webber. 2001. Effects of sampling standardization on estimates of Phanerozoic marine diversification. Proc Natl Acad Sci USA 98(11):6261‐6.

123

Alroy, J., M. McKinney, and J. Drake. 1998. Equilibrial diversity dynamics in North American mammals. Pp. 233‐287. In J. A. D. Michael L. McKinney, ed. Biodiversity Dynamics: Turnover of Populations, Taxa, and Communities. Columbia University Press, New York. Amthor, J. E., J. P. Grotzinger, S. Schroder, S. A. Bowring, J. Ramezani, M. W. Martin, and A. Matter. 2003. Extinction of Cloudina and Namacalathus at the Precambrian‐Cambrian boundary in Oman. Geology 31(5):431‐434. Araujo, M. S., D. I. Bolnick, and C. A. Layman. 2011. The ecological causes of individual specialisation. Ecology letters 14(9):948‐958. Astashkin, V. A., and A. I. U. r. e. Rozanov. 1991. The Cambrian system on the Siberian platform: correlation chart and explanatory notes. International Union of Geological Sciences. Ausich, W. I., and D. J. Bottjer. 1982. Tiering in suspension‐feeding communities on soft substrata throughout the phanerozoic. Science 216(4542):173‐4. Babcock, L., R. Robison, and P. Shanchi. 2011. Cambrian stage and series nomenclature of Laurentia and the developing global chronostratigraphic scale. Mus North Ariz Bull 67:12‐26. Babcock, L. E. 2003. Trilobites in Paleozoic predator‐prey systems, and their role in reorganization of early Paleozoic ecosystems. Pp. 55‐92. In K. M. H. Kelley P. H., T. A., ed. Predator—prey interactions in the fossil record. Kluwer/Plenum, New York. Babcock, L. E. 2005. Interpretation of biological and environmental changes across the Neoproterozoic–Cambrian boundary: developing a refined understanding of the radiation and preservational record of early multicellular organisms. Palaeogeography, Palaeoclimatology, Palaeoecology 220(1):1‐5. Babcock, L. E., and S. C. Peng. 2007. Cambrian chronostratigraphy: Current state and future plans. Palaeogeography Palaeoclimatology Palaeoecology 254(1‐2):62‐66. Bambach, R. K. 1977. Species Richness in Marine Benthic habitats through the Phanerozoic. Paleobiology 3(2):152‐167. Bambach, R. K. 1983. Ecospace utilization and guilds in marine communities through the Phanerozoic. Pp. 719‐746. In P. L. M. Michael J. S. Tevesz, ed. Biotic interactions in recent and fossil benthic communities. Springer, New York. Bambach, R. K. 1985. Classes and adaptive variety: the ecology of diversification in marine faunas through the Phanerozoic. Pp. 191‐253. In J. E. Valentine, ed. Phanerozoic diversity patterns: Profiles in macroevolution. Princeton University Press, Princeton, NJ. Bambach, R. K. 1993. Seafood through time: changes in biomass, energetics, and productivity in the marine ecosystem. Paleobiology 19(03):372‐397. Bambach, R. K. 1999. Energetics in the global marinefauna: A connection between terrestrial diversification and change in the marine biosphere. Geobios 32(2):131‐144. Bambach, R. K., A. M. Bush, and D. H. Erwin. 2007. Autecology and the filling of ecospace: Key metazoan radiations. Palaeontology 50(1):1‐22. Bambach, R. K., A. H. Knoll, and J. J. Sepkoski, Jr. 2002. Anatomical and ecological constraints on Phanerozoic animal diversity in the marine realm. Proc Natl Acad Sci U S A 99(10):6854‐ 9. Bambach, R. K., A. H. Knoll, and S. C. Wang. 2004. Origination, extinction, and mass depletions of marine diversity. Paleobiology 30(4):522‐542. Begon, M., C. R. H. Townsend, L. John, R. T. Colin, and L. H. John. 2006. Ecology: from individuals to ecosystems. Blackwell Publishing, United Kingdom. Bekker, A., H. D. Holland, P. L. Wang, D. Rumble, 3rd, H. J. Stein, J. L. Hannah, L. L. Coetzee, and N. J. Beukes. 2004. Dating the rise of atmospheric oxygen. Nature 427(6970):117‐20. Bengtson, S. 1977. Aspects of problematic fossils in the early Palaeozoic. Acta Universitatis

124

Upsaliensis 415:1‐71. Bengtson, S. 2002. Origins and early evolution of predation. Paleontological Society Papers 8:289‐318. Benton, M. J. 1995. Diversification and extinction in the history of life. Science 268(5207):52‐8. Berkner, L. V., and L. Marshall. 1965. On the origin and rise of oxygen concentration in the Earth's atmosphere. Journal of the atmospheric sciences 22(3):225‐261. Bottjer, D. J., and D. Jablonski. 1988. Paleoenvironmental patterns in the evolution of post‐ Paleozoic benthic marine invertebrates. Palaios:540‐560. Brasier, M. 1982. Sea‐level changes, facies changes and the Late Precambrian—Early Cambrian evolutionary explosion. Precambrian Research 17(2):105‐123. Brasier, M. D. 1992. Nutrient‐enriched waters and the early skeletal fossil record. Journal of the Geological Society (London) 149:621‐629. Brasier, M. D., and J. B. Antcliffe. 2009. Evolutionary relationships within the Avalonian Ediacara biota: new insights from laser analysis. Journal of the Geological Society 166(2):363‐384. Brasier, M. D., and J. F. Lindsay. 2001. Did supercontinental amalgamation trigger the “Cambrian Explosion”? Pp. 69‐89. In A. Y. Zhuravlev, and R. Riding, eds. The Ecology of the Cambrian Radiation. Columbia University Press, New York. Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27(4):325‐349. Brennan, S. T., T. K. Lowenstein, and J. Horita. 2004. Seawater chemistry and the advent of biocalcification. Geology 32(6):473‐476. Budd, G. E. 2008. The earliest fossil record of the animals and its significance. Philos Trans R Soc Lond B Biol Sci 363(1496):1425‐34. Bush, A. M., R. K. Bambach, and D. H. Erwin. 2011. Ecospace utilization during the Ediacaran radiation and the Cambrian eco‐explosion. Pp. 111‐133. In M. Laflamme, Schiffbauer, J.D., Dornbos, S.Q., ed. Quantifying the Evolution of Early Life: Numerical Approaches to the Evaluation of Fossils and Ancient Ecosystems. : Topics in Geobiology. Springer, Berlin. Butterfield, N. J. 2003. Exceptional fossil preservation and the cambrian explosion. Integr Comp Biol 43(1):166‐77. Butterfield, N. J. 2007. Macroevolution and macroecology through deep time. Palaeontology 50(1):41‐55. Butterfield, N. J. 2009. Macroevolutionary turnover through the Ediacaran transition: ecological and biogeochemical implications. Geological Society, London, Special Publications 326(1):55‐66. Butterfield, N. J. 2011. Animals and the invention of the Phanerozoic Earth system. Trends Ecol Evol 26(2):81‐7. Campbell, I. H., and R. J. Squire. 2010. The mountains that triggered the Late Neoproterozoic increase in oxygen: The Second Great Oxidation Event. Geochimica et Cosmochimica Acta 74(15):4187‐4206. Canfield, D. E. 2005. The early history of atmospheric oxygen: homage to Robert M. Garrels. Annu. Rev. Earth Planet. Sci. 33:1‐36. Canfield, D. E., S. W. Poulton, A. H. Knoll, G. M. Narbonne, G. Ross, T. Goldberg, and H. Strauss. 2008. Ferruginous conditions dominated later neoproterozoic deep‐water chemistry. Science 321(5891):949‐52. Canfield, D. E., S. W. Poulton, and G. M. Narbonne. 2007. Late‐Neoproterozoic deep‐ocean oxygenation and the rise of animal life. Science 315(5808):92‐5. Canfield, D. E., and A. Teske. 1996. Late Proterozoic rise in atmospheric oxygen concentration inferred from phylogenetic and sulphur‐isotope studies. Nature 382(6587):127‐32.

125

Carpenter, S. R., J. F. Kitchell, and J. R. Hodgson. 1985. Cascading Trophic Interactions and Lake Productivity. BioScience 35(10):634‐639. Catling, D. C., C. R. Glein, K. J. Zahnle, and C. P. McKay. 2005. Why O2 Is Required by Complex Life on Habitable Planets and the Concept of Planetary" Oxygenation Time". Astrobiology 5(3):415‐438. Cawood, P. A., P. J. A. McCausland, and G. R. Dunning. 2001. Opening Iapetus: Constraints from the Laurentian margin in Newfoundland. Geological Society of America Bulletin 113(4):443‐453. Cawood, P. A., A. A. Nemchin, R. Strachan, T. Prave, and M. Krabbendam. 2007. Sedimentary basin and detrital zircon record along East Laurentia and Baltica during assembly and breakup of Rodinia. Journal of the Geological Society 164(2):257‐275. Chiarucci, A., G. Bacaro, D. Rocchini, and L. Fattorini. 2008. Discovering and rediscovering the sample‐based rarefaction formula in the ecological literature. Community ecology 9(1):121‐123. Clapham, M. E., and G. M. Narbonne. 2002. Ediacaran epifaunal tiering. Geology 30(7):627‐630. Clapham, M. E., G. M. Narbonne, and J. G. Gehling. 2003. Paleoecology of the oldest known animal communities: Ediacaran assemblages at Mistaken Point, Newfoundland. Paleobiology 29(4):527‐544. Collins, A. S. 2006. Madagascar and the amalgamation of Central Gondwana. Gondwana Research 9(1‐2):3‐16. Condie, K. C. 2003. Supercontinents, superplumes and continental growth: the Neoproterozoic record. Geological Society, London, Special Publications 206(1):1‐21. Condie, K. C. 2013. Plate tectonics & crustal evolution. Butterworth‐Heinemann, Oxford, UK. Conway Morris, S. 1993. Ediacaran‐Like Fossils in Cambrian Burgess Shale‐Type Faunas of North‐ America. Palaeontology 36:593‐635. Conway Morris, S., and S. Bengtson. 1994. Cambrian Predators ‐ Possible Evidence from Boreholes. Journal of Paleontology 68(1):1‐23. Conway Morris, S., and J. B. Caron. 2007. Halwaxiids and the early evolution of the lophotrochozoans. Science 315(5816):1255‐8. Cook, P. J., and J. H. Shergold. 1984. Phosphorus, Phosphorites and Skeletal Evolution at the Precambrian Cambrian Boundary. Nature 308(5956):231‐236. Crampton, J. S., M. Foote, A. G. Beu, P. A. Maxwell, R. A. Cooper, I. Matcham, B. A. Marshall, and C. M. Jones. 2009. The ark was full! Constant to declining Cenozoic shallow marine biodiversity on an isolated midlatitude continent. 32(04):509‐532. Dalziel, I. W. 2014. Cambrian transgression and radiation linked to an Iapetus‐Pacific oceanic connection? Geology:G35886. 1. Dalziel, I. W. D. 1997. Neoproterozoic‐Paleozoic geography and tectonics: review, hypothesis, environmental speculation. Geological Society of America Bulletin 109(1):16‐42. Darroch, S. A., E. A. Sperling, T. H. Boag, R. A. Racicot, S. J. Mason, A. S. Morgan, S. Tweedt, P. Myrow, D. T. Johnston, and D. H. Erwin. 2015. Biotic replacement and mass extinction of the Ediacara biota. P. 20151003. Proceedings of the Royal Society of London. Darroch, S. A. F., M. Laflamme, and M. E. Clapham. 2013. Population structure of the oldest known macroscopic communities from Mistaken Point, Newfoundland. Paleobiology 39(4):591‐608. Des Marais, D. J., H. Strauss, R. E. Summons, and J. M. Hayes. 1992. Carbon isotope evidence for the stepwise oxidation of the Proterozoic environment. Nature 359(6396):605‐9. Droser, M. L., J. G. Gehling, and S. Jensen. 1999. When the worm turned: Concordance of Early Cambrian ichnofabric and trace‐fossil record in siliciclastic rocks of South Australia.

126

Geology 27(7):625‐628. Droser, M. L., J. G. Gehling, and S. R. Jensen. 2006. Assemblage palaeoecology of the Ediacara biota: The unabridged edition? Palaeogeography, Palaeoclimatology, Palaeoecology 232(2):131‐147. Droser, M. L., S. Jensen, and J. G. Gehling. 2002. Trace fossils and substrates of the terminal Proterozoic‐Cambrian transition: implications for the record of early bilaterians and sediment mixing. Proc Natl Acad Sci U S A 99(20):12572‐6. Dzik, J., and G. Krumbiegel. 1989. The oldest ‘onychophoran’Xenusion: a link connecting phyla? Lethaia 22(2):169‐181. Erwin, D. H. 2015. Was the Ediacaran–Cambrian radiation a unique evolutionary event? Paleobiology 41(01):1‐15. Erwin, D. H., M. Laflamme, S. M. Tweedt, E. A. Sperling, D. Pisani, and K. J. Peterson. 2011. The Cambrian conundrum: Early divergence and later ecological success in the early history of animals. Science 334(6059):1091‐7. Erwin, D. H., and J. W. Valentine. 2012. The Cambrian explosion: the construction of animal biodiversity. Roberts & Company, Colorado, USA. Farquhar, J., H. Bao, and M. Thiemens. 2000. Atmospheric influence of Earth's earliest sulfur cycle. Science 289(5480):756‐9. Fike, D. A., J. P. Grotzinger, L. M. Pratt, and R. E. Summons. 2006. Oxidation of the Ediacaran ocean. Nature 444(7120):744‐7. Fischer, W. W. 2016. Breathing room for early animals. Proceedings of the National Academy of Sciences 113(7):1686‐1688. Foote, M. 1994. Temporal Variation in Extinction Risk and Temporal Scaling of Extinction Metrics. Paleobiology 20(4):424‐444. Foote, M. 1997. The evolution of morphological diversity. Annual Review of Ecology and Systematics 28:129‐152. Foote, M. 2000. Origination and Extinction Components of Taxonomic Diversity: General Problems. Paleobiology 26(4):74‐102. Foote, M. 2005. Pulsed origination and extinction in the marine realm. Paleobiology 31(1):6‐22. Foote, M. 2009. Origination and extinction components of taxonomic diversity: general problems. Paleobiology 26:74‐102. Foote, M., and D. M. Raup. 1996. Fossil preservation and the stratigraphic ranges of taxa. Paleobiology 22(2):121‐40. Freeman, G., and J. W. Lundelius. 2005. The transition from planktotrophy to lecithotrophy in larvae of Lower Palaeozoic Rhynchonelliform brachiopods. Lethaia 38(3):219‐254. Freeman, G., and J. W. Lundelius. 2007. Origin of planktotrophy—evidence from early molluscs: a response to Nützel et al.(2006). Evolution & development 9(4):307‐310. Gehling, J. G., S. Jensen, M. L. Droser, P. M. Myrow, and G. M. Narbonne. 2001. Burrowing below the basal Cambrian GSSP, Fortune Head, Newfoundland. Geological Magazine 138(02):213‐218. Geyer, G., and E. Landing. 2004. A unified lower–middle Cambrian chronostratigraphy for West Gondwana. Acta Geologica Polonica 54(2):179‐218. Geyer, G., and J. S. Peel. 2011. The Henson Gletscher Formation, North Greenland, and its bearing on the global Cambrian Series 2–Series 3 boundary. Bulletin of Geosciences 86(3):465‐534. Geyer, G., and J. Shergold. 2000. The quest for internationally recognized divisions of Cambrian time. Episodes 23(3):188‐195. Ginsburg, S., and E. Jablonka. 2010. The evolution of associative learning: A factor in the

127

Cambrian explosion. J Theor Biol 266(1):11‐20. Glaessner, M. F. 1985. The dawn of animal life: a biohistorical study. Cambridge University Press, Cambridge. Golonka, J. 2002. Plate‐tectonic maps of the Phanerozoic. Pp. 21‐75. In W. Kiessling, E. Flügel, and J. Golonka, eds. Phanerozoic reef patterns. Society for Sedimentary Geology. Gould, S. J. 1989. Wonderful life: the Burgess Shale and the nature of history. New York: Norton. Gould, S. J., and C. B. Calloway. 1980. Clams and Brachiopods ‐ Ships That Pass in the Night. Paleobiology 6(4):383‐396. Grazhdankin, D. 2004. Patterns of distribution in the Ediacaran biotas: facies versus biogeography and evolution. Paleobiology 30(2):203‐221. Grazhdankin, D. V., U. Balthasar, K. E. Nagovitsin, and B. B. Kochnev. 2008. Carbonate‐hosted Avalon‐type fossils in arctic Siberia. Geology 36(10):803‐806. Green, J. L., A. J. Holmes, M. Westoby, I. Oliver, D. Briscoe, M. Dangerfield, M. Gillings, and A. J. Beattie. 2004. Spatial scaling of microbial eukaryote diversity. Nature 432(7018):747‐50. Grotzinger, J., E. Adams, and S. Schröder. 2005. Microbial–metazoan reefs of the terminal Proterozoic Nama Group (c. 550–543 Ma), Namibia. Geological Magazine 142(05):499‐ 517. Grotzinger, J. P., S. A. Bowring, B. Z. Saylor, and A. J. Kaufman. 1995. Biostratigraphic and geochronologic constraints on early animal evolution. SCIENCE‐NEW YORK THEN WASHINGTON‐:598‐598. Grotzinger, J. P., W. A. Watters, and A. H. Knoll. 2000. Calcified metazoans in thrombolite‐ stromatolite reefs of the terminal Proterozoic Nama Group, Namibia. Paleobiology 26(3):334‐359. Hagadorn, J. W., R. H. Dott, and D. Damrow. 2002. Stranded on a Late Cambrian shoreline: Medusae from central Wisconsin. Geology 30(2):147‐150. Hagadorn, J. W., C. M. Fedo, and B. M. Waggoner. 2000. Early Cambrian Ediacaran‐type fossils from California. Journal of Paleontology 74(4):731‐740. Hagadorn, J. W., and B. Waggoner. 2000. Ediacaran fossils from the southwestern Great Basin, United States. Journal of Paleontology 74(2):349‐359. Hairston, N. G., F. E. Smith, and L. B. Slobodkin. 1960. Community structure, population control, and competition. American naturalist:421‐425. Harper, D. A. T. 2006. The Ordovician biodiversification: Setting an agenda for marine life. Palaeogeography Palaeoclimatology Palaeoecology 232(2‐4):148‐166. Hautmann, M. 2014. Diversification and diversity partitioning. Paleobiology 40(2):162‐176. Hautmann, M. 2015. Diversity partitioning in Jurassic level‐bottom communities. Historical Biology:1‐9. Hoffman, P. F. 1991. Did the breakout of Laurentia turn Gondwanaland inside‐out? Science 252(5011):1409‐1412. Hoffman, P. F. 1999. The break‐up of Rodinia, birth of Gondwana, true polar wander and the snowball Earth. Journal of African Earth Sciences 28(1):17‐33. Hofmann, H. J., S. J. O'Brien, and A. E. King. 2008. Ediacaran biota on Bonavista Peninsula, Newfoundland, Canada. Journal of Paleontology 82(1):1‐36. Holland, H. 1994. Early Proterozoic atmospheric change. Early life on Earth:237‐244. Holland, H. D. 2006. The oxygenation of the atmosphere and oceans. Philos Trans R Soc Lond B Biol Sci 361(1470):903‐15. Holland, S. M. 2010. Additive diversity partitioning in palaeobiology: revisiting Sepkoski’s question. Palaeontology 53(6):1237‐1254. Hopkins, M. J., C. Simpson, and W. Kiessling. 2014. Differential niche dynamics among major

128

marine invertebrate clades. Ecology letters 17(3):314‐323. Horner‐Devine, M. C., M. Lage, J. B. Hughes, and B. J. Bohannan. 2004. A taxa–area relationship for bacteria. Nature 432(7018):750‐753. Hortal, J., N. Roura‐Pascual, N. J. Sanders, and C. Rahbek. 2010. Understanding (insect) species distributions across spatial scales. Ecography 33(1):51‐53. Hua, H., Z. Chen, X. L. Yuan, L. Y. Zhang, and S. H. Xiao. 2005. Skeletogenesis and asexual reproduction in the earliest biomineralizing animal Cloudina. Geology 33(4):277‐280. Hua, H., B. R. Pratt, and L. Y. Zhang. 2003. Borings in Cloudina shells: Complex predator‐prey dynamics in the terminal neoproterozoic. Palaios 18(4‐5):454‐459. Huntley, J. W., and M. Kowalewski. 2007. Strong coupling of predation intensity and diversity in the Phanerozoic fossil record. Proc Natl Acad Sci U S A 104(38):15006‐10. Jablonski, D. 2005. Evolutionary innovations in the fossil record: the intersection of ecology, development, and macroevolution. J Exp Zool B Mol Dev Evol 304(6):504‐19. Jablonski, D., and D. J. Bottjer. 1990. Onshore‐offshore trends in marine invertebrate evolution. Causes of evolution: a paleontological perspective. University of Chicago Press, Chicago:21‐75. Jablonski, D., and R. A. Lutz. 1983. Larval Ecology of Marine Benthic Invertebrates ‐ Paleobiological Implications. Biol Rev Camb Philos Soc 58(1):21‐89. Jablonski, D., J. J. Sepkoski, Jr., D. J. Bottjer, and P. M. Sheehan. 1983. Onshore‐offshore patterns in the evolution of phanerozoic shelf communities. Science 222(4628):1123‐5. Jensen, S. 2003. The proterozoic and earliest cambrian trace fossil record; patterns, problems and perspectives. Integr Comp Biol 43(1):219‐28. Johnston, D., S. Poulton, N. Tosca, T. O'Brien, G. Halverson, D. Schrag, and F. Macdonald. 2013. Searching for an oxygenation event in the fossiliferous Ediacaran of northwestern Canada. Chemical Geology 362:273‐286. Jost, L. 2010. The relation between evenness and diversity. Diversity 2(2):207‐232. Khomentovsky, V. V., and G. A. Karlova. 1993. Biostratigraphy of the Vendian Cambrian Beds and the Lower Cambrian Boundary in Siberia. Geological Magazine 130(1):29‐45. Kiessling, W. 2005. Habitat effects and sampling bias on Phanerozoic reef distribution. Facies 51(1‐4):24‐32. Kiessling, W. 2008. Sampling‐standardized expansion and collapse of reef building in the Phanerozoic. Fossil Record 11(1):7‐18. Kiessling, W., and M. Aberhan. 2007. Environmental determinants of marine benthic biodiversity dynamics through Triassic‐Jurassic times. Paleobiology 33(3):414‐434. Kiessling, W., M. Aberhan, and L. Villier. 2008. Phanerozoic trends in skeletal mineralogy driven by mass extinctions. Nature Geoscience 1(8):527‐530. Kiessling, W., E. Flügel, and J. Golonka. 2002. Phanerozoic reef patterns. SEPM Press, Tulsa, Oklahoma. Kiessling, W., and Á. T. Kocsis. 2015. Biodiversity dynamics and environmental occupancy of fossil azooxanthellate and zooxanthellate scleractinian corals. Paleobiology FirstView(Supplement ‐1):1‐13. Kiessling, W., C. Simpson, and M. Foote. 2010. Reefs as Cradles of Evolution and Sources of Biodiversity in the Phanerozoic. Science 327(5962):196‐198. Kimura, H., and Y. Watanabe. 2001. Oceanic anoxia at the Precambrian‐Cambrian boundary. Geology 29(11):995‐998. Kirschvink, J. L., and T. D. Raub. 2003. A methane fuse for the Cambrian explosion: carbon cycles and true polar wander. Comptes Rendus Geoscience 335(1):65‐78. Knoll, A. H. 2003. Life on a young planet: the first three billion years of evolution on earth.

129

Princeton University Press, Princeton, New Jersey. Knoll, A. H., and S. B. Carroll. 1999. Early animal evolution: Emerging views from comparative biology and geology. Science 284(5423):2129‐2137. Korovnikov, I. 2006. Lower‐Middle Cambrian boundary in open shelf facies of the Siberian Platform. Palaeoworld 15(3):424‐430. Kowalewski, M. 2002. The fossil record of predation: an overview of analytical methods. Paleontological Society Papers 8:3‐42. Kruse, P. D., J. B. Jago, and J. R. Laurie. 2009. Recent developments in Australian Cambrian biostratigraphy. J Stratigr 33(1):35‐47. Kump, L. R. 2008. The rise of atmospheric oxygen. Nature 451(7176):277‐8. Laflamme, M., S. A. F. Darroch, S. M. Tweedt, K. J. Peterson, and D. H. Erwin. 2013. The end of the Ediacara biota: Extinction, biotic replacement, or Cheshire Cat? Gondwana Research 23(2):558‐573. Laflamme, M., J. D. Schiffbauer, G. M. Narbonne, and D. E. G. Briggs. 2011. Microbial biofilms and the preservation of the Ediacara biota. Lethaia 44(2):203‐213. Landing, E. 1991. Upper Precambrian through Lower Cambrian of Cape‐Breton Island ‐ Faunas, Paleoenvironments, and Stratigraphic Revision. Journal of Paleontology 65(4):570‐595. Landing, E. 1992. Lower Cambrian of southeastern Newfoundland: epeirogeny and Lazarus faunas, lithofacies–biofacies linkages, and the myth of a global chronostratigraphy. Pp. 283‐309. In J. H. Lipps, Signor, P.W. , ed. Origin and early evolution of the metazoa. Plenum Press, New York. Landing, E., G. Geyer, M. D. Brasier, and S. A. Bowring. 2013. Cambrian Evolutionary Radiation: Context, correlation, and chronostratigraphy‐Overcoming deficiencies of the first appearance datum (FAD) concept. Earth‐Science Reviews 123:133‐172. Leprieur, F., P. Descombes, T. Gaboriau, P. F. Cowman, V. Parravicini, M. Kulbicki, C. J. Melian, C. N. de Santana, C. Heine, D. Mouillot, D. R. Bellwood, and L. Pellissier. 2016. Plate tectonics drive tropical reef biodiversity dynamics. Nat Commun 7:11461. Levin, L. A. 2003. Oxygen minimum zone benthos: adaptation and community response to hypoxia. Oceanography and Marine Biology: An Annual Review 41:1‐45. Li, G. X., M. Steiner, X. J. Zhu, A. H. Yang, H. F. Wang, and B. D. Erdtmann. 2007. Early Cambrian metazoan fossil record of South China: Generic diversity and radiation patterns. Palaeogeography Palaeoclimatology Palaeoecology 254(1‐2):229‐249. Li, G. X., X. Zhao, A. Gubanov, M. Y. Zhu, and L. Na. 2011. Early Cambrian Mollusc Watsonella crosbyi: A Potential GSSP Index Fossil for the Base of the Cambrian Stage 2. Acta Geologica Sinica‐English Edition 85(2):309‐319. Li, Z. X., S. V. Bogdanova, A. S. Collins, A. Davidson, B. De Waele, R. E. Ernst, I. C. W. Fitzsimons, R. A. Fuck, D. P. Gladkochub, J. Jacobs, K. E. Karlstrom, S. Lu, L. M. Natapov, V. Pease, S. A. Pisarevsky, K. Thrane, and V. Vernikovsky. 2008. Assembly, configuration, and break‐up history of Rodinia: A synthesis. Precambrian Research 160(1‐2):179‐210. Liang, M. C., H. Hartman, R. E. Kopp, J. L. Kirschvink, and Y. L. Yung. 2006. Production of hydrogen peroxide in the atmosphere of a Snowball Earth and the origin of oxygenic photosynthesis. Proc Natl Acad Sci U S A 103(50):18896‐9. Lindsay, J. F., M. D. Brasier, D. Dorjnamjaa, R. Goldring, P. D. Kruse, and R. A. Wood. 1996. Facies and sequence controls on the appearance of the Cambrian biota in southwestern Mongolia: Implications for the Precambrian‐Cambrian boundary. Geological Magazine 133(4):417‐428. Liu, J., M. Steiner, J. A. Dunlop, H. Keupp, D. Shu, Q. Ou, J. Han, Z. Zhang, and X. Zhang. 2011. An armoured Cambrian lobopodian from China with arthropod‐like appendages. Nature

130

470(7335):526‐30. Losos, J. B. 2010. Adaptive radiation, ecological opportunity, and evolutionary determinism. American Society of Naturalists E. O. Wilson award address. Am Nat 175(6):623‐39. Luchinina, V. A., I. V. Korovnikov, N. V. Novozhilova, and D. A. Tokarev. 2013. Benthic Cambrian biofacies of the Siberian Platform (hyoliths, small shelly fossils, archeocyaths, trilobites and calcareous algae). Stratigraphy and Geological Correlation 21(2):131‐149. Madin, J. S., J. Alroy, M. Aberhan, F. T. Fursich, W. Kiessling, M. A. Kosnik, and P. J. Wagner. 2006. Statistical independence of escalatory ecological trends in Phanerozoic marine invertebrates. Science 312(5775):897‐900. Maloof, A. C., S. M. Porter, J. L. Moore, F. O. Dudas, S. A. Bowring, J. A. Higgins, D. A. Fike, and M. P. Eddy. 2010. The earliest Cambrian record of animals and ocean geochemical change. Geological Society of America Bulletin 122(11‐12):1731‐1774. Mangano, M. G., and L. A. Buatois. 2014. Decoupling of body‐plan diversification and ecological structuring during the Ediacaran‐Cambrian transition: evolutionary and geobiological feedbacks. Proc Biol Sci 281(1780):20140038. Marenco, K. N., D. J. Bottjer, K. Cuddington, J. Byers, W. Wilson, and A. Hastings. 2007. Ecosystem engineering in the fossil record: early examples from the Cambrian period. Pp. 163‐184. In K. Cuddington, B. J. E, W. W. G, and A. Hastings, eds. Ecosystem engineers: plants to protists. Elesevier, London. Marshall, C. R. 2006. Explaining the Cambrian “explosion” of animals. Annu. Rev. Earth Planet. Sci. 34:355‐384. Martiny, J. B., J. A. Eisen, K. Penn, S. D. Allison, and M. C. Horner‐Devine. 2011. Drivers of bacterial beta‐diversity depend on spatial scale. Proc Natl Acad Sci USA 108(19):7850‐4. McCartney, M. A., G. Keller, and H. A. Lessios. 2000. Dispersal barriers in tropical oceans and speciation in Atlantic and eastern Pacific sea urchins of the genus Echinometra. Mol Ecol 9(9):1391‐400. Meert, J. G. 2003. A synopsis of events related to the assembly of eastern Gondwana. Tectonophysics 362(1‐4):1‐40. Micheli, F., C. H. Peterson, L. S. Mullineaux, C. R. Fisher, S. W. Mills, G. Sancho, G. A. Johnson, and H. S. Lenihan. 2002. Predation structures communities at deep‐sea hydrothermal vents. Ecological Monographs 72(3):365‐382. Mikhalsky, E. V., J. W. Sheraton, and K. Hahne. 2006. Charnockite composition in relation to the tectonic evolution of East Antarctica. Gondwana Research 9(4):379‐397. Miller, A. I. 1997. A New Look at Age and Area: The Geographic and Environmental Expansion of Genera During the Ordovician Radiation. Paleobiology 23(4):410‐419. Miller, A. I., M. Aberhan, D. P. Buick, K. V. Bulinski, C. A. Ferguson, A. J. W. Hendy, and W. Kiessling. 2009. Phanerozoic trends in the global geographic disparity of marine biotas. Paleobiology 35(4):612‐630. Miller, A. I., and M. Foote. 1996. Calibrating the Ordovician radiation of marine life: implications for Phanerozoic diversity trends. Paleobiology 22(02):304‐309. Mittelbach, G. G., and D. W. Schemske. 2015. Ecological and evolutionary perspectives on community assembly. Trends Ecol Evol 30(5):241‐247. Mount, J. F., and P. W. Signor. 1985. Early Cambrian Innovation in Shallow Subtidal Environments ‐ Paleoenvironments of Early Cambrian Shelly Fossils. Geology 13(10):730‐ 733. Murdoch, W. W. 1969. Switching in general predators: experiments on predator specificity and stability of prey populations. Ecological Monographs:335‐354. Na, L., and W. Kiessling. 2015. Diversity partitioning during the Cambrian radiation. Proc Natl

131

Acad Sci U S A 112(15):4702‐4706. Naimark, E., Y. Shabanov, and I. Korovnikov. 2011. Cambrian trilobite Ovatoryctocara Tchernysheva, 1962 from Siberia. Bulletin of Geosciences 86(3):405‐422. Narbonne, G., S. Xiao, G. Shields, F. Gradstein, J. Ogg, M. Schmitz, and G. Ogg. 2012. The Ediacaran Period. Geologic Timescale:427‐449. Narbonne, G. M. 2005. The Ediacara biota: Neoproterozoic origin of animals and their ecosystems. Annu. Rev. Earth Planet. Sci. 33:421‐442. Narbonne, G. M., and J. D. Aitken. 1990. Ediacaran Fossils from the Sekwi Brook Area, Mackenzie Mountains, Northwestern Canada. Palaeontology 33(4):945‐980. Narbonne, G. M., and J. G. Gehling. 2003. Life after snowball: The oldest complex Ediacaran fossils. Geology 31(1):27‐30. Narbonne, G. M., B. Z. Saylor, and J. P. Grotzinger. 1997. The youngest Ediacaran fossils from southern Africa. J Paleontol 71(6):953‐67. Nekola, J. C., and P. S. White. 1999. The distance decay of similarity in biogeography and ecology. Journal of Biogeography 26(4):867‐878. Newell, N. D. 1959. The nature of the fossil record. Proceedings of the American Philosophical Society:264‐285. Nürnberg, S., and M. Aberhan. 2015. Interdependence of specialization and biodiversity in Phanerozoic marine invertebrates. Nat Commun 6:6602. Nützel, A., O. Lehnert, and J. í. Frýda. 2006. Origin of planktotrophy—evidence from early molluscs. Evolution & development 8(4):325‐330. Ogg, J. G., G. Ogg, and F. M. Gradstein. 2016. A Concise Geologic Time Scale: 2016. Elsevier, New York. Paine, R. T. 1966. Food web complexity and species diversity. American naturalist:65‐75. Paine, R. T. 1971. Short‐Term Experimental Investigation of Resource Partitioning in a New‐ Zealand Rocky Intertidal Habitat. Ecology 52(6):1096‐&. Palmer, A. R. 1984. The Biomere Problem ‐ Evolution of an Idea. Journal of Paleontology 58(3):599‐611. Palmer, A. R. 1998a. A proposed nomenclature for stages and series for the Cambrian of Laurentia. Canadian Journal of Earth Sciences 35(4):323‐328. Palmer, A. R. 1998b. Why is intercontinental correlation within the Lower Cambrian so difficult. Revista Espanola de Paleontologia n extr. Homenaje al Prof. Gonzalo Vidal:17‐21. Palumbi, S. R. 1994. Genetic divergence, reproductive isolation, and marine speciation. Annual Review of Ecology and Systematics 25(1):547‐572. Pelechaty, S. M. 1996. Stratigraphic evidence for the Siberia‐Laurentia connection and early Cambrian rifting. Geology 24(8):719‐722. Peng, S. 2003. Chronostratigraphic subdivision of the Cambrian of China. Geologica Acta 1(1):135. Peng, S., L. Babcock, and R. Cooper. 2012. The Cambrian Period. Pp. 37‐46. In G. O. J. G. Ogg, F. M. Gradstein, ed. The Concise Geological Timescale. Cambridge University Press, Cambridge. Peng, S. C., and R. A. Robison. 2000. Agnostoid biostratigraphy across the Middle‐Upper Cambrian boundary in Hunan, China. Journal of Paleontology 74(4):1‐104. Penny, A. M., R. Wood, A. Curtis, F. Bowyer, R. Tostevin, and K. H. Hoffman. 2014. Ediacaran metazoan reefs from the Nama Group, Namibia. Science 344(6191):1504‐6. Peters, S. E. 2005. Geologic constraints on the macroevolutionary history of marine animals. Proc Natl Acad Sci U S A 102(35):12326‐31. Peters, S. E., and R. R. Gaines. 2012. Formation of the 'Great Unconformity' as a trigger for the

132

Cambrian explosion. Nature 484(7394):363‐366. Peters, S. E., and D. P. Loss. 2012. Storm and fair‐weather wave base: A relevant distinction? Geology 40(6):511‐514. Peters, S. E., and M. McClennen. 2015. The Paleobiology Database application programming interface. Paleobiology FirstView:1‐7. Peterson, K. J. 2005. Macroevolutionary interplay between planktic larvae and benthic predators. Geology 33(12):929‐932. Pimm, S. L. 1991. The balance of nature?: ecological issues in the conservation of species and communities. University of Chicago Press, Chicago. Porter, J. W. 1972. Predation by Acanthaster and Its Effect on Coral Species Diversity. American naturalist 106(950):487‐+. Porter, S. M. 2004. Closing the phosphatization window: testing for the influence of taphonomic megabias on the pattern of small shelly fossil decline. Palaios 19(2):178‐183. Porter, S. M. 2007. Seawater chemistry and early carbonate biomineralization. Science 316(5829):1302. Powell, C. M., and G. M. Young. 1995. Are Neoproterozoic glacial deposits preserved on the margins of Laurentia related to the fragmentation of two supercontinents?: Comment and Reply. Geology 23(11):1053‐1055. Pratt, B. R., B. R. Spincer, R. A. Wood, A. Y. Zhuravlev, A. Zhuravlev, and R. Riding. 2001. Ecology and evolution of Cambrian reefs. Pp. 254‐274. In Z. A. I. U. R. R., ed. The ecology of the Cambrian radiation. Columbia University Press, New York. Raup, D. M. 1972. Taxonomic Diversity during the Phanerozoic. Science 177(4054):1065‐71. Raup, D. M. 1979. Biases in the fossil record of species and genera. Bulletin of the Carnegie Museum of Natural History 13:85‐91. Raup, D. M. 1991. A kill curve for Phanerozoic marine species. Paleobiology 17(1):37‐48. Reich, P. B., I. J. Wright, J. Cavender‐Bares, J. M. Craine, J. Oleksyn, M. Westoby, and M. B. Walters. 2003. The evolution of plant functional variation: Traits, spectra, and strategies. International Journal of Plant Sciences 164(3):S143‐S164. Rhoads, D. C., and J. W. Morse. 1971. Evolutionary and ecologic significance of oxygen‐ deficient marine basins. Lethaia 4(4):413‐428. Ronov, A. B., V. E. Khain, A. N. Balukhovsky, and K. B. Seslavinsky. 1980. Quantitative‐Analysis of Phanerozoic Sedimentation. Sedimentary Geology 25(4):311‐325. Rosenzweig, M. L., and R. D. Mccord. 1991. Incumbent Replacement ‐ Evidence for Long‐Term Evolutionary Progress. Paleobiology 17(3):202‐213. Rowland, S. M. 2002. Reef patterns and environmental influences in the Cambrian and earliest Ordovician. Pp. 95‐128. In W. Kiessling, E. Flügel, and J. Golonka, eds. Phanerozoic reef patterns. SEPM Press, Tulsa, Oklahoma. Ruban, D. A. 2010. Palaeoenvironmental setting (glaciations, sea level, and plate tectonics) of Palaeozoic major biotic radiations in the marine realm. Pp. 143‐158. Annales de Paléontologie. Elsevier. Runnegar, B. 1982a. The Cambrian Explosion ‐ Animals or Fossils. Journal of the Geological Society of Australia 29(3‐4):395‐411. Runnegar, B. 1982b. Oxygen requirements, biology and phylogenetic significance of the late Precambrian wormDickinsonia, and the evolution of the burrowing habit. Alcheringa: An Australasian Journal of Palaeontology 6(3):223‐239. Saltzman, M. R., R. L. Ripperdan, M. D. Brasier, K. C. Lohmann, R. A. Robison, W. T. Chang, S. C. Peng, E. K. Ergaliev, and B. Runnegar. 2000. A global carbon isotope excursion (SPICE) during the Late Cambrian: relation to trilobite extinctions, organic‐matter burial and sea

133

level. Palaeogeography Palaeoclimatology Palaeoecology 162(3‐4):211‐223. Saltzman, M. R., S. A. Young, L. R. Kump, B. C. Gill, T. W. Lyons, and B. Runnegar. 2011a. Pulse of atmospheric oxygen during the late Cambrian. Proc Natl Acad Sci USA 108(10):3876‐81. Saltzman, M. R., S. A. Young, L. R. Kump, B. C. Gill, T. W. Lyons, and B. Runnegar. 2011b. Pulse of atmospheric oxygen during the late Cambrian. Proc Natl Acad Sci U S A 108(10):3876‐81. Sanders, H. L. 1968. Marine benthic diversity: a comparative study. American naturalist:243‐282. Schluter, D. 2000. The ecology of adaptive radiation. Oxford University Press, Oxford, UK. Scotese, C. R. 2001. Paleomap Project. http://www.scotese.com/. Scotese, C. R. 2009. Late Proterozoic plate tectonics and palaeogeography: a tale of two supercontinents, Rodinia and Pannotia. Geological Society, London, Special Publications 326:67‐83. Seilacher, A. 1989. Vendozoa ‐ Organismic Construction in the Proterozoic Biosphere. Lethaia 22(3):229‐239. Sepkoski, J. J. 1981. A Factor Analytic Description of the Phanerozoic Marine Fossil Record. Paleobiology 7(1):36‐53. Sepkoski, J. J., R. K. Bambach, D. M. Raup, and J. W. Valentine. 1981. Phanerozoic Marine Diversity and the Fossil Record. Nature 293(5832):435‐437. Sepkoski, J. J., Jr. 1974. Quantified coefficients of association and measurement of similarity. Journal of the International Association for Mathematical Geology 6(2):135‐152. Sepkoski, J. J., Jr. 1978. A kinetic model of Phanerozoic taxonomic diversity I. Analysis of marine orders. Paleobiology:223‐251. Sepkoski, J. J., Jr. 1979. A kinetic model of Phanerozoic taxonomic diversity II. Early Phanerozoic families and multiple equilibria. Paleobiology 5(3):222‐251. Sepkoski, J. J., Jr. 1984. A Kinetic Model of Phanerozoic Taxonomic Diversity. III. Post‐Paleozoic Families and Mass Extinctions. Paleobiology 10(2):246‐267. Sepkoski, J. J., Jr. 1988. Alpha, Beta, or Gamma: Where does all the diversity go? Paleobiology 14(3):221‐234. Sepkoski, J. J., Jr. 1992. Proterozoic–Early Cambrian diversification of metazoans and metaphytes. The Proterozoic Biosphere:553‐564. Sepkoski, J. J., Jr. 1993a. Ten years in the library: new data confirm paleontological patterns. Paleobiology:43‐51. Sepkoski, J. J., Jr. 1993b. Ten years in the library: new data confirm paleontological patterns. Paleobiology 19(1):43‐51. Sepkoski, J. J., Jr. 1995. The Ordovician radiations: diversification and extinction shown by global genus‐level taxonomic data. Pp. 393‐396. In J. D. Cooper, et al., ed. Ordovician odyssey: Short papers for the Seventh International Symposium on the Ordovician System. Pacific Section, Society for Sedimentary Geology (SEPM), Fullerton, California. Sepkoski, J. J., Jr. 1996. Patterns of Phanerozoic extinction: a perspective from global data bases. Pp. 35‐51. Global events and event stratigraphy in the Phanerozoic. Springer. Sepkoski, J. J., Jr. 1997. Biodiversity: past, present, and future. J Paleontol 71(4):533‐9. Sepkoski, J. J., Jr. 2002. A compendium of fossil marine animal genera. Bulletins of American Paleontology 363:1‐563. Sepkoski, J. J., Jr., and P. M. Sheehan. 1983. Diversification, faunal change, and community replacement during the Ordovician radiations. Pp. 673‐717. Biotic interactions in recent and fossil benthic communities. Springer. Servais, T., D. A. Harper, A. Munnecke, A. W. Owen, and P. M. Sheehan. 2009. Understanding the Great Ordovician Biodiversification Event (GOBE): Influences of paleogeography, paleoclimate, or paleoecology. GSa Today 19(4):4‐10.

134

Shields‐Zhou, G., and L. Och. 2011. The case for a Neoproterozoic oxygenation event: geochemical evidence and biological consequences. GSa Today 21(3):4‐11. Shu, D. G., S. C. Morris, J. Han, Y. Li, X. L. Zhang, H. Hua, Z. F. Zhang, J. N. Liu, J. F. Guo, Y. Yao, and K. Yasui. 2006. Lower Cambrian vendobionts from China and early diploblast evolution. Science 312(5774):731‐734. Signor, P. W., and G. J. Vermeij. 1994. The Plankton and the Benthos ‐ Origins and Early History of an Evolving Relationship. Paleobiology 20(3):297‐319. Skovsted, C. B. 2003. Small shelly fauna from the late Early Cambrian of North‐East Greenland. Skovsted, C. B., G. A. Brock, A. Lindstrom, J. S. Peel, J. R. Paterson, and M. K. Fuller. 2007. Early Cambrian record of failed durophagy and shell repair in an epibenthic mollusc. Biol Lett 3(3):314‐7. Smith, A. B. 2001. Large‐scale heterogeneity of the fossil record: implications for Phanerozoic biodiversity studies. Philos Trans R Soc Lond B Biol Sci 356(1407):351‐67. Smith, A. B. 2007a. Intrinsic versus extrinsic biases in the fossil record: contrasting the fossil record of echinoids in the Triassic and early Jurassic using sampling data, phylogenetic analysis, and molecular clocks. Paleobiology 33(2):310‐323. Smith, A. B. 2007b. Marine diversity through the Phanerozoic: problems and prospects. Journal of the Geological Society 164(4):731‐745. Smith, A. B., and A. J. McGowan. 2005. Cyclicity in the fossil record mirrors rock outcrop area. Biol Lett 1(4):443‐5. Smith, A. B., and B. Stockley. 2005. The geological history of deep‐sea colonization by echinoids: roles of surface productivity and deep‐water ventilation. Proc Biol Sci 272(1565):865‐9. Smith, M. P., and D. A. T. Harper. 2013. Causes of the Cambrian Explosion. Science 341(6152):1355‐1356. Sperling, E. A., C. A. Frieder, A. V. Raman, P. R. Girguis, L. A. Levin, and A. H. Knoll. 2013. Oxygen, ecology, and the Cambrian radiation of animals. Proc Natl Acad Sci USA 110(33):13446‐ 13451. Sperling, E. A., C. J. Wolock, A. S. Morgan, B. C. Gill, M. Kunzmann, G. P. Halverson, F. A. Macdonald, A. H. Knoll, and D. T. Johnston. 2015. Statistical analysis of iron geochemical data suggests limited late Proterozoic oxygenation. Nature 523(7561):451‐4. Squire, R. J., I. H. Campbell, C. M. Allen, and C. J. L. Wilson. 2006. Did the Transgondwanan Supermountain trigger the explosive radiation of animals on Earth? Earth and Planetary Science Letters 250(1‐2):116‐133. Stanley, S. M. 1976. Fossil data and the Precambrian‐Cambrian evolutionary transition. American Journal of Science 276(1):56‐76. Stanley, S. M. 2007. An analysis of the history of marine animal diversity. Paleobiology 33(4):1‐ 55. Stanley, S. M. 2008. Predation defeats competition on the seafloor. Paleobiology 34(1):1‐21. Steiner, M., G. X. Li, Y. Qian, M. Y. Zhu, and B. D. Erdtmann. 2007. Neoproterozoic to early Cambrian small shelly fossil assemblages and a revised biostratigraphic correlation of the Yangtze Platform (China). Palaeogeogr Palaeoclim Palaeocl 254(1‐2):67‐99. Terborgh, J., and J. A. Estes. 2013. Trophic cascades: predators, prey, and the changing dynamics of nature. Island Press, Washington,DC. Thayer, C. W. 1983. Sediment‐mediated biological disturbance and the evolution of marine benthos. Pp. 479‐625. Biotic interactions in recent and fossil benthic communities. Springer. Tobler, W. R. 1970. A computer movie simulating urban growth in the Detroit region. Economic geography 46(sup1):234‐240.

135

Tomašových, A., J. D. Kennedy, T. J. Betzner, N. B. Kuehnle, S. Edie, S. Kim, K. Supriya, A. E. White, C. Rahbek, and S. Huang. 2016. Unifying latitudinal gradients in range size and richness across marine and terrestrial systems. P. 20153027. Proc. R. Soc. B. The Royal Society. Torsvik, T., M. Smethurst, J. G. Meert, R. Van der Voo, W. McKerrow, M. Brasier, B. Sturt, and H. Walderhaug. 1996. Continental break‐up and collision in the Neoproterozoic and Palaeozoic—a tale of Baltica and Laurentia. Earth‐Science Reviews 40(3):229‐258. Torsvik, T. H., and E. F. Rehnstrom. 2001. Cambrian palaeomagnetic data from Baltica: Implications for true polar wander and Cambrian palaeogeography. Journal of the Geological Society 158:321‐329. Trotter, J. A., I. S. Williams, C. R. Barnes, C. Lécuyer, and R. S. Nicoll. 2008. Did cooling oceans trigger Ordovician biodiversification? Evidence from conodont thermometry. Science 321(5888):550‐554. Valentine, J. W. 1969. Patterns of taxonomic and ecological structure of the shelf benthos during Phanerozoic time. Palaeontology 12(4):684‐709. Valentine, J. W. 1970. How many marine invertebrate fossil species? A new approximation. Journal of Paleontology:410‐415. Valentine, J. W. 1971. Plate Tetonics and Shallow Marine Diversity and Endemism, an Actualistic Model. Systematic Zoology 20(3):253‐264. Valentine, J. W. 1994. Late Precambrian bilaterians: grades and clades. Proc Natl Acad Sci U S A 91(15):6751‐7. Valentine, J. W. 2002. Prelude to the Cambrian explosion. Annual Review of Earth and Planetary Sciences 30(1):285‐306. Valentine, J. W. 2004. On the origin of phyla. University of Chicago Press, Chicago. Valentine, J. W., and D. H. Erwin. 1983. Patterns of diversification of higher taxa: a test of macroevolutionary paradigms. Modalités, rythmes et mécanismes de l’évolution biologique, Éditions du CNRS, Paris:219‐223. Valentine, J. W., T. C. Foin, and D. Peart. 1978. A provincial model of Phanerozoic marine diversity. Paleobiology:55‐66. Valentine, J. W., D. Jablonski, and D. H. Erwin. 1999. Fossils, molecules and embryos: new perspectives on the Cambrian explosion. Development 126(5):851‐9. Valentine, J. W., and E. M. Moores. 1970. Plate‐tectonic regulation of faunal diversity and sea level: a model. Nature 228(5272):657‐9. Valentine, J. W., and E. M. Moores. 1972. Global tectonics and the fossil record. The Journal of Geology 80(2):167‐184. Vannier, J., M. Steiner, E. Renvoise, S. X. Hu, and J. P. Casanova. 2007. Early Cambrian origin of modern food webs: evidence from predator arrow worms. Proc Biol Sci 274(1610):627‐ 33. Veech, J. A., K. S. Summerville, T. O. Crist, and J. C. Gering. 2002. The additive partitioning of species diversity: recent revival of an old idea. Oikos 99(1):3‐9. Vermeij, G. J. 1987. Evolution and escalation: an ecological history of life. Princeton University Press, Princeton, New Jersey. Vermeij, G. J. 1989. The origin of skeletons. Palaios 4(6):585‐589. Vermeij, G. J. 1994. The Evolutionary Interaction among Species ‐ Selection, Escalation, and Coevolution. Annual Review of Ecology and Systematics 25:219‐236. Waggoner, B. 2003. The ediacaran biotas in space and time. Integr Comp Biol 43(1):104‐13. Wagner, P. J., M. A. Kosnik, and S. Lidgard. 2006. Abundance distributions imply elevated complexity of post‐Paleozoic marine ecosystems. Science 314:1289‐1292. Walcott, C. D. 1890. The fauna of the Lower Cambrian or Olenellus zone. US Government

136

Printing Office. Walker, L. J., B. H. Wilkinson, and L. C. Ivany. 2002. Continental drift and Phanerozoic carbonate accumulation in shallow‐shelf and deep‐marine settings. The Journal of Geology 110(1):75‐87. Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. Ecol Monographs 30(3):279‐338. Whittaker, R. H. 1972. Evolution and measurement of species diversity. Taxon 21(2‐3):213‐251. Wille, M., T. F. Nagler, B. Lehmann, S. Schröder, and J. D. Kramers. 2008. Hydrogen sulphide release to surface waters at the Precambrian/Cambrian boundary. Nature 453(7196):767‐769. Wood, R., and A. Curtis. 2015. Extensive metazoan reefs from the Ediacaran Nama Group, Namibia: the rise of benthic suspension feeding. Geobiology 13(2):112‐22. Wood, R., A. Y. Zhuravlev, and F. Debrenne. 1992. Functional biology and ecology of Archaeocyatha. Palaios 7(2):131‐156. Wood, R. A. 2011. Paleoecology of the earliest skeletal metazoan communities: Implications for early biomineralization. Earth‐Science Reviews 106(1‐2):184‐190. Wu, W., M. Zhu, and M. Steiner. 2014. Composition and tiering of the Cambrian sponge communities. Palaeogeography, Palaeoclimatology, Palaeoecology 398(0):86‐96. Xiao, S., and M. Laflamme. 2009. On the eve of animal radiation: phylogeny, ecology and evolution of the Ediacara biota. Trends Ecol Evol 24(1):31‐40. Yin, Z., M. Zhu, E. H. Davidson, D. J. Bottjer, F. Zhao, and P. Tafforeau. 2015. Sponge grade body fossil with cellular resolution dating 60 Myr before the Cambrian. Proceedings of the National Academy of Sciences 112(12):E1453‐E1460. Yin, Z., M. Zhu, P. Tafforeau, J. Chen, P. Liu, and G. Li. 2013. Early embryogenesis of potential bilaterian animals with polar lobe formation from the Ediacaran Weng’an Biota, South China. Precambrian Research 225:44‐57. Young, G. C., and J. R. Laurie. 1996. An Australian Phanerozoic Timescale. Oxford University Press, Oxford. Zhan, R., and J. Jin. 2014. Early‐Middle Ordovician brachiopod dispersal patterns in South China. Integr Zool 9(2):121‐40. Zhang, X., and L. Cui. 2016. Oxygen requirements for the Cambrian explosion. Journal of Earth Science 27(2):187‐195. Zhao, Y. L., J. L. Yuan, S. Peng, L. E. Babcock, J. Peng, Q. J. Guo, J. P. Lin, T. S. Tai, R. D. Yang, and Y. X. Wang. 2008. A new section of Kaili Formation (Cambrian) and a biostratigraphic study of the boundary interval across the undefined Cambrian Series 2 and Series 3 at Jianshan, Jianhe County, China with a discussion of global correlation based on the first appearance datum of Oryctocephalus indicus (Reed, 1910). Progress in Natural Science 18(12):1549‐1556. Zhu, M. Y., L. E. Babcock, and S. C. Peng. 2006. Advances in Cambrian stratigraphy and paleontology: integrating correlation techniques, paleobiology, taphonomy and paleoenvironmental reconstruction. Palaeoworld 15(3‐4):217‐222. Zhu, M. Y., G. X. Li, J. m. Zhang, and M. Steiner. 2001. Early Cambrian stratigraphy of east Yunnan, southwestern China: a synthesis. Acta Palaeontologica Sinica 40:4‐29. Zhu, M. Y., J. Vannier, H. Van Iten, and Y. L. Zhao. 2004. Direct evidence for predation on trilobites in the Cambrian. Proc Biol Sci 271 Suppl 5(Suppl 5):S277‐80. Zhuravlev, A. Y. 2001. Biotic diversity and structure during the Neoproterozoic‐Ordovician transition. Pp. 173‐199. The ecology of the Cambrian radiation. Zhuravlev, A. Y., and E. B. Naimark. 2005. Alpha, beta, or gamma: Numerical view on the Early

137

Cambrian world. Palaeogeography Palaeoclimatology Palaeoecology 220(1‐2):207‐225. Zhuravlev, A. Y., E. B. Naimark, and R. A. Wood. 2015. Controls on the diversity and structure of earliest metazoan communities: Early Cambrian reefs from Siberia. Earth‐Science Reviews 147:18‐29. Zhuravlev, A. Y., and R. Riding. 2001. The ecology of the Cambrian radiation. Columbia University Press, New York. Zhuravlev, A. Y., and R. A. Wood. 1996. Anoxia as the cause of the mid‐early Cambrian (Botomian) extinction event. Geology 24(4):311‐314.

138

Appendix: [R] CODE

#Install packages library(vegan) # for diversity library(Hmisc) # for error bars library(sp) # for world map library(MASS) # for non-metric scaling library(rgl) # for scatterplot library(geosphere) # for great circle distance calculation library(ggplot2) #for plot library(nlme) #for linear regression library(MuMIn) # for model selection

Get and clean data setwd(choose.dir()) camb<- read.csv(file="occurrence(23052014).csv", header=TRUE, sep=";") # Load time scale stages <- read.csv(file="time_scale_new.csv", header=TRUE, sep=";", quote="\"") stage.age <- read.csv(file="stage_age_new.csv", header=TRUE, sep=";") age <- sort(stage.age$Age.m, decreasing=T) age.legend <- c(age, 482) bases2 <- sort(stage.age$Age.b, decreasing=T) bases.legend <- c(bases2, 478.6) st.lab <- rev(stage.age$stage.s) nbox <- length(age.legend)-1 # number of boxes to be drawn for figures # Load age assingments of collections colage<- read.csv(file="collections_new(230514).csv", header=T, sep=";", quote="\"") # Link collection ages with occurrences col <- read.csv("LNa-collections.csv", header = T, sep = ";", quote = "\"")

139

col.age<-subset(col, select=c(collection_no, gp_mid_lat, gp_mid_lng, localsection, localbed, regionalsection, regionalbed))

col.age. <- merge(col, colage, by="collection_no") col.age.. <- subset(col.age., select=c(collection_no, gp_mid_lat, gp_mid_lng, localsection, localbed, regionalsection, regionalbed, stage.num, Probability.s)) data <- merge(camb, col.age.., by="collection_no") data$slc <-data$stage.num+3 camb <- rbind(data) #Exclude Acritarchs and eggs excl <- c("Parapandorina", "sphaeridium", "spaeridium", "phycus", "Megasphaera") acr1 <- numeric() for (i in 1:length(excl)) { acr <- grep(excl[i], camb$occurrence.genus_name) acr1 <- c(acr1, acr) } camb <- camb[-acr1,] # Raw data counts temp <- unique(subset(camb, select=c(collection_no, occurrence.genus_name, slc))) ocs <- table(temp$slc) # number of occurrences coll <- unique(subset(temp, select=c(collection_no, slc))) cols <- table(coll$slc) # number of collections gene <- unique(subset(temp, select=c(occurrence.genus_name, slc))) gens <- table(gene$slc) #number of genera

Diversity estimates

#Gamma diversity #Use sqs () function provided online by John Alroy

140

sq <- numeric() sd.sq <- numeric() for (i in 1:14){ x <- subset(camb., slc==i) t <- table(factor(x$occurrence.genus_name)) sq[i] <- sqs(t, 0.7, 100)[3] sd.sq[i] <- sqs(t, 0.7, 100)[16] }

SqDiv <- sq # gamma diversity sd.sq<-sd.sq # standard deviation

#Estimates of alpha, beta, extinction, and origination rate using by-list unweighted #subsampling temp <- unique(subset(camb., select=c(collection_no, occurrence.genus_name, slc))) coll <- unique(subset(temp, select=c(collection_no, slc))) cols <- table(coll$slc) # Subsampling approach for collections (unweighted lists, UW) ss <- min(cols) vl<-14 trial <- 100 ssmat<-data.frame(); smat<- matrix(ncol=trial, nrow=vl) rich<-smat; cols<-smat; ocs<-smat; fext.<-smat;fori.<-smat; pext <-smat;pori <- smat thr<-numeric();ex<-numeric();or<-numeric();sg<-numeric() tot<- numeric();s1<-numeric();s2<-numeric();ns<-numeric() attach(temp) # Start big loop for (k in 1:trial) { for (i in 1:nbox) { if (length(levels(factor(collection_no[slc==i])))>=ss) { x <- sample(levels(factor(collection_no[slc==i])), ss)

141

y <-subset(temp, collection_no %in% as.numeric(x)) cols[i,k]<-length(levels(factor(y$collection_no))) # number of collections in each bin ocs[i,k]<-nrow(y) # number of occurrrences in bin rich[i,k]<-length(levels(factor(y$occurrence.genus_name))) #generic richness in each #bin } ssmat<-rbind(ssmat, y) } #calculate the number of four fundamental classes of taxa for (i in m.sq){ rfl<-levels(factor(ssmat$occurrence.genus_name[ssmat$slc==i])) #All genera in bin rfupl<-levels(factor(ssmat$occurrence.genus_name[ssmat$slc>i])) #All genera in #younger bins rfdol<-levels(factor(ssmat$occurrence.genus_name[ssmat$slc

s1[i] <- length(rfl [rfl %in% rfupl]) #Survivors from bin (Ft) s2[i] <- length(rfl [rfl %in% rfdol]) #Survivors to bin (bL) ns[i] <- length(rfl [rfl %in% rfotl]) thr[i] <- length(rfdol [rfdol %in% rfupl]) tot[i] <- length(rfl) sg[i]<-length(rfl)-ns[i] ex[i]<- length(rfl)-s1[i] or[i]<- length(rfl)-s2[i] fext.[i, k]<- -log(thr[i]/(thr[i]+ex[i]-sg[i])) fori.[i, k]<- -log(thr[i]/(thr[i]+or[i]-sg[i]))

142

pori[i, k]<-(sg[i]+s1[i])/tot[i] pext[i, k]<-(sg[i]+s2[i])/tot[i] } } # end big loop detach(temp) alpha.uw <- ocs/cols sd.alpha.uw<-apply(alpha.uw, 1, sd, na.rm=T) alpha.uw<-apply(alpha.uw, 1, mean) b <- rich/alpha.uw beta.uw <- apply(b, 1, mean) sd.beta.uw <- apply(b, 1, sd, na.rm=T) bad <- rich -alpha.uw beta.ad.uw <- apply(bad, 1, mean) sd.beta.ad<- apply(bad, 1, sd,na.rm=T) source("plot_alpha_new.R") #Foote’s per-capita origination and extinction rates fext.uw<-apply(fext., 1, mean) fori.uw<-apply(fori., 1,mean) sd.fext<-apply(fext., 1, sd, na.rm=T) sd.fori<-apply(fori., 1,sd, na.rm=T) #Per-taxon origination and extinction rates pext.uw<-apply(pext, 1, mean) pori.uw<-apply(pori, 1,mean) sd.pext<-apply(pext, 1, sd, na.rm=T) sd.pori<-apply(pori, 1,sd, na.rm=T)

Origination/extinction rates among onshore-offshore gradient

Ext <- read.csv(file="Ext.csv", header=TRUE, sep=";") Ori <- read.csv(file="Ori.csv", header=TRUE, sep=";")

143

windows(h=6, w=8) par(mfrow=c(2,2)) par(mar = c(5,5,2,2)) #plot extinction rates first plot(Ext$zone2, Ext$zone3, type="p", xlim=c(-0.2, 1), ylim=c(-0.2, 1), xlab="", ylab="",xaxt="n", yaxt="n") axis(1, cex.axis=1.3) axis(2, cex.axis=1.3) mtext(side = 1, line=3.6, "Extinction rate in Zone 2", col="black", cex=1.5) mtext(side = 2, line=3.3, "Extinction rate in Zone 3", col="black", cex=1.5) plotCI(Ext$zone2, Ext$zone3, ui=Ext$zone3 + Ext$sd.zone3, li=Ext$zone3 - Ext$sd.zone3,xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="y", pch=18, cex=1.5, add=T,xaxt="n", yaxt="n") par(new=T) plotCI(Ext$zone2, Ext$zone3, ui=Ext$zone2 + Ext$sd.zone2, li=Ext$zone2 - Ext$sd.zone2, xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="x", pch=18, cex=1.8, add=T,xaxt="n", yaxt="n") text(Ext$zone2, Ext$zone3, labels=Ext$stage, cex= 0.8, pos=3, offset=1) segments(-0.2, -0.2, 1, 1, col="black") abline(v=0, lty=2);abline(h=0, lty=2) #second one plot(Ext$zone3, Ext$zone4, type="p", xlim=c(-0.2, 1), ylim=c(-0.2, 1), xlab="", ylab="",xaxt="n", yaxt="n") axis(1, cex.axis=1.3) axis(2, cex.axis=1.3) mtext(side = 1, line=3.6, "Extinction rate in Zone 3", col="black", cex=1.5) mtext(side = 2, line=3.3, "Extinction rate in Zone 4", col="black", cex=1.5) plotCI(Ext$zone3, Ext$zone4, ui=Ext$zone4 + Ext$sd.zone4, li=Ext$zone4 - Ext$sd.zone4,xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="y", pch=18, cex=1.5, add=T,xaxt="n", yaxt="n")

144

par(new=T) plotCI(Ext$zone3, Ext$zone4, ui=Ext$zone3 + Ext$sd.zone3, li=Ext$zone3 - Ext$sd.zone3, xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="x", pch=18, cex=1.8, add=T,xaxt="n", yaxt="n") text(Ext$zone3, Ext$zone4, labels=Ext$stage, cex= 0.8, pos=3, offset=1) segments(-0.2, -0.2, 1, 1, col="black") abline(v=0, lty=2);abline(h=0, lty=2) #third one plot(Ext$zone4, Ext$zone5, type="p", xlim=c(-0.2, 1), ylim=c(-0.2, 1), xlab="", ylab="",xaxt="n", yaxt="n") axis(1, cex.axis=1.3) axis(2, cex.axis=1.3) mtext(side = 1, line=3.6, "Extinction rate in Zone 4", col="black", cex=1.5) mtext(side = 2, line=3.3, "Extinction rate in Zone 5", col="black", cex=1.5) plotCI(Ext$zone4, Ext$zone5, ui=Ext$zone5 + Ext$sd.zone5, li=Ext$zone5 - Ext$sd.zone5,xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="y", pch=18, cex=1.5, add=T,xaxt="n", yaxt="n") par(new=T) plotCI(Ext$zone4, Ext$zone5, ui=Ext$zone4 + Ext$sd.zone4, li=Ext$zone4 - Ext$sd.zone4, xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="x", pch=18, cex=1.8, add=T,xaxt="n", yaxt="n") text(Ext$zone4, Ext$zone5, labels=Ext$stage, cex= 0.8, pos=3, offset=1) segments(-0.2, -0.2, 1, 1, col="black") abline(v=0, lty=2);abline(h=0, lty=2) #fourth one plot(Ext$zone5, Ext$zone6, type="p", xlim=c(-0.2, 1), ylim=c(-0.2, 1), xlab="", ylab="",xaxt="n", yaxt="n") axis(1, cex.axis=1.3) axis(2, cex.axis=1.3) mtext(side = 1, line=3.6, "Extinction rate in Zone 5", col="black", cex=1.5)

145

mtext(side = 2, line=3.3, "Extinction rate in Zone 6", col="black", cex=1.5) plotCI(Ext$zone5, Ext$zone6, ui=Ext$zone6 + Ext$sd.zone6, li=Ext$zone6 - Ext$sd.zone6,xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="y", pch=18, cex=1.5, add=T,xaxt="n", yaxt="n") par(new=T) plotCI(Ext$zone5, Ext$zone6, ui=Ext$zone5 + Ext$sd.zone5, li=Ext$zone5 - Ext$sd.zone5, xlim=c(-0.2, 1), ylim=c(-0.2, 1), err="x", pch=18, cex=1.8, add=T,xaxt="n", yaxt="n") text(Ext$zone5, Ext$zone6, labels=Ext$stage, cex= 0.8, pos=3, offset=1) segments(-0.2, -0.2, 1, 1, col="black") abline(v=0, lty=2);abline(h=0, lty=2) #origination plot same as extinction plot, apply “Ori” dataset

Geodisparity

#Estimates of dissimilarity of assemblage composition among geographic grids geo2<-subset(camb, select=c(collection_no, occurrence.genus_name, paleolatdec, paleolngdec, gp_mid_lat, gp_mid_lng, slc)) #Aggregate data by grids by 5 * 5 gr <- 5 # determine size grids by 5 * 5 start.lat <- 90-gr/2 end.lat <- -90+gr/2 lats <- seq(from=start.lat, to=end.lat, by=-gr) start.lon <- -180 + gr/2 end.lon <- 180-gr/2 lons <- seq(from=start.lon, to=end.lon, by=gr) lat <- geo2$paleolatdec lon <- geo2$paleolngdec gri <- numeric(nrow(geo2)) la <- numeric(nrow(geo2)) lo <- numeric(nrow(geo2)) geodist <- numeric(nrow(geo2))

146

gr. <- 1 for (y in lats) { for (x in lons) { gr.x <- which(lat=y-gr/2 & lon>x-gr/2 & lon<=x+gr/2) la[gr.x] <- y lo[gr.x] <- x gri[gr.x] <- gr. gr. <- gr. +1 } } temp <- cbind(geo2, gri, la, lo)

#start loop for geodisparity for (ii in 4:14){ temp1 <- subset(temp, slc==ii) x <- table(factor(temp1$gri)) #set threshold of 10 occurrences for each grid t <- x [which (x>=10)] temp2 <- subset(temp1, gri %in% names(t)) # genus level gens <- levels(factor(temp2$occurrence.genus_name)) res <- matrix(data=0, nrow=length(gens), ncol=length(t)) res <- as.data.frame(res) row.names(res) <- gens colnames(res) <- names(t) z <- names(t) for (y in 1:length(t)) { oc <- table(factor(temp2$occurrence.genus_name[temp2$gri==names(t[y])])) res[names(oc),y] <- oc }

147

for (i in 1:ncol(res)) res[,i] <- res[,i]/sum(res[,i]) s <- matrix(nrow=ncol(res), ncol=ncol(res), dimnames = list(names(res),names(res))) for (j in 1 : (ncol(res)-1)) { for (i in j : (ncol(res)-1)) { s [i+1, j] <- sum(pmin(res[, j], res[, i+1])) } } dat <- as.dist(1-s) a<-1 b<-0 d <- levels(factor(temp2$gri)) geodist <- matrix(nrow=length(d), ncol=length(d)) colnames(geodist)<-d rownames(geodist)<-d for (h in d) { b<-b+1 a<-1 for (l in d) { lox <- as.numeric(levels(factor(temp2$lo[temp2$gri==h]))) lax <- as.numeric(levels(factor(temp2$la[temp2$gri==h]))) loy <- as.numeric(levels(factor(temp2$lo[temp2$gri==l]))) lay <- as.numeric(levels(factor(temp2$la[temp2$gri==l]))) geodist[a,b] <- distHaversine(c(lox, lax), c(loy, lay))/1000 #geodist.s2[a,b] <- distVincentyEllipsoid(c(lox, lax), c(loy, lay))/1000 a<-a+1 } } #geodispariy geodist <- as.dist(geodist) # matrix of geodisparity

148

} }

Ecodisparity ecod <- numeric() ecsd <- numeric() for (ii in 4:14){ a <- subset(ecoD, slc==ii) x <- table(factor(a$t)) x <- x[x>=10] b <- subset(a, t %in% names(x)) gens <- levels(factor(b$occurrence.genus_name)) res <- matrix(data=0, nrow=length(gens), ncol=length(x)) res <- as.data.frame(res) row.names(res) <- gens colnames(res) <- names(x) for (y in 1:length(x)) { oc <- table(factor(b$occurrence.genus_name[b$t==names(x[y])])) res[names(oc),y] <- oc } for (i in 1:ncol(res)) res[,i] <- res[,i]/sum(res[,i]) s <- matrix(nrow=ncol(res), ncol=ncol(res), dimnames = list(names(res),names(res))) for (j in 1:(ncol(res)-1)) { for (i in j:(ncol(res)-1)) { s [i+1, j] <- sum(pmin(res[, j], res[, i+1])) } } dis <- as.dist(1-s) aa<-1 bb<-0

149

e <- levels(factor(b$t)) ecodist <- matrix(data=0, nrow=length(e), ncol=length(e)) colnames(ecodist)<-e rownames(ecodist)<-e #ecodist<-as.dist(ecodist) l <- length(e) for (ff in 1:l){ for (kk in 1:l){ tes1 <- subset(b, t %in% e[ff]) tes2 <- subset(b, t %in% e[kk]) lith1 <- levels(factor(tes1$lithnow)) bath1 <- levels(factor(tes1$bathnow)) lith2 <- levels(factor(tes2$lithnow)) bath2 <- levels(factor(tes2$bathnow)) re1 <- c(lith1, bath1) re2 <- c(lith2, bath2) ecodist[ff,kk] <- 2-length(intersect(re1,re2)) ##return 0 when there are two flips ##return 1 when there is there is one flip ##return 2 when two vectors are the same } } ecodist<-as.dist(ecodist) Edis <- numeric() Esd <- numeric() st<-seq(1:2) for (k in 1:2){ o <- dis[which(ecodist==st[k])] Edis[k] <- mean(o) Esd[k] <- sd(o)/sqrt(length(o)) }

150

ecod[ii] <- Edis[1] ecsd[ii] <- Esd[1] }

Escalation

#Test escalatory ecological trends in the Cambrian invertebrates #Proportions of genera for taxa classified as carnivorous temp1 <- subset(camb., select=c(collection_no, class_name, occurrence.genus_name, composition1, feeding, habit, locomotion, attatched, slc)) temp <- subset(temp1, habit!="planktonic") NonCarni<-unique(subset(temp, (composition1=="calcite"|composition1=="aragonite"|composition1=="high Mg calcite" |composition1=="hydroxyapatite"|composition1=="low Mg calcite"|composition1=="phosphatic" |composition1=="silica") & feeding!="carnivore")) All <- unique(subset(temp, select=c(occurrence.genus_name, slc))) Allo <- unique(subset(temp, select=c(collection_no, occurrence.genus_name, slc))) AllGen <- table(All$slc) Allocs <- table(Allo$slc) Carni <- unique(subset(temp, feeding=="carnivore")) CarniAocs <- unique(subset(Carni, select=c(collection_no, occurrence.genus_name, slc))) Carniocs <- table(CarniAocs$slc) ProCarniocs <- Carniocs/Allocs[-c(1, 2, 3)] CarniAll <- unique(subset(Carni, select=c(occurrence.genus_name, slc))) CarniGen <- table(CarniAll$slc) ProCarni <- CarniGen/AllGen[-c(1, 2, 3)] NonCarniAll <- unique(subset(NonCarni, select=c(occurrence.genus_name, slc))) NonCarniAllo <- unique(subset(NonCarni, select=c(collection_no, occurrence.genus_name, slc)))

151

NonCarniSkel <- table(NonCarniAll$slc) NonCarniSkelocs <- table(NonCarniAllo$slc)

#Proportions of genera for taxa classified as non-carnivorous infaunal NoCarni <- unique(subset(temp, feeding!="carnivore")) NoCarniInf <- unique(subset(NoCarni, habit=="infaunal")) NoCarniInfAll <- unique(subset(NoCarniInf, select=c(occurrence.genus_name, slc))) NoCarniInfGen <- table(NoCarniInfAll$slc) NoCarniInfAllo <- unique(subset(NoCarniInf, select=c(collection_no, occurrence.genus_name, slc))) NoCarniInfocs <- table(NoCarniInfAllo$slc)

#Proportions of non-carnivorous infaunal in all non-carnivorous taxa NoCarniAll <- unique(subset(NoCarni, select=c(occurrence.genus_name, slc))) NoCarniGen <- table(NoCarniAll$slc) ProNoCarniInf <- NoCarniInfGen/NoCarniGen[-c(1, 2)]

#Proportions of occurrences NoCarniAllo <- unique(subset(NoCarni, select=c(collection_no, occurrence.genus_name, slc))) NoCarniocs <- table(NoCarniAllo$slc) ProNoCarniocs <- NoCarniInfocs/NoCarniocs[-c(1, 2)] ProNoCarniInfocs <- NoCarniInfocs/NoCarniocs[-c(1, 2)]

#Proportions of non-carnivorou infaunal in all taxa ProNoCarniInfinAll <- NoCarniInfGen/AllGen[-c(1, 2)]

#Proportions of genera for taxa calssified as non-carnivorous mobile (actively and #facultatively) NoCarniMob <- unique(subset(NoCarni, locomotion=="actively mobile" | locomotion=="facultatively mobile" |

152

locomotion=="fast-moving" | locomotion=="slow-moving")) NoCarniMobAll <- unique(subset(NoCarniMob, select=c(occurrence.genus_name, slc))) NoCarniMobGen <- table(NoCarniMobAll$slc) ProNoCarniMob <- NoCarniMobGen[-c(1)]/NoCarniGen[-c(1, 2, 3)] ProNoCarniMobinAll <- NoCarniMobGen[-c(1)]/AllGen[-c(1, 2, 3)] NoCarniMobAllo <- unique(subset(NoCarniMob, select=c(collection_no, occurrence.genus_name, slc))) NoCarniMobocs <- table(NoCarniMobAllo$slc) ProNoCarniMobocs <- NoCarniMobocs[-c(1)]/NoCarniocs[-c(1, 2, 3)]

#Proportions of genear for taxa classified as skeletal prey Skel <- unique(subset(NoCarni, (composition1=="calcite" | composition1=="aragonite" | composition1=="high Mg calcite" | composition1=="hydroxyapatite" | composition1=="low Mg calcite" | composition1=="phosphatic" | composition1=="silica"))) SkelPrey <- unique(subset(Skel, locomotion=="facultatively mobile" | locomotion=="stationary")) SkelPreyAll <- unique(subset(SkelPrey, select=c(occurrence.genus_name, slc))) SkelPreyGen <- table(SkelPreyAll$slc) ProSkelPrey <- SkelPreyGen/NoCarniGen SkelPreyAllo <- unique(subset(SkelPrey, select=c(collection_no, occurrence.genus_name, slc))) SkelPreyocs <- table(SkelPreyAllo$slc) ProSkelPreyocs <- SkelPreyocs/NoCarniocs #Cross-correlations among ecological groups library(car) library(kimisc)

153

xCar <- as.numeric(logit(ProCarni)) yInf <- as.numeric(logit(ProNoCarniInf[c(2:12)])) yMob <- as.numeric(logit(ProNoCarniMob)) ySke <- as.numeric(logit(ProSkelPrey[c(4:14)])) xBio <- as.numeric(logit(ProBiot)) yImm <- as.numeric(logit(ProImmEpi[c(5:14)])) #first difference d.xCar <- diff(xCar) d.yInf <- diff(yInf) d.yMob <- diff(yMob) d.ySke <- diff(ySke) d.xBio <- diff(xBio) d.yImm <- diff(yImm) #generalized differnce g.xCar <- gdiff(xCar) g.yInf <- gdiff(yInf) g.yMob <- gdiff(yMob) g.ySke <- gdiff(ySke) g.xBio <- gdiff(xBio) g.yImm <- gdiff(yImm) x <- cor.test(xCar, yInf, method="spearman") x <- cor.test(xCar, yMob, method="spearman") x <- cor.test(xCar, ySke, method="spearman") x <- cor.test(xBio, yImm, method="spearman")

Model selection z <- lm(Div.sq ~ alpha.uw + beta.uw + CarniDiv + ProCarniocs + ProCarni, data = CC) z <- lm(Div.sq ~ alpha.uw + beta.uw + ProCarniocs + ProCarni, data = CC) z <- lm(Div.sq ~ alpha.uw + beta.uw + ProCarni, data = CC) z <- lm(Div.sq ~ alpha.uw + beta.uw, data = CC) z <- lm(Div.sq ~ beta.uw, data = CC)

154

AIC(z) AICc(z)

Paleomaps

# Show occurrences on paleogeographic maps library(sp) # for world map library(maptools) library(rgdal) # also s540 <- readOGR("s540.shp", "s540") # warnings s520 <- readOGR("s520.shp", "s520") # warnings s500 <- readOGR("s500.shp", "s500") # warnings grids <- readOGR("latlong.shp", "latlong") # warnings grid10 <- grids[grids$DEGREE10=="Y",] grid20 <- grids[grids$DEGREE20=="Y",] grid30 <- grids[grids$DEGREE30=="Y",] # Plot one bin # Tommotian plot(s540, col="grey80", main="Tommotian") plot(grid30, add=T) temp <- subset(camb., slc==4) points(temp$collections.paleolngdec, temp$collections.paleolatdec , pch=19, col="red", cex=1.5) # Stage3 plot(s500, col="grey80", main="Tommotian") plot(grid30, add=T) temp <- subset(camb., slc==4) points(temp$collections.paleolngdec, temp$collections.paleolatdec , pch=19, col="red", cex=1.5) windows(h=6, w=8) par(mfrow = c(1, 1),omi=c(0,0,0,0)) op <- par(xaxs="i", yaxs="i")

155

plot(s540, col="grey80") plot(grid30, add=T) for (i in 1: nrow(res3)){ inter<-gcIntermediate( c(res3$lo.x[i],res3$la.x[i]), c(res3$lo.y[i], res3$la.y[i]), n=1000, breakAtDateLine=T) if(is.list(inter)== FALSE)lines(inter, lwd=res3$V3[i]*3, col="red") if(is.list(inter)== TRUE)lines(inter[[1]], lwd=res3$V3[i]*3, col="red") if(is.list(inter)== TRUE)lines(inter[[2]], lwd=res3$V3[i]*3, col="red") } par(op) op <- par(xaxs="i", yaxs="i") plot(s540, col="grey80") plot(grid30, add=T) i<-nrow(res3.) for (i in 1: k){ inter<-gcIntermediate( c(res3.$lo.x[i],res3.$la.x[i]), c(res3.$lo.y[i], res3.$la.y[i]), n=1000, breakAtDateLine=T) if(is.list(inter)== FALSE)segments(as.numeric(res3.$lo.x[i]),as.numeric(res3.$la.x[i]), as.numeric(res3.$lo.y[i]), as.numeric(res3.$la.y[i]), lwd=res3.$V3[i]*3, col="green") if(is.list(inter)== TRUE)segments(as.numeric(res3.$lo.x[i]),as.numeric(res3.$la.x[i]), as.numeric(inter[[1]][nrow(inter[[1]]),1]), as.numeric(inter[[1]][nrow(inter[[1]]),2]), lwd=res3.$V3[i]*3, col="green") if(is.list(inter)== TRUE)segments(as.numeric(inter[[2]][1, 1]), as.numeric(inter[[2]][1, 2]), as.numeric(res3.$lo.y[i]), as.numeric(res3.$la.y[i]), lwd=res3.$V3[i]*3, col="green") } par(op)

156