The Biology of the Evolution of Viviparity in Asterinid Sea Stars

Home , Asterinidae, Matrotrophy, Parvulastra vivipara, Sea daisy, Viviparity

THE BIOLOGY OF THE EVOLUTION OF VIVIPARITY IN ASTERINID SEA STARS

Mohammad Sadequr Rahman Khan

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in the School of Life and Environmental Sciences, Faculty of Science, at The University of Sydney

2020

Dedicated to my dear parents and beloved wife

DECLARATION

I hereby declare that this work is my own, except where otherwise acknowledged. I also declare that this thesis has not been submitted in any form for another degree or diploma at any university or other institution. I consent to this thesis being made available for photocopying and loan under the appropriate Australian copyright laws.

Mohammad Sadequr Rahman Khan Date: 10th February 2020

AUTHORSHIP ATTRIBUTION

Chapter 2 of this thesis is published: Khan, M.S.R., Whittington, C.M., Thompson, M.B., & Byrne, M. (2019) Intragonadal incubation of progeny in three viviparous asterinid sea stars that differ in offspring provisioning, lecithotrophy vs matrotrophy. Marine Biology, 166(6), 81.

I analysed data and wrote the draft of the manuscript. MB Thompson and M Byrne assisted in sample collection. All co-authors contributed to manuscript preparation, revision and approved the final version of the manuscript.

Chapter 3 of this thesis is partially published at Zoosymposia (2019), 15, 71–82 (see Appendix).

I designed the study, collected and analysed data and wrote the draft of the manuscript. All co-authors assisted in research design, manuscript preparation and revision and approved the final version.

Chapter 4

I did microscopy, analysed images, and wrote the manuscript. M Byrne prepared the histological slides. All co-authors assisted in manuscript preparation and revision.

Chapter 5 of this thesis is partly published: Khan, M.S.R., Whittington, C.M., Thompson, M.B. & Byrne, M. (2019) Arrangement and size variation of intra-gonadal offspring in a viviparous asterinid sea star. Zoosymposia, 15, 71–82 (see Appendix). I did microscopy, analysed data and wrote the draft of the manuscript. All co-authors assisted in research design, manuscript preparation and revision of the manuscript and approved the final version for submission.

As supervisor for the candidature upon which this thesis is based, we confirm that the authorship attribution statements above are correct.

Prof. Michael B. Thompson Date: 10th February 2020

Prof. Maria Byrne Date: 10th February 2020

Dr. Camilla Whittington

Date: 10th February 2020

iii

ABSTRACT

Viviparous asterinids exhibit great diversity in reproductive and offspring provisioning strategies, which raises fascinating life-history questions. This thesis investigates the biology of parent-offspring size variation, offspring release, nutrient provisioning and morphological adaptions in three viviparous asterinid sea stars, Cryptasterina hystera, Parvulastra vivipara and P. parvivipara. These species have contrasting lecithotrophic and matrotrophic provisioning of developing offspring in the gonads. In C. hystera (lecithotrophic), the juveniles (655 µm diameter) develop from large eggs (440 µm diameter). In P. vivipara and P. parvivipara, juveniles vary greatly in diameter (500–5000 µm) and develop from small eggs (84–150 µm diameter) through sibling cannibalism (matrotrophy). In these species, larger parents had greater reproductive output and produced more, but not larger, offspring. The species with matrotrophic offspring provisioning had a higher reproductive output than the lecithotrophic species.

Parvulastra parvivipara released juveniles in 1–5 cohorts and exhibits continuous reproduction. Cryptasterina hystera retained a few large offspring in the gonad after 30 days of synchronous release. The degree of parental investment measured as matrotrophy index (the ratio of juvenile to egg dry mass) ranged from 597–55082 (P. parvivipara) and 1.7–6.2 (C. hystera), indicating a continuum in offspring provisioning. Potential specializations for viviparity and provisioning of nutrients for offspring were investigated using confocal microscopy and histology. The early larvae were closely associated with the inner gonad wall, supported by thin processes from somatic cells. The arrangement of P. parvivipara progeny in the gonads was observed three-dimensionally using micro-computed tomography. The juveniles were orally opposite to each other, presumably as a defensive strategy to protect themselves from being eaten. Confocal microscopy revealed 2–6 developmental stages in each gonad. The size variation of offspring intensifies when siblings start cannibalism post-metamorphosis.

ACKNOWLEDGEMENTS

It would have not been possible for me to write this thesis or publish my articles without the support of my talented supervisory team: Prof. Mike Thompson, Prof. Maria Byrne, and Dr. Camilla Whittington. My heartiest gratitude to Mike for choosing me as a PhD student and guiding me throughout this tough journey. Thanks, Mike, for your constructive criticism, inspiration, suggestions and numerous revisions of my writing. You were a great mentor for me. I have met very few people as smart as you are. Your financial supports were really helpful, as they allowed me to introduce myself to renowned researchers in national and international conferences. I wish you sound health.

I am proud and honoured to have Maria Byrne as one of my supervisors. Her suggestions and active support in research design, data collection and innumerable revisions to prepare my manuscripts made it possible to submit this thesis on time. Maria, someone verily said that you are the grandma in the field of echinoderm research. Your expertise, supervision, data interpretation, and article writing pattern overwhelmed me and taught me a lot. There are very few supervisors who care for their students as much as you do. I wish you sound health and a long life.

Camilla Whittington, my special thanks for your inspiration, support and guidance. You continuously pushed me towards achieving my goal. I am motivated by your enthusiasm and devotion for research. You were very kind to consider my food restrictions in every lab party. I am sure you will reach the peak of your expectations by dint of your quality, devotion and expertise.

I express my gratitude to Liz McTaggart, Senior Natural Resources Officer, Department of Environment, Water and Natural Resources, South Australia, and to Dr. William Figueira and Mr. Geoff Prestedge for their assistance in sample collection. I would also like to thank Dr. Christopher Friesen, Dr. Mathew Crowther, and Dr. James Van Dyke for their assistance in data analyses. Thanks, also, to the Ministry of Agriculture, Fisheries and Forestry, Government of South Australia for the Ministerial Exemption to collect samples.

I also acknowledge the Advance Microscopy Facility at the Bosch Institute and the Australian Centre for Microscopy and Microanalysis at the University of Sydney for allowing

v me to use their equipment, and for the assistance provided by the staff of these facilities. I want to thank Prof. Chris Murphy, Matthew Foley, Sadaf Kalam, and Nasir Uddin for their assistance with lab techniques and research ideas.

Obviously, the three and half years long journey would not be possible without continuous support, caring and feedback from my fellow lab members, especially Jacqueline Herbert, Jessica Dudley, Charles Foster, Claudia Santori, Alice Buddle, Zoe Skalkos, Melanie Laird, Monty Oldroyd, Josh Kemsley, Henrique Braz, and Oliver Griffith. I also thank Claudia Santori for taking care of the sea stars in my absence, and discussing a lot of issues. I will not forget the help that I received from Dr. Paula Cisternas, Januar Harianto, Dione Deaker and Hamish Campbell.

I would like to thank HDRAC, the University Sydney for providing me with a fully funded University Sydney International Scholarship, without which my desire to carry out a PhD would have not been possible. Also thanks for awarding me PRSS grants for participating in conferences.

I could not have felt more welcome in Sydney. Some people who did not know me before were really helpful and kindly opened their doors and let me stay in their safe and comfortable home. Of them, I want to mention Abdul Muktadir and Umme Salma. Also thanks to people that made my life abroad enjoyable.

Last but not least, I would not be here if my parents had not supported me from my childhood. They are the best parents I could have ever asked for. My beloved wife was always beside me during most of my PhD and helped me a lot mentally by encouraging and supporting me in stressful situations. Every day, she cooked tasty foods even though she was pregnant and doing her own PhD works. Thanks Mansura for choosing me as your husband and staying with me in my woes and happiness.

TABLE OF CONTENTS

DECLARATION...... I AUTHORSHIP ATTRIBUTION ...... II ABSTRACT ...... IV ACKNOWLEDGEMENTS ...... V TABLE OF CONTENTS ...... VII LIST OF FIGURES ...... X LIST OF TABLES ...... XII CHAPTER 1: GENERAL INTRODUCTION ...... 1 1.1 Reproductive strategies in animals ...... 1 1.2 Reproductive strategies in marine invertebrates ...... 4 1.3 Viviparity-driven conflict and size variation of offspring ...... 6 1.4 Life-history strategies in echinoderms ...... 9 1.5 Viviparity in echinoderms ...... 11 1.5.1 Holothuroidea ...... 13 1.5.2 Ophiuroidea ...... 14 1.5.3 Echinoidea: ...... 16 1.5.4 Crinoidea ...... 17 1.5.5 Asteroidea ...... 17 1.5 Maternal-offspring relationships in viviparous asterinids ...... 20 1.5.1 Cryptasterina hystera ...... 21 1.5.2 Parvulastra vivipara ...... 22 1.5.3 Parvulastra parvivipara ...... 22 1.6 Thesis outline ...... 22

CHAPTER 2: INTRAGONADAL INCUBATION OF PROGENY IN THREE VIVIPAROUS ASTERINID SEA STARS THAT DIFFER IN OFFSPRING PROVISIONING ...... 25 2.1 Abstract ...... 25 2.2 Introduction ...... 26 2.3 Materials and methods ...... 31 2.3.1 Data collection ...... 31 2.3.2 Statistical analyses ...... 31 2.4 Results ...... 33 2.4.1 Cryptasterina hystera – lecithotrophic viviparity ...... 33 2.4.2 Parvulastra vivipara - matrotrophic viviparity ...... 36 2.4.3 Parvulastra parvivipara - matrotrophic viviparity ...... 38 2.4.4 Allometry of brooding ...... 40 2.4.5 Comparison of reproductive investment in the viviparous asterinids ...... 41 2.5 Discussion ...... 45

vii

CHAPTER 3: TEMPORAL PATTERN OF OFFSPRING RELEASE AND DEGREE OF PARENTAL INVESTMENT IN TWO VIVIPAROUS ASTERINID SEA STARS ...... 53 3.1 Abstract ...... 53 3.2 Introduction ...... 54 3.3 Materials and methods ...... 55 3.3.1 Sample collection ...... 55 3.3.2 Offspring release and retention dynamics ...... 56 3.3.3 Matrotrophy index ...... 56 3.3.4 Statistical analyses ...... 57 3.4 Results ...... 58 3.4.1 Temporal pattern of juvenile release and retention in Parvulastra parvivipara ...... 58 3.4.2 Offspring release and retention in Cryptasterina hystera ...... 66 3.5 Discussion ...... 69

CHAPTER 4: THE DUAL FUNCTION OF THE GONADS OF VIVIPAROUS ASTERINID SEA STARS IN GAMETOGENESIS AND OFFSPRING INCUBATION ...... 73 4.1 Abstract ...... 73 4.2 Introduction ...... 73 4.3 Materials and methods ...... 75 4.3.1 Sample collection ...... 75 4.3.2 Dissection and confocal microscopy ...... 76 4.3.3 Histology ...... 76 4.4 Results ...... 76 4.4.1 Cryptasterina hystera ...... 76 4.4.2 Parvulastra parvivipara ...... 82 4.5 Discussion ...... 86

CHAPTER 5: ARRANGEMENT AND SIZE VARIATION OF INTRA-GONADAL OFFSPRING IN A VIVIPAROUS ASTERINID SEA STAR, PARVULASTRA PARVIVIPARA ...... 90 5.1 Abstract ...... 90 5.2 Introduction ...... 91 5.3 Materials and methods ...... 92 5.3.1 Sample collection ...... 92 5.3.2 Micro-computed tomography and confocal microscopy ...... 93 5.4 Results ...... 94 5.4.1 Arrangement of young in gonads ...... 94 5.5 Discussion ...... 99

CHAPTER 6: GENERAL DISCUSSION ...... 102 6.1 Offspring investment, size, fecundity and allometry ...... 102 6.2 Offspring release and retention dynamics ...... 104 6.3 The dual role of the gonads in viviparous asterinids ...... 105

viii

6.4 Arrangement of intragonadal offspring in P. parvivipara ...... 105 6.5 Future directions ...... 106 6.6 Conclusion ...... 108

REFERENCES ...... 110 APPENDIX ...... 128

LIST OF FIGURES

Figure 1.1: Schematic diagram showing the potential parent-offspring, sexual and siblings conflicts in viviparous species...... 7 Figure 1.2: Schematic diagram showing the evolutionary pathways of the development of parity mode, larval development and egg size in echinoderms...... 10 Figure 1.3: Asterinid relationships showing egg diameter, reproductive mode, development mode and larval type...... 19 Figure 1. 4: The map showing the short range distribution of three viviparous asterinids, Cryptasterina hystera, Parvulastra vivipara and P. parvivipara in Australia...... 21 Figure 2.1: Scatterplots (untransformed data) showing the relationship between the number of juveniles, total juvenile diameter with adult size...... 34 Figure 2.2: Cryptasterina hystera, linear regression of the relationship between mean juvenile weight vs parent radius ...... 36 Figure 2.3: Size variation of juveniles dissected from an individual C. hystera, P. vivipara, and P. parvivipara...... 37 Figure 2.4: Coefficients of variation (CV) in offspring size within a parent and among parents of C. hystera, P. vivipara, and P. parvivipara ...... 38 Figure 2.5: Scatterplot showing the relationship between the total juvenile weights and mean juvenile weight and the number of juveniles in three viviparous sea stars ...... 39 Figure 3.1: Parvulastra parvivipara...... 59 Figure 3.2: Juveniles released by Parvulastra parvivipara over 14 days...... 60 Figure 3.3: The size of released and retained juveniles over 14 days in 36 Parvulastra parvivipara adults ...... 61 Figure 3.4: The size frequency of the juveniles of Parvulastra parvivipara monitored over 14 days ...... 62 Figure 3.5: The diameter of released and retained progeny from 20 Parvulastra parvivipara adults over 60 days...... 63 Figure 3.6: Mean dimeter of the juveniles released at different times as a cohort by twelve Parvulastra parvivipara adults over 60 days...... 64 Figure 3.7: Released and retained offspring in Parvulastra parvivipara adults monitored over 60 days ...... 66 Figure 3.8: Cryptasterina hystera: ...... 66 Figure 3.9: The size frequency of Cryptasterina hystera juveniles at peak release ...... 67

Figure 3.10: The relationship between mean juvenile diameter and mean freeze dry weight of the juveniles...... 68 Figure 4.1: Cryptasterina hysteria pre-incubation gonads…………………………………..78 Figure 4.2: Cryptasterina hystera gonad during incubation ...... 80 Figure 4.3: Gonad histology of Cryptasterina hystera (hematoxylin-eosin) ...... 81 Figure 4.4: Parvulastra parvivipara gonads...... 83 Figure 4.5: Gonad histology of Parvulstra parvivipara ...... 85 Figure 5.1: Parvulastra parvivipara...... 95 Figure 5.2: The size of gonads, embryos and juveniles in 15 Parvulastra parvivipara adults ...... 96 Figure 5.3: Confocal microscopy of Parvulastra parvivipara gonads ...... 97 Figure 5.4: Micro-computed tomography of Parvulastra parvivipara ...... 98

LIST OF TABLES

Table 1.1: Glossary of terminology ...... 2 Table 1.2: The size of eggs and metamorphosed juveniles in viviparous echinoderms that incubate their progeny...... 11 Table 2.1 Linear regression between the offspring parameters and the parent radius in three viviparous sea star……………………………………………………………………………35 Table 2.2: Allometric exponents for the relationship between adult radius and total diameter in three viviparous sea star species ...... 41 Table 2.3: ANCOVA testing the effects of species and parent size on fecundity in three viviparous species ...... 42 Table 2.4: Pairwise Comparisons in the fecundity of three viviparous species ...... 43 Table 2.5: List of hypotheses considered for three viviparous asterinid sea stars ...... 44 Table 3.1: Matrotrophy index for the juveniles across different size classes in Cryptasterina hystera and Parvulastra parvivipara………………………………………………………...69 Table 4.1: The width of the gonad wall and coelomic sinus in non-incubating and incubating gonads at different phases of gonadal development in Cryptasterina hystera and Parvulastra parvivipara…………………………………………………………………………………...84

xii

CHAPTER 1: GENERAL INTRODUCTION

1.1 Reproductive strategies in animals

The wide range of reproductive modes and degree of parental care across the Metazoa has provided the basis for much evolutionary theory (Blackburn 1999; Mercier et al. 2016; Wu et al. 2017). Reproductive mode and embryonic nutrition are two important traits that have profound influences on the life-history of an organism and the survival and fitness of both parent and offspring (Marshall & Keough 2006; Pollux et al. 2009; Ostrovsky et al. 2015). The mode of reproduction varies from release of fertilized or unfertilized eggs (oviparity), to different forms of embryonic incubation within the parent either internally (e.g. gonad, body cavity) or in external chambers (e.g. bursae in ophiuroids) thus resulting in live-birth of offspring, a mode of reproduction called viviparity (Byrne 1996; Frick 1998; Blackburn 2000; Byrne 2005; Ostrovsky et al. 2015; Mercier et al. 2016). Although oviparity is common among animals, viviparity is also widespread (Blackburn 1999; Ostrovsky et al. 2015). Viviparity has evolved independently and repeatedly more than 160 times in animals, including in all vertebrate classes except birds, and in many invertebrates including marine species (Wourms 1981; Frick 1998; Blackburn 1999; Byrne 2005; Sewell et al. 2006; Ostrovsky et al. 2015). Incubation of offspring within the body of the parent probably evolved as a ‘safe harbour’ in providing nutrients, and protection from the external extreme environment, thus improving survival and generating fitter offspring at the expense of reduced fecundity and energetic cost incurred by the parent (Shine 1978; Wourms 1981; Wourms & Lombardi 1992; Blackburn 1999; Avise 2013). Though several other drivers of the evolution of viviparity have been proposed [e.g. cold-climate hypothesis (Shine 1983), maternal manipulation hypothesis (Webb et al. 2006), reduction of maternal investment in eggs (Guimaraes 1977), and protection from egg predation (Fitch 1970)], the safe harbour hypothesis is widely applicable to animal lineages (Shine 1978; Kalinka 2015).

In terms of embryonic nutrition, oviparous species are typically lecithotrophic, meaning that embryos do not get nourishment from the mother other than what is provided in the egg (Blackburn 1992; Blackburn 1999). In viviparous species, embryos may receive nutrients entirely from the egg (lecithotrophy) (Table 1.1), or may be provided with extra embryonic nutrients (matrotrophy) (Table 1.1), for instance by placentation or by diverse aplacental maternal-offspring nutrient transfer methods (Musick & Ellis 2005; Sewell et al. 2006;

Orlando et al. 2015; Ostrovsky et al. 2015). There may be a small amount of mother to embryo nutrient transfer in mainly lecithotrophic viviparous species, a phenomenon known as incipient matrotrophy (Blackburn 1992, 2015; Orlando et al. 2015). Thus, embryonic nutritional mode across viviparous species exhibits a continuum of nutrient provisioning along the lecithotrophy-matrotrophy spectrum.

Table 1.1: Glossary of terminology

Term Definition Asynchronous Development of several stages at the same time development Brooding Any type of parental care (providing shelter, nutrition, supplying oxygen, supporting with larval development) of progeny (egg, embryo, juvenile) either internally or externally Facultative Larvae that are able to feed but do not need to feed to complete feeding larvae metamorphosis Lecithotrophy Offspring development is fully supported by egg nutrients Matrotrophy A mode of development where the parent provides additional (extra- embryonic) nutrients to offspring through various means such as placenta, histotrophy, histophagy, oophagy, adelphophagy/embryophagy in addition to nutrients reserved in egg. Oviparity Parity mode in which females lay unfertilized or developing eggs that complete their development and hatch in the external environment. Planktotrophy Possession of a small egg with limited nutrients for offspring that development through a feeding planktonic larva. Offspring settle and metamorphosis into the juvenile Superfetation Incubation of more than one developing stages of embryo/progeny at the same time within an adult Synchronous Development mode where only one stage of progeny is present at development any one time Viviparity Incubation of progeny internally (e.g. in gonad, coelom) or external indentation of the body (e.g. bursa or brood pouch). Offspring emerge from parent as juveniles.

Matrotrophy varies depending on the manner in which the progeny receive nutrients from the parent. Developing embryos can receive nutrients from the mother through secretions (histotrophy), eating maternal tissues or secretions (histophagy/matrophagy), placental transfer (placentotrophy), or through the consumption of unfertilized eggs (oophagy/ovatrophy) or even cannibalism of siblings (embryophagy/adelphophagy) (Dulvy & Reynolds 1997; Blackburn 2015; Ostrovsky et al. 2015). Some species exhibit different modes of matrotrophy at different times during offspring incubation, e.g. oophagy at early stages and embryophagy at later stages, as in the sand tiger shark (Carcharias taurus), and ranging from histotrophy to placentotrophy across gestation in some bryozoans (Blackburn et al. 1985; Gilmore 1993; Blackburn 2015; Ostrovsky et al. 2015). While viviparous matrotrophy is widespread in many vertebrate and invertebrate taxa, oviparous matrotrophy is extremely rare in vertebrates, with a single controversial example of the monotreme in which eggs may receive nutrients just prior to oviposition (Hughes & Carrick 1978). However, oviparous matrotrophy is present in many marine invertebrate species that brood their young after oviposition and provide extra-embryonic nutrients to their young during this period (Chia 1966; McClary & Mladenov 1990; Gillespie & McClintock 2007).

Viviparity has evolved independently from ancestral species with oviparity, and matrotrophy has evolved independently from ancestral species with lecithotrophy, many times across the Metazoa (Blackburn 1992; Blackburn 1999). Modifications of reproduction in the switch from oviparity to viviparity in most animals are associated with morphological, physiological and structural modification of the reproductive tract or adaptations to support the embryo with nutrient transfer (in matrotrophic species), oxygen and waste removal (Wourms et al. 1988; Wourms & Lombardi 1992; Byrne & Cerra 1996; Ostrovsky 2013b; Griffith et al. 2015; Ostrovsky et al. 2015; Whittington et al. 2015). The adaptations related to the evolution of viviparity are diverse across animals and include internal fertilization, placenta or placenta-like attachment (placental analogue, i.e. embryophore in bryozoa), absorptive nutritive structures, reduction in the number of eggs, reduction or loss of an egg envelope, simultaneous development of sperm and ovum, secondary loss of egg size, loss of larval stages, modification of larvae, and compartmentalization of embryos (Hendler 1975; Wourms 1981; Blackburn 1999; Byrne 2006; Ostrovsky 2013b; Van Dyke et al. 2014; Blackburn & Starck 2015; Griffith et al. 2015; Ostrovsky et al. 2015). The generally accepted pathway involved in the evolution of viviparity and matrotrophy is that oviparous lecithotrophy was the ancestral condition from which internal fertilization developed; following that, fertilized 3 eggs were gradually retained for a longer period to support embryonic development within incubation site, eventually resulting in live-birth (Wourms et al. 1988; Blackburn 1992; Byrne 1996; Blackburn 1999). The embryonic nutrition at this stage is entirely lecithotrophic. Incipient matrotrophy then evolved in some taxa, followed by substantial matrotrophy with an increase in the amount of extra-embryonic support (Blackburn 1992; Ostrovsky 2013b).

1.2 Reproductive strategies in marine invertebrates

Marine invertebrates are renowned for their remarkable diversity in reproductive modes that ranges from ancestral broadcasting of eggs to internal or external incubation of progeny (Vance 1973b; Strathmann 1978, 1985; Marshall & Keough 2008). Viviparous marine invertebrates exhibit a great diversity in incubation site (gonad, coelom, body cavity, and bursae), mode of parental provisioning (lecithotrophy, matrotrophy) and larval development (planktonic, lecithotrophic and direct developing larvae) (Vance 1973b; Chia 1974; Christiansen & Fenchel 1979; Heath 1979; Walker & Lesser 1989; Byrne 2006; Kamel et al. 2010b). The maternal-offspring nutritional relationship in viviparous species is a determinant of offspring size, number, survival and species population dynamics (Wourms 1981; Blackburn et al. 1985; Blackburn 1992; Byrne & Cerra 1996). Understanding the evolution of viviparity in marine species requires investigation of morphological and developmental diversity in maternal-offspring relationships.

The most widely acceptable definition of viviparity, a reproductive mode where female retains eggs inside their reproductive tract and give birth to live offspring, is vertebrate- centric (Wourms 1981; Blackburn 1999; Blackburn & Starck 2015). Marine invertebrates that internally fertilize their eggs may retain embryos in a range of locations including in their body (e.g. body cavity, coelom, gonads) and outside the body, such as the bursa of ophiuroids and the brood pouch of echinoids (e.g Abatus sp.) (Hendler 1975; Pearse & Mcclintock 1990; Schinner & McClintock 1993; Byrne 1996; Frick 1998; Sewell et al. 2006). There has been some discussion as to which species can be considered viviparous. For example, Hendler (1979) noted 60 viviparous ophiuroid species based on the presence of offspring in the bursae and the gonads. However Komatsu et al. (1990) suggested that viviparity should apply to just the species that incubate their offspring in the gonads. She suggested that ophiuroids that incubate their young in bursae should be called brooders. Bursal incubation in ophiuroids is widely termed in the literature to be viviparity and/or brooding (Mortensen 1920; Hendler

1975; Hendler 1979; Byrne 1989). The offspring in the bursae are topologically internally positioned and may obtain extra-embryonic nutrients from the bursal wall (e.g. Amphipholis squamata) (Ferguson 1964; Walker & Lesser 1989). In viviparous ophiuroids, the eggs and embryos develop inside the bursae and are never released before they born as birth juveniles. So, the development of young is not post-paritive. Thus, in this thesis bursal incubation of young is considered to be a form of viviparity.

Embryo incubation sites in marine invertebrates are complex and diverse, and may serve several roles, such as the bursae in ophiuroids, which are also respiratory structures (Frick 1998; Ostrovsky et al. 2015). The vertebrate definition of viviparity appears too narrow to incorporate the great diversity in invertebrates (Frick 1998; Sewell et al. 2006). Frick (1998) suggested viviparity should be defined as: “the retention of eggs in the parental body followed by release of offspring to the exterior”. This includes incubation within non- reproductive structures (e.g. coelom, bursae). For example, the self-fertilizing holothuroid Synaptula hydriformis has internal fertilization and incubates young in the perivisceral coelom, and thus should be considered a viviparous species (Frick et al. 1996; Frick 1998). If the eggs or embryos are released and then taken back into the parent’s body, they do not exhibit viviparous incubation, rather it is considered brooding (Frick 1998).

The terms “viviparity” and “brooding” have been used interchangeably in the invertebrate literature, even in the case of intraovarian incubation (Hansen 1968; Byrne 1988, 1989, 1991; Sewell 1993; Sewell & Chia 1994; Sewell 1994). Ostrovsky et al. (2015) defined viviparity as the development of young pre-parition and defined brooding as the development of young post-parition. Further, he defined viviparity in a wider context as “an incubational mode, where embryonic development occurs within the reproductive system (ovary or sexual duct), body cavity (coelom, pseudocoel or haemocoel) or parental tissues or tissue-like layers (parenchyma, mesohyl, mesoglea), resulting in live birth”. However, this definition does not cover the site of incubation in some echinoderms, for example the bursa in ophiuroids and brood pouches in echinoids.

In marine invertebrates, “brooders” has been used to describe species with viviparous reproduction (Fell 1946; Hendler 1979; Turner & Dearborn 1979; Byrne 1989; Walker & Lesser 1989; Byrne 1991; Sewell & Chia 1994). Brooding has been used to indicate any type of parental care (providing shelter, nutrition, supplying oxygen, supporting with larval 5 development) of progeny (egg, embryo, juvenile) either internally or externally. The term brooding may apply to oviparous or viviparous modes of parental care. The definitions of embryonic nutrition and parity modes used in this thesis are provided in the glossary (Table 1.1).

1.3 Viviparity-driven conflict and size variation of offspring

The size variation of offspring is a fundamental life-history trait in marine invertebrates (McEdward & Janies 1993; Moran & Emlet 2001; Marshall & Keough 2006), with among- population and intraspecific (within and among adults) variation in offspring size (Byrne 1996; George 1996; Bingham et al. 2004; Marshall & Keough 2008). Theory suggests that a parent has a finite amount of resources for reproduction, and selection should favour both maternal and offspring fitness (Vance 1973b; Smith & Fretwell 1974). Thus, a parent should balance the cost of producing a few well-provisioned large offspring against producing a larger number of a small offspring (Vance 1973a; Smith & Fretwell 1974; Marshall et al. 2010). For broadcast spawning species, the causes of offspring size variation among species and populations is suggested to be due to genetic as well as non-genetic maternal effects (Allen et al. 2008). Maternal size, age, nutrition and the maternal environment can influence the variation in the size of the juveniles produced (George 1996; Mousseau & Fox 1998; Sakai & Harada 2001; Marshall & Keough 2008). The size variation of offspring within a population is greatest for brooders (>14 %) compared to planktotrophic (4 %) or lecithotrophic (6 %) species (Marshall & Keough 2008). Viviparous echinoderms show great intraspecific offspring size variation (i.e. > 40% in P. vivipara) compared to other modes of reproduction (Marshall & Keough 2008; Khan et al. 2019a). Within-parent size variation of offspring and the potential for parent-offspring conflict and sibling rivalries have been shown for species that incubate their offspring (Kamel & Williams 2017).

The close proximity between parent and offspring during incubation in viviparous species creates potential for an evolutionary conflict of interest (Crespi & Semeniuk 2004; Schrader & Travis 2009; Kamel et al. 2010a; Pollux & Reznick 2011; Kalinka 2015; Kamel & Williams 2017). Theory suggests that a parent has limited resources for reproduction, for which the offspring may compete (Vance 1973b; Smith & Fretwell 1974). A parent is equally related to all of its offspring, so a parent should have equal allocation for all of its offspring, but an individual offspring is more related to itself than the other siblings (Parker & Begon

1986). Thus offspring would benefit from securing resources at a level greater than equal distribution. Interactions between siblings and parents can influence the size variation of offspring in viviparous fishes and some marine invertebrates (Frick 1998; Schrader & Travis 2009; Kamel et al. 2010b; Ostrovsky et al. 2015; Mercier et al. 2016; Kamel & Williams 2017). Competition is especially intense in viviparous species with extra-embryonic nutrition (matrotrophy) (Schrader & Travis 2009; Schrader & Travis 2012). In some cases, offspring may secure the greatest share of resources at the expense of sibling disadvantage or mortality (i.e. embryophagy, adelphophagy) (Byrne 1996; Schrader & Travis 2012).

Figure 1.1: Schematic diagram showing the potential parent-offspring, sexual and siblings conflicts in viviparous species. Siblings compete with each other for parental investment (sibling conflict) to secure more resources than the equal allocation by their parent(s) (parent- offspring conflict). In parallel, there is male-female conflict (sexual conflict) with respect to the amount of investment in progeny in gonochoric species. There should be no sexual conflict in parents that are simultaneous hermaphrodites and have self-fertilization (e.g. Parvulastra parvivipara). Modified from Parker et al. (2002) and Kamel et al. (2010a).

Size differences among offspring may arise due to asymmetric competition due to the initial difference in competitive ability or size (Schrader & Travis 2012; Kamel & Williams 2017). The equal amount of initial parental investment per offspring may limit the possibility of asymmetric competition among siblings as seen in lecithotrophic viviparous species (Khan et al. 2019a). Since the energy required for embryonic nutrition is pre-packaged before

7 fertilization, there will be a low level of sibling competition and high parental control of nutrient provisioning (Trexler 1997; Byrne 2005; Schrader & Travis 2012; Khan et al. 2019b). Lecithotrophic viviparous species produce many small juveniles with low size variation (Khan et al. 2019a). However, there is also a possibility that some offspring may get a slightly more investment than the others and so might outcompete others, creating asymmetric competition (Kamel & Williams 2017). Kamel and Williams (2017) proposed a model that showed that offspring size may vary due to sibling competition even when there is no disparity in initial competitive abilities, and parental investment per offspring is fixed. They noted that offspring competition should increase with decreasing genetic relatedness. However, it is also evident that in viviparous asterinids where there is self-fertilization and the juveniles are genetically identical to each other, great size variation of offspring may still occur (Puritz et al. 2012; Keever et al. 2013; Khan et al. 2019a). In contrast, matrotrophic species produce a few large juveniles at the expense of fecundity and may exhibit a great size variation of offspring due to intense competition among siblings (Trexler 1997; Schrader & Travis 2009; Collin & Spangler 2012; Khan et al. 2019 a, b).

The large size of juveniles released by many viviparous echinoderms may convey a selective advantage compared with the small post-larvae of planktonic developers (Hendler 1975; Byrne & Cerra 1996). For example, viviparous ophiuroids give rise to crawl-away juveniles that range from 600–5000 µm diameter, whereas in species with planktonic development the newly settled post larvae may reach up to 900 µm diameter (Hendler 1975).Viviparous asterinids give release to juveniles that are nearly reproductively mature at birth (Byrne 1996). Viviparous parents with limited space in the incubation site face the evolutionary trade-off between offspring size and fecundity (Vance 1973a; Smith & Fretwell 1974; Strathmann & Strathmann 1982; Marshall & Keough 2008). Across a wide range of taxa, larger mothers commonly produce larger offspring, as they are able to provide resources to offspring more efficiently or they have more access to resources (Hendry et al. 2001; Sakai & Harada 2001). Overall, maternal investment per offspring typically increases with increasing maternal size (Fox & Czesak 2000; Lim et al. 2014). This is however influenced by maternal provisioning strategy. For instance, with increasing parent size, lecithotrophic viviparous asterinid sea stars do not produce larger offspring and instead produce a greater number of similarly sized small offspring (Byrne et al. 2003; Byrne 2005; Khan et al. 2019a). Interestingly, matrotrophic viviparous asterinids do not show a relationship between parent

8 size and offspring size due to the influence of offspring competition and sibling cannibalism (Byrne 1996; Khan et al. 2019a).

1.4 Life-history strategies in echinoderms

Among marine invertebrates, echinoderms are well studied as a model system to generate insights into life-history evolution due to their diversity in parity mode, larval development and mode of embryonic nutrition (Wray 1995; Hart 2002; Villinski et al. 2002; Raff & Byrne 2006). Parity mode ranges from oviparity to viviparity, and larval development modes include planktotrophy, planktonic lecithotrophy, benthic lecithotrophy and direct development of larvae within an incubation site on the parent (Strathmann 1978, 1985; Byrne 2006; Raff & Byrne 2006). The ancestral form of development in echinoderms is thought to be feeding planktotrophic larva, and this trait probably evolved at least 500 mya in oviparous species (lower Ordovician or Pre-Cambrian) (Strathmann 1978, 1985). Planktotrophic larvae depend on adequate food supply in the environment and are subject to high mortality (Vance 1973b; Strathmann 1978, 1985). In echinoderms, non-feeding larvae have evolved in all five classes from an ancestor that had planktotrophic larvae (Strathmann 1978; McEdward & Miner 2001).

Many groups of echinoderms have evolved large eggs and non-feeding larvae where development is supported by egg nutrients (planktonic lecithotrophy) (Vance 1973b; Hart 2002; Raff & Byrne 2006). Non-feeding larvae have no feeding structures and are simpler than the planktotrophic larvae (Fig. 1.2). Planktonic lecithotrophic development was likely favoured when planktonic food was rare and planktonic predation was low; the increase in egg size that would have been supported by increase in lipid content would have been at the expense of reduced fecundity (Marshall et al. 2012; Byrne & Sewell 2019). Many of the eggs of these echinoderms are lipid-rich, and this conveys buoyancy to the eggs and larvae (Byrne et al. 1999; Byrne & Sewell 2019). In some clades with lecithotrophic development, there has been a change from the ancestral planktonic reproduction to deposition of egg masses with development of offspring in the benthos form as seen in the asterinid sea stars (Fig. 1.2). Asteroids with this form of benthic development often have negatively buoyant eggs due to an increase in protein content (Strathmann 1985; McEdward & Janies 1997; Prowse et al. 2008; Byrne & Sewell 2019) and the larvae have benthic attachment organs (Byrne et al. 1999; Byrne 2006). Lecithotrophic larvae may have evolved from planktotrophic larvae

9 through an evolutionary intermediate of unstable facultative feeding larvae that had the ability to feed on plankton but depended on egg reserve for development (Wray 1996; Prowse et al. 2008; Marshall et al. 2012; Byrne & Sewell 2019).

Figure 1.2: Schematic diagram showing the evolutionary pathways of the development of parity mode, larval development and egg size in echinoderms. Adapted from (Hendler 1975; Christiansen & Fenchel 1979; Chia & Walker 1991; McEdward & Janies 1993; Wray 1996; Byrne 2006; Raff & Byrne 2006) (See Table 1.1 for definition of terms).

Viviparous species that incubate their offspring exhibit the most derived mode of reproduction in echinoderms (Hendler 1975; Byrne 1996; Gillespie & McClintock 2007). Maternal provisioning in the form of extra-embryonic nutrients, may have evolved in parallel with viviparity (Blackburn 1992; Byrne 1996). Offspring provisioning has shifted from pre- fertilization investment in a large egg (lecithotrophic viviparity) to post-fertilization allocation during incubation (matrotrophic viviparity) through a secondary reduction in egg size (Hendler 1975; Turner & Dearborn 1979; Byrne 1996). Reduction in egg size in 10 viviparous echinoderms would have been supported by the adaptation of other forms of embryonic nourishment (nurse eggs, oophagy, embryophagy) (Hendler 1975; Byrne 1996). Therefore, life histories among closely related species of echinoderms are diverse, which provides a useful comparative model for understanding the evolution of viviparity and offspring provisioning (Byrne 2006; Raff & Byrne 2006).

1.5 Viviparity in echinoderms

Echinoderms exhibit a diversity in internal incubation, larval development and nutritional adaptations. Those that incubate their offspring are listed in Table 1.2.

Table 1.2: The size of eggs and metamorphosed juveniles in viviparous echinoderms that incubate their progeny. Data are adjusted to uniform measure (i.e. µm). G, Gonad; C, Coelom; B, Bursae; SH, Simultaneous hermaphrodite; Gn, Gonochoric, PH, Protandric hermaphrodite; AS, Asynchronous development; S, Synchronous development, MM, Metamorphic; M, Matrotrophy, L, lecithotrophy; those with no data are indicated by an empty space.

Class and Incubatio Egg Newly Juvenile Increase of Mod References species n location size MM size (µm) offspring e of and sex of (µm) juvenile at size with provi incubating release respect to sioni (µm) parent egg size ng

Asteroidea

Parvulastra G, SH ~150 360–500 1000– M Byrne 1996; vivipara 5000 Khan et al.

2019a

P. parvivipara G, SH, AS ~84– 200–400 500– 600–9000 M Khan et al. 135 3000 2019a

Cryptasterina G, SH, AS 440 600 650–850 > 0.7–5.2 L, M Byrne 2005; hystera fold Khan et al. 2019a

C. pacifica G, SH, AS 450 450–650 ~900 2 fold L Komatsu et al. 1990

Xyloplax G, SH, AS ~150 560 - > 4 fold M Rowe et al. medusiformis 1988

Xyloplax G Mah, 2006 janetae

Crinoidea

Comatilia G ~150– 1000– >7 fold M Messing 1984 iridometriformis 250 1400

Holothuroidea

Leptosynapta G 200– 1000– > 5–15 fold M Sewell 1994 clarki 240 2000

Oneirophanta G 600– 7000– 12–50 fold M Hansen 1968 mutabilis affinis 900 30000

Taeniogyrus G ~3000 20–30 fold M Sewell 2006; contortus Hansen 1968

Trachythyone G mira

Synaptula C, SH, AS ~250 500– 2–40 fold M Frick 1998 hydriformis 10,000

Chiridota C ~250 < 3000 M Clark, 1910 rotifera

Staurothyone C, S 500– M Materia et al. inconspicua 6000 1910

Neoamphicyclus C, S, GC 500 500– M Materia et al. lividus 8000 1910

Ophiuroidea

Ophionotus G 200 480 8000 2200 fold M Turner & hexactis Dearborn 1979

Amphiura G Mortensen, microplax 1936

Amphiura G Mortensen, monorima 1936

Astrochlamys G Byrne, 2005 sol

Stegophiura B 140 360 Hendler 1975 sculpta

Amphipholis B, SH, 100 240 250– M Hendler 1975 squamata 1000

Amphiura B, Gn, 200 ~8000 M Hendler, 2001 carchara B, AS

Ophioderma B, Gn 3500– M wahlbergii 6000

Ophiopeza B, SH, AS 280 283 Byrne, 2008 spinosa

Ophiolepis B, SH, AS 400 480 L Byme, 1988 paucispina

Ophionereis B, PH 400 480 L olivacea

Sigsbeia B 760 800 L Byrne, 1991 conifera

Ophiurochaeta B 680 1000 L, M Byrne, 1991 sp. A

Ophiacantha B 600 1500– M vivipara 2500

1.5.1 Holothuroidea Four species of sea cucumber show intraovarian incubation of their young (Table 1.2). In Leptosynapta clarki and Taeniogyrus contortus, nutrition for developing pentactulae is provided by the resorption of unfertilized eggs that break into droplets ("yolky spheres") or the body tissues of dead larvae that are resorbed in the ovary and may provide further nutrition for the remaining offspring (Hansen 1968; Sewell 1993; Sewell & Chia 1994). In L. clarki and Oneirophanta mutabilis affinis, myoepithelial extensions of the visceral peritoneum and the enlargement of genital haemal sinus during intraovarian incubation are associated with nutrient transfer to the developing young (Hansen 1975; Sewell et al. 2006). In O. mutabilis affinis, vacuolated nutritive tissues within the gonad store nutrition for the developing embryo (Hansen 1975).

In Synaptula hydriformis, the juveniles incubated in the perivisceral coelom are 100-fold greater in dry weight compared with egg weight (Frick et al. 1996; Frick 1998). Frick (1998) experimentally demonstrated that matrotrophy occurs in echinoderms. Synaptula hydriformis juveniles exhibit tentacular feeding in the adult coelom. Experimentally introduced radiolabelled microsphere particles and 14C-palmitic acid were ingested by the juveniles (Frick 1998). The possibility that nutrients are taken up through the body wall, and trans- epidermal absorption of nutrients during incubation was also suggested (Frick 1998). In S. hydrifomis, one or more juveniles with body lengths of 15–20 mm length are sometimes retained within the parental coelom, along with numerous small embryos and larvae after the adult has released dozens of juveniles of intermediate size (3000–5000 µm length). Coelomocytes found in coelomic fluid can concentrate molecules present in the adult coelomic fluids, which may act to package nutrient molecules for the young (Frick 1998). Chridota rotifera incubate their young in the body cavity and the juveniles may reach up to 3000 µm in size (from the eggs ~250 µm diameter) (Clark 1910). Cucumariids, Staurothyone inconspicua and Neoamphicyclus lividus, incubate juveniles that are up to 8000 µm length in the coelom. In S. inconspicua, some juveniles remain free in the adult coelom but some may remain firmly attached to the tip of the gonadal tubule through a calcareous ring (Materia et al. 1991). Juveniles may receive nutrients directly from the gonad as the gonads shrink in size (Materia et al. 1991).

1.5.2 Ophiuroidea

Viviparity is reported for many ophiuroids, and most of them are hermaphrodites (Mortensen 1920). Hendler (1979) noted 57 viviparous ophiuroids in cold waters and 8 species from tropical waters. Four species incubate their offspring in the gonads (Table 1.2) (Byrne 2005). Intrabursal incubation has been documented in ~ 50 species. The bursae are sac-like invaginations at the base of each arm (Fell 1946; Hendler & Tran 2001; Gillespie & McClintock 2007). In oviparous species, mature gametes pass from the gonads through the bursae and out through the bursal slits. In viviparous ophiuroids, gametes are retained within the bursae, which serve as an incubation site, or in some case the gametes develop in sacs that extend from the bursae (Gillespie & McClintock 2007). Ophiuroid bursae are densely

14 ciliated, thin-walled body wall invaginations used for respiration, which is an exaptation for the evolution of viviparity (Hendler 1979; Gillespie & McClintock 2007).

The shift from broadcasting to viviparous reproduction in ophiuroids would have been facilitated initially by the retention of large yolky eggs (400–900 µm egg diameter) (Schoener 1972; Byrne 1991). For example, the lecithotrophic viviparous species Ophionereis olivacea, has 400 µm diameter eggs and the lecithotrophic larvae are 480 µm in length. In O. olivacea, the larvae are morphologically similar to the planktonic lecithotrophic larvae of the closely related species but with reduction in the ciliary bands and larval skeleton (Byrne 1991).

Some viviparous ophiuroids (Ophionotus hexactis, Amphiura carchara, Amphipholis squamata) have small eggs (100–200 µm) and vestigial larvae. In these species, extensive nutrient provisioning (matrotrophy) produces large young (Fell 1946; Fontaine & Chia 1968). The matrotrophic ophiuroid A. carchara retains different developmental stages of offspring, usually in different bursae (Hendler & Tran 2001). In this species, almost all late-stage juveniles are positioned with their mouths and arms pressed against the bursal wall, suggesting that juveniles acquire some of their nutrients from the parent. In A. carchara, the late-stage embryos are transparent due to exhaustion of yolk reserves, and they receive nutrients from the bursal wall through their mouths (Hendler & Tran 2001). This bursal association was suggested as physiological necessity and as evidence of matrotrophy (Hendler & Tran 2001).

Greater protection of larvae, large young size and greater dispersal capacity are suggested to be the advantages of viviparity over oviparity in echinoderms (Hendler 1975; Byrne 1996; Trumbo 1996; Gillespie & McClintock 2007). Hendler (1975) stated that the post-larvae of viviparous ophiuroids are born with a size advantage, with a larger disc diameter (600–5000 µm) than that of the planktotrophic species (200–700 µm). In the intraovarian incubator O. hexactis, the largest juvenile stage increases in the organic content from egg by 100,000- times via utilizing nurse eggs and parental fluids (Turner & Dearborn 1979). Only one embryo usually develops per ovary in O. hexactis (Turner & Dearborn 1979). In A. squamata, juveniles (1000 µm) depend on extraembryonic nutrients for development from a very small egg (100 μm diameter) (Fell 1946; Walker & Lesser 1989). The embryonic attachment organ (outgrowth of bursal wall) that appears in early embryonic stages is thought to provide embryonic nourishment, but this connection disappears soon after metamorphosis (Fell 15

1946). The developing juveniles obtain extraembryonic nutrients from the haemal sinus in the bursal wall by applying their mouth and stomach to the sinuses of the bursal wall (Walker & Lesser 1989). They also appear to absorb dissolved organic molecules to supplement their nutritional requirements (Fontaine & Chia 1968). Ophionereis olivacea juveniles also maintain a similar position within the bursae (Byrne 1991).

Small eggs of intraovarian, intracoelomic, and some intrabursal brooders appear to have insufficient yolk to account for the presumed nutritional requirements of development and growth before birth (Mortensen 1920; Fell 1946; Chia 1974; Hendler 1975). Adelphophagy and ingestion of nurse eggs appear to be frequent nutritive mechanisms for offspring in the incubation chamber (Turner & Dearborn 1979). The embryos may also absorb dissolved nutrients from parental body fluids (Turner & Dearborn 1979). The specialized bursal sinuses associated with viviparous development are not present in ovoviviparous brooders O. olivacea and O. paucispina (Byrne 1988, 1989), both of which release juveniles (480 µm) that are very similar in size to their eggs (400 µm). Thus embryonic nutrition in both of these viviparous species probably depends on egg reserves (Byrne 1991).

1.5.3 Echinoidea:

Viviparity has not been reported for echinoids. However, brooding is prevalent and has evolved independently at least fifteen times, with about 7 % of the approximately 900 known extant species exhibiting this reproductive mode (Gillespie & McClintock 2007). However, the occurrence of brooding was estimated to be about 20% among all extant echinoids and may be up to 72% for Antarctic and sub-Antarctic species (Poulin & Féral 1996). Commonly, echinoids brood in adoral or aboral pouches that are considered external structures. The exception are the holasteroids Urechinus mortenseni and Plexechinus nordenskjoldi, which brood their young within a complex internal brood pouch that is a deeply invaginated extension of the body wall suspended from the interior edges of the female's apical plates; this species gives birth of juveniles through a birth canal (David & Mooi 1990; Schinner & McClintock 1993). The brood pouches in these species maintain constant communication with the sea (David & Mooi 1990). Thus, David and Mooi (1990) described the juveniles position as external to the mother’s body surface anatomically but internal topologically, similar to bursal incubation in ophiuroids. Following Mortensen’ (1936) definition of viviparity, David and Mooi (1990) described both of these species as non-viviparous

16 brooding species. Antarctic brooding spatangoids Abatus nimrodi and A. shackletoni incubate their embryos within the depressed aboral brood chamber or marsupia and release juveniles from 2000–4300 µm and 1600–2800 µm (from the egg 1970 and 1280 µm in diameter, respectively) (Pearse & Mcclintock 1990; Schinner & McClintock 1993). Thus, in these brooding species, the embryos and juveniles may receive nutrient, oxygen and other support from the parent. The position of the embryos in the brood chamber, nutrient support and releasing of offspring as juveniles external to the chamber raises the question whether these brooding species should be classed as viviparous.

1.5.4 Crinoidea

The deep-water tropical feather star Comatilia iridometriformis is the only viviparous crinoid that has intragonadal development (Messing 1984). Oophagy of nurse eggs and embryophagy were suggested as possible sources of nutrients for the developing young (1400 µm), which develop from ~200 µm egg diameter (Messing 1984). The production of giant embryos was suggested as being due to paedomorphosis; accelerations of sexual maturation and viviparity may have evolved to produce large sexually mature juveniles in a unstable environment that had a unidirectional flow of the current that might otherwise wash away the young (Messing 1984).

1.5.5 Asteroidea Asteroids have most diverse life-history strategies seen in the Echinodermata in terms of larval forms and development, ranging from the ancestral life history mode of feeding larvae to non-feeding planktonic and benthic lecithotrophic larvae to intragonadal incubation (Fig. 1.3). Viviparity occurs in six species (five species of asterinid sea star, and one species of peripodid sea daisy), all of which incubate their young in the gonads (Table 1.2).

Asterinid viviparity is the focus of this thesis and is reported to occur in two genera, Cryptasterina and Parvulastra (Table 1.2). In asterinids, viviparity evolved independently three times: twice in Cryptasterina and once in Parvulastra (Byrne et al. 2003). Until viviparity was discovered in Parvulastra vivipara, this species was thought to be a morph of Parvulastra exigua (Byrne et al. 2003; Dartnall et al. 2003; Hart et al. 2003). Subsequently, additional viviparous species were discovered (Dartnall et al. 2003). Parvulastra vivipara and P. parvivipara have reduced intragonadal larvae that develop from very small eggs (135–

150 µm diameter) with minimal lipid stores, and produce few large juveniles (500–5000 µm) with great size variation (Byrne & Cerra 1996; Byrne et al. 2003; Khan et al. 2019a). Byrne (1996) observed the undigested parts of juveniles and ossicles in the mouth of surviving juveniles and suggested that sibling cannibalism supports substantial post-metamorphic development. Degenerating oocytes, sperm and unfertilized eggs may also be a source of nutrition for these juveniles (Byrne 1996). Juveniles in P. vivipara position their oral face opposite to each other while in the gonad, a strategy that indicates that they prey on their siblings (Byrne 1996).

Matrotrophic P. vivipara and P. parvivipara have benthic vestigial non-functional larvae (Komatsu et al. 1990; Byrne 2005). The eggs in the viviparous Parvulastra species are smaller than those of planktotrophic species (Fig. 1.3). Intragonadal development in Parvulastra is suggested to have evolved through retention of large eggs (Byrne 1996, 2006). Viviparity in Parvulastra species is estimated to have evolved 1–3 Mya from a P. exigua-like ancestor that had a large egg and non-feeding benthic larvae. This would have been followed by a secondary reduction in egg size (Fig. 1.2) associated with the evolution of sibling cannibalism (matrotrophic) post-metamorphosis to support the production of large offspring (Byrne 1996; Hart et al. 1997).

Three species of Cryptasterina are viviparous (C. hystera, C. pacifica, and Cryptasterina sp.; Table 1.2). In C. hystera and C. pacifica, hundreds of juveniles develop synchronously provisioned by nutrients from a large egg (Table 1.2) (Komatsu et al. 1990; Byrne 2005). Juvenile development is suggested to be fully supported by the nutrient reserves in the egg (Komatsu et al. 1990; Byrne 2006). In C. hystera, the presence of few large juveniles (900– 4000 µm) in the gonads after the synchronous release of most offspring indicates that these juveniles receive some form of nourishment in the gonads, perhaps by sibling cannibalism (Byrne 2005; Khan et al. 2019a, b). Other sources of nutrients include ingestion of sperm, eggs and gonadal fluid. Cryptasterina hystera and C. pacifica have the planktonic type non- feeding larvae in the gonad. The presence of positively buoyant eggs and functional brachiolaria larvae in these asterinids suggests that intragonadal development evolved from species that had planktonic larvae (Byrne 2005). Several molecular phylogenies suggest that viviparity in Cryptasterina evolved very recently (6000 ya–0.5 mya) from a C. pentagona- like oviparous ancestor that had a planktonic brachiolaria larvae (Fig. 1.3) (Byrne et al. 1999; Byrne 2006; Puritz et al. 2012; Byrne 2013). The convergent evolution of ovotestis gonad 18 and potential for self-fertilization is a common trait in the evolution of viviparity in asterinids (Byrne et al. 2003). In Cryptasterina and Parvulastra, the gonads are morphologically similar to their oviparous congeners except for the presence of sperm and oocyte within the same gonad (Byrne 1996; Byrne et al. 1999). There appear to be no morphological or anatomical constraints in the evolution of viviparity in this group (Byrne & Cerra 1996; Byrne et al. 2003).

Figure 1.3: Asterinid relationships showing egg diameter, reproductive mode, development mode and larval type. The tree branch lengths are not proportional to time (see Hart et al. 1997; Puritz et al. 2012 for molecular phylogeny). The oviparous species Patiriella regularis is considered to have the ancestral-type life history with a small egg and dispersive planktotrophic larva (bipinnaria). Species with planktonic lecithotrophic larvae evolved separately in the Meridiastra and Cryptasterina clades. Viviparity in Cryptasterina pacifica and in C. hystera evolved from an oviparous C. pentagona-like ancestor that had a large egg with a planktonic lecithotrophic brachiolaria larva (Hart et al. 1997). Viviparity in Parvulastra evolved from an oviparous P. exigua-like ancestor that had a large egg and benthic larvae (Hart et al. 1997). Intragonadal development in Parvulastra is associated with a secondary reduction in egg size and vestigial non-feeding larvae that metamorphose into minute juveniles that cannibalise siblings. Ovi, oviparous; Viv, viviparous; Pt, planktotroph; Bl, benthic lecithotroph; Pl, planktonic lecithotroph; Bip, Bipinnaria; Brach, brachiolaria (after Byrne 2006: Raff and Byrne 2006).

In Xyloplax, three species have so far been identified. Xyloplax medusiformis is a gonochoric viviparous species in which juveniles develop from a small oocyte (Table 1.2). Juvenile development relies on coelomic dissolved nutrients or nutrient resorption from the tissues of the vestigial gut (Rowe et al. 1988). Still, the mechanism of nutrient transfer to the embryo is not known for this species. Xyloplax janetae n. sp. also have intraovarian embryonic development and the embryos vary in size (Mah 2006). Xyloplax turnerae is an oviparous brooder.

1.5 Maternal-offspring relationships in viviparous asterinids

The evolution of viviparity develop a variety of maternal-offspring relationships such as nutritional, morphological, developmental, physiological and immunological relationship (Wourms 1981; Blackburn et al. 1985; Blackburn 2015). The most important relationship is the trophic or nutritional relationship along with larval development and morphological changes as they effect on the size and number of offspring at release, which influences population dynamics of a species. Intragonadal offspring produced by the lecithotrophic species are smaller than the offspring produced by the matrotrophic species (Byrne 1996, 2005). In viviparous asterinids, larval development and provisioning modes clearly influence the offspring size-number trade-off that are central to the evolutionary ecology of offspring size variation in marine invertebrates.

Many studies on the reproductive strategies on marine animals address the question of whether parents should produce many small or few large well-provisioned offspring (Vance 1973b; Smith & Fretwell 1974; Moran & Emlet 2001; Marshall et al. 2008; Marshall & Keough 2008). However, our understanding of the selection pressures that favour different offspring size in viviparous marine invertebrates remains incomplete. Asterinids provide an opportunity to compare how closely related species with diverse parental provisioning differ in the development pattern and size variation in offspring (Byrne 2006). There are very few studies on the maternal-offspring nutritional relationship and morphological adaptations in viviparous asterinids (Byrne 1996, 2005). The sources of nutrition for asterinid offspring include the egg, ingestion of unfertilized eggs (oophagy), embryo or siblings (cannibalism), and gonad fluids (Byrne 1996, 2005). The larger size of the embryo than the eggs suggests histotrophy may be present (Table 1.2). However, the mechanisms by which offspring receive nutrient within the gonads is not known.

This thesis investigates maternal-offspring relationships in viviparous asterinid sea stars with a focus on the correlation between parent size and offspring population characteristics (e.g. size and number), dynamics of offspring release, and offspring provisioning mechanisms. The three species investigated, C. hystera, P. parvivipara and P. vivipara, have a small range and are endemic to Australia (Fig. 1.4) (Prestedge 1998; Byrne 2005; Liversage & Byrne 2018).

Figure 1. 4: The map showing the short range distribution of three viviparous asterinids, Cryptasterina hystera, Parvulastra vivipara and P. parvivipara in Australia.

1.5.1 Cryptasterina hystera

Cryptasterina hystera occurs in Central Queensland from Mission beach to Yeppoon and on One Tree Island in the Great Barrier Reef, along 100 km of coast line (Dartnall et al. 2003;

Byrne & Walker 2007). The habitat of this species is mostly composed of intertidal coral rubble and cobble boulders. This species may reach up to 12 mm arm radius and is gravid from September to November. In December, the gonads mostly contain juveniles (Dartnall et al. 2003). The juveniles develop simultaneously and are released synchronously at ~ 650 µm diameter (Byrne et al. 2003). The juveniles are nearly 5 % of the parent radius. However, occasionally a few larger juveniles are seen in the gonads (Byrne 2005).

1.5.2 Parvulastra vivipara

Parvulastra vivipara is the first known viviparous asterinid and is endemic to southeast Tasmania, distributed in 7 locations along 100 km of coastline (Prestedge 1998). It has a maximum arm radius of 15 mm (average 10 mm) and can produce up to 50 juveniles (Byrne 1996; Prestedge 1998). This species can live in captivity for 8–10 years and is able to give rise to progeny in isolation (Prestedge 1998). Parvulastra vivipara was declared endangered in July 1998 under the Tasmanian Threatened Species Protection Act, 1995 (Prestedge 1998). The main period of reproduction for P. vivipara is December to February, however they can reproduce year round (Prestedge 1998).

1.5.3 Parvulastra parvivipara

Parvulastra parvivipara is the smallest known sea star, with a maximum arm radius of 5 mm. It inhabits the intertidal granite pools along 200 km of coastline of the Eyre Peninsula, South Australia (Keough & Dartnall 1978; Roediger & Bolton 2008) and reproduces year-round. In the wild, the recorded smallest size of the juveniles is 2 mm diameter (Byrne 1996). However, the size at which they emerge from the parent is not known. Parvulastra parvivipara adults are suggested to die soon after releasing the juveniles (Keough & Dartnall 1978).

1.6 Thesis outline This thesis contains four data chapters. The research addresses gaps in knowledge of the reproductive biology of viviparous sea stars, which represent the most derived life-history strategy in the Echinodermata. This thesis contributes to our understanding of the evolution of offspring size variation, offspring provisioning modes and associated life-history strategies, and the potential adaptions related to the switch to viviparity in asterinids.

Chapter 2 investigates the relationships between parent and offspring size, parent size and offspring output, and offspring size-number trade-off in three intragonadal viviparous asterinids and the different strategies associated with lecithotrophic and matrotrophic provisioning. The relationships between the number, diameter and mass of the intragonadal juveniles and the parent size were analysed to test life-history questions: 1) Do larger parents produce larger offspring or more offspring compared to smaller adults? 2) Is there a trade-of between offspring size and number? 3) Is gonad size a constraint with respect to the number of juveniles that can be incubated at once? 4) Do matrotrophic parents have a greater investment in offspring production than lecithotrophic parents?

Chapter 3 investigates the pattern of juvenile release to understand the dynamics of offspring release and retention in C. hystera and P. parvivipara with lecithotrophic and matrotrophic life-history, respectively. The matrotrophy index (embryo dry mass ÷ egg dry mass), which shows the degree of maternal provisioning in addition to the egg, was used to estimate the level of matrotrophy. Several questions specific to each species were addressed. These include: 1) What is the size range of offspring released by P. parvivipara? 2) Do P. parvivipara release all juveniles at the same time or in cohorts? 3) Do P. parvivipara parents die after releasing juveniles? 4) Is retention of some juveniles in the gonad after peak release the norm in C. hystera? 5) Is C. hystera strictly lecithotrophic? 6) What is the level of maternal provisioning in offspring for juveniles of different size classes?

Chapter 4 investigates the dual function of the gonad for gametogenesis and offspring incubation in C. hystera and P. parvivipara. Potential morphological specializations of the gonad for viviparity were investigated using confocal microscopy and histology. The focus was placed on the gonad wall, haemal layer, coelomic layer and somatic cells during gamete development and embryo incubation. The questions addressed were: 1) What gonadal changes occur during gamete development and embryo incubation? 2) Is there embryonic contact with the gonad epithelium that may play a role in nutrient transport? 3) Is there a common strategy for nutrient support of gametes and offspring in the two species?

Chapter 5 documents the arrangement of offspring in the gonads of P. parvivipara in the context of post-metamorphic sibling competition and cannibalism. The position of the

23 juveniles was observed three-dimensionally in situ using micro-computed tomography. Confocal microscopy was used to image early developing stages. The juveniles not only vary within an adult but also within a single gonad. The extent to which offspring size varies within and among gonads was observed. The potential causes of offspring size variation within a gonad was addressed. The questions addressed are: 1) How much size variation of offspring may occur within a gonad and among gonads of an individual P. parvivipara? 2) At what size do gonads start incubation? 3) What are the potential causes of offspring size variation within a gonad? 4) How many development stages are present in a single gonad? 5) How are the juveniles positioned/arranged within the gonad? 6) Do larger adults have more incubating gonads?

CHAPTER 2: INTRAGONADAL INCUBATION OF PROGENY IN THREE VIVIPAROUS ASTERINID SEA STARS THAT DIFFER IN OFFSPRING PROVISIONING1

2.1 Abstract

In marine invertebrates that care for their young, the number of offspring is often correlated with adult size. The number, size and mass of progeny relative to parent size was investigated in three asterinid sea star species that incubate their young in the gonads. Cryptasterina hystera has intragonadal planktonic-type lecithotrophic larvae with development supported by large eggs (440 µm diameter), and the juveniles are similar in size (655 µm diameter; coefficient of variation, CV = 6.89 %). By contrast, Parvulastra vivipara and P. parvivipara have small vestigial larvae and small eggs (135–150 µm diameter) with offspring development supported by sibling cannibalism (matrotrophy). The juveniles vary in size within a parent (500–5000 µm diameter, CV = 63.87 and 53.27 %, respectively). All three species show a positive relationship between parent size and the number and size of juveniles. The allometry of brooding hypothesis, that the number of progeny that can be cared for is paradoxically constrained in large adults due to space limitation, was tested. In all species, the number of progeny increased with adult size, indicating that there are no allometric constraints on offspring incubation. To compare parental investment across the two modes of provisioning, the juvenile weight of C. hystera was used as a pro rata progeny unit. The matrotrophs (Parvulastra spp.) had a higher reproductive output than similarly sized C. hystera. Of the hypotheses proposed to explain the evolution of parental care in marine invertebrates, none are broadly applicable to viviparous asterinids because of the marked differences in their reproductive strategies.

1This chapter is published and has been reformatted here. Therefore, there is a small amount of repetition between the introduction of this chapter, and Chapter 1.

Khan, M.S.R., Whittington, C.M., Thompson, M.B., & Byrne, M. (2019) Intragonadal incubation of progeny in three viviparous asterinid sea stars that differ in offspring provisioning, lecithotrophy vs matrotrophy. Marine Biology, 166(6), 81.

2.2 Introduction

Marine invertebrates exhibit many reproductive modes, larval types and mechanisms for maternal provisioning of offspring (McEdward & Janies 1993; Moran & Emlet 2001; Byrne et al. 2003; Marshall & Keough 2006; Ostrovsky 2013a; Allen & Marshall 2014). These traits are central to fundamental concepts in life history evolution and theory (Byrne et al. 2003; Sun et al. 2012). Reproductive mode varies from gamete release in broadcast spawners (oviparity) with no subsequent investment in offspring, to different forms of embryonic incubation on or in the parent’s body associated with a variety of offspring provisioning (Byrne 1991, 2006; Gillespie & McClintock 2007; Royle et al. 2012). Viviparous species incubate their offspring internally and give rise directly to crawl-away juveniles (Wourms 1981; Blackburn 1992; Mercier et al. 2016). In some viviparous marine invertebrates, the progeny are wholly supported by nutrients from the egg, a mode of provisioning called lecithotrophy (Wourms 1981; Morrison et al. 2017), whereas in other species extra- embryonic nutrition involves the transfer of nutrients to developing offspring, a mode of provisioning called matrotrophy (Wourms 1981; Frick 1998; Musick & Ellis 2005; Sewell et al. 2006; Ostrovsky et al. 2015). Parental care through incubation within the parent’s body may have evolved as a ‘safe harbour’ for offspring (Shine 1978), ensuring food supply and protection from the external environment and predation, all of which should enhance survival of individual offspring (Royle et al. 2012; Avise 2013). Diverse modes of parental care of offspring have evolved independently in marine invertebrates (McClary & Mladenov 1990; Byrne 1991; Levin & Bridges 1995; Gillespie & McClintock 2007; Ostrovsky 2013a; Ostrovsky et al. 2015), which suggests strong selection for the evolution of parental care in response to different environmental pressures among taxa.

The diversity in modes of parental care of developing offspring in marine invertebrates makes it difficult to identify the selection pressures or driving forces underlying evolution of this life history trait (Trumbo 1996; Gillespie & McClintock 2007). Some species incubate externally fertilized eggs, whereas in viviparous species fertilization is internal followed by retention of post-zygotic stages (Blackburn 1992; Gillespie & McClintock 2007; Kalinka 2015). In viviparous species, siblings may have limited space and nutrients, and competition may be intense (Kamel et al. 2010b; Kamel & Williams 2017). The offspring may also compete with the parent to gain greater control over resources (Trivers 1974; Crespi & Semeniuk 2004; Kalinka 2015; Kamel & Williams 2017). Due to this competition,

26 matrotrophs are predicted to have a great variation among offspring (Schrader & Travis 2009; Collin & Spangler 2012; Schrader & Travis 2012; Oyarzun & Brante 2014; Mercier et al. 2016) as seen in the adelphophagic (sibling cannibalistic) gastropod Crepidula onyx (Collin & Spangler 2012).

Size variation of offspring in marine invertebrates is a fundamental life-history trait (McEdward & Janies 1993; Moran & Emlet 2001; Marshall & Keough 2006; Collin & Spangler 2012; Sun et al. 2012). Variation in offspring/egg size may be due to genetic and to non-genetic maternal effects (Allen et al. 2008; Kamel & Williams 2017). Maternal size, age, nutrition and maternal environment can influence the variation in egg size (George 1996; Sakai & Harada 2001; Steer et al. 2004). For species with lecithotrophic development, egg size is used as a proxy for the size of the metamorphosed offspring (Marshall & Keough 2008). Across a wide range of taxa, including some marine invertebrates, maternal size is positively correlated with offspring size; larger mothers produce larger offspring, as they are able to support offspring more efficiently or have more access to resources (Hendry et al. 2001; Sakai & Harada 2001; Cameron et al. 2016; Rollinson & Rowe 2016). Larger mothers are also predicted to have a higher reproductive investment (Fox & Czesak 2000; Lim et al. 2014), and so produce more eggs and thereby offspring than their smaller counterparts (Trexler 1997; Marshall et al. 2010; Cameron et al. 2016). This feature is seen in some marine invertebrates including sea stars (Leptasterias aequalis), gastropods (Crepidula dilatata and Buccinum undatum) and a bryozoan (Bugula neritina) (Chaparro et al. 1999; Marshall et al. 2003; Bingham et al. 2004; Nasution et al. 2010). Thus, life history theory predicts that, in species that care for their young, larger adults should have higher fecundity, larger offspring, and higher reproductive output compared with smaller adults (Lim et al. 2014; Cameron et al. 2016; Rollinson & Rowe 2016).

Parental care incurs costs associated with providing nutrients, oxygen and space to young, resulting in a trade-off between parental reproductive investment and the number of progeny that can be produced (Smith & Fretwell 1974; Bosch & Slattery 1999; Fernandez et al. 2000; Fernandez et al. 2006; Kalinka 2015; Rollinson & Rowe 2016). As species with lecithotrophic development supply the majority of the nutrients for offspring in the egg, their eggs tend to be larger than those species with extra-embryonic provisioning (Musick & Ellis 2005; Ostrovsky 2013a; Ostrovsky et al. 2015). While production of a large egg in lecithotrophs incurs a greater initial cost, the total cost per offspring may be greater for the 27 matrotrophs, as the parent may provide more space, time and energy throughout incubation (Trexler & DeAngelis 2003). However, matrotrophy also frees the parent from the constraints to have a fixed initial investment in offspring as they can provide nutrition on demand to young (Wourms & Lombardi 1992). Thus, matrotrophy may allow production of large individual offspring as in some sharks and marine invertebrates, where the developing progeny achieve large size through eating nurse eggs (oophagy) or siblings (adelphophagy/embryophagy) within the incubation site (Wourms & Lombardi 1992; Byrne 1996; Chaparro et al. 1999; Collin & Spangler 2012).

It is not known whether closely related species of a similar size with contrasting lecithotrophic and matrotrophic provisioning have similar overall investment per progeny. Since a parent has finite resources, there is a need to balance the cost of producing a few well-provisioned large offspring or a large number of small offspring (Vance 1973a; Smith & Fretwell 1974; Marshall et al. 2010). Thus, in marine invertebrates, mothers with limited space for embryo incubation face the evolutionary trade-off between offspring size and number (Vance 1973a; Smith & Fretwell 1974; Christiansen & Fenchel 1979; Strathmann & Strathmann 1982; Marshall & Keough 2008; Ostrovsky et al. 2015). I address this trade-off in three viviparous asterinid sea stars with contrasting forms of maternal provisioning (lecithotrophy vs matrotrophy).

In many marine invertebrate taxa, parental care is associated with small adult size and hermaphroditism (e.g. octopuses, asteroids, ophiuroids, bivalves and gastropods) (Heath 1977; Hendler 1979; Strathmann & Strathmann 1982; Kabat 1985; Eernisse 1988; Byrne 1991; Sewell 1994; Berecoechea et al. 2017). The reason why care of progeny is prevalent in small-bodied species is not known, with many hypotheses proposed. The allometry of brooding hypothesis posits that, within species that care for their young, larger individuals face a spatial limitation because space (area) to accommodate young increases with the square of parent length (body size), while reproductive output (fecundity) increases with the cube of parent length and linearly with gonad volume (Strathmann & Strathmann 1982; Strathmann et al. 1984; Kabat 1985; Peters 1986; Sewell 1994). Therefore, as parent size increases, the capacity to produce eggs is greater than the spatial capacity to accommodate young (Strathmann & Strathmann 1982). There are also constraints on oxygen supply to embryos in larger adults resulting in delayed development and embryo mortality (Hess 1993; Strathmann & Strathmann 1995; Baeza & Fernandez 2002). If parental care is associated with 28 constraints in respect to parent size, then the allometry hypothesis predicts that larger adults within a species will have proportionately fewer young than smaller adults.

In many marine invertebrates, care of offspring externally or internally is a derived mode of reproduction that arose through the evolution of fertilization of eggs in or near the incubation space and loss of feeding planktonic larvae (Strathmann 1985; Byrne 1991; Wray 1995). Asteroids in the family Asterinidae have a diverse range of reproductive strategies with varying levels of parental care up to the crawl away juvenile stage. These sea stars, in the genera Cryptasterina and Parvulastra, provide an opportunity to test theories related to life- history evolution (Byrne 2006). While they are closely related species, viviparity has independently evolved in these two lineages as indicated by the relationships determined from molecular phylogeny (Fig. 1.3). Cryptasterina hystera has a large egg and lecithotrophic viviparous development. The intragonadal offspring develop through a planktonic-type lecithotrophic asteroid larva that when removed from the gonad exhibits the traits of a functional larva (e.g. exploratory behaviour, attachment and metamorphosis to the juvenile) characteristics of a non-feeding larva similar to that of its oviparous sister species C. pentagona (Hart et al. 1997; Byrne et al. 2003; Byrne 2006). These larvae can also develop to the juvenile when removed from the parent. Lecithotrophic viviparity in C. hystera appears to have evolved from a C. pentagona-like ancestor that had dispersive lecithotrophic larvae. These two species are estimated to have diverged 6000 years ago, with C. hystera on the southern fringe of the distribution of C. pentagona (Puritz et al. 2012). Parvulastra vivipara and P. parvivipara are matrotrophic viviparous species that have a small egg secondarily reduced in size (135–150 µm diameter) and a highly derived (matrotrophic) form of extraembryonic provisioning where the juveniles grow through sibling cannibalism (adelphophagy) (Byrne & Cerra 1996; Dartnall et al. 2003). Viviparity in Parvulastra species evolved from an oviparous P. exigua-like ancestor that had large eggs and benthic non- feeding larvae with an estimated divergence time of ca-3 million years (Fig. 1.3) (Byrne 2006).

The viviparous asterinids are self-fertile, simultaneous hermaphrodites with ovotestes. In C. hystera, P. vivipara and P. parvivipara, most progeny are genetically identical to the parent indicating extensive self-fertilization (Puritz et al. 2012; Keever et al. 2013). A study of P. vivipara showed that this species reproduced in isolation over 8 years (Prestedge 1998). The juveniles are released through the gonopores (Byrne 1996; Byrne & Cerra 1996). 29

Cryptasterina hystera cares for hundreds of synchronously developing juveniles of relatively similar size (~ 600–800 µm diameter) within the gonads over a 2–3 month reproductive season (Byrne et al. 2003; Dartnall et al. 2003). In contrast, P. parvivipara and P. vivipara have a small number (1–46) of large offspring (up to 30% of the parent diameter) in the gonads that show marked variation in size (0.5–5 mm diameter) with the incubation of progeny throughout the year (Byrne & Cerra 1996).

All three species have special evolutionary and conservation significance as they are all small range endemics (Prestedge 1998; Byrne et al. 2003; Roediger & Bolton 2008; Liversage & Byrne 2018). Parvulastra parvivipara with a radius of 5 mm is one of the smallest known sea stars (Keough & Dartnall 1978; Byrne & Cerra 1996; Roediger & Bolton 2008) and P. vivipara is listed as endangered (Prestedge 1998), with conservation concerns for both species (Liversage & Byrne 2018). As the parent in these self-fertile species serves the role of mother and father, the mode of care is more aptly termed ‘parentotrophy’, but for our description I use ‘matrotrophy’ as this is the term commonly used in the literature (Blackburn & Starck 2015; Ostrovsky et al. 2015). In a recent review, matrotrophy is described in a wider context as ‘continuous parental extra-vitelline nutrient supply during gestation’ (Ostrovsky et al. 2015).

I investigated the biology of three asterinid species to assess their reproductive traits with respect to their viviparous life-history. I address the predictions that: 1) larger parents have a larger number of offspring in the gonad, 2) larger parents produce larger juveniles, and 3) that there is a trade-off between offspring number and size. I also address the prediction that the matrotrophic species (P. vivipara and P. parvivipara) have greater offspring size variation than the lecithotrophic species (C. hystera). I tested the allometry hypothesis to determine whether the number of young that can be incubated in the gonad is constrained in larger adults due to space limitations. To address the predictions that similarly sized individuals of the lecithotrophic and matrotrophic species have similar investment in juvenile production, I compared the reproductive output between the less derived (lecithotrophic) mode of viviparity in C. hystera with a highly derived matrotrophic mode (sibling cannibalism) in Parvulastra species. The reproductive output in the three species as indicated by the number of progeny in the gonad was compared using the weight of individual juveniles of C. hystera as a pro rata unit of progeny size. I use these data to address the question, does a matrotrophic parent with viviparity invest as much in offspring as a parent with lecithotrophic viviparity? 30

Offspring size in Parvulastra species was used as a cannibalism index, as a larger offspring have consumed more siblings.

2.3 Materials and methods

2.3.1 Data collection

Cryptasterina hystera were collected from Statue Bay (23o15ʹS; 150o45ʹE) (n = 20) and Lammermoor (23o16ʹS; 150o77ʹE) (n = 20), Queensland, Australia in November 2007. Parvulastra vivipara adults were collected from Midway Point (42o48ʹS; 147o32ʹE) (n = 57) and Tessellated Pavement (43o31ʹS; 147o56ʹE) (n = 65), Tasmania in December 1991 and January 1992, respectively. Parvulastra parvivipara adults were collected from Thevenard (32o14ʹS; 133o64ʹE) (n = 16) and Smooth Pool (32o92ʹS; 134o08ʹE) (n = 14) on Eyre Peninsula, South Australia in December 1991 and 1994, respectively. Additional (n = 42) P. parvivipara were collected from the Smooth Pool site in December 2017. Parent arm radius (R) was measured from the centre of the mouth to the tip of one arm. The sea stars were dissected to isolate the progeny in the gonads under a dissection microscope, as very small embryos and juveniles are difficult to detect in fixed gonads. The offspring were counted and photographed using a digital camera connected to the microscope. From these images, the total area (mm2) and the total diameter (µm) of the juveniles were measured using Image J. All offspring from each parent were preserved in 70 % ethanol in separate Eppendorf tubes. The total weight of offspring was measured using a microbalance (Mettlar H35AR) after allowing the ethanol to evaporate for 5 min for 31, 37, and 42 C. hystera, P. vivipara and P. parvivipara individuals, respectively. The mean area, diameter and weight of the offspring were calculated from the respective total area, diameter, and weight respectively in each parent.

2.3.2 Statistical analyses

The relationships between adult size (R) and the number and total size (total diameter and area) of offspring were analysed by linear regression. Linear regression analysis was also used to examine the relationship between total offspring diameter and the number of juveniles in the parent. The relationship between the total and mean weight of offspring and parent radius was investigated by linear regression to determine if larger parents had larger total weight of offspring. For these analyses, the data were transformed by either log or

31 square root to achieve normality and homogeneity of variance as determined with the Shapiro-Wilk test and Levene’s test, respectively. However, untransformed data were represented graphically to show the actual distribution of data points to better convey the biology of viviparity in these species. The coefficient of variation (CV) in the diameter of offspring was calculated (CV = standard deviation/mean × 100) to estimate offspring size variation within a parent and the CVs were compared among species through one-way analysis of variance (ANOVA). Where a significant difference was indicated, the data were compared by Duncan Post Hoc Test. The CV of offspring size among parents was calculated from all the juveniles across all parents.

The data from the two populations for each species were compared using independent sample t-test. Where two populations differed the data were analysed by One-way Analysis of Covariance (ANCOVA) to determine whether population or adult radius (covariate) or the interaction between population and adult radius influenced reproductive output in the two populations. To assess the overall pattern of reproduction across as broad a size range of adults as possible, data from both populations were used for each species.

Since the total diameter of offspring best represents the space they occupy, the allometric exponents were calculated from the relationships between adult radius and total offspring diameter in each species. The relationships were analysed by Reduced Major Axis (RMA) type-II regression (GenStat 18.1) to obtain the slope and 95% confidence interval (McArdle

1988). The data were transformed using natural log (loge). The slope and 95% confidence intervals for RMA were compared with the slope predicted by isometry (b = 1) using Student’s t-test following the formula described following Clarke (1980) and McArdle (1988). A slope < 1 indicates scaling constraints on the number of offspring that can be accommodated. The relationship between the number of offspring and adult size was also investigated using untransformed data through non-linear (power) regression following Kabat (1985). To address the prediction that larger P. vivipara and P. parvivipara have larger individual juveniles in the gonads, I determined the number of large juveniles (≥ 2000 µm for P. vivipara and ≥ 700 µm for P. parvivipara) across the size range of adults as an index of cannibalism (CI). This index was calculated using the formula: number of larger juveniles adult-1 (species specific size)/total number of juveniles adult-1) × 100. The relationship between CI and adult size was analysed using linear regression.

The individual juvenile weight of C. hystera was used as a pro rata unit of progeny size to compare the reproductive investment between matrotrophic and lecithotrophic provisioning. The putative equivalent clutch size was computed by dividing the total mass of offspring in each P. vivipara and P. parvivipara by the mean individual weight of a C. hystera juvenile (20.60 µg). Thus, the total weight of offspring in the matrotrophic sea stars was converted to a putative fecundity (with respect to C. hystera juvenile weight) to produce an estimate of the number of equivalent offspring in the individual P. vivipara and P. parvivipara. Parent radius was used as a covariate. Prior to analyses, the data (fecundity) were log transformed to meet the assumptions of normality and homogeneity of variance that were tested with Shapiro- Wilk test and Levene’s test, respectively. The number of offspring of C. hystera was compared with the putative number of offspring of P. vivipara and P. parvivipara based on the pro rata using individual size of a C. hystera juvenile (see above) using One-way Analysis of Covariance (ANCOVA) with species as a fixed factor, parent radius as a covariate and number of offspring (log) as a dependent variable. The effect of population within species was estimated by a nested model. In a second analysis, this comparison was repeated between C. hystera and P. vivipara with similar size range (R = 5–8 mm) as a more direct comparison of investment in progeny with respect to adult size. Significant differences among species were assigned using a Bonferroni multiple pairwise comparisons test. All data except RMA were analysed using IBM SPSS statistics 24 and significance was assigned at P = 0.05%.

2.4 Results

2.4.1 Cryptasterina hystera – lecithotrophic viviparity

Cryptasterina hystera (mean R = 7.45, SE = 0.24 mm, range = 5.0–10.5 mm, n = 39) had a mean number of 366.7 offspring (SE = 56.1, range = 48–1622, n = 39) in the gonads. Although the Lammermoor population had a higher mean number of offspring (Mean = 464, SE = 98.68, n = 19) than the Statue Bay population (Mean = 274, SE = 50.83, n = 20), the difference was not significant (Independent t test, t37 = -1.734, P > 0.05). The individual with the largest number of offspring was from Lammermoor (adult R 10.5 mm) while the individual with the smallest number of offspring was from Statue Bay (adult R 5.0 mm) (Fig. 2 2.1a). There was a positive relationship (r = 0.772, F(1,37) = 125.35, P < 0.05) between adult size and number of offspring (Table 2.1). Similarly, parent radius was significantly positively

33 correlated with the total diameter of juveniles (r = 0.879, n = 39, P < 0.05) and the total area of juveniles (r = 0.882, n = 39, P < 0.05) (Fig. 2.1b, Table 2.1).

Figure 2.1: Scatterplots (untransformed data) showing the relationship between the number of juveniles, total juvenile diameter with adult size (radius) (a, b, d, e, g, h), and the relationship between the number of juveniles and the total juvenile diameter (c, f, i) in three viviparous sea stars: Cryptasterina hystera (a–c), Parvulastra vivipara (d–f), and P. parvivipara (g–i). Two populations were investigated for each species as indicated by the different symbols (see boxes in the panels a, d, g).

Table 2.1 Linear regression between the offspring parameters (number, diameter, area and weight) and the parent radius (R) in three viviparous sea stars. Juvenile’s data were either log transformed (log10) or square root (sqrt) to achieve normality (n = number of adults observed)

Log (Independent-Y) vs ANOVA dependant (R) Regression equation r2 df F Sig. n Cryptasterina hystera Number vs R Y = 0.0002x + 0.7704 0.772 (1, 37) 125.35 < 0.001 39 Total diameter vs R Y = 0.0002x + 3.5566 0.773 (1, 37) 126.23 < 0.001 39 Mean diameter vs R Y = 4E-06x + 2.7862 0.04 (1, 37) 1.528 .224 39 Total area vs R Y = 0.0002x + 6.2456 0.779 (1, 37) 130.18 < 0.001 39 Mean area vs R Y = 5E-06x + 5.4753 0.017 (1, 37) 0.646 .427 39 Total weight vs R Y = 0.0001x + 2.9504 0.372 (1, 29) 17.178 < 0.001 31 Mean weight vs R Y = -7E-05x + 1.8523 0.219 (1, 29) 8.114 .008 31 Total diameter vs Y = 0.9801x - 2.712 0.994 (1, 37) 6196.95 < 0.001 39 offspring number Parvulastra vivipara Number (sqrt) vs R Y = 0.0003x + 0.6805 0.265 (1, 116) 41.72 < 0.001 118 Tot. diam. (sqrt) vs R Y = 0.0073x + 46.297 0.09 (1, 116) 11.49 .001 118 Mean diameter vs R Y = -2E-05x + 3.1537 0.024 (1, 116) 2.831 .095 118 Total area vs R Y = 0.0002x + 2.5065 0.312 (1, 51) 23.10 < 0.001 53 Mean area vs R Y = 8E-05x + 2.7003 0.075 (1, 51) 4.12 .048 53 Total weight vs R Y = 3E-05x + 3.6356 0.021 (1, 35) 0.76 .389 37 Mean weight vs R Y = -6E-05x + 3.2018 0.055 (1, 35) 2.03 .163 37 Total diameter vs Y = 0.0066x + 0.2465 0.708 (1, 116) 290.87 < 0.001 118 offspring number Parvulastra parvivipara Number vs R Y = 1E-04x + 0.3738 0.151 (1, 26) 4.615 .041 28 Total diameter vs R Y = 1.0017x + 2.6061 0.247 (1, 26) 8.546 .007 28 Mean diameter vs R Y = 4E-05x + 2.3783 0.13 (1, 26) 3.872 .060 28 Total area vs R Y = 0.0003x + 4.974 0.542 (1,43) 49.785 < 0.001 44 Mean area vs R Y = 3E-05x + 5.8027 0.012 (1,43) 0.526 .472 44 Total weight vs R Y = 0.0004x + 1.7526 0.592 (1,41) 57.946 < 0.001 42 Mean weight vs R Y = -0.0052x + 212.99 .002 (1,41) 0.069 .759 42 Total diameter vs Y = 0.8463x - 2.0598 0.848 (1, 26) 144.71 < 0.01 28 offspring number

The mean juvenile diameter in C. hystera was 655 µm (SE = 6.90 µm, n = 39) and area was 0.32 mm2 (SE = 0.006 mm2, n = 39). Juvenile size was uniform and so mean offspring diameter was not related to adult radius (n = 31 adults) (Fig. 2.2 and 2.3a, Table 2.1). The coefficient of variation (CV) in juvenile size within a parent and among parents was 6.89 % (SE = 0.29, range = 3–11 %, n = 40) and 9.49 % (n = 800 offspring, 20 offspring per adult) (Fig. 2.3, 2.4). Larger adults did not have larger offspring in their gonads (Fig. 2.2 and 2.3a). There was a positive relationship between total juvenile diameter and number of juveniles in 2 the gonads (r = 0.994, F(1,37)= 6196.95, P < 0.001) (Fig. 2.1c, Table 2.1). The total juvenile mass was significantly positively correlated (r = 0.61, n = 31, P < 0.001) with adult radius and number of juveniles (Fig. 2.5a, Table 2.1). Therefore, larger C. hystera produce significantly higher total offspring biomass. The mean weight of an individual juvenile was 20.6 µg (SE = 1.39, range = 17.75–23.45 µg, n = 37) and was significantly negatively related to parent size (Fig. 2.2, Table 2.1).

Figure 2.2: Cryptasterina hystera, linear regression of the relationship between mean juvenile weight (Juv wt) (log) vs parent radius, and mean juvenile diameter (Juv diam) (log) vs parent radius (n = 31). Correlation coefficients (r) given with significant relationship (P < 0.05) indicated by asterisk.

2.4.2 Parvulastra vivipara - matrotrophic viviparity

Parvulastra vivipara (mean R = 7.85, SE = 0.16 mm, range = 4.0–12.0 mm, n = 118) had a mean number of 10.7 juveniles (SE = 0.62, range = 1–26, n = 118) in the gonads. The P.

vivipara at Midway Point (R, 5.5–12 mm) were larger than those from Tessellated Pavement

(R, 4–10 mm) (Independent t test, t116 = 6.033, P < 0.05) (Fig. 2.1d, 2.1e). The parent radius

(covariate) had greater influence on the number (ANCOVA, F(1,116) = 19.406, P < 0.05, 2 2 partial η = 0.145) and total diameter (ANCOVA, F(1,116) = 5.873, P < 0.05, partial η = 0.049) of offspring than source population (partial η2 = 0.000 and 0.007, respectively). The mean number of juveniles adult-1 for the Midway Point population was 13.4 (SE = 0.95, n = 56) and for the Tessellated Pavement population was 8.2 (SE = 0.69, n = 62). The mean diameter of the juveniles did not differ between the two populations (1089 ± 69 µm, n = 56 and 1189 ± 61 µm, n = 62, respectively) (Independent t test, t116 = -1.083, P > 0.05). The CV of offspring diameter within a single parent was 41.48 % (SE = 2.08 %, range = 10–82 %, n = 114 parent) and among parents was 53.27 % (n = 1369 offspring) (Fig. 2.3b, 2.4). The average total juvenile mass was 7353 µg (SE = 730 µg, n = 15) and 8031 µg (SE = 782 µg, n = 22) for the two populations respectively, which did not differ (P > 0.05). The mean juvenile weight of P. vivipara was 739 µg (SE = 100 µg, range = 535–943 µg, n = 37).

Figure 2.3: Size variation of juveniles dissected from an individual C. hystera (a), P. vivipara (b), and P. parvivipara (c). Scale bar = 2 mm.

There was a positive relationship, albeit with low r2 value (0.265), between the number of offspring and parent radius (F(1,116) = 41.726, P < 0.001) (Table 2.1). In both populations, the number of offspring increased with increasing parent radius up to 9 mm (Fig. 2.1d). Larger adults differed from this trend in having fewer offspring (Fig 2.1d), a pattern that was similar for the relationship between total offspring diameter and parent radius (r2 = 0.101) (Fig. 2.1e, Table 2.1). The mean diameter of offspring was variable among adults and there was no relationship between juvenile size (total and mean offspring size) and adult size. Total juvenile area was positively correlated with the number of juveniles (r = 0.786, n = 53, P < 0.001). Total juvenile mass was not related to adult radius (Table 2.1) and brood size (Fig. 2.5b). There was a positive relationship between total juvenile diameter and number (r2 =

0.786, F(1,116) = 290.870, P < 0.001) (Fig. 2.1f). 37

Figure 2.4: Coefficients of variation (CV) in offspring size within a parent (white bars with SE) and among parents (filled bars) of C. hystera (n = 39), P. vivipara (n = 114), and P. parvivipara (n = 30). Letters denote significant differences in CV in offspring size (Duncan Post Hoc Test)

2.4.3 Parvulastra parvivipara - matrotrophic viviparity

Parvulastra parvivipara (mean R = 6.04, SE = 0.24 mm, range = 3.0–8.0 mm, n = 28) had a mean number of 11.1 offspring (SE = 1.24, range = 1–32, n = 28) in the gonads. The Smooth

Pool and Thevenard populations did not differ in offspring number (Independent t test, t26 =

1.210, P > 0.05) and total offspring diameter (Independent t test, t26 = 0.222, P > 0.05). The mean number of offspring in these populations was 13.8 (SE = 2.11, n = 13) and 8.8 (SE = 1.18, n = 15), and the mean total offspring diameter was 4654 µm (SE = 711 µm, n = 13) and 4440 µm (SE = 658 µm, n = 15), respectively. The mean juvenile diameter for the two populations was 424 µm (SE = 25 µm, n = 28). The Thevenard population had a mean diameter of 503 µm, (SE = 34 µm, n = 15). The Smooth Pool population sample collected in 1994 had a mean offspring diameter 367 µm (SE = 21 µm, n = 13) and in 2017 was 1050 µm (SE = 41 µm, n = 42). The Smooth Pool samples collected in 1996 had many recently metamorphosed juveniles (range 200–300 µm diameter) while the sample collected in 2017 had mostly large juveniles (> 700 µm diameter). This resulted in a lower mean offspring diameter in the 1996 samples compared with those taken in 2017 (Independent t test, t53 = - 9.112, P < 0.05).

Figure 2.5: Scatterplot showing the relationship between the total juvenile weights (Tot juv wt) and mean juvenile weight (Mn juv wt) and the number of juveniles in three viviparous sea stars: C. hystera (a), P. vivipara (b) and P. parvivipara (c) using log transformed data to show wide range of values in C. hystera. For each relationship, correlation coefficient (r) given with significant relationship (P < 0.05) indicated by asterisk.

2 There was a weak positive relationship (r = 0.151, F(1, 26) = 4.615, P < 0.005) between the number of offspring and parent radius (Fig. 2.1g, Table 2.1). The relationship did not improve even after discarding outliers (adults that contained only one juvenile). Similarly, total juvenile diameter was positively related to parent size with low r2 value 0.247 (Table 2.1, Fig. 2.1h). The CV of offspring size within a parent was 41.47 % (SE = 3.19 %, range = 9–28 %, n = 30 parent) and among parents was 53.27 % (n = 442 offspring) (Fig. 2.3c and 2.4). The mean brood mass was 196 µg (SE = 14 µg, n = 42). The total juvenile mass was positively related to adult size and number of offspring (Table 2.1, Fig. 2.5c). There was a 2 positive relationship (r = 0.848, F(1, 26) =144.709, P < 0.001) between total diameter and number of offspring (Fig. 2.1i). Similar to P. vivipara, the number of juveniles decreased in the largest adult of P. parvivipara (R > 7 mm) (Fig. 2.1g). The largest number of offspring (50 juveniles) in a parent (R = 4.7 mm) was dominated by small juveniles (418–800 µm diam, n = 30) and had a lower total juvenile mass than the next largest clutch (30 juveniles) in a parent with larger juveniles (> 800 µm, n = 18) (Fig. 2.5c).

2.4.4 Allometry of brooding

2.4.4.1 Lecithotrophic incubation in C. hystera

The relationship between the total diameter of the juveniles and incubation space (adult radius) as determined by RMA analysis indicated that the allometric exponent for C. hystera was 4.29, with 95 % confidence intervals > 1 (Table 2.2). The slope is significantly greater than the isometric slope b = 1 (T = 8.057, df = 39, P < 0.001), which suggests that there is no decrease in the number of offspring that can be accommodated as adult size increases in C. hystera. The regression (power) equation of offspring number on adult radius is: number of 3.691 2 offspring = 1.418E-12 (adult radius) (r = 0.765, F(1,37) = 120.175, P < 0.001). As the exponent in this relationship is greater than 3 (i.e. b = 3.69), offspring number increased as a function of body volume or weight.

2.4.4.2 Matrotrophic incubation in P. vivipara and P. parvivipara

For P. vivipara, the allometric exponent was 4.28 with 95 % confidence interval larger than slope b = 1 (T = 6.688, df = 118, P < 0.001) (Table 2.2). In P. parvivipara, the RMA slope was 3.34, significantly > 1 (T = 2.778, df = 28, P < 0.01). The allometric exponents suggest that total juvenile diameter is not proportionately smaller in larger adults in these

matrotrophs. The exponents from power regressions in the relationship between number of offspring and parent size were 1.76 and 1.15 in P. parvivipara and P. vivipara, respectively. Thus fecundity did not increase as a function of the square or cube of adult length. Variation in the size of offspring made it difficult to estimate allometric exponents in P. parvivipara and P. vivipara.

Table 2.2: Allometric exponents for the relationship between adult radius (R) and total diameter (TD) in three viviparous sea star species (C. hystera, P. parvivipara, and P. vivipara) using reduced major axis (RMA) regression on log natural (loge) transformed data (n = number of adults observed)

95% confidence Standard Constant Slope interval Ln TD vs Ln R r2 value error of n Ln(a) (b) 2 Upper b (s β) Lower (b) (b)

C. hystera 0.767 -26.16 4.29** 0.34 3.69 4.90 39

P. vivipara 0.125 -29.23 4.28** 0.43 3.58 5.33 118

P. parvivipara 0.226 -20.77 3.34* 1.06 1.89 5.95 28

** indicates significantly different from slope, b = 1 at .001 significance level

* indicates significantly different from slope, b = 1 at .01 significance level

With respect to the question of whether larger adults have proportionately greater number of larger individual juveniles, the linear regression model could not explain the relationship between the proportion of larger juveniles or cannibalism index (CI) and parent size for Parvulastra spp. due to the very low r2 value. There was no relationship between CI and 2 parent radius (r = 0.004, F(1,22) = 0.091, P > 0.05) in P. parvivipara, but there was a negative 2 relationship in P. vivipara (r = 0.076, F(1,112) = 9.194, P < 0.05).

2.4.5 Comparison of reproductive investment in the viviparous asterinids

In estimating the putative fecundity (number of offspring) in the Parvulastra species based on progeny weight of C. hystera as the pro rata juvenile unit, the output of the matrotrophs

was greater across adult size (ANCOVA, F (2,117) = 21.966, P < 0.05, species × parent radius 41

model) (Table 2.3). A Bonferroni pairwise comparison (P = 0.05) revealed a significant difference in the fecundity among species (P. parvivipara > P. vivipara > C. hystera) (Table 2.4). Both P. vivipara and P. parvivipara had significantly higher fecundity than C. hystera (Table 2.4). Comparison of the number of offspring of similarly sized C. hystera and P. vivipara adults revealed that P. vivipara had higher fecundity than C. hystera (Table 2.4). The ANCOVA revealed that, if each species produced offspring of the same weight, P. vivipara and P. parvivipara parents would have a greater reproductive output than a C. hystera of the same size.

Table 2.3: ANCOVA testing the effects of species and parent size on fecundity in three viviparous species: C. hystera (n = 39), P. parvivipara (n = 42), and P. vivipara (n = 37). The fecundity in P. parvivipara and P. vivipara were calculated from their total juvenile’s weight dividing by C. hystera mean offspring weight

Sum of Partial Squares Mean eta Source df Square F P value squared

Corrected Model 21.013a 7 3.002 70.788 < 0.001 .818 Intercept 4.426 1 4.426 104.380 < 0.001 .487 Species 1.559 2 .780 18.386 < 0.001 .251 Parent radius 4.736 1 4.736 111.686 < 0.001 .504 Population(species) .279 2 .139 3.289 .041 .056 Species × Parent radius 1.824 2 .912 21.504 < 0.001 .281 Error 4.665 110 .042 Total 610.216 118 Corrected Total 25.678 117 a R Squared = .818 (Adjusted R Squared = .807)

Table 2.4: Pairwise Comparisons in the fecundity (number of juveniles) of three viviparous species C. hystera (n = 39), P. parvivipara (n = 42), and P. vivipara (n = 37) using ANCOVA. Dependent Variable: log (number of juveniles)

(I) species (J) species Mean Std. Sig.b 99% Confidence Difference Error Interval for Differenceb (I-J) Lower Upper Bound Bound

C. hystera P. parvivipara -.747 .130 < 0.001 -1.064 -.430

P. vivipara -.396 .078 < 0.001 -.585 -.207

P. parvivipara C. hystera .747 .130 < 0.001 .430 1.064

P. vivipara .351 .136 .034 .020 .682

P. vivipara C. hystera .396 .078 < 0.001 .207 .585

P. parvivipara -.351 .136 .034 -.682 -.020

Based on estimated marginal means b Adjustment for multiple comparisons: Bonferroni

Table 2.5: List of hypotheses considered for three viviparous asterinid sea stars

Hypothesis Cryptasterina Parvulastra P. parvivipara Sl. hystera vivipara

1 Larger parents have a larger Yes Yes Yes number of offspring

2 Larger parents produce larger No No No juveniles

3 Larger adults have greater Yes Yes Yes reproductive output

4 There is a trade-off between Some support, mean No support, No support, offspring number and size in juvenile weight offspring size is offspring size is species that incubate their young. decreased as the highly variable highly variable number of juveniles within a parent within a parent increased, but no size relationship

5 Offspring size variation (CV) is Yes, CV (6.89 %) is Yes, CV (41.48 %) Yes, CV (41.47 %) greater for matrotrophic than lower than is higher than is higher than lecithotrophic incubation. matrotrophic species lecithotrophic lecithotrophic species species

6 Allometry - the number of young No constraints, No constraints, No constraints, that can be incubated in the larger adults have larger adults have larger adults have gonad is constrained in larger more young more young more young adults due to space limitations

7 Similarly sized individuals of No, Lecithotrophic No, Lecithotrophic No, Lecithotrophic lecithotrophic and matrotrophic < Matrotrophic < Matrotrophic < Matrotrophic species have similar investment in juvenile production

8 In matrotrophs, larger adults CI not related to CI not related to should have more larger juveniles adult size adult size as indicative of cannibalism index (CI)

2.5 Discussion

The mechanism of nutrient support to offspring differs markedly between the species with lecithotrophic (C. hystera) and matrotrophic (P. parvivipara and P. vivipara) viviparity. Similarly, different strategies of maternal provisioning have evolved across a number of groups that have internal incubation of progeny including other invertebrates, fishes and reptiles (Blackburn 1992; Thompson & Speake 2006; Schrader & Travis 2009; Collin & Spangler 2012; Ostrovsky et al. 2015). In all three asterinid species, fecundity and total size of offspring (total diameter and total area) increased with parent size. Thus, larger adults had a greater maternal investment for offspring production than the smaller adults. Thus maternal size is indicative of maternal investment in reproduction, similar to several other marine invertebrate and vertebrate species that incubate their young (Bingham et al. 2004; Ilano et al. 2004; Lim et al. 2014).

In the lecithotrophic viviparous species, C. hystera, adult size did not influence individual offspring size. In this species, the nutrients devoted to the juveniles are relatively fixed in the egg itself. The large lipid-rich eggs provide sufficient energy for embryonic development to the juvenile stage (Byrne et al. 1999; Baeza & Fernandez 2002; Byrne 2005). For C. hystera, offspring provisioning remains unchanged during incubation, although cannibalism occasionally occurs (Byrne 2005). Thus, while intragonadal cannibalism is not common in this species, the evolution of this behaviour, as seen in P. vivipara and P. parvivipara, may have been a fairly simple change. Occasional cannibalism in C. hystera indicates that this behaviour might be a pre-adapted trait for the evolution of this form of matrotrophy. For C. hystera, the hypothesis that larger mothers would produce larger offspring was not supported, as is also the case for the capsule brooding neogastropod, Buccinum species (Miloslavich & Dufresne 1994; Valentinsson 2002), and the isopod, Ceratoserolis polita (Clarke 1992). For C. hystera, the reproductive strategy is to produce a greater number of offspring with investment (number of offspring) increasing with increasing parental size. Reproduction in C. hystera supports the hypothesis that larger parents should produce more but not larger offspring (Hendry et al. 2001).

For the matrotrophic viviparous species, P. vivipara and P. parvivipara, mean offspring size and weight varied greatly and there was no relationship with parent size. Thus, the hypothesis that larger adults would have larger offspring in the gonads was also not supported for these

species. In addition, larger parents did not have larger individual offspring. Viviparous Parvulastra species give rise to crawl-away juveniles (0.5–5.0 mm diameter) that start development supported by a small egg (135–150 µm diameter) with metamorphosis at 300 µm diameter followed by extra-embryonic nutrition through sibling cannibalism (Byrne 1996). This mode of matrotrophy evolved in parallel to a decrease in egg size (Byrne & Cerra 1996) as is the case for other matrotrophic vertebrates and invertebrates (Ostrovsky et al. 2015). Juveniles obtain significant nourishment from cannibalizing their siblings, an unpredictable source of food as juveniles leave the gonad at different sizes (Keough & Dartnall 1978; Byrne 1996). This results in great variation in the size of juveniles recruiting to the environment. As for C. hystera, larger individuals of the two Parvulastra species had a higher number and weight of offspring. This also supports the hypothesis that reproductive investment increases with size (Hendry et al. 2001), but does not support the hypothesis that bigger parents should produce bigger offspring (Mah 2006; Marshall et al. 2010).

In C. hystera and P. parvivipara, larger parents had higher reproductive output with respect to total offspring mass. The total mass of offspring increased significantly with increasing fecundity (number of offspring) in P. parvivipara. For these cannibalistic species, where the juveniles appear to leave at any stage and/or the parent moves them out, the relationship between parent size and the number of intragonadal offspring is not straightforward. For the matrotrophs, the pattern of prolonged reproduction, sibling cannibalism and the variation in juvenile size complicates our ability to apply parental allocation theory to model this reproductive strategy with respect to parent size.

The matrotrophic species had higher among- and within- individual CV in offspring size than the lecithotrophic species, C. hystera, similar to Crepidula species, where the CV in offspring size is greater for adelphophagic species (12.6 %) than for lecithotrophic species (7.9 %) (Collin & Spangler 2012). Our findings show that viviparous species have the greatest offspring size variation among marine invertebrates, as found for other species that care for their young internally (see Marshall and Keough 2008)

Individual variation in offspring size in P. vivipara and P. parvivipara due to intragonadal cannibalism may be an adaptation to an unpredictable external environment (Chia 1974; Marshall & Keough 2008). In a variable environment, parents may produce offspring of variable sizes, so that some of the offspring may be of optimum size (Mousseau & Fox 1998; 46

Marshall & Keough 2008). For example, in a competitive environment, the bryozoan Bugula neritina adjusts offspring size to maximize fitness by producing larger offspring with higher dispersal capacity to overcome overcrowding (Allen et al. 2008). For P. vivipara and P. parvivipara, in contrast, predation pressure in the gonad and sibling competition features of the internal environment are likely to influence the variance in the timing that the juveniles evacuate the gonad. This may be the driving mechanism rather than the mother “adjusting” the size of released offspring. In the Parvulastra species, some juveniles leave the gonad before others likely to avoid being eaten. The feedback from the external and internal (gonad) environments may influence selection for this unusual reproductive strategy with respect to offspring fitness, but the parental (internal-external interaction) factors are difficult to identify. Juveniles within a parent are also likely to be different ages since fertilization due to multiple bouts of egg maturation and self-fertilization, and this needs investigation. Parvulastra vivipara and P. parvivipara can give rise to juveniles throughout the year (Prestedge 1998; Roediger & Bolton 2008). Parvulastra parvivipara produces the highest numbers of small offspring in spring and few larger offspring in summer and autumn (Roediger & Bolton 2008; Roediger 2011). This seasonal variation in the timing of resource allocation strategy may be environmentally controlled as in other sea stars (e.g. day length, temperature) (Pearse et al. 1986), as well as being influenced by intragonadal sibling cannibalism (Byrne 1996). In summer, the juveniles appear to remain in the gonads for a longer time and grow to a larger size.

The evolution of lecithotrophic viviparity in C. hystera involves parental investment before fertilization, whereas, the evolution of matrotrophy in P. vivipara and P. parvivipara subsumes the major parental provisioning to the post-fertilization phase, as for other marine invertebrates with matrotrophic provisioning (Frick 1998; Ostrovsky et al. 2009; Ostrovsky 2013b, a; Ostrovsky et al. 2015; Mercier et al. 2016). Asymmetric sibling competition (cannibalism) and parent-offspring competition may arise due to sharing of common limited parental resources during the post-fertilization phase of development, as in poeciliid fish, sea anemones and polychaetes (Schrader & Travis 2009; Pollux & Reznick 2011; Collin & Spangler 2012; Schrader & Travis 2012; Mercier et al. 2016). Therefore, in the Parvulastra species and other invertebrates and vertebrates with matrotrophic incubation, sibling interactions may limit the parent’s ability to control offspring size or release (Schrader & Travis 2009; Pollux & Reznick 2011). For P. vivipara and P. parvivipara, it was not possible to identify potential trade-offs between offspring size and number due to the high coefficient 47

of variation in offspring size. It appears that matrotrophy in P. vivipara and P. parvivipara is an adaptation that allows for smaller egg size and greater offspring size through sibling cannibalism, similar to live-bearing fish, Poecilia latipinna (Trexler 1997) and adelphophagic molluscs, Crepidula species (Collin & Spangler 2012).

The allometric exponents relating to parent size and the total diameter of progeny indicate that there are no scaling constraints on the number of juveniles that can be accommodated in the three asterinids. Thus, allometry cannot explain why viviparity/parental care in these sea stars is restricted to small-bodied species. The allometry hypothesis was developed in a study of a brooding asterinid, Asterina phylactica, which cares for its young externally under its oral surface and for this species there is an adult size constraint on the number of young (Strathmann et al. 1984). Brooding capacity in A. phylactica depends on the planar surface area of the substratum on to which the eggs are deposited, while fecundity depends on gonad volume. Thus, the space available to accommodate offspring constrains the number of offspring that can be cared for (Byrne 1991). In contrast, Cryptasterina and Parvulastra species incubate their young in a distensible organ that expands to accommodate increasing size of the offspring mass (Byrne 1996; Byrne et al. 2003). However, in C. hystera fecundity increases more than parent weight or volume. Understanding why the capacity to accommodate young is not constrained in these species, especially in C. hystera, and why bigger bodied sea stars do not have intragonadal viviparity requires three-dimensional examination of the gonad space and how the offspring are arranged. Similarly, allometric constraint is not evident for holothuroids, asteroids, ophiuroids, and polychaetes that incubate offspring within their distensible structures (Daly 1972; Menge 1975; Byrne 1991; Hess 1993; Sewell 1994), or in a bivalve that packages its young three-dimensionally on the gill surface (Kabat 1985). Thus, the correlation between small adult size and incubation of offspring may not be applicable to taxa that have the embryos in a distendable structure.

There are several explanations as to why small adult size is associated with offspring care and why larger animals do not have this life history mode, including the hypothesis that smaller adults cannot produce enough young to compensate for planktonic mortality and unpredictable recruitment (Menge 1975; Strathmann & Strathmann 1982). The energetic hypothesis proposes that small species do not have sufficient energy to produce enough eggs to ensure recruitment through dispersal (Chia 1974). Though this theory is applicable to these small sea stars, it is not readily testable and does not explain why larger species do not care 48

for their young (Byrne 1991). The time constraint hypothesis suggests that slow development rates may be disadvantageous because stage-specific mortality will be higher for species with a longer incubation period and in larger species (Hess 1993). Although it was not possible to determine stage-specific mortality for the asterinids that incubate their young, cannibalism undoubtedly results in significant loss of progeny. With the trade-off that larger offspring may be more successful, the presence of single, very large juveniles in some gonads indicates that these individuals attained this size by preying on many siblings.

The ventilation of progeny hypothesis predicts that removal of waste and supply of oxygen to offspring arranged in a closely packed mass in the incubation space may explain the decrease in fecundity in larger adults or alternatively why parental care is rare in large bodied adults (Strathmann & Chaffee 1984; Naylor et al. 1999; Baeza & Fernandez 2002). Fewer offspring in the largest P. vivipara and P. parvivipara raises the question - are larger mothers unable to provide enough nutrients or oxygen for all of the embryos, thus constraining embryonic biomass? Alternatively, the largest juveniles dominate the brood space and continue to grow through predation. In P. vivipara there is evidence of a moderate decrease in total weight of offspring in the largest parents (R > 9 mm), and that reproductive output is maximal in intermediate sized parents. This might be taken to support the allometry hypothesis, although it was not supported by RMA analyses. The reduced weight of offspring in the largest parents might be due to space constraint to accommodate young or the fact that the remaining juveniles in the gonad have a high cannibalism potential, causing siblings to flee to avoid being eaten. Sibling cannibalism may be a mechanism to reduce potential constraints on oxygen due to expansion of the distensible gonad to facilitate gas exchange between the gonad and surrounding fluid. In the snail, Acanthina monodon, sibling cannibalism increases with decreasing oxygen concentration, which is indicative of competition mediated by oxygen demand (Lardies & Fernandez 2002). For the asterinids, oxygen has to diffuse through the parent’s body wall to supply the embryos. There does not appear to be any specific oxygen provisioning mechanisms, although fluid circulation driven by the coelomic ciliated epithelial lining is likely to be important in internal oxygen exchange.

According to the sex allocation hypothesis, simultaneous hermaphrodites that care for their young should produce fewer eggs than gonochoristic species in order to allocate some resources to male function (Heath 1977). Self-fertilizing simultaneous hermaphrodites do not need a partner. Thus, natural selection may select for production of just sufficient sperm to 49

fertilize the eggs, reserving more energy to produce eggs (Heath 1977). This hypothesis is supported by the three species investigated here, as they have a minute allocation to male function, which is sufficient for fertilizing all of the ova (Byrne 1996, 2005).

The individual offspring of P. vivipara and P. parvivipara are ~ 10–35 times heavier than those of C. hystera. Based on the pro rata C. hystera juvenile unit, P. vivipara and P. parvivipara invest more in offspring than C. hystera of the same size. It appears that the evolution of matrotrophy is associated with enhanced juvenile investment compared with the lecithotrophic strategy, as suggested for the poecilid fish Poecilia latipinna (Trexler 1997). Though there is no direct comparison of maternal investment between matrotrophic and lecithotrophic marine invertebrates, Collin and Spangler (2012) found that the adelphophagic marine gastropod Crepidula cf. onyx produces hatchlings of similar size to those of a lecithotrophic congener which develops from a small egg. In P. parvivipara, the largest juvenile (3000 µm) weighed 3510 µg, equating to 170 C. hystera juveniles. Matrotrophy reduces fecundity in favour of increased offspring size. Parvulastra vivipara and P. parvivipara have asynchronous gamete production and fertilization and incubate embryos year-round (Byrne 1996; Byrne et al. 2003), so the reproductive outputs determined here for these species are likely to be an underestimate. Thus the level of parental care influences offspring number, size, weight and fitness in viviparous sea stars, similar to other marine invertebrates (Kamel et al. 2010b).

Since a mother can allocate only limited resources to its progeny, selection should favour fitness of the mother to employ a reproductive strategy that supports maximum reproductive output (Vance 1973; Smith and Fretwell 1974). Matrotrophy is energetically more costly to the parent than lecithotrophy, as parents supply energy continuously and may retain the offspring for a longer period to produce large offspring (Trexler & DeAngelis 2003). However, larger offspring are considered to be fitter and of higher quality than the smaller offspring. These more fit offspring may serve to offset the reduced fecundity in matrotrophic species (Vance 1973a; Moran & Emlet 2001; Bingham et al. 2004; Marshall & Keough 2006; Marshall et al. 2008). It would be expected that the larger offspring of Parvulastra species would be of better quality than the smaller offspring of C. hystera in a similar environment. Therefore, matrotrophy would be expected to be more common among marine invertebrates that incubate their young, but this is not the case.

A greater number of offspring in larger adults of C. hystera supports the hypothesis that parents with higher reproductive output produce more, but not larger, offspring (Hendry et al. 2001). Both the smaller offspring of lecithotrophic sea stars and larger offspring of matrotrophic sea stars survive in the intertidal environment. The question arises as to why P. vivipara and P. parvivipara produce large offspring at the expense of fecundity, and why their offspring differ in size to those of C. hystera? The difference in offspring size and provisioning pattern may be linked to the evolutionary pathways to viviparity. The evolution of viviparity in Parvulastra was from a P. exigua-like ancestor with a benthic (attached) larva (Fig. 1.3) and a switch to a non-functional larva in parallel with a secondary reduction of egg size (Byrne 2006; Raff & Byrne 2006). Sibling cannibalism would have occurred parallel to a reduction in egg size to ensure the emergence of large juveniles (Byrne and Cerra 1996). In contrast, C. hystera evolved from an ancestor that had large eggs and a planktonic non- feeding larva. For this species the only evolutionary change was retention of internally fertilized eggs.

Viviparity in these sea stars is suggested to have been selected for due to their harsh intertidal environment and the stressful conditions for early embryos and larvae (Byrne et al. 2003), as noted for the evolution of parental care in other echinoderms (Trumbo 1996; Lawrence & Herrera 2000). In stressful environmental conditions larger offspring size is beneficial (Marshall and Keough 2006; Allen et al. 2008). The large juveniles released by P. vivipara and P. parvivipara are likely to be more robust in their intertidal habitat which is characterized by marked environmental extremes (e.g. temperature range (2.0–37.5o C) (Roediger 2011). Cryptasterina hystera inhabits tropical intertidal shores and possesses comparatively smaller juveniles in an environment with less temperature variation (15.5– 41.5oC) (Roediger 2011). The lecithotrophic strategy of C. hystera that gives rise to many small juveniles is the opposite strategy to that employed by the Parvulastra species. If the environmental hypothesis applies, one would predict that the external environment of C. hystera is more benign than that of the Parvulastra species. Selection has favoured higher numbers of small offspring in the gonad in lecithotrophic C. hystera, but large offspring size in its matrotrophic counterparts. Therefore, different selection pressures may act in the different environments where viviparity evolved (e.g. temperate: P. vivipara and P. parvivipara; tropical: C. hystera).

Several hypotheses have been proposed to explain the evolution of parental care in marine invertebrates with respect to correlation to maternal/parental size and variable offspring size (Table 2.5). However, none of these is broadly applicable because of the great variation in reproductive strategies across species. The rationale for the size variation in the released juveniles of the Parvulastra species is still enigmatic. Further research is needed on these viviparous sea stars as models for greater understanding of the evolutionary ecology related to viviparity and offspring size variation in marine invertebrates.

CHAPTER 3: TEMPORAL PATTERN OF OFFSPRING RELEASE AND DEGREE OF PARENTAL INVESTMENT IN TWO VIVIPAROUS ASTERINID SEA STARS1

3.1 Abstract

The temporal pattern of juvenile release by two species of viviparous asterinid sea stars that incubate their young in the gonads was documented. Two groups of Parvulastra parvivipara were monitored in captivity for 14 days and 60 days to log the pattern of offspring release. After this time their gonads were dissected to compare the size of released and retained offspring. Juveniles (0.4–3.0 mm diameter) were released in 1–5 cohorts. Released juveniles were larger than the retained juveniles. For Cryptasterina hystera most of the offspring were released in one large clutch of similarly sized juveniles (732 µm diameter). After this initial release the C. hystera were monitored for 30 days and the gonads were dissected. The presence of large juveniles (944 µm diameter) in these gonads indicates that they are supported by matrotrophy, potentially through sibling cannibalism. The degree of parental investment additional to the egg in both species was estimated by using a matrotrophy index (MI, the ratio of juvenile to egg dry mass). As the eggs of P. parvivipara and C. hystera cannot be isolated, the eggs of their free spawning congeners (P. exigua and C. pentagona, respectively) were used as a proxy to estimate this index. The data from these eggs were scaled to the size of the eggs of the two viviparous species to determine MI. The MI ranges from 597–55082 in P. parvivipara and 1.7–6.2 in C. hystera for juveniles across the different size classes. Similarly sized juveniles do not differ in dry mass between the two species.

1This chapter is partially published in Zoosymposia (see Appendix for the article).

3.2 Introduction

Intraspecific variation (within individual and within clutch) in offspring size is important in the evolutionary ecology of marine invertebrates as it influences performance of progeny and thereby population dynamics (Allen et al. 2008; Marshall & Keough 2008; Kamel et al. 2010b; Sun et al. 2012). For oviparous species with large eggs and lecithotrophic larvae, size variation of offspring may be due to physiological constraints that prevent mothers from producing similarly sized offspring and is suggested to be a bet-hedging strategy in unpredictable environments (Einum & Fleming 2004; Marshall & Keough 2008). Intraspecific offspring size variation also occurs in brooding and viviparous species (Frick 1998; Collin & Spangler 2012; Sun et al. 2012). The mechanisms driving intraspecific offspring size variation are complex in viviparous species that have extra-embryonic parental investment (Ostrovsky et al. 2015; Kamel & Williams 2017). For these species, conflict may arise between parents and offspring and among siblings due to their close proximity (Schrader & Travis 2009; Kalinka 2015). Intense interactions between siblings and parents influence size variation of offspring (Frick 1998; Schrader & Travis 2009; Mercier et al. 2016; Kamel & Williams 2017). For instance, most echinoderms that have the viviparous life-history exhibit great size variation of offspring within the incubation site (Table 1.2). These include asteroids (e.g. Parvulstra vivipara), holothuroids (e.g. Oneirophanta mutabills and Synaptula hydriformis) and crinoids (e.g. Comatilia iridometriformis).

Among marine invertebrates, the asterinid sea stars have a particularly high diversity of life history modes, including viviparous species that incubate their young in the gonads such as Parvulstra parvivipara and Crypterasterina hystera (Byrne 2006). Parvulastra parvivipara produces tiny eggs and exhibits sibling cannibalism (a form of matrotrophy, see Table 1.1 for definition) to provide significant nourishment for juveniles that vary in size (Chapter 2). This species releases juveniles year round and is reported to die after releasing their juveniles (Keough & Dartnall 1978). In contrast, C. hystera has seasonal reproduction and releases juveniles of similar size (~650–850 µm diameter) synchronously over a few hours (Byrne 2005). Development to the juvenile stage is supported by the energy stored in large eggs (Byrne 2005; Khan et al. 2019a). However, some juveniles remain in the gonads for longer and grow up to 4-mm diameter (Byrne 2005), suggesting that there is some form of extra- embryonic nutrition. Previous studies (Byrne & Cerra 1996; Roediger & Bolton 2008; Roediger 2011) determined the size of the incubated juveniles by measuring them after

release or by dissecting the gonads. As the juveniles in the gonads of the viviparous asterinids can vary in size within a clutch and over time, the dynamics of offspring release and retention are of interest to understand the reproductive patterns of these species.

Parental investment in individual offspring differs markedly between the lecithotrophic (C. hystera) and matrotrophic (P. parvivipara) mode of provisioning (Byrne 2006). The actual amount of provisioning beyond the egg reserve is not known. This investment can be estimated by calculating a matrotrophy index (MI), which is the ratio of offspring dry mass at birth/release to egg dry mass at fertilization. This matrotrophy index is a useful metric of parental investment that is commonly used for vertebrates (e.g. reptiles, fishes) (Reznick et al. 2002; Thompson & Speake 2006)) but has not been applied to marine invertebrates. In viviparous fishes and reptiles, the MI ranges from a little or no maternal provisioning (MI = 0.6–0.7) to moderate (MI = 0.8–2.0) and extensive (MI > 5.0) provisioning post-fertilization (Reznick et al. 2002; Thompson & Speake 2006). In this study, I used MI to compare parental investment per offspring in P. parvivipara and C. hystera, which appear to have different modes of parental investment in progeny.

Offspring release, retention and juvenile size in P. parvivipara and C. hystera were monitored in the laboratory and checked whether adult size has effect on offspring size at release and retention. Two groups of P. parvivipara were monitored for 14 days and 60 days, respectively followed by dissection of the gonads to compare the size of released and retained offspring and to know whether parents release all juveniles at the same time or in cohorts. Mortality of the parent after offspring release was determined through daily monitoring. I also investigated whether retention of some juveniles after peak release occurs in C. hystera and to what extent of matrotrophic provisioning may be involved. The dry mass of the juveniles was determined in both viviparous species to compare the degree of parental investment (MI) in individual juveniles relative to provisioning mode.

3.3 Materials and methods

3.3.1 Sample collection

Parvulastra parvivipara (n = 65) was collected from Smooth Pool (32ʹ54ʹʹS, 134ʹ04ʹʹE), Eyre Peninsula, South Australia in December 2017. Parvulastra exigua (n = 30) was collected

from Clovelly (33ʹ91ʹʹS, 151ʹ27ʹʹE), New South Wales, in April–May, 2017. Cryptasterina hystera (n = 15) was collected from One Tree Island (23ʹ 15ʹʹS, 150ʹ45ʹʹE) and C. pentagona (n = 10) was collected from Orpheus Island (18ʹ62ʹS, 146ʹ50ʹE), Queensland in October 2018. All sea stars were transported to the University of Sydney in plastic containers filled with sea water and rocks collected from the sample site and maintained in a constant room temperature (20 oC) with regular exchange of sea water every few days at 35 ppt salinity. The arm radius (R) was measured from the centre of the mouth to the tip of one arm using photographs taken by a camera connected to a dissection microscope (Olympus SXZ10).

3.3.2 Offspring release and retention dynamics

a) Parvulastra parvivipara Parvulastra parvivipara were kept in separate plastic pots filled with filtered sea water (1 µm) and a small rock from the collection site. The pots were set under photosynthetically active light tubes to support growth of algal biofilm. Forty-five P. parvivipara were observed for 14 days (group 1) and 20 were observed daily for 60 days (group 2) to count and measure the diameter of the juveniles as they were released. The juveniles were photographed using a camera connected to a dissection microscope (Olympus SZX10) to measure their diameter using Micropublisher 3.0. They were blotted with tissue paper and weighed using a microbalance (Mettler H35AR) to µg and then preserved in 70 % ethanol. Group 1 and 2 adults were dissected on days 15 and 61, respectively. The offspring in the gonad (embryos and juveniles) were counted and measured.

b) Cryptasterina hystera Cryptasterina hystera released juveniles during transport and soon after being placed in aquaria. Forty six of these juveniles were measured immediately. Nine and three C. hystera were dissected after 30 and 60 days, respectively, to determine the presence of retained juveniles, and these were counted and measured.

3.3.3 Matrotrophy index

Intragonadal viviparity in P. parvivipara evolved from a P. exigua-like oviparous ancestor and in C. hystera from a C. pentagona-like ancestor (Byrne & Cerra 1996; Hart et al. 1997; Byrne 2005, 2006). As the eggs could not be obtained from P. parvivipara and C. hystera, I

used the eggs of their free spawning congeners P. exigua and C. pentagona, scaled to the size of P. parvivipara and C. hystera eggs, as proxies to calculate the MI of the viviparous species. The eggs of P. exigua (n =30 females) and C. pentagona (n = 3 females) were obtained by injections of ovulatory hormone 1-Methyladenine (0.5 M in sea water). The eggs were photographed using a digital camera (as above) to measure egg diameter. The eggs were then rinsed gently with distilled water to remove sea water and then blotted using tissue paper to remove excess water. They were counted and freeze dried for 19 h (main drying at 40 oC, 0.12 mbar for 18 h 20 min and final drying at 60 oC, 0.011 mbar for 40 min) in a CHRIST ALPHA 2-4 LD plus freeze dryer and weighed to the nearest 1 µg (as above).

I assumed that P. exigua and P. parvivipara eggs have similar densities. To approximate the dry mass of a P. parvivipara egg, the volume of an egg for P. parvivipara (84 µm diameter, Chapter 4) and P. exigua was measured assuming the egg as a sphere (formula: volume of sphere = 4/3×π×r3, r = mean radius of the egg). Then the mass of P. exigua was scaled according to the volume of a P. parvivipara egg to determine the putative mean egg mass of P. parvivipara using the following formula: mean mass of �. �� egg × mean volume of �. �� mean volume of �. �� egg

Juvenile P. parvivipara were divided into 14 size classes (Table 3.1) and were freeze dried in a CHRIST ALPHA 2-4 LD plus freeze dryer for 19 hours to determine their dry mass. The Mean dry mass of juvenile matrotrophy index (MI) was calculated using: Putative dry mass of egg

The MI for C. hystera was calculated similarly assuming the density of C. hystera and C. pentagona egg is similar, using the volume and mass of an egg of C. pentagona and scaling the mass to the volume of C. hystera egg to determine putative mass of C. hystera egg.

3.3.4 Statistical analyses

The relationship between diameter and mean dry mass of juveniles was estimated using a non-linear power regression model. The coefficient of variation (CV) in offspring size was measured for released and retained juveniles separately using the formula: CV = Standard deviation/Mean × 100. The size of the released and retained juveniles were compared through an independent sample t-test. As the data were not normally distributed and violated the

assumption of homogeneity of variance for t-tests, values from “equal variances not assumed” are reported. The mean dry mass of juveniles was compared between the two species using ANCOVA with species as a fixed factor, juvenile diameter as a covariate and mean dry mass of juveniles (log) as the dependent variable. Similarly, the MI data were compared between two species using ANCOVA. Normality and homogeneity of variance were tested with Shapiro-Wilk and Levene’s test, respectively. As the juvenile mass (log) in two species violated the assumption of homogeneity of variance, significance was assigned at higher level (1%). The relationship between the released and retained offspring diameter and parent radius was analysed using linear regression. Analyses were done using IBM SPSS statistics 24 and significance was assigned at the 5% level except for MI.

3.4 Results

3.4.1 Temporal pattern of juvenile release and retention in Parvulastra parvivipara

Parvulastra parvivipara (adult R = 3.4 mm, range = 1.8–5.2 mm, n = 65) released juveniles that emerged through the gonopores over a few hours, folding their body as they exited (Fig. 3.1a). On dissection, the spent gonads were reddish or dark in colour (Fig. 3.1b, d). Dark reddish debris was observed in the mouth of some juveniles (n = 13) (Fig. 3.1i). Several juveniles were orally opposite to each other within the gonad and had amorphous material in their mouths (Fig. 3.1i).

a. Group 1, 14 day monitoring:

The juveniles released over 14 days had a mean diameter 1352 µm (SE = 32.52, range = 732– 2514 µm, n =125 juveniles from 36 adults; CV = 26.9 %). Among 36 adults, 25 % released juveniles on the first day in captivity (Fig. 3.2a), and most (78 %) started releasing juveniles between 2–4 days (Fig. 3.2b). The total number and diameter of juveniles that the parent released was highest on days 2–4 post-capture (Fig. 3.2b, c). The mean diameter of the juveniles released was variable and was not related to the day they were released (Fig. 3.2d). Of the 36 adults that released juveniles (Fig. 3.3, 3.4), nine released juveniles twice and two released 3–4 different cohorts over 14 days, and two adults released all of the juveniles they were brooding in a single cohort (mean = 1776 µm diameter, SD = 397, range = 1080–2514 µm). The released juveniles in the smallest adults (R = 2.2–3.2 mm) had a mean diameter of 1327 µm (SE = 52, range = 916–1853 µm, n = 25 juveniles from 12 adults). The diameter of the released juveniles was not related to adult size (r2 = 0.022, P > 0.05). 58

Figure 3.1: Parvulastra parvivipara: (a) a juvenile (arrow) emerging from the gonopore by folding its body; (b) dark spent gonads (arrow) post juvenile release; (c) gonads (arrows) and pyloric caeca post juvenile release; (d) the pyloric caeca was destroyed and the aboral body wall of a parent was ruptured (arrow) by an emerging juvenile, juveniles were present in other gonads; (e) the pyloric caeca was destroyed (arrow) by an emerging juvenile; (f) a juvenile eating maternal pyloric caeca (arrow); (g) retained progeny in an adult includes pre- metamorphic (PJ), metamorphic (MJ) and fully developed juveniles; (h) five retained juveniles are much smaller than the two released juveniles (arrows); (i) juveniles orally opposite to each other in the gonads (circle) had amorphous material (arrow) in the mouth (inset). J, Juvenile; P, Pyloric caeca; S, Spent gonad.

Figure 3.2: Juveniles released by Parvulastra parvivipara over 14 days: a) percentage of adults that released juveniles; b) total number of juveniles released by the days; c) the total diameter (± SE) of the juveniles released by the adults; d) the mean diameter (± SE) of juveniles released by adults; e) size frequency of released juveniles day 1–5; f) size frequency of released juveniles day 6–14.

Figure 3.3: The size of (a) released and (b) retained juveniles over 14 days in 36 Parvulastra parvivipara adults, and (c) the size of juveniles and embryos in the gonads in nine P. parvivipara that did not release at all.

Figure 3.4: The size frequency of the juveniles of Parvulastra parvivipara monitored over 14 days showing released (n = 125) and retained (n = 147) juveniles, and the size of progeny (n = 143) in the gonads in nine adults that did not release at all.

b. Group 2, 60 day monitoring:

Juveniles released over 60 days were 1316 µm diameter (SE = 44, range = 473–3014 µm, n = 143; CV = 40%) (Fig. 3.5). All adults released juveniles at least once, and twelve adults released juveniles in 2–5 cohorts (Fig. 3.6). Three adults released all of their juveniles (mean = 2062 µm diameter, SE = 116, range = 1146–3014 µm, n = 22) in a single cohort. In most parents that released juveniles over a long period, the earlier released juveniles were larger in diameter than the later released ones (Fig. 3.6, 3.7).

On day 60, seven P. parvivipara had juveniles (mean = 643, SE = 29, 321–895 µm, n = 35; CV = 26 %) and embryos (mean 217, SD = 29, n = 5) in the gonads. The retained juveniles were smaller than the released juveniles (t = 13.118, df = 158.219, P < 0 .001). The ontogenetic stages in the gonads ranged from a single brachiolaria to metamorphic juveniles to a fully developed juvenile.

Figure 3.5: The diameter of released (n = 143) and retained (n = 41) progeny (juveniles and embryos) from 20 Parvulastra parvivipara adults over 60 days.

Three small adults (R = 1.8, 2.2, 3.2 mm) died after releasing all of their juveniles (mean = 1062 µm diameter, SE = 96, range = 495–1972 µm, n = 21). In two of these, the juveniles (1398 and 1443 μm diameter) ruptured the body wall during their emergence and were 37 % and 18 % of the parent size (Fig. 3.1d, e). The pyloric caeca of the parents was destroyed or exhausted (Fig. 3.1d, f). One juvenile was observed eating the pyloric caeca (Fig. 3.1f).

On day 14, the size of the offspring retained in the gonads was variable (mean = 776 µm diameter, SE = 25.85, range = 112–1897 µm, n = 147 progeny from 34 adults, CV = 40.4%,

Fig. 3.3b, 4) and was not related to the parent size (r2 = 0.086, P > 0.05). The developmental stages in the gonads ranged from embryos (mean = 217 µm diameter, SD = 31, n = 6) to pre- metamorphic juveniles (mean = 217 µm diameter, SD = 31, n = 6) and fully formed juveniles (n = 136) (Fig. 3.1g, 3.3a). Among the 45 adults monitored, nine (20 %) did not release offspring but had fully developed juveniles (mean = 829 μm diameter, SE = 28, range = 277– 2190 μm, CV = 39.7%, n = 138, Fig. 3.3c, 3.4), larvae and pre- metamorphic juveniles in their gonads (Fig. 3.3b). The released juveniles were significantly larger than those that were retained (t = 11.639, df= 250.309, P < 0.001, Fig. 3.4).

Three adults died, one on day 5 and two on day 8. Two of these (R = 2.5 and 3.1 mm) had released a single juvenile and had juveniles left in the gonads (mean = 1095, range = 760– 1612 µm). The third individual (R = 2.5) had released three juveniles and had juveniles and embryos left in the gonad (146–448 µm diameter, n = 4).

Figure 3.6: Mean dimeter (± SD) of the juveniles (µm) released at different times as a cohort by twelve Parvulastra parvivipara adults over 60 days. Individual adults are identified by their size (arm radius as µm) in the figure legend. Points without an error bar indicates that a single juvenile was released.

Figure 3.7: Released and retained offspring in Parvulastra parvivipara adults monitored over 60 days: a) number of adults and respective times they released juveniles; b–e) size frequency of offspring released: day 1–5, 6–25 and 26–60 days and retained juveniles at day 60 in 12 parents that released juveniles in several cohorts.

3.4.2 Offspring release and retention in Cryptasterina hystera

Juveniles released by C. hystera (adult R = 9.0 mm, range = 7.4–11.4 mm, n = 12) were 732 µm in diameter (SE = 11, range = 624–907 µm, n = 46 juveniles, CV = 10.0%). On day 30, juveniles (n = 1–37) were present in their gonads (mean = 861 µm diameter, SE = 18, range = 647–1281, n = 87 juveniles from 9 parents, CV = 19.6 %) (Fig. 3.8, 3.9). The retained juveniles were larger than the released juveniles (t = - 6.744, df = 126.409, P < 0 .001) (Fig. 3.8d, f, 3.9) and were not related to parent size (r2 = 0.014, P > 0.05). Adults dissected after 60 days had no juveniles.

Figure 3.8: Cryptasterina hystera: (a) an adult has aboral gonads (arrows); (b) gonads incubating fully developed juveniles (arrows); (c) released juveniles are of relatively similar size from a parent; (d) gonads post-release juvenile; (e) retained juveniles within the gonads (arrows) after 30 days; (f) retained juveniles that differ in size were dissected from a parent; dark spot (arrow) shows where the aboral side of a juvenile was damaged by another juvenile. G, gonad.

Figure 3.9: The size frequency of Cryptasterina hystera juveniles at peak release (a) and the juveniles that were dissected after 30 days (b), and (c) the relationship between the diameter of the retained juveniles with parent size (n = 9).

3.3.3 Length-mass relationship and matrotrophy index for the juveniles:

Parvulastra parvivipara juveniles (510–3000 µm) were 618 µg in dry mass (SE = 151, range = 36–3305 µg, n = 26) (Table 3.1). Cryptasterina hystera juveniles (650–1200 µm) were 62 µg in dry mass (SE = 8, range = 33–125 µg, n = 12). There were strong relationships (P < 0.05) between juvenile diameter and dry mass in both species (Fig. 3.10).

The mean diameter and dry mass of P. exigua eggs was 395 µm (SE = 4.9, range = 350–440 µm, n = 30 adults) and 16.2 µg (SE = 1.1, range = 7.6–34.3 µg, n = 30 adults). Using the eggs of P. exigua as a pro rata proxy unit to scale to the egg mass of P. parvivipara, the eggs of the viviparous species (84 µm diameter) would have an estimated mean dry mass of 0.06 µg. The matrotrophy index for P. parvivipara juveniles therefore ranged from 597–55082 (Table 3.1).

The mean diameter and dry mass of eggs of C. pentagona was 398 µm (SD = 23, range = 350–445 µm, n = 3 adults) and 15.2 µg (SD = 4, range = 11.9–19.4 µg, n = 3 adults). Using the eggs of C. pentagona egg as a pro rata proxy unit to scale to the egg mass of C. hystera, the eggs of viviparous species (440 µm diameter) would have an estimated dry mass of 20.4 µg. For C. hystera juveniles, the MI values therefore ranged from 1.7–6.2 (Table 3.1).

Figure 3.10: The relationship between mean juvenile diameter and mean freeze dry weight of the juveniles in (a) Cryptasterina hystera (n =12 size classes, 97 juveniles) and (b) Parvulastra parvivipara (n =14 size classes, 65 juveniles).

Juveniles of P. parvivipara and C. hystera of similar diameter did not vary in mean dry mass

(ANCOVA, F(1,37) = 2.197, P = 0.147). The MI values for the juveniles were significantly higher in P. parvivipara than C. hystera (ANCOVA, F(1,37)= 16.657, P < 0.01).

Table 3.1: Matrotrophy index for the juveniles across different size classes in Cryptasterina hystera and Parvulastra parvivipara

Cryptasterina hystera Parvulastra parvivipara Juvenile Mean dry MI n Juvenile Mean dry MI n diameter mass of index diameter mass of index (µm) juvenile (µg) (µm) juvenile (µg) 650 33 1.7 12 510 36 598 7 660 41 2.0 15 616 45 750 8 700 44 2.2 9 800 113 1889 3 727 47 2.3 10 1000 150 2333 13 748 48 2.4 12 1100 147 2444 6 793 49 2.4 10 1200 153 2548 7 850 48 2.4 5 1300 233 3875 4 882 67 3.3 9 1500 276 4604 6 950 70 3.5 6 1700 410 6833 3 1000 75 3.7 2 2000 544 9065 13 1100 100 5.0 3 2200 570 9500 3 1200 125 6.2 4 2400 720 12000 2 2800 2135 35583 2 3000 3305 55083 2

3.5 Discussion

In P. parvivipara, the released juveniles were larger than the retained juveniles. The parents release cohorts of juveniles over time, indicating continuous reproduction as previously reported (Byrne 1996). After 60 days, most adults had a few juveniles remaining in the gonad and early juvenile stages (pre-metamorphic) were present. Adults that released all juveniles in a single cohort over a few hours survived. Even the smallest adult (R= 1.8 mm) released juveniles that were > 1000 µm diameter. Thus parents, regardless of size, can release large 69

juveniles. However, several adults retained juveniles that were 1.5–2.0 times larger than the average size of released juveniles. These very large juveniles were up to 1897 µm diameter and ~28 % of the parent diameter. This observation prompts the question, why are these large juveniles retained within the gonad? These juveniles secured their nutrient supply in the gonads and are likely to have attained this size through consuming many siblings (Chapter 2). It is not known why these large juveniles did not emerge or indeed if emergence was even possible due to their large size.

Juvenile release is probably a combined response of the parent and juvenile, as the parent softens the area adjacent to gonopore and juveniles emerge, although offspring release within a day of collection may have been influenced by disturbance (Byrne 1996). The smallest juvenile released by P. parvivipara in the laboratory was 473 µm diameter. The smallest juveniles found in nature are 2,000 µm diameter (Byrne 1996). For most P. parvivipara the process of juvenile emergence did not usually cause death or damage to the parent, although some small adults died. In a rare case, the juveniles consumed the parent’s pyloric caeca (the nutrient storage organ), perhaps indicating lack of other food sources. Adult mortality may be associated with destruction or deterioration of nutrient storage organ and thus poor condition. The presence of embryos and juveniles in the gonads after 60 days suggests that this species has continuous fertilization and embryogenesis, similar to that found for P. vivipara, which can reproduce in isolation over 8 years (Prestedge 1998).

The size variation in intragonadal progeny in P. parvivipara is due to post-metamorphic sibling cannibalism (Byrne 1996). Juveniles may vacate the parent to avoid cannibalism or competition within the gonads. This small sea star species might not have enough energy to invest in developing all the gonads simultaneously (Chia 1974); the evolution of continuous incubation and offspring provisioning may therefore facilitate utilization of limited space and energy resources to produce large juveniles, similar to other matrotrophic species (Olivera- Tlahuel et al. 2015).

Cryptasterina hystera released most of their juveniles in one large cohort, but a few large juveniles were retained in the gonads and these were ~1.5–2 times larger in size than the released ones. The viviparous congener C. pacifica also synchronously releases a large cohort of juveniles (~900 µm diameter) and may retain a few juveniles that are released later (Komatsu et al. 1990). How the retained juveniles of C. hystera achieved their large size is 70

not known, but it suggests sibling cannibalism or use of eggs, sperms and gonadal fluid as a nutrient source (Byrne et al. 2003). The larger size and higher CV for the retained juveniles compared with those released supports the hypothesis that prolonged internal care and post- zygotic provisioning increase the size variation in offspring in matrotrophic species (Cubillos et al. 2007; Schrader & Travis 2009; Mercier et al. 2016; Kamel & Williams 2017). It is not known why C. hystera retain some juveniles and show occasional cannibalism (Byrne 2005). Retention of some juveniles points to an adaptation for matrotrophy in an incubation pattern similar to P. parvivipara and P. vivipara.

Intragonadal incubation in asterinids is likely to support better survival of progeny compared with the ancestral free spawning life-history (Byrne 1996, 2005). In C. hystera, irrespective of parent size, the peak released juveniles are of relatively similar size with low CV. In contrast, the juveniles in P. parvivipara are usually much larger than C. hystera and vary in size due to variable matrotrophic provisioning (Chapter 2). A comparison of juveniles of the same size in P. parvivipara and C. hystera indicated similar investment. Larger parents of C. hystera do not produce larger offspring but produce more offspring, whereas P. parvivipara does not exhibit such a relationship (Chapter 2). Larger offspring may be selected for in P. parvivipara by the harsh temperate intertidal habitat, whereas the greater fecundity in C. hystera may be a result of their less variable tropical habitat (Chapter 2).

The MI of P. parvivipara suggests that juveniles gain substantial dry mass (MI > 500) through extra-embryonic nourishment. Though a standard value for the loss of dry mass in juveniles compared with the egg is not available for echinoderms, lecithotrophic vertebrates lose 30–40% of the egg dry mass during development due to metabolism, resulting in an MI of 0.6~0.7 (Reznick et al. 2002; Thompson & Speake 2006). For C. hystera, the released juveniles (650–850 µm diameter) gain 70–140 % in dry mass over that of the egg (MI = 1.7– 2.4), which indicates a small amount of matrotrophy. As I did not measure the ash-free dry weight of the juveniles and eggs and analysed proximal biochemical constituents, I am unable to determine the contribution of organic and inorganic matter to the total juvenile composition. However, MI from the dry weight in the retained juveniles (> 850 µm–4 mm diameter) indicates extensive matrotrophy (Byrne et al. 2003; Byrne 2005). This pattern of dual provisioning of juveniles, mostly with large eggs and low levels of other forms of extra- embryonic nourishment, indicates the evolution of incipient matrotrophy, which is similar to some live-bearing fish, squamate reptiles and bryozoa (Stewart 1989; Blackburn 1992; 71

Trexler 1997; Ostrovsky 2013b). This pattern of dual provisioning may also occur in C. pacifica, in which 900 µm juveniles develop from a large egg (450 µm diameter) (Komatsu et al. 1990), and in other viviparous echinoderms where the newly metamorphic juveniles are much larger than the eggs (Table 1.2). This dual provisioning may have evolved to permit flexibility for the parents in the distribution of stored energy before fertilization and recently acquired energy post-fertilization, similar to several fish and reptiles (Stewart 1989; Trexler 1997). Thus, in viviparous species, maternal provisioning may not be strictly lecithotrophic or matrotrophic, but is rather a continuum of degree of provisioning (Riesch et al. 2010).

Matrotrophy and size variation of offspring can be inferred in 17 echinoderm species that care for their young (Table 1.2). In the ophiuroid Ophionotus hexactis, juveniles exhibit a 2,200-fold increase in dry mass, and in the crinoid Comatilia iridometriformis, juveniles increase 7-fold in diameter compared to egg size (Table 1.2) (Turner & Dearborn 1979; Messing 1984)). In the holothuroid Leptosynapta clarki, juveniles in the gonad are 1–3 mm long whereas the eggs are 200–240 µm diameter (Sewell et al. 2006). In C. hystera, P. parvivipara and other viviparous echinoderms, post-metamorphic juveniles may uptake nutrients from gonadal fluid and/or cannibalize siblings after developing their digestive system (Hansen 1968; Byrne & Cerra 1996; Byrne 2005; Sewell et al. 2006). Newly metamorphosed juveniles in most viviparous echinoderms are much larger than their eggs (Table 1.2), and there should be a cost involved with the development of embryo from an egg, suggesting nutrient transfer to the embryo, but the mechanisms are not known. Size variation of offspring at release may be characteristic of viviparous echinoderms, but this phenomenon is poorly reported.

CHAPTER 4: THE DUAL FUNCTION OF THE GONADS OF VIVIPAROUS ASTERINID SEA STARS IN GAMETOGENESIS AND OFFSPRING INCUBATION

4.1 Abstract

In viviparous asterinids, the gonads function in gamete development and embryonic incubation. The potential specialization of the gonad for viviparity and provisioning of nutrients for the gametes and embryos were investigated using confocal microscopy and histology in two species that have contrasting patterns of embryonic provisioning. In Cryptasterina hystera, the gametogenic and offspring incubation phases are temporarily distinct following a seasonal reproductive pattern. The juveniles develop relatively synchronously from a large egg. In contrast, Parvulastra parvivipara has asynchronous gonad development with continuous year-round release of juveniles that develop from a tiny egg (84 µm diameter). The presence of an early gastrula at 86 µm diameter, which is smaller than the largest egg observed (134 µm diameter), suggests that the terminal egg size may vary. In both species, the haemal and coelomic layers of the gonad wall enlarge in association with oocyte development. The width of the gonad wall decreases during embryo incubation in C. hystera but not in P. parvivipara. In both species, the early larvae were closely associated with the inner gonad wall and were supported by thin processes from somatic cells. In C. hystera, a layer of amorphous appearing material surrounds the gastrula and may facilitate association of the embryos with the gonad wall. In P. parvivipara, the dense accumulation of gonad fluid was observed where pre-metamorphic stages are juxtaposed with the gonad wall. In both species, the gastrulae and larvae may obtain nutrients where they are apposed to the gonad wall. Somatic cells may also facilitate nutrient transfer. The embryos and larvae may be pre-adapted to absorb nutrients while in the gonad. The uptake of nutrients by the embryo warrants further investigation to understand the extent of parental provisioning in offspring and the evolution of viviparity in these sea stars.

4.2 Introduction

In intragonadal viviparous marine invertebrates, the gonads function as the site for gamete development as well as an incubation site for embryo and juvenile development (Chia & Walker 1991; Sewell et al. 2006). In many viviparous species, the gonads exhibit morphological, physiological and structural modifications to support the embryo with

nutrients (in matrotrophic species) and in gas exchange and waste removal (Komatsu et al. 1990; Blackburn 1992; Van Dyke et al. 2014; Griffith et al. 2015; Whittington et al. 2015; Swift et al. 2016). The interaction between developing offspring and the incubation site is important to understand the evolution of viviparity and particularly the adaptations related to maternal-offspring nutrient transfer. To date, the focus has largely been on vertebrates (Blackburn & Starck 2015; Ostrovsky et al. 2015).

In marine invertebrates, echinoderms exhibit a great diversity of parity modes ranging from ancestral oviparity to derived viviparity (Vance 1973a, b; Strathmann 1978, 1985). Offspring nutrient investment ranges from lecithotrophy to matrotrophy (Strathmann 1978; Byrne 1996; Ostrovsky et al. 2009; Sun et al. 2012; Ostrovsky 2013b; Ostrovsky et al. 2015). Intragonadal viviparity was reported in 14 species of echinoderm (Table 1.2) that incubate their offspring in the gonads (Byrne 2005; Ostrovsky et al. 2015). In echinoderm gonads, the genital haemal layer of the gonad wall serves as a nutrient reservoir to support gametogenesis (Chia 1968; Walker 1979, 1980; Beijnink et al. 1984b; Byrne 1988). The haemal layer enlarges with nutritive substances during gametogenesis (Walker, 1979). In viviparous species, the haemal sinus may also provide nutrients to the embryos (Byrne 1996).

Very little is known about the morphological changes of the gonad during embryo incubation and the potential nutrient transfer mechanisms in viviparous echinoderms. There is evidence that in viviparous holothuroids (Leptosynapta clarki and Oneirophanta mutabilis affinis) and the ophiuroid (Amphiopholis squamata) that the offspring incubation site (gonad/bursae) undergoes morphological changes in association with maternal-offspring nutrient transfer (Fell 1946; Sewell et al. 2006). For ophiuroids, the haemal sinus is an important source of nutrients for embryos (Fontaine & Chia 1968; Walker & Lesser 1989; Byrne 1991). The morphological changes in the gonads of viviparous asterinids associated with gametogenesis and embryonic incubation are not well understood.

Asterinid sea stars exhibit both lecithotrophic and matrotrophic viviparity. Viviparity has evolved independently and recently (3~6 million years) several times from lecithotrophic broadcasting congeners (Fig. 1.3) (Hart et al. 1997; Byrne 2006; Puritz et al. 2012). These sea stars offer an opportunity to investigate the gonadal morphological changes that occur during gametogenesis and embryonic incubation and to compare these changes between the two embryonic nourishment modes in Cryptasterina hystera and Parvulastra parvivipara. 74

Both species are simultaneous hermaphrodites. Cryptasterina hystera has an annual reproductive cycle and the juveniles (~ 650–800 µm) develop from a large egg (440 µm diameter) that provides most of the energy for post-metamorphic development (lecithotrophy, Chapter 2) (Byrne et al. 2003; Byrne 2005). In contrast, P. parvivipara has a year-round reproduction with juveniles (up to 3.0 mm) developing from a small egg (~135 µm diameter), and sibling cannibalism provides substantial nutrition post-metamorphosis (matrotrophy, Chapters 3 and 5) (Byrne 1996). The intragonadal juveniles of C. hystera and P. parvivipara are both larger than the egg (Chapter 1) and obtain extra-embryonic nutrients from the parent (Table 1.2).

The potential specialization of the gonads of C. hystera and P. parvivipara for viviparity and provisioning of nutrients for the gametes and embryo was investigated using confocal microscopy and histology. Confocal microscopy has not been used with the gonads of viviparous echinoderms previously. Confocal sections of whole mount gonads do not have the artefacts associated with processing for histology sections and offer an excellent opportunity to examine development and changes in the layers of the gonad wall to determine the potential of nutrient sources for progeny. Histology was used to assist in the interpretation of the confocal images. Emphasis was placed on the changes in the gonad wall structure, haemal layer, genital coelom and somatic cells of the non-incubating and incubating gonads. In both species, embryonic contact with the gonadal wall was investigated. Potential sources of nutrients to the intragonadal young in viviparous asterinids are considered.

4.3 Materials and methods

4.3.1 Sample collection

Cryptasterina hystera were collected from One Tree Island (23o15ʹ S, 150o45ʹ E) in September 2016 (n = 8), April 2017 (n = 10) and October 2018 (n = 8) and transported in plastic containers filled with sea water and rocks collected from sampling site. Parvulastra parvivipara were collected from the Smooth Pool (32ʹ54ʹʹS, 134ʹ04ʹʹE), Eyre Peninsula, South Australia (Ministerial Exemption from the Ministry of Agriculture, Fisheries and Forestry, Government of South Australia- ME9902902) in October 2016 (n = 10), September 2017 (n = 10) and December 2017 (n = 10).

4.3.2 Dissection and confocal microscopy At every collection, three sea stars were dissected under dissection microscope Olympus SZX10 and the gonads were photographed with a digital camera connected to the microscope. For confocal microscopy, entire C. hystera and P. parvivipara adults were fixed in 2.5% glutaraldehyde in 1 µm filtered sea water (FSW) for 24 hours, and then stored in 70% ethanol. The gonads were then dissected and dehydrated in an ethanol series (70–95 %) and whole mounts were made cleaning the tissue with benzyl alcohol and benzyl benzosate, 2:1 or 1: 2 (v/v) (Sigma-Aldrich Co.). Glutaraldehyde-conveyed auto-fluorescence was used for confocal imaging using a Leica SPE-2 or LSM 510 or ZEISS LSM 800 Confocal Microscopes. The samples were excited using a 488 nm argon laser. Images were taken from different focal planes either at 512×512 or 1024×1024 pixel array. Three-dimensional video and images were reconstructed from Z-stacks.

4.3.3 Histology For histology, dissected gonads of C. hystera and P. parvivipara were fixed and decalcified in Bouin’s fixative for 24 hours followed by 3 ×10 minute rinses in distilled water and dehydration in an ethanol series (50–100 %) for 10 minutes. This was followed by paraffin embedding and sectioning (5 µm thick). Sections were stained in hematoxylin-eosin and examined and photographed under Olympus BX10 using a camera connected to the microscope.

The width of the gonad wall and genital coelomic sinus, length of the gonads, and the diameter of oocytes and lipid droplets were measured using Leica online assistance connected to the confocal microscope and Micropublisher 3.0 connected to the light microscope. Images were also analysed with software Image J.

4.4 Results

4.4.1 Cryptasterina hystera

Anatomically, the gonad is organized in lobes, with 9–15 lobes per gonad (Fig. 4.1a, b). The gonads are formed by outer and inner tissue layers separated by the genital coelom (Fig. 4.1c, e) (See Byrne 2005). For C. hystera, there are two distinct gonad phases: gamete development and offspring incubation. During early gametogenic phases (April), the ovaries

contained eggs at different stages of development including pre- (~50–80 µm) and mid- vitellogenic (232–255 µm) oocytes (4.1c, d). The two-sac structure of the gonad wall is not clearly recognizable. The gonad wall is ~ 5 µm (± 2 µm, n = 4) in width (Fig. 4.1c, d; Table 4.1). Yolk granules in mid-vitellogenic oocytes are 2.9–4.2 µm diameter (n = 4). An accumulation of large lipid droplets (5.6–6.6 µm dimeter, n = 4) was seen within a large area of mid-vitellogenic oocytes (Fig. 4.1c).

As the gonads approached the period of peak vitellogenesis in September, they were dominated by late-vitellogenic (~300–360 µm) oocytes closely attached to the germinal epithelium through a basal stalk (Fig. 4.1e, f). With confocal imaging the two-sac structure of the gonad wall was evident. The gonad wall is 4–31 µm (x̅ = 15, SD = 10, n = 11) in width (Fig. 4.1e–i). The genital coelomic sinus measured 5 µm (SD = 3, range = 1–9 μm, n = 8) in width (Fig. 4.1e, f). The haemal layer formed a prominent layer filled with amorphous material at the base and periphery of the vitellogenic oocytes (Fig. 4.1g–i). The haemal sinus expanded near the basal stalk at the oocytes (Fig. 4.1g, h). Developing sperm were present (Fig. 4.1h).

In late September, offspring at different stages of development were present in the gonads (1000–1200 µm length) (Fig. 4.2). Late stage oocytes (301–417 µm), gastrulae (380–470 µm), brachiolaria, and pre-metamorphic juveniles were present (Fig. 4.2d, i). Within an adult the development stages were relatively similar or slightly asynchronous (Fig. 4.2 d). The yolk granules in the late stage oocytes measured 10–12 µm in diameter. The gonad wall width ranged 4–9 µm (x̅ = 6 µm, n = 8) and the coelomic sinus width was ~1–3 µm. Some gonad lobes with only vitellogenic oocytes had wider gonad wall (12–16 µm, n = 10) and coelomic sinus (~5 µm, n = 10) than the gonads that were incubating embryos (Fig. 4.3e).

Figure 4.1: Cryptasterina hysteria pre-incubation gonads: a–b) a dissected adult has ten gonads (black circle) that are branched into gonadal lobes (L); c) a three-dimensional confocal image of a gonadal lobe containing early- (O) and mid-vitellogenic (M) oocytes; the two-sac structure of the gonad wall cannot be recognized (arrow); d) a mid-vitellogenic oocyte juxtaposed with the germinal epithelium which has a uniform florescent layer which

may be haemal sinus (arrow); the oocyte has an accumulation of lipid droplets (D); e) in a vitellogenic gonad, oocytes are surrounded by a dense accumulation of what appears to be haemal fluid; the two-sac structure (outer and inner sac) is evident with enlarged coelomic sinus in the middle (empty arrow); f) accumulation of haemal fluid at the vitellogenic oocyte base; coelom (empty arrow) is enlarged; g) larger view of a vitellogenic oocyte stalk showing close association with haemal fluid; h) enlarged haemal sinus layer adjacent to a vitellogenic oocyte that is surrounded by follicle cells (arrows), and developing sperm (S); i) a vitellogenic oocyte that is closely associated with the inner gonad wall (arrow), and coelomic layer (empty arrow) separates the two sacks. N, nucleus; L, lumen; IS, inner sac; OS, outer sac; O, oocyte; H, haemal fluid.

Early larvae were surrounded by an amorphous/acellular layer on their gonad lumen side and their other side was juxtaposed with the germinal epithelium where they were surrounded by thin processes from somatic cells (Fig. 4.3a–d). The amorphous layer separated larvae from the oocytes and sperm (Fig. 4.3b). The somatic cells appeared to provide support for the embryos and larvae near the epithelium (Fig. 4.3c, d). The surrounding matrix (amorphous layer) had degenerated around pre-metamorphic juveniles (Fig. 4.1.h). Gastrulae, brachiolaria and pre-metamorphic larvae remained closely associated with the germinal epithelium, often supported by somatic cells (Fig. 4.2a, b, c and 4.3d, f, g, h). The larval ectoderm was composed of columnar cells that contained lipid droplets (5–7.5 µm diameter) (Fig. 4.2e). During development, these lipid droplets are extruded into the blastocoel (Fig. 4.2b, c, e). Similar lipid droplets were seen in somatic cells adjacent to the embryo (Fig. 4.2e), but it is not known if these lipids are used to support the embryo or are due to resorption of degenerated oocytes. Small vesicles and somatic cells were seen where the embryo was juxtaposed with the germinal layer (Fig. 4.2c, e). Some adults with large gonadal tubules (> 1500 µm in length) contained juveniles with two pairs of tube feet (~ 550 µm) (Fig. 4.2i). In these cases, the gonad wall reduces in width (3–6 µm, n = 3) as it stretches to accommodate juveniles, and it lacks a clearly recognizable two-sac structure.

Figure 4.2: Cryptasterina hystera gonad during incubation: a) an early embryo associated with the germinal layer and somatic cells (empty arrow); b) a brachiolaria (Br) is closely associated with the inner wall (empty arrow) supported by many somatic cells (S); lipid droplets (D) in the blastocoel; c) association of embryo, somatic cells and gonad wall (empty arrow); vesicles seen adjacent to somatic cells; d) fully developed oocytes (O), brachiolaria (B) and pre-metamorphic (M) juveniles in a gonadal lobe; e) a brachiolaria epithelial cells in the process of basal shunting of lipid droplets (empty arrow); same lipid droplets are also seen adjacent to the gonad wall (arrow); f) somatic cells (S) and vesicles adjacent to the inner gonad wall and brachiolaria (Br); g) a three-dimensional view of the larva closely associated to gonad wall (empty arrow); h) a pre-metamorphic juvenile adjacent to the gonad wall surrounded by sperm (empty arrow); i) two gonadal lobes with developing juveniles (J) with two pairs of tube feet and fully developed oocytes (O); note gonad wall has stretched and reduced in width. L, lumen.

Figure 4.3: Gonad histology of Cryptasterina hystera (hematoxylin-eosin): a–b) a layer of acellular membrane (arrows) surrounds the early embryos (E) and acts as a support to the embryo association with the gonad wall (white arrow); c–d) a gastrula (E) is surrounded by the acellular membrane (empty arrows) except the basal stalk (white arrow); e) the gonad wall and coelomic layer adjacent to the vitellogenic oocyte (O) is wider than the gonad wall (empty arrow) that incubating embryo (E); the visceral cells (black arrow) of the outer wall are denser in a vitellogenic gonad; f) The close connection (arrow) between gastrula (E) epithelium and inner sac of gonad through filamentous processes (arrow); g) the larval epithelial (E) and gonad wall are connected (arrow) with a layer of somatic cells (S); h) the acellular membrane has degenerated (empty arrow) around the metamorphic juveniles (M) that are closely associated with the gonad wall (white arrow).

4.4.2 Parvulastra parvivipara

As for C. hystera, the gonads of P. parvivipara have a two-sac structure and are organized in lobes (Fig. 4.4a, b). For this species, gamete development is asynchronous and embryo incubation occurs through most of the year. Thus, the gonad does not have two discrete phases (Fig. 4.4a). The gonads within an adult ranged from predominantly male (sperm only), to vitellogenic, incubating and spent gonads (Fig. 4.4a). The mean oocyte size was 84 µm (SD = 21, range = 54–134 µm, n = 43) and the largest oocyte observed was 134 µm diameter.

In vitellogenic gonads, oocytes (25–90 µm diameter) were attached to the germinal layer (Fig. 4.4b, 4.5a). Developing sperm and somatic cells occur around the oocytes (Fig. 4.4b, c, e). The outer and inner epithelial layers of gonad were prominent, and the gonad ranged in width from 7–27 µm (x̅ = 14, SD = 10, n = 7). The genital coelomic sinus becomes prominent (2–8 µm, n = 7) at this stage. The yolk granules in vitellogenic and fully developed oocytes were ~1.5 µm diameter and do not increase in size through development.

In incubating gonads, developing progeny ranged from gastrula to brachiolaria, pre- metamorphic juveniles and fully developed juveniles (Fig. 4.4). Early gastrulae (~80 µm in diameter) were closely positioned with the inner gonad wall through an accumulation of somatic cells (Fig. 4.4d, 4.5d). Gastrulae and brachiolariae were surrounded by somatic cells and were juxtaposed with the gonad wall in a small basal area (Fig. 4.4e–h, 4.5e–h). This basal region had an accumulation of amorphous material and was often supported by thin processes from associated somatic cells (Fig. 4.5f, g, h). The gonad wall may surround the embryos and larvae (Fig. 4.5f). Pre-metamorphic juveniles were closely apposed to the gonad wall, where accumulation of haemal-like fluid was evident (Fig. 4.4 i, j, k). Histology revealed that pre-metamorphic juveniles may have a close association with the inner gonad wall through somatic micro-filamentous process (Fig. 4.5g, h). The gonad wall and the genital coelom were clearly recognized and were 7–38 (x̅ = 22, n = 12), and 2–8 µm (x̅ = 4, n = 11) in width, respectively (Fig. 4.4e, j).

Gonads that were incubating post-metamorphic juveniles or fully formed juveniles (with two pairs of tube feet) had eggs at different stages of development (Fig. 4.5 i) and the gonad wall was thinner (Table 4.1) as it stretched to accommodate large juveniles. The size of the

juveniles was variable within the same gonad (Fig. 4.4m). Juveniles had their mouth extruded adjacent to fluids and amorphous materials in the gonad (Fig. 4.5i).

Figure 4.4: Parvulastra parvivipara gonads: a) an adult containing undeveloped (U), predominantly male (S), vitellogenic (D) and incubating gonads (I and J); b) a vitellogenic gonad that has enlarged two sac structure (white arrow); oocytes of different stages of

development (early, EO; vitellogenic, VO) are attached with the inner gonad wall; c) a gonad incubating gastrula (G) closely apposed to the gonad wall; d) the association of an early larvae with the gonad wall and presence of somatic cells at the association base (empty arrow); e) an early gastrula surrounded by somatic cells, and fully developed oocytes attached to the gonad wall; f) acellular materials (A) seen between an early larvae and inner gonad wall; g) acellular material (A) is also seen adjacent to brachiolaria (B); h) the attachment of brachiolaria and inner gonad wall is supported with acellular material (empty arrow) that can also be seen around the larva; i) two metamorphic juveniles (M) are seen closely attached with the inner gonad wall (empty arrow); the two-sac structure of the wall is highly enlarged with accumulation of fluid (white arrow) at the base; j) dense accumulation of fluid at the attachment base (empty arrow) of a metamorphic juvenile (M); k) the epithelial layer of a larvae apposed to the gonad wall; l) a large juvenile close to the gonad wall and free sperm (S); m) a gonad that has three juveniles (J1, J2, J3) and oocyte; the gonad wall has attenuated to accommodate large juveniles and the gonadal sacs are not visible. EO, early oocyte; VO, vitellogenic oocyte; O, oocyte; IS, inner sac; OS, outer sac.

Table 4.1: The width of the gonad wall and coelomic sinus in non-incubating and incubating gonads at different phases of gonadal development in Cryptasterina hystera and Parvulastra parvivipara. Data represented as mean (range, n).

Gonad stages Cryptasterina hystera Parvulastra parvivipara

Gonad wall Coelom Gonad wall Coelom (µm) (µm) (µm) (µm)

Non-incubating:

Early gametogenic 5 (3–8, 5) Not evident 3 (3–4, 3) Not evident

Vitellogenic 15 (4–31, 11) 5 (1–9, 8) 14 (7–27, 7) 4 (2–8, 7)

Incubating:

Pre-metamorphic 6 (4–9), 8 2 (1–3, 8) 22 (7–38, 12) 4 (2–8, 11)

Post-metamorphic 4 (3–6, 4) Not evident 5 (3–8, 10) Not evident

Figure 4.5: Gonad histology of Parvulstra parvivipara: a) vitellogenic oocytes developing from germinal epithelium of the inner sac (IS) that is separated from outer sac (OS) by a layer of coelomic sinus (empty arrow); b) a gastrula (G) surrounded by amorphous somatic materials (A); c) degenerating oocytes (D) and remnants of oocytes (arrow) in a gonad that incubating metamorphic juveniles (M); d–e) a metamorphic juvenile (M) within a gonad, and a brachiolaria (B) is adjacent to the inner gonad wall supported through somatic cells, (inset, magnified in e); arrow indicates expanded genital coelom; f) the gonad wall has an invagination (arrow) towards the attachment to the brachiolaria; g–h) a pre-metamorphic (M) larvae is connected with the gonad wall through micro-filamentous process of somatic cells, (inset, magnified in h); i) a juvenile has extruded its mouth (arrow). S, somatic cells; B, brachiolaria; Gd, gonoduct; O, oocyte.

4.5 Discussion

The ovaries of viviparous C. hystera and P. parvivipara observed are morphologically typical of asteroids in having a two-sac structure (Walker 1979, 1980; Beijnink et al. 1984b; Byrne 1989). These ovaries exhibited changes during vitellogenesis and incubation related to gamete and embryonic provisioning modes. Developing oocytes are closely associated with the haemal sinus at their basal stalk zone, similar to other asteroids (Schoenmakers et al. 1981; Beijnink et al. 1984b).

In C. hystera, the oocytes undergo synchronous vitellogenesis and are similar in size and lipid rich profile as those of the free spawning congeners (e.g. C. pentagona) that have seasonal reproductive cycle (Byrne 1992). While development was largely synchronous, there were some differences between offspring stages indicating slight asynchrony. In the matrotrophic P. parvivipara, several stages of oocyte and offspring were found developing simultaneously in a single gonad, indicating asynchronous gametogenesis and fertilization. In P. parvivipara, the presence of an early gastrula at 86 µm diameter, which is smaller than the largest egg observed (134 µm diameter), suggests that the terminal egg size may vary. Asynchronous gametogenesis is also evident in the viviparous ophiuroid Ophiolepis paucispina and holothuroid Synaptula hydriformis that have matrotrophic provisioning (Byrne 1989; Frick et al. 1996). Asynchronous development of several ontogenetic stages or superfetation commonly occurs in live-bearing fishes and was thought to be restricted to fishes (Wourms 1981; Frick 1998). Frick (1998) first used the term ‘superfetation’ in the marine invertebrate literature to describe viviparity in the holothuroid Synaptula hydriformis that cares for multiple cohorts of embryos in the perivisceral coelom. Asynchronous viviparity may be evolutionarily linked to matrotrophy (sibling cannibalism) in P. parvivipara, where producing a small egg (84 μm) freed from fixed energy investment at the onset of embryo development supports continuous reproduction and large offspring to maximize energetic efficiency (Travis et al. 1987).

The intimate connection observed here between the haemal sinus and developing oocytes, and the apparent accumulation of fluid in the haemal layer adjacent to the basal stalk of the oocyte, indicates that the haemal sinus may provide materials for oocyte development in C. hystera and P. parvivipara, as reported for other asteroids (Beijnink et al. 1984a; Beijnink et al. 1984b; Byrne 1989). The haemal sinus and genital coelom were prominent during

vitellogenesis in both species, indicating that dynamic changes in gonad wall morphology may be related to the transfer of nutrients to the oocytes. In asteroids, the haemal layer enlarges during vitellogenesis with significant proteinaceous and carbohydrate substances that supply nutrients to the oocytes (Beijnink et al. 1984a; Beijnink et al. 1984b; Byrne 1989). This includes vitellogenin, which passes through the basal lamina to the oocyte where it is taken up by endocytosis (Beijnink et al. 1984a; Ferrand 1984; Smiley 1988; Byrne 1989; Smiley 1990; Prowse & Byrne 2012; Zazueta-Novoa et al. 2016). Vitellogenin is synthesized in extragonadal tissues (pyloric caeca in asterinids, and the intestine, coelom and gonad in echinoids) and in the gonad somatic cells and is cleaved into protein (Voogt et al. 1985; Harrington & Ozaki 1986). Similar to other asteroids, the basal attachment of the oocyte in C. hystera and P. parvivipara likely forms a pathway through which nutrients may pass from haemal sinus to oocyte (Beijnink et al. 1984b). The haemal sinus can also sequester nutrients from the coelomic fluid through endocytosis of coelomically-derived substances and the action of flagellated-collar cells of the visceral peritoneum (outer gonad wall) (Beijnik et al 1984, Ferguson 1982).

In C. hystera, the number and size of yolk granules and lipids increase during egg development, as is typical of echinoderms (Smiley 1990). However, this was not the case for P. parvivipara, which has a small egg. During development of C. hystera eggs, lipids are shunted into the blastocoel, similar to the oviparous sea star Pateriella brevispina and the sea urchin Heliocidaris erythrogramma (Emlet & Hoegh‐Guldberg 1997; Byrne & Cerra 2000). The lipids are stored in the blastocoel and utilized for post-metamorphic growth in these species (Emlet & Hoegh‐Guldberg 1997; Byrne 2005).

I found that in C. hystera, the gonad wall reduces in width after vitellogenesis as most of the nutrients required for development has already been packaged in the egg. In contrast, the gonad wall and haemal sinus remain expanded during incubation in P. parvivipara, indicating the potential role of this structure in providing nutrients for offspring as they develop from a very small egg. Byrne (1996) also observed that the gonad wall adjacent to pre-metamorphic juveniles was thicker in P. parvivipara compared to non-incubating gonads. In the viviparous holothuroid Leptosynapta clarki, enlargement of the haemal sinus and the extension of myoepithelial cells of the visceral peritoneum around the embryos was suggested to be a significant change in the gonad morphology associated with brooding (supply of nutrients and oxygen to embryo) (Hansen 1975; Sewell et al. 2006). In another viviparous holothuroid, 87

Oneirophanta mutabilis affinis, vacuolated nutritive tissue develops in the ovarian wall, which supports extra-embryonic nutrition (Hansen 1975). In the matrotrophic ophiuroid Amphipholis squamata, the haemal sinus in the bursal wall produces nutritive material for the embryo (Fell 1946; Walker & Lesser 1989). The nutrient-laden haemal sinus of the gonadal wall is suggested to serve a common role in viviparous species, supplying nutrient to gametes during gametogenesis and providing extra-embryonic nutrition (Beijnink et al. 1984b; Byrne 1988, 1989; Sewell et al. 2006). In viviparous asterinids, the reduction in the gonadal wall width during post-metamorphic juvenile development is due to stretching of the gonad wall to accommodate the growing juveniles. The reduction of the width of the gonad wall may facilitate respiratory gas exchange with the juveniles. In C. hystera, the acellular layer around the early embryo is a fertilization membrane that has no supportive function, but is a by- product of fertilization.

In C. hystera and P. parvivipara, larvae and pre-metamorphic juveniles were seen closely associated with the inner gonad in a small area where somatic cells densely accumulated, and the accumulation of haemal fluid was also evident in this region. A similar embryonic association site is evident in the holothuroid Staurothyone inconspicua (Materia et al. 1991). In this species, some juveniles develop in the coelom while others remain attached to the gonadal tubule, which may allow them to uptake nutrients (Materia et al. 1991). The embryonic-parental attachment organ disappears in viviparous Amphipholis squamata soon after metamorphosis (Fell 1946). Similarly, in both viviparous asterinids observed, the basal stalk of the oocyte may remain as an attachment site for the embryo and may support offspring provisioning, which is indirect evidence of matrotrophy. This suggestion should be explored using other methods such as transmission electron microscopy and immuno- histochemistry to identify the presence and cellular localisation of lysosomal and enzymatic activity of somatic cells to show endo- and exocytosis in juxtaposed parental and embryonic cells that may assist in the process of nutrient uptake. Simultaneous development of both gamete types was suggested a common exaptation for the evolution of internal fertilization in asterinids (Byrne 2005). Simultaneous hermaphroditism is also evident in the viviparous holothuroid Synaptula hydriformis and ophiuroid Leptosynapta clarki (Eckelbarger & Young 1992; Frick et al. 1996).

Excess or degenerated oocytes in C. hystera and P. parvivipara may be reabsorbed or be utilized as a local nutrient source for oogenesis, as occurs in other echinoderms (Pearse 1965; 88

Byrne 1996), and also may be used to support embryonic development. Degenerating eggs provide nutrition for embryos in viviparous crinoid, Comatilia iridometriformis (Messing 1984). In the matrotrophic ophiuroid, Ophionotus hexactis, embryos are nourished initially by nurse eggs and later by parental body fluids (Mortensen, 1921; Turner and Dearborn, 1979). The presence of degenerated oocytes and sperm around the developing offspring observed here in P. parvivipara and C. hystera suggest that these gametes may be utilized as a source of nutrients by embryos once their digestive system has developed (Byrne 1989, 1996; Byrne & Cerra 1996; Byrne 2005). The pre-metamorphic juveniles in both species are much bigger and heavier than the eggs (Table 1.2, Chapter 3), indicating that they obtain extra-embryonic nutrients in the gonad.

How viviparous echinoderm embryos uptake nutrients within the gonad is not known. Echinoderm embryos are able to uptake nutrients from sea water, for example the echinoplutei of sea urchin Strongylocentrotus purpuratus, and larvae of sand dollar Dentraster excentricus electively concentrate amino acids from seawater (Manahan et al. 1983; Wright & Ahearn 1997; Vidavsky et al. 2014). Similarly, external epithelial tissues of viviparous Amphipholis squamata embryos can incorporate radioactive amino-acids (Fontaine & Chia 1968). Walker and Lesser (1989) suggested that absorption of dry organic matter directly from seawater may be a possible source of nutrients for intra-bursal embryos in viviparous ophiuroids. Parvulastra parvivipara and C. hystera larvae have cilia and microvilli around outer epithelial cell (Byrne & Cerra 1996; Byrne et al. 2003) that are the probable absorptive organelles. The presence of lipid droplets around the gonad wall and embryo epithelium observed here suggests that lipids that are produced or carried in the gonad wall may be taken up by the embryo. Developing embryos are also likely to uptake nutrients from the haemal layer at the attachment zone or through endocytosis, as somatic cells may control gonadal micro-environment and nutrient recycling (Walker 1979, 1980). This suggestion requires further investigation. For C. hystera, transmission electron microscopy shows that small vesicles are abundant in the apical cytoplasmic region of the embryo epithelium, indicating that the embryo may uptake nutrients within the gonad through endocytosis (Byrne 2005). Thus, embryos in viviparous asterinids may also be able to incorporate nutrients from the gonadal fluid and the haemal layer of the gonad wall, which would mean that echinoderm embryos may be pre-adapted to matrotrophy.

CHAPTER 5: ARRANGEMENT AND SIZE VARIATION OF INTRA-GONADAL OFFSPRING IN A VIVIPAROUS ASTERINID SEA STAR, Parvulastra parvivipara1

5.1 Abstract

Sibling competition and developmental asynchrony may greatly influence the arrangement and size of offspring of marine invertebrates that care for their young. The size variation and the arrangement of progeny within the gonads were observed in a matrotrophic asterinid sea star, Parvulastra parvivipara, which incubates its young in the gonads. Sibling cannibalism supports post-metamorphic development in this species. The progeny vary in size not only between adults but also within (coefficient of variation, CV = 22.6 %) and among (CV = 17.7%) the gonads. Confocal microscopy was used to visualize early embryos, revealing that this species incubates 2–6 developmental stages in individual gonads, which may initiate asymmetric competition. Embryo incubation starts when the gonads reach ~400 µm in length. Sibling competition intensifies once the digestive tract is functional in the tiny juveniles which then start to consume siblings. The three-dimensional arrangement of the juveniles in the gonads was observed using micro-computed tomography. The juveniles were oriented with their oral surface facing each other, presumably as a defensive strategy to protect themselves from being eaten. The size variation in the gonads of P. parvivipara initiates due to asynchronous development at early developing stages and intensifies when siblings start cannibalism. Larger adults had a greater allocation to female function than smaller adults.

1This chapter is partly published

Khan, M.S.R., Whittington, C.M., Thompson, M.B. & Byrne, M. (2019) Arrangement and size variation of intra-gonadal offspring in a viviparous asterinid sea star. Zoosymposia, 15, 71–82. (See appendix for the article)

5.2 Introduction

Since there are limitations in the space and resources that can be provided to offspring (Bernardo 1996; Kamel et al. 2010a), parents in marine invertebrates that care for their young internally adopt diverse strategies to arrange or package their offspring to maximize reproductive output and offspring fitness. For example, early developmental stages in the sea urchin, Amphipneustes lorioli, are partitioned in the deeper recesses of the brood pouches to reduce overcrowding, an adaptation to limited space (Galley et al. 2005). In the clam, Transennella tantilla, the smallest embryos are positioned dorsally adjacent to the oviduct and, as they develop, progressively move towards the ventral surface of the gill, which facilitates continuous reproduction and hatching (Kabat 1985). The maternal provisioning strategy may play a role in determining the arrangement and size of the offspring in the incubation space (Carrasco & Phillips 2014).

For species that care for their young internally, offspring interact with the parent and may compete with siblings for limited resources; these interactions may influence size variation at birth and offspring arrangement (Schrader & Travis 2009; Kamel et al. 2010b). Parent- offspring conflict and sibling competition are more intense in species that provide extra- embryonic nutrition (Frick 1998; Schrader & Travis 2009; Ostrovsky et al. 2015; Mercier et al. 2016; Kamel & Williams 2017). Extreme sibling competition may result in cannibalism where the progeny ingest eggs (oophagy) or siblings (embryophagy or adelphophagy) (Lesoway et al. 2014; Ostrovsky et al. 2015). Sibling cannibalism greatly increases size variation of offspring in some gastropods, sharks and sea stars (Rivest 1983; Smith & Reay 1991; Byrne 1996; Collin & Spangler 2012). Thus, adelphophagic matrotrophic species provide an opportunity to assess how sibling competition mediates offspring size variation and how the young are arranged in the incubation space.

Non-invasive imaging techniques such as micro-computed tomography (µ-CT) have been used to observe offspring inside the brood chamber non-destructively in situ in ophiuroids (Landschoff & Griffiths 2015). The asterinid sea star, Parvulastra parvivipara, exhibits a highly derived mode of matrotrophic incubation, sibling cannibalism, and a high level of offspring size variation (Byrne 1996; Khan et al. 2019a). This species cares for its young within the gonads to an advanced juvenile stage that varies in size from 0.4–3.0 mm in diameter (Byrne 1996; Khan et al. 2019a). Adelphophagy provides significant nourishment

for the juveniles, and gonad histology shows that cannibalism may mediate the position of juveniles within the gonads (Byrne 1996).

The position of juveniles in P. parvivipara was observed three-dimensionally using µ-CT to investigate whether post-metamorphic sibling cannibalism mediates the arrangement of juveniles. Parvulastra parvivipara exhibit a great size variation of offspring not only within and among adults (Khan et al. 2019a) but also within and among the gonads (Chapter 4). The extent to which the offspring size varies within and among gonads is not known. In P. parvivipara, the gonads are at different stages of development with individual adults, and most of them contain juveniles. It is not known at what size the gonads initiate incubation (Byrne 1996). There are several features of the reproductive biology of this species that are yet to be determined including: 1) at what size gonads start incubation 2) how many development stages are present in a single gonad 3) whether the larger parents have more incubating gonads than smaller parents or whether larger adults have more allocation for female (brooding) function than smaller parents. These questions are addressed here. The potential causes of offspring size variation within gonads are also considered.

5.3 Materials and methods

5.3.1 Sample collection

Parvulastra parvivipara (n = 52) were collected from the Smooth Pool (32ʹ54ʹʹS, 134ʹ04ʹʹE), Eyre Peninsula, South Australia (Ministerial Exemption from the Ministry of Agriculture, Fisheries and Forestry, Government of South Australia- ME9902902) and transported to the University of Sydney. A total of 50 adults were dissected under a dissection microscope (Olympus SZX10) to count the number of incubating gonads and predominantly male gonads (testes). From these dissected adults, 15 were fixed for confocal microscopy and 22 were used to determine the coefficient of variation (CV) in offspring size within and among gonads. To determine CV, the gonads were isolated and dissected and juveniles were counted, measured and photographed using a camera connected to that dissection microscope (Olympus SZX10). Adults and juveniles were measured using the software Micropublisher 3.0. The arm radius (R) of the parent was measured from the centre of the mouth to the tip of one arm. The CV in offspring size was calculated from the gonads with more than one developing young using the following formula: CV = Standard deviation/mean × 100. The

relationship between within gonad CV and number of offspring was analysed using linear regression to determine whether the number of offspring influences CV in offspring size within a gonad.

The relationship between the number and size of gonads that were incubating progeny per adult and adult size was analysed using linear regression. The relationship between the number and size of the male gonads (which only had sperm) and the parent size was also analysed. These data were used to determine relative investment in the male (testes) and female (brooding) function using linear regression. Prior to analysis, data were tested for normality with the Shapiro-Wilk test and for homogeneity of variance with Levene’s test. All linear regressions were analysed using IBM SPSS statistics 24 and significance was assigned at the 5% level.

5.3.2 Micro-computed tomography and confocal microscopy

One P. parvivipara was preserved in 70% ethanol. Micro-CT scans of this individual were performed using Skyscan 1172 with isotropic voxel resolution 2.94 µm. The specimen was firmly positioned within an Eppendorf tube using polystyrene foam and then scanned. A scan of 11 hours at 23 kV and 200 µA resulted in 1170 images that visualized the skeleton. The images were analysed using the software Avizo 9.5 version. Manual “draw tool” and “magic wand” were applied to each image stack to isolate each of the juveniles. A different colour was applied to each juvenile within a single gonad to visualize them separately. The position and orientation of the juveniles were visualized by applying three-dimensional (3-D) reconstruction and by making the parent’s body virtually transparent.

Confocal microscopy was done following the same procedure mentioned in Chapter 4. However, emphasis was given to the juvenile position and number of development stages in gonads. Fifteen dissected P. parvivipara were fixed in 2.5% glutaraldehyde in 1 µm filtered seawater (FSW) for 24 hours and then stored in 70% ethanol. Each gonad was isolated from individual sea stars and dehydrated in an ethanol series (70–95 %). The gonads were cleared with benzyl alcohol and benzyl benzoate, 2:1 (v/v) (Sigma-Aldrich Co.) and mounted on a glass slide. The glutaraldehyde auto-fluorescence was used as a signal for analysis using a Leica SPE-2 Confocal Laser Microscope or ZEISS LSM 800 plus Spectral Confocal Microscope. The samples were excited at 488 nm wavelength with an argon laser contrasting

separate gain and offset values for each sample. Images were taken from the different focal planes at 1024×1024 pixel array with a frame average that varied from 4 to 12 based on sample response to excitation. A series of images were stacked for viewing 3D view along the Z axis (Z-stack). The length of the gonad and the diameter of egg and embryo were measured using confocal microscope online assistance. The development stages in relation to the size (length) of the gonad were observed for 77 gonads isolated from 15 P. parvivipara adults to find the size of incubating and non-incubating gonads. The shape of these gonads was determined using a measure of circularity through Image J. Circularity is the measure of roundness which ranges from 0–1, with a value of 1 representing a perfectly round shape.

5.4 Results

Parvulastra parvivipara with young in the gonads were a range of sizes (R = 1.8–5.2 mm, 0.0094–0.1234 g). The gonads varied in size and shape (Fig. 5.1a, b). In gonads with one juvenile, the gonad shape reflected the juvenile form (Fig. 5.1b). Only two of the 50 adults had juveniles in all ten of their gonads (Fig. 5.1a). The gonads ranged in size from 120–2630 µm (x̅ = 941, SD = 624, n = 49) due to the wide size range of juveniles. Predominantly male gonads were small (x̅ = 422, SD = 18, Range = 147–940 µm, n = 56 gonads from 27 adults) with a mean of 2.11 testes per individual (SD = 0.96, Range = 0–4, n = 28). The number of 2 male gonads was not related to adult size (r = 0.079, F(1,26) = 2.242, P > 0.05), but the size of 2 these gonads was larger in the bigger adults (r = 0.388, F(1,25) = 16.156, P < 0.05). The number of gonads that were incubating progeny was positively related to adult size (r2 =

0.466, F(1,38) = 33.107, P < 0.05).

Offspring within a gonad ranged from developing brachiolaria, and metamorphic juvenile to fully developed juveniles (Fig. 5.1d). The CV in offspring size within an individual gonad was 24.7 % (SD = 24.18, Range = 0.14–112.57, n = 64 gonads from 22 adults). Among- gonad CV within an adult was 19.5 % (SD = 14.56, Range = 1.91–39.01, n = 22 adults 64 gonads). The largest number of juveniles in a single gonad was nine.

5.4.1 Arrangement of young in gonads i) Confocal Microscopy

Among 63 gonads observed from 13 adults, four were predominantly testes (Fig. 5.3f). The other 32 gonads contained juveniles, five with embryos only, four with embryos and juveniles, and 18 had developing oocytes. All gonads except four male gonads (Fig. 5.3d) had oocytes of variable stages and sizes. The size of the gonads at which incubation starts varied among individuals, with the smallest gonad containing embryos being ~ 450 µm in length (Fig. 5.2a). Gonads with developing oocytes were tubular or dome-shaped and become oval or round with increasing size and number of juveniles (Fig. 5.2b). The profile of gonads that had only one large juvenile reflected the shape of the juvenile (Fig. 5.2b).

Figure 5.1: Parvulastra parvivipara: a) ten gonads with juveniles (white arrows); b) asynchronous gonad development with juveniles (white arrows) variable in size; predominantly male gonads (black arrows); c) an adult incubating nine juveniles in a gonad (arrow), and three juveniles in another gonad (white arrow) where juveniles are orally opposite to each other (inset, 1.5 mm scale); d) pre-metamorphic (black arrow), newly metamorphic (upper top) and fully developed (white arrow) juveniles from a gonad. Scale: a and b = 2.0 mm, c = 1.5 mm, and d = 0.5 mm.

The gonads contained a range of developmental stages (1–6) within a gonad, including gastrula, brachiolaria, pre-metamorphic and fully developed juveniles (Fig. 5.3a, d). The mean diameter of gastrulae was 91 μm (SD = 4, n = 3), early brachiolaria was 112 μm (SD = 6, n = 4), late brachiolaria was 145 μm (SD = 13, n = 3), and pre-metamorphic juveniles was 218 μm (SD = 4, n = 3). Juveniles remain orally apposed to each other, with amorphous material in between (Fig. 5.3a, b). In some gonads, the smaller juveniles or embryos were seen adjacent to the mouth of the larger juveniles (Fig. 5.3 c, e).

Figure 5.2: a) The size of gonads, embryos and juveniles in 15 Parvulastra parvivipara adults (n = 77 gonads). All gonads except predominantly male gonads (that contain sperm) had oocytes. b) The relationship between size and shape of the gonads; shape is indicated by circularity (n = 60 gonads).

Figure 5.3: Confocal microscopy of Parvulastra parvivipara gonads: a) a 3-D view of a small juvenile positioned at the mouth of a large juvenile; b) a 3-D view of two juveniles with their oral faces (arrow) opposite to each other; c) oral face of a large juvenile with two pairs of tube feet in each arm contrasting in size with a small metamorphosing juvenile (arrow); d) gonad containing eggs, early gastrula (arrow), gastrula, brachiolaria, late brachiolaria, and metamorphic juvenile and juvenile; e) juvenile with stomach extruded (arrow); f) sperm in a predominantly male gonad. E, egg; G, gastrula; B, brachiolaria; LB, late brachiolaria; MJ, metamorphic juvenile; J, Juvenile; M, mouth; S, sperm; T, tube feet.

Figure 5.4: Micro-computed tomography of Parvulastra parvivipara: a) 3-D image showing the position of 13 juveniles (coloured differently) within the gonads (A–J), gonad B, I and J had no juveniles; b) juveniles (coloured differently) with parent body removed virtually; separate numbers indicate juveniles from separate gonads (A–H); c) two juveniles orally apposed to each other in a gonad; d) position of five juveniles (coloured differently) within a gonad oriented with their oral surfaces opposed. Scale: a and b = 1.0 mm, c and d = 0.5 mm. ii) µ-CT

The 3-D tomographic video of P. parvivipara shows 13 juveniles (392–1620 μm diameter) in the gonads (Fig. 5.4a–b). Four gonads each contained one juvenile, two gonads each contained two juveniles, one gonad had five juveniles and three gonads had no juveniles (Fig. 5.4a–b). In gonads with two juveniles, these were facing orally opposite to each other (Fig. 5.4c). In the gonad that had five juveniles, the largest juvenile was 926 μm diameter and the

smallest juvenile was 721 μm diameter. All five juveniles were oriented with their oral face opposite to each other, and they had their oral face towards the centre of the gonad (Fig. 5.4d). The position and orientation of juveniles had no specific pattern in the gonads that had a single juvenile. The largest juvenile was more than four times larger than the smallest solitary juvenile.

5.5 Discussion

Parvulastra parvivipara exhibits a unique spatial arrangement of its young in the gonads that is influenced by its maternal provisioning strategy and sibling cannibalism. The juveniles usually position their oral surface opposing one another, an orientation that may defend them from being preyed upon by siblings (Byrne 1996). Juveniles that achieve an advantageous position and face the aboral surface of another juvenile may successfully cannibalize their sibling (Byrne 1996). Large juveniles appear to be relatively stationary in the gonads, potentially to prevent rupture of the gonad wall. Juveniles often had their stomachs extruded, indicative of predation and perhaps using gonadal fluid and eggs as sources of nutrient.

The 3-D reconstruction of P. parvivipara offspring adds a new dimension in observing brooding in marine invertebrates. There is one previous three-dimensional reconstruction (µ- CT) of the offspring in the brood chamber of echinoderms, those of the ophiuroids, Ophioderma wahlbergii and Amphipholis squamata (Landschoff & Griffiths 2015). In these ophiuroids, the juveniles had their oral surface pressed against the bursal sinus, which provides nutrients to developing young (Byrne 1991; Hendler & Tran 2001). In P. parvivipara, amorphous material in the gonad may be haemal fluid and is a potential source of nutrients for the juveniles. For this species, however, adelphophagy is the main source of nutrients, which supports substantial post-metamorphic growth.

Studies that relate sibling competition to the arrangement of offspring in the incubation chamber in marine invertebrates are rare. This unusual mode of maternal provisioning is also reported in the holothuroid Leptosynapta clarki and the sea star Pteraster militaris, and Parvulastra vivipara (McClary & Mladenov 1990; Byrne & Cerra 1996; Sewell et al. 2006). Sibling cannibalism is common in marine gastropods and polychaetes (Cable & Tinsley 1991; Cubillos et al. 2007; Carrasco & Phillips 2014). Sibling cannibalism has evolved to support extra-embryonic nutrition and generation of large offspring size in sharks and

molluscs (Wourms 1981; Smith & Reay 1991; Strathmann 1995; Cubillos et al. 2007). For example, in sand tiger sharks, Carcharias taurus, 9–10 embryos begin development in each oviduct, but only one gigantic offspring survives (Wourms 1981). Similarly, several molluscs place hundreds of embryos within a single egg case to support few large hatchlings that grow through cannibalism (Strathmann & Strathmann 1995; Carrasco & Phillips 2014). As the juveniles of P. parvivipara develop from a very small secondarily reduced egg (84 µm diameter), intragonadal sibling cannibalism may have evolved in parallel with a decrease in egg size to support post-metamorphic nutrition and produce large offspring sizes (Byrne 1996; Byrne & Cerra 1996).

Sibling cannibalism may be an adaptation to reduce the number of embryos competing for oxygen, as in the mollusc Acanthina monodon, where adelphophagy increases with increasing oxygen demand (Lardies & Fernandez 2002). It is not known whether sibling cannibalism in P. parvivipara decreases embryonic development time as it does in several muricid gastropods (e.g. Crepipatella fecunda) (Spight 1975; Cubillos et al. 2007). Adelphophagy should be favoured if structural and energetic constraints limit maximum egg size (Strathmann 1995; Sakai & Harada 2001). This may be the case for P. parvivipara, which is the smallest known sea star. The offspring produced by P. parvivipara are much larger than the juveniles released by the closely related asterinid brooders that have large eggs and lecithotrophic development (Byrne & Cerra 1996). Sibling cannibalism in P. parvivipara may be an adaptation to produce large offspring size (>1000 µm) to enhance their success in the harsh high intertidal environment occupied by this species (Byrne 1996; Trumbo 1996; Gillespie & McClintock 2007). The largest juveniles produced by P. parvivipara are nearly reproductively mature (Byrne 1996). Large offspring may be selected for early reproductive maturity, high survival and stress tolerance (Rivest 1983; Moran & Emlet 2001; Olivera- Tlahuel et al. 2015).

Most of the gonads in P. parvivipara started incubating offspring at ~400 µm in length and incubate several development stages at a time. This asynchronous development may allow them to produce offspring continuously. Continuous production of offspring may be a mechanism to reduce overcrowding and utilize the limited space available to accommodate offspring in this small-bodied sea star, similar to the sea urchin A. lorioli (Galley et al. 2005). According to the energetic hypothesis, this tiny sea star may not have sufficient energy to produce more and larger eggs to invest all of its resources at a time (Chia 1974). However, 100

larger adults of P. parvivipara are more capable of allocating a greater amount of energy at a time for female functioning or offspring care than smaller adults.

Parvulastra parvivipara exhibits substantial among- and within gonad-variation in offspring size. In species with asynchronous viviparity, early hatched larger embryos/juveniles outcompete less-developed siblings (asymmetric competition) due to their physical superiority or greater competitive ability (Olivera-Tlahuel et al. 2015). Hence, they will be larger at birth, as occurs in some live-bearing fish (Schrader & Travis 2012). In P. parvivipara, asynchronous viviparity initiates asymmetric competition among early developing embryos, and dynamic sibling cannibalism reduces parental control on offspring size (Pollux & Reznick 2011) and intensifies the size variation of offspring at the post- metamorphic stage. Within-brood variation in offspring size in P. parvivipara may also be an adaptation to an unpredictable intertidal environmental (bet-hedging) (Marshall & Keough 2008), or some offspring may simply obtain slightly more resources in the early development stages (Rivest 1983; Moran & Emlet 2001).

101

CHAPTER 6: GENERAL DISCUSSION

The strategy of reproduction and provisioning of offspring profoundly influences the reproductive output, fitness, and population dynamics of a species (Vance 1973b; Christiansen & Fenchel 1979; Byrne 1996; Marshall & Keough 2008; Sun et al. 2012; Lim et al. 2014; Mercier et al. 2016). Viviparous animals exhibit diversity in reproductive strategies and offspring provisioning modes that raises fascinating life-history questions about the selective pressures, constraints and adaptations associated with this reproductive mode (Hendler 1975; Christiansen & Fenchel 1979; Wourms 1981; Byrne 1996; Blackburn 2005; Blackburn 2015; Kalinka 2015; Ostrovsky et al. 2015). Among marine invertebrates, viviparous asterinid sea stars exhibit remarkably diverse offspring development, provisioning and size variation, traits that are the central to the concepts of life-history evolution (McEdward & Janies 1993; Byrne 2006; Khan et al. 2019a). Asterinid sea stars are therefore excellent models for comparing reproductive and provisioning strategies to investigate the potential selective forces involved in the evolution of viviparity in marine invertebrates.

This thesis used three asterinid species, Cryptasterina hystera, Parvulastra vivipara and P. parvivipara, as models to address questions about the biology and evolution of echinoderm viviparity. This thesis specifically investigates several maternal-offspring relationships in species that have contrasting lecithotrophic and matrotrophic provisioning in order to understand the evolution of viviparity and nutrient provisioning in echinoderms. I investigated: 1) offspring size and maternal size correlations, 2) evolutionary trade-offs between size and number of offspring, 3) the pattern of offspring release and retention, 4) the level of parental investment in offspring in addition to the egg, 5) offspring provisioning strategies, and 6) potential adaptations in gonad morphology.

6.1 Offspring investment, size, fecundity and allometry

A major question in life history evolution is whether the parent size influences the amount of investment in offspring, and whether larger parents produce larger offspring (Bingham et al. 2004; Ilano et al. 2004; Lim et al. 2014). This issue was addressed in Chapter 2. In viviparous asterinids, the level of parental care influences fecundity, size and reproductive output. For lecithotrophic C. hystera, the hypothesis that larger mothers produce larger offspring was not

102

supported; instead larger adults produced more similarly sized offspring (Clarke 1992; Miloslavich & Dufresne 1994; Hendry et al. 2001; Valentinsson 2002; Khan et al. 2019a). Cryptasterina hystera invests almost all the energy required for offspring in the egg (Byrne et al. 2003; Byrne 2005; Khan et al. 2019a). For P. vivipara and P. parvivipara, intragonadal growth of offspring in the two species with matrotrophic provisioning depends on sibling cannibalism, an unpredictable source of nutrients. This situation increases sibling-sibling and parent-offspring conflict. An asymmetric competition between offspring arises due to asynchronous development in P. parvivipara and P. vivipara (Khan et al. 2019a, b). Sibling cannibalism results in size variation of offspring as in other matrotrophic species (Schrader & Travis 2009; Pollux & Reznick 2011; Collin & Spangler 2012; Schrader & Travis 2012). The parent exerts little or no parental control over offspring size, with juveniles >1000 µm diameter regardless of parent size. Thus, the hypothesis that larger adults would have larger offspring in the gonads was not supported for these species. The second question focused on allometry of brooding hypothesis, to test whether the number of progeny that can be cared for is constrained in large adults due to space limitation, which has been shown to apply to many marine invertebrate groups (Strathmann & Strathmann 1982; Strathmann et al. 1984). In contrast, this hypothesis did not apply to viviparity in the asterinid sea stars.

The juveniles in these species are genetically identical to each other (Keever et al. 2013) and so according to theory sexual conflict should not produce offspring size variation (Kamel et al. 2010a; Kamel et al. 2010b; Kalinka 2015). However, this is not the case for the viviparous asterinids and indeed does not apply to the matrotrophic species (P. vivipara and P. parvivipara) in general. Current hypotheses on the size variation offspring are based on species with equal amount of provisioning to each offspring and therefore do not apply to the matrotrophic asterinids (Vance 1973a, b; Smith & Fretwell 1974). This situation suggests, in cases where siblings can compete for parental resources, that selection may favour a parent to produce offspring that are variable at birth, which may be beneficial in a variable environment (Cameron et al. 2017).

Comparison of offspring biomass indicated that the matrotrophic parent would have greater investment then the lecithotrophic parents of the same size (Chapter 2). This is the first study in which matrotrophy index was used to estimate the level of parental investment in any echinoderm. The individual offspring of P. vivipara and P. parvivipara are larger and heavier than those of C. hystera. It appears that matrotrophy may have evolved to support larger 103

juvenile size at the expense of fecundity, compared with the lecithotrophic strategy (Trexler 1997; Khan et al. 2019a). The tropical and temperate distribution of viviparous asterinids is counter to Thorson’s hypothesis that species with parental care are distributed in higher latitudes (Mileikovsky 1971; Clarke 1992).

6.2 Offspring release and retention dynamics

In viviparous asterinids, the juveniles can vary in size within a clutch and over time. The dynamics of offspring release and retention is of interest to understand whether P. parvivipara and C. hystera release juveniles all at once or in several cohorts, and what are the level of maternal provisioning in offspring in different size clutches. In captivity, P. parvivipara released juveniles over a long period in several cohorts. The suggestion that birth is a cause of death in parent P. parvivipara (Keough & Dartnall 1978) was addressed. It is clear that all P. parvivipara do not die after releasing the juveniles, contrary to suggestions of Keough and Dartnall (1978), although there was some mortality in very small parents. There may be a trade-off in nature between the degree and duration of offspring care versus mortality. In nature, a sufficient number of adults survived this bottleneck (Chapter 3). The smallest adults were able to release juveniles that are similar in size to those released by the larger adults (Chapter 3). Thus parents, regardless of size, can release large juveniles. Parvulastra parvivipara juveniles received an extensive amount of matrotrophy on the basis of dry weight gain. Ash-free dry weight measurement would support important information on how much inorganic ions (i.e. calcium) the juveniles receive from the parent during development. My results also indicated that this species can reproduce in isolation and has continuous reproduction, similar to P. vivipara (Prestedge 1998).

Cryptasterina hystera gave birth to its young in a large cohort, however, a few large juveniles were retained. This observation indicates that larger juveniles are supported by extra- embryonic provisioning potentially through sibling cannibalism (Byrne 2005). The retention of few juveniles may be an exaptation to a matrotrophy similar to P. parvivipara and P. vivipara. Cryptasterina hystera juveniles increased in dry weight 1.7 to 6.2-fold compared to eggs, indicating that offspring provisioning is not strictly lecithotrophic but instead there is a continuum of extra-embryonic provisioning in viviparous asterinids (Chapter 3) (Riesch et al. 2010). A small amount of nutrient transfer to the embryos in lecithotrophic viviparous species

104

may a consequence of the physiological intimacy between parent and embryo produced by viviparity (Blackburn 1992; Thompson et al. 2000).

6.3 The dual role of the gonads in viviparous asterinids

A major question in echinoderm reproduction is whether the gonad plays a common role in both gamete production and offspring incubation (Sewell et al. 2006). This question was addressed in Chapter 4. The gonad histology and confocal microscopy suggests that the haemal layer of the gonad provides nutrients for gamete development and for developing embryos in C. hystera and P. parvivipara, like the situation in the viviparous holothuroid Leptosynapta clarki (Sewell et al. 2006). The haemal layer expands during gametogenesis due to an accumulation of nutritive material (Walker 1979). In addition to the haemal layer, the genital coelom also enlarged, indicating that the coelom may also contribute nutritional material. The dynamic changes in the gonad wall are associated with the provisioning modes and timing of energy allocation in C. hystera and P. parvivipara.

In these two viviparous asterinids, the early embryos and larvae remained closely associated with the inner gonad wall and were supported by thin processes from somatic cells. This association may support the embryo by temporary attachment, as in the ophiuroid Amphipholis squamata (Fell 1946). It may also provide nutrients to the embryos, but this possibility and the underlying mechanisms remain to be investigated. This attachment may be one of the potential adaptions for viviparity in asterinids (Byrne 2005). The absorption of molecules through embryo epithelium may be possible in both species, as was suggested for the holothuroid Synaptula hydriformis and the ophiuroid Amphipholis squamata (Fontaine & Chia 1968; Frick 1998). Somatic cells and its phagocytic properties may play a key role in controlling the nutrient recycling, a general feature of echinoderm gonads (Walker 1979, 1980).

6.4 Arrangement of intragonadal offspring in P. parvivipara

The position of juveniles was observed three-dimensionally using micro-computed tomography. In P. parvivipara, the progeny varied in size not only within adults but also within and among gonads (Chapter 5). The intragonadal juveniles of P. parvivipara remained orally opposite to each other, probably as a part of their defensive strategy to avoid being

105

preyed upon by siblings (Byrne 1996). Micro-computed tomography is an important tool that can be applied to observe offspring arrangement pattern and reproductive biology in other viviparous or brooding echinoderms. The viviparous ophiuroids Amphipholis squamata and Ophioderma wahlbergii) were found to press their mouths against the bursal wall, which provides nutrients (Landschoff & Griffiths 2015). The size variation of offspring in the gonads of P. parvivipara starts at early developing stages due to asynchronous development that initiates asymmetric competition for parental resources, as in superfetating fishes (Scrimshaw 1944; Schrader & Travis 2009; Schrader & Travis 2012). Size variation of juveniles within a gonad intensifies when siblings start cannibalism (Byrne 1996).

6.5 Future directions

As highlighted in chapters 1, 3 and 4, the embryos and pre-metamorphic juveniles are larger than the eggs in most viviparous echinoderms and they do not have a well-developed digestive system (Table 1.2). Whether the embryos receive nutrients and the mechanisms underlying potential nutrient transfer are still not known. Calcium is an important ion for biological functions in echinoderms, and its absence may obstruct skeletal formation in embryos (Okazaki 1956; Whitaker 2008; Vidavsky et al. 2016). The embryo in oviparous species may uptake nutrients (amino acids, calcium) from sea water (Manahan 1990), but in viviparous species, the gonad may supply these nutrients. Use of radio-labelled nutritive materials (e.g. palmitic acid, glycine, glucose, and calcium markers such as calcein), and florescent microspheres containing polystyrene latex would be a useful approach to track nutrient transport, as has been done for Synaptula hydriformis, where the embryonic absorption of molecules from the parent coelom was observed (Manahan et al. 1983; Manahan 1990; Frick 1998; Vidavsky et al. 2016). A pulse of calcium signal to mark the skeleton could also be used as a marker to determine time of incubation (Whitaker 2008).

One of the possible routes of nutrient transfer may be epidermal absorption associated with specialization such as receptor mediated endocytosis (Walker & Lesser 1989; Frick 1998). This possibility could be explored using labelled tracers and transmission electron microscopy. Serial transmission electron microscopy and/or serial block-face scanning electron microscopy will be useful to determine if there are morphological specializations associated with incubation of offspring. In particular, it would be interesting to investigate

106

whether the attachment between embryo and gonad wall facilitates nutrient transport. Comparison of the morphology in the gonads of oviparous and viviparous species may provide understanding into morphological adaptations associated with the evolution of viviparity. In free spawners, the egg jelly coat contains triggering substances to stimulate contraction of the ovarian wall in the presence of calcium ions (Kishimoto et al. 1984). The egg jelly coat is hydrated on contact with sea water and has sperm attractants (Shirai et al. 1981; Kishimoto et al. 1984; Whitaker 2008). However, eggs in viviparous asterinids are internally fertilized and should not come in contact with sea water. One aspect that requires in depth investigation is the change in gametes associated with intragonadal fertilization.

Lysosomes are organelles that can degrade intracellular macromolecules (lipids, glucose and proteins) into smaller particles and secrete material extracellularly (apocrine secretion) via membrane-bound vesicles (Blott & Griffiths 2002). Apocrine mechanisms underlythe nutrient transport across uterine plasma membrane in a viviparous skink (Biazik et al. 2009). Lysosomal vacuoles and lysosomal enzymes are present in the yolk vesicles of oviparous Asterias rubens (Ferrand 1980). The lysosome-like vacuoles in the basal region of the ooplasm in Asterias play a role in active transport of nutrient from germinal epithelium to oocytes (Beijnink et al. 1984b). It is not known whether lysosomes are present in the embryo epithelium and in the gonad wall in viviparous asterinids or whether lysosomes assist in transporting nutrients from the gonad wall to the embryo. Lysosomal enzymes, especially alkaline phosphatase (AP), aid in active transfer of molecules through the plasma membrane (Biazik et al. 2009, 2010). Alkaline phosphatase belongs to a membrane-bound protein family which shows pronounced activity in secretory sites of epithelial cells (Biazik et al. 2009). In sea urchins, there is greater activity and abundance of this enzyme in larval stages than in oocytes (Gustafson & Hasselberg 1950). It was suggested that AP activity increases with increasing protein demand in gastrula larvae in the sea urchin Psammechinus miliaris (Gustafson & Hasselberg 1950). Alkaline phosphatase was also suggested to be associated with skeletal formation and absorptive function in the digestive tract of embryo in sea urchin (Hsiao & Fujii 1963; Pfohl 1975). Thus, lysosome and lysosomal enzymes may also be associated with any translocation of nutrients from parent to the embryo within the gonad in viviparous sea stars. This possibility should be examined through enzyme histochemistry and electron microscopy.

107

The supply of respiratory gas exchange to the intragonadal offspring may be a limiting factor for brooding/viviparous marine invertebrates (Strathmann & Chaffee 1984; Strathmann 1995). The gonad wall in viviparous asterinids decreases in width and stretches to accommodate large juveniles, which may allow easy diffusion of respiratory gases for the intragonadal juveniles. It is not known how the viviparous embryos receive respiratory gases. To determine this, gonads with live juveniles could be examined using an oxygen microoptode and respirometry experimental set up, as was used to study the pattern of oxygen supply to offspring in brooding crabs (Fernandez et al. 2000; Fernández et al. 2003). These techniques may allow us to identify whether the gas exchange to juveniles is supplied via diffusion through the gonad wall or whether the coelomic and haemal fluid carries oxygen to the juveniles in viviparous asterinids.

6.6 Conclusion

This thesis addresses knowledge gaps in the biology and life-history evolution of viviparous asterinids and sheds light on the potential adaptations for embryonic nutrition in echinoderms. The findings of this thesis contribute to our understanding of the evolutionary ecology of offspring size variation in marine invertebrates that have parental care of offspring. My research found that, in viviparous asterinids, the parent size is indicative of the total reproductive output but not indicative of offspring size, irrespective of parental provisioning modes. The greater size variation of offspring in matrotrophic species than lecithotrophic species suggests that post-fertilization parental provisioning increases sibling competition. Sibling cannibalism increases size variation in offspring and may influence offspring arrangement in the gonads. My research emphasized that almost all viviparous echinoderms exhibit size variation of offspring, and matrotrophic species show greater variation than any other reproductive mode observed in marine invertebrates. However, the size variation in offspring among viviparous species is poorly reported. The dynamics of offspring release and retention observed here in viviparous species enhance our understanding of the evolutionary ecology of offspring size variation in marine invertebrates. Selection seems to have favoured larger offspring in matrotrophic species but greater fecundity in lecithotrophic species. This result suggests that the selective pressures involved in the evolution of viviparity are diverse due to the diversity in reproductive patterns exhibited by viviparous asterinids, and may also be related to the habitat of the species. My comparisons of reproductive investment suggest that matrotrophic parents may overall invest

108

more in reproduction than lecithotrophic species. My morphological observations of the gonad wall suggest that the close association between the embryo and the gonad wall is a potential adaptation associated with the evolution of viviparity in asterinids. The haemal layer and the genital coelomic sinus of the gonad wall undergo morphological changes and play a common role in providing nutrients for the gametes and embryos in viviparous species. Further studies on the mechanisms of extra-embryonic nutrition and gonadal ultrastructural changes in asterinids are suggested for greater understanding of the evolution of viviparity. The comparison of reproductive and provisioning strategies among closely related species as in the three species investigated here is an important approach to enhance our understanding of how and why viviparity evolved.

109

REFERENCES

Allen, R.M., Buckley, Y.M. & Marshall, D.J. (2008) Offspring size plasticity in response to intraspecific competition: an adaptive maternal effect across life-history stages. American Naturalist, 171, 225–237. Allen, R.M. & Marshall, D. (2014) Egg size effects across multiple life-history stages in the marine annelid Hydroides diramphus. PLoS One, 9, e102253. Avise, J.C. (2013) Evolutionary perspectives on pregnancy. Columbia University Press, New York. Baeza, J.A. & Fernandez, M. (2002) Active brood care in Cancer setosus (Crustacea: Decapoda): the relationship between female behaviour, embryo oxygen consumption and the cost of brooding. Functional Ecology, 16, 241–251. Beijnink, F.B., Broertjes, J.J.S., Brands, F. & Voogt, P.A. (1984a) Immunocytochemical demonstration of vitellogenic substances in the haemal system of the sea star, Asterias rubens. Marine Biology Letters, 5, 303–313. Beijnink, F.B., Walker, C.W. & Voogt, P.A. (1984b) An ultrastructural study of relationships between the ovarian hemal system, follicle cells, and primary oocytes in the sea star, Asterias rubens - implications for oocyte nutrition. Cell and Tissue Research, 238, 339–347. Berecoechea, J.J., Brogger, M.I. & Penchaszadeh, P.E. (2017) New evidence of brooding in the deep-sea brittle star Astrotoma agassizii Lyman, 1876 from a South Western Atlantic Canyon. Deep-Sea Research Part I Oceanographic Research Papers, 127, 105–110. Bernardo, J. (1996) The particular maternal effect of propagule size, especially egg size: patterns, models, quality of evidence and interpretations. American Zoologist, 36, 216–236. Biazik, J.M., Thompson, M.B. & Murphy, C.R. (2009) Lysosomal and alkaline phosphatase activity indicate macromolecule transport across the uterine epithelium in two viviparous skinks with complex placenta. Journal of Experimental Zoology Part B: Molecular and Developmental Evolution, 312, 817–826. Biazik, J.M., Thompson, M.B. & Murphy, C.R. (2010) Paracellular and transcellular transport across the squamate uterine epithelium. Herpetological Conservation and Biology, 5, 257–262.

110

Bingham, B.L., Giles, K. & Jaeckle, W.B. (2004) Variability in broods of the sea star Leptasterias aequalis. Canadian Journal of Zoology, 82, 457–463. Blackburn, D.G. (1992) Convergent evolution of viviparity, matrotrophy, and specializations for fetal nutrition in reptiles and other vertebrates. American Zoologist, 32, 313–321. Blackburn, D.G. (1999) Viviparity and oviparity: evolution and reproductive strategies. Encyclopedia of reproduction. (ed. N.J.D. E. Knobil), pp. 994–1003. Academic Press, San Diego, Calif. Blackburn, D.G. (2000) Classification of the reproductive patterns of amniotes. Herpetological Monographs, 14, 371–377. Blackburn, D.G. (2005) Evolutionary origins of viviparity in fishes. Viviparity in Fishes (eds H. Grier & M. Uribe), pp. 303– 317. FL: New Life Publications, Homestead. Blackburn, D.G. (2015) Evolution of vertebrate viviparity and specializations for fetal nutrition: a quantitative and qualitative analysis. Journal of Morphology, 276, 961– 990. Blackburn, D.G., Evans, H.E. & Vitt, L.J. (1985) The evolution of fetal nutritional adaptations. Fortschritte Der Zoologie, 30, 437–439. Blackburn, D.G. & Starck, J.M. (2015) Morphological specializations for fetal maintenance in viviparous vertebrates: an introduction and historical retrospective. Journal of Morphology 276, E1–16. Blott, E.J. & Griffiths, G.M. (2002) Secretory lysosomes. Nature Reviews Molecular Cell Biology, 3, 122–131. Bosch, I. & Slattery, M. (1999) Costs of extended brood protection in the Antarctic sea star, Neosmilaster georgianus (Echinodermata: Asteroidea). Marine Biology, 134, 449– 459. Byrne, M. (1988) Evidence for endocytotic incorporation of nutrients from the haemal sinus by the oocytes of the brittlestar Ophiolepis paucispina. Echinoderm Biology. Proceedings of the Sixth International Echinoderm Conference (ed. R. Burke, Lambert, P. V., Parsley, R. L.), pp. 557–563. Victoria, 23–28 August 1987, Balkema, Rotterdam. Byrne, M. (1989) Ultrastructure of the ovary and oogenesis in the ovoviviparous ophiuroid Ophiolepis paucispina (Echinodermata). Biological Bulletin, 176, 79–95. Byrne, M. (1991) Reproduction, development and population biology of the Caribbean ophiuroid Ophionereis olivacea, a protandric hermaphrodite that broods its young. Marine Biology, 111, 387–399. 111

Byrne, M. (1992) Reproduction of sympatric populations of Patiriella gunnii, P. calcar and P. exigua in New South Wales, asterinid seastars with direct development. Marine Biology, 114, 297–316. Byrne, M. (1996) Viviparity and intragonadal cannibalism in the diminutive sea stars Patiriella vivipara and P parvivipara (family Asterinidae). Marine Biology, 125, 551–567. Byrne, M. (2005) Viviparity in the sea star Cryptasterina hystera (Asterinidae)- conserved and modified features in reproduction and development. Biological Bulletin, 208, 81– 91. Byrne, M. (2006) Life history diversity and evolution in the Asterinidae. Integrative and Comparative Biology, 46, 243–254. Byrne, M. (2013) Asteroid evolutionary developmental biology and ecology. Starfish: biology and ecology of the Asteroidea (ed. J.M. Lawrence), pp. 51–58. The Johns Hopkins University Press, Baltimore, Maryland. Byrne, M. & Cerra, A. (1996) Evolution of intragonadal development in the diminutive asterinid sea stars Patiriella vivipara and P. parvivipara with an overview of development in the Asterinidae. Biological Bulletin, 191, 17–26. Byrne, M. & Cerra, A. (2000) Lipid dynamics in the embryos of Patiriella species (Asteroidea) with divergent modes of development. Development, Growth & Differentiation, 42, 79–86. Byrne, M., Cerra, A. & Villinski, J.T. (1999) Oogenic strategies in the evolution of development in Patiriella (Echinodermata: Asteroidea). Invertebrate Reproduction and Development, 36, 195–202. Byrne, M., Hart, M.W., Cerra, A. & Cisternas, P. (2003) Reproduction and larval morphology of broadcasting and viviparous species in the Cryptasterina species complex. Biological Bulletin, 205, 285–294. Byrne, M. & Sewell, M.A. (2019) Evolution of maternal lipid provisioning strategies in echinoids with non-feeding larvae: selection for high-quality juveniles. Marine Ecology Progress Series, 616, 95-106. Byrne, M. & Walker, S.J. (2007) Distribution and reproduction of intertidal species of Aquilonastra and Cryptasterina (Asterinidae) from One Tree Reef, Southern Great Barrier Reef. Bulletin of Marine Science, 81, 209–218.

112

Cable, J. & Tinsley, R.C. (1991) Intra-uterine larval development of the polystomatid monogeneans, Pseudodiplorchis americanus and Neodiplorchis scaphiopodis. Parasitology, 103 Pt 2, 253–266. Cameron, H., Monro, K., Malerba, M., Munch, S. & Marshall, D. (2016) Why do larger mothers produce larger offspring? a test of classic theory. Ecology, 97, 3452–3459. Cameron, H., Monro, K. & Marshall, D.J. (2017) Should mothers provision their offspring equally? a manipulative field test. Ecology Letters, 20, 1025–1033. Carrasco, S.A. & Phillips, N.E. (2014) Encapsulation and development of three New Zealand neogastropods with contrasting embryo packaging and maternal provisioning. New Zealand Journal of Zoology, 41, 171–186. Chaparro, O., Oyarzun, R., Vergara, A. & Thompson, R. (1999) Energy investment in nurse eggs and egg capsules in Crepidula dilatata Lamarck (Gastropoda, Calyptraeidae) and its influence on the hatching size of the juvenile. Journal of Experimental Marine Biology and Ecology, 232, 261–274. Chia, F.-S. (1966) Brooding behavior of a six-rayed starfish, Leptasterias hexactis. The Biological Bulletin, 130, 304–315. Chia, F.-S. (1974) Classification and adaptive significance of developmental patterns in marine invertebrates. Thalassia Jugosl, 10, 121–130. Chia, F.S. (1968) Some observations on the development and cyclic changes of the oocytes in a brooding starfish, Leptasterias hexactis. Journal of Zoology, 154, 453–461. Chia, F.S. & Walker, C.W. (1991) Echinodermata: Asteroidea. Reproduction of Marine Invertebrates. (eds A.C. Giese, J.S. Pearse & V.B. Pearse), pp. 301–353. Boxwood Press, Pacific Grove, California. Christiansen, F. & Fenchel, T. (1979) Evolution of marine invertebrate reproductive patterns. Theoretical Population Biology, 16, 267–282. Clark, H.L. (1910) The development of an apodous holothurian (Chiridota rotifera). Journal of Experimental Zoology, 9, 497–516. Clarke, A. (1992) Reproduction in the cold - Thorson revisited. Invertebrate Reproduction and Development, 22, 175–184. Clarke, M.R.B. (1980) The reduced major axis of a bivariate sample. Biometrika, 67, 441– 446. Collin, R. & Spangler, A. (2012) Impacts of adelphophagic development on variation in offspring size, duration of development, and temperature-mediated plasticity. Biological Bulletin, 223, 268–277. 113

Crespi, B. & Semeniuk, C. (2004) Parent-offspring conflict in the evolution of vertebrate reproductive mode. American Naturalist, 163, 635–653. Cubillos, V.M., Chaparro, O.R., Montiel, Y.A. & Veliz, D. (2007) Unusual source of food: impact of dead siblings on encapsulated embryo development of Crepipatella fecunda (Gastropoda : Calyptraeidae). Marine and Freshwater Research, 58, 1152–1161. Daly, J. (1972) The maturation and breeding biology of Harmothoe imbricata (Polychaeta: Polynoidae). Marine Biology, 12, 53–66. Dartnall, A.J., Byrne, M., Collins, J. & Hart, M.W. (2003) A new viviparous species of asterinid (Echinodermata, Asteroidea, Asterinidae) and a new genus to accommodate the species of pantropical exiguoid sea stars. Zootaxa, 359, 1–14. David, B. & Mooi, R. (1990) An echinoid that “gives birth”: morphology and systematics of a new Antarctic species, Urechinus mortenseni (Echinodermata, Holasteroida). Zoomorphology, 110, 75–89. Dulvy, N.K. & Reynolds, J.D. (1997) Evolutionary transitions among egg-laying, live- bearing and maternal inputs in sharks and rays. Proceedings of the Royal Society B- Biological Sciences, 264, 1309–1315. Eckelbarger, K.J. & Young, C.M. (1992) Ovarian ultrastructure and vitellogenesis in ten species of shallow-water and bathyal sea cucumbers (Echinodermata: Holothuroidea). Journal of the Marine Biological Association of the United Kingdom, 72, 759–781. Eernisse, D.J. (1988) Reproductive patterns in six species of Lepidochitona (Mollusca: Polyplacophora) from the Pacific Coast of North America. Biological Bulletin, 174, 287–302. Einum, S. & Fleming, I.A. (2004) Environmental unpredictability and offspring size: conservative versus diversified bet-hedging. Evolutionary Ecology Research, 6, 443– 455. Emlet, R.B. & Hoegh‐Guldberg, O. (1997) Effects of egg size on postlarval performance: experimental evidence from a sea urchin. Evolution, 51, 141–152. Fell, H.B. (1946) The embryology of the viviparous ophiuroid Amphipholis squamata Delle Chiaje. Transactions of the Royal Society of New Zealand, 419–464. Ferguson, J.C. (1964) Nutrient transport in starfish. I. Properties of the coelomic fluid. Biological Bulletin, 126, 33–53. Fernandez, M., Bock, C. & Portner, H.O. (2000) The cost of being a caring mother: the ignored factor in the reproduction of marine invertebrates. Ecology Letters, 3, 487– 494. 114

Fernandez, M., Calderon, R., Cifuentes, M. & Pappalardo, P. (2006) Brooding behaviour and cost of brooding in small body size brachyuran crabs. Marine Ecology Progress Series, 309, 213–220. Fernández, M., Ruiz-Tagle, N., Cifuentes, S., Pörtner, H.-O. & Arntz, W. (2003) Oxygen- dependent asynchrony of embryonic development in embryo masses of brachyuran crabs. Marine Biology, 142, 559–565. Ferrand, J. (1980) Hydrolases lysosomiques en vitellus des ovocytes d'Asterina gibbosa et d'Asterias rubens (Echinodermes, Asterides). Echinoderms: present and past. Balkema, Rotterdam, 393–397. Ferrand, J.G. (1984) Lysosomes and yolk formation in the oocytes of Asterina gibbosa (Echinodermata, Asteroidea). International Journal of Invertebrate Reproduction and Development, 7, 143–147. Fitch, H.S. (1970) Reproductive cycles in lizards and snakes. Miscellaneous Publications of the University of Kansas Museum of Natural History, 52, 1–247. Fontaine, A. & Chia, F.-S. (1968) Echinoderms: an autoradiographic study of assimilation of dissolved organic molecules. Science, 161, 1153–1155. Fox, C.W. & Czesak, M.E. (2000) Evolutionary ecology of progeny size in arthropods. Annual Review of Entomology, 45, 341–369. Frick, J.E. (1998) Evidence of matrotrophy in the viviparous holothuroid echinoderm Synaptula hydriformis. Invertebrate Biology, 117, 169–179. Frick, J.E., Ruppert, E.E. & Wourms, J.P. (1996) Morphology of the ovotestis of Synaptula hydriformis (Holothuroidea, Apoda): an evolutionary model of oogenesis and the origin of egg polarity in echinoderms. Invertebrate Biology, 115, 46–66. Galley, E.A., Tyler, P.A., Clarke, A. & Smith, C.R. (2005) Reproductive biology and biochemical composition of the brooding echinoid Amphipneustes lorioli on the Antarctic continental shelf. Marine Biology, 148, 59–71. George, S.B. (1996) Echinoderm egg and larval quality as a function of adult nutritional state. Oceanologica Acta, 19, 297–308. Gillespie, J.M. & McClintock, J.B. (2007) Brooding in echinoderms: how can modern experimental techniques add to our historical perspective? Journal of Experimental Marine Biology and Ecology, 342, 191–201. Gilmore, R.G. (1993) Reproductive biology of lamnoid sharks. Environmental Biology of Fishes, 38, 95–114.

115

Griffith, O.W., Blackburn, D.G., Brandley, M.C., Van Dyke, J.U., Whittington, C.M. & Thompson, M.B. (2015) Ancestral state reconstructions require biological evidence to test evolutionary hypotheses: a case study examining the evolution of reproductive mode in squamate reptiles. Journal of Experimental Zoology Part B: Molecular and Developmental Evolution, 324, 493–503. Guimaraes, J.H. (1977) A systematic revision of the Mesembrinellidae, stat. Nov. (Diptera, Cyclorrhapha). Arquivos de Zoologia (Sao Paulo), 29, 1–109. Gustafson, T. & Hasselberg, I. (1950) Alkaline phosphatase activity in developing sea urchin eggs. Experimental Cell Research, 1, 371–375. Hansen, B. (1968) Brood-protection in a deep-sea Holothurian, Oneirophanta mutabills Théel. Nature, 217, 1062–1063. Hansen, B. (1975) Systematics and biology of the deep-sea Holothurians: Part 1: Elasipoda: Danish and English Summary. Scandinavian Science Press. Harrington, F.E. & Ozaki, H. (1986) The major yolk glycoprotein precursor in echinoids is secreted by coelomocytes into the coelomic plasma. Cell Differentiation, 19, 51–57. Hart, M., Byrne, M. & Johnson, S. (2003) Cryptic species and modes of development in Patiriella pseudoexigua. Journal of the Marine Biological Association of the United Kingdom, 83, 1109–1116. Hart, M.W. (2002) Life history evolution and comparative developmental biology of echinoderms. Evolution & Development, 4, 62–71. Hart, M.W., Byrne, M. & Smith, M.J. (1997) Molecular phylogenetic analysis of life-history evolution in asterinid starfish. Evolution, 51, 1848–1861. Heath, D.J. (1977) Simultaneous hermaphroditism; cost and benefit. Journal of Theoretical Biology, 64, 363–373. Heath, D.J. (1979) Brooding and the evolution of hermaphroditism. Journal of Theoretical Biology, 81, 151–155. Hendler, G. (1975) Adaptational significance of the patterns of ophiuroid development. American Zoologist, 15, 691–715. Hendler, G. (1979) Sex-reversal and viviparity in Ophiolepis kieri, n. sp., with notes on viviparous brittlestars from the Caribbean (Echinodermata: Ophiuroidea). Proceedings of the Biological Society of Washington, 92, 783–795. Hendler, G. & Tran, L.U. (2001) Reproductive biology of a deep-sea brittle star Amphiura carchara (Echinodermata : Ophiuroidea). Marine Biology, 138, 113–123.

116

Hendry, A.P., Day, T. & Cooper, A.B. (2001) Optimal size and number of propagules: allowance for discrete stages and effects of maternal size on reproductive output and offspring fitness. American Naturalist, 157, 387–407. Hess, H.C. (1993) The evolution of parental care in brooding spirorbid polychaetes - the effect of scaling constraints. American Naturalist, 141, 577–596. Hsiao, S. & Fujii, W. (1963) Early ontogenic changes in the concentration of alkaline phosphatase in Hawaiian sea urchins. Experimental Cell Research, 32, 217–231. Hughes, R. & Carrick, F. (1978) Reproduction in female monotremes. Australian Zoologist, 20, 233–253. Ilano, A.S., Fujinaga, K. & Nakao, S. (2004) Mating, development and effects of female size on offspring number and size in the neogastropod Buccinum isaotakii (Kira, 1959). Journal of Molluscan Studies, 70, 277–282. Kabat, A.R. (1985) The allometry of brooding in Transennella tantilla (Gould) (Mollusca, Bivalvia). Journal of Experimental Marine Biology and Ecology, 91, 271–279. Kalinka, A.T. (2015) How did viviparity originate and evolve? of conflict, co‐option, and cryptic choice. BioEssays, 37, 721–731. Kamel, S.J., Grosberg, R.K. & Marshall, D.J. (2010a) Family conflicts in the sea. Trends in Ecology & Evolution, 25, 442–449. Kamel, S.J., Oyarzun, F.X. & Grosberg, R.K. (2010b) Reproductive biology, family conflict, and size of offspring in marine invertebrates. Integrative and Comparative Biology, 50, 619–629. Kamel, S.J. & Williams, P.D. (2017) Resource exploitation and relatedness: implications for offspring size variation within broods. Oikos, 126, 1219–1226. Keever, C.C., Puritz, J.B., Addison, J.A., Byrne, M., Grosberg, R.K., Toonen, R.J. & Hart, M.W. (2013) Shallow gene pools in the high intertidal: extreme loss of genetic diversity in viviparous sea stars (Parvulastra). Biology Letters, 9, 20130551. Keough, M.J. & Dartnall, A.J. (1978) A new species of viviparous asterinid asteroid from Eyre Peninsula South Australia. Records of the South Australian Museum (Adelaide), 17, 406–416. Khan, M.S.R., Whittington, C.M., Thompson, M.B. & Byrne, M. (2019a) Intragonadal incubation of progeny in three viviparous asterinid sea stars that differ in offspring provisioning, lecithotrophy vs matrotrophy. Marine Biology, 166, 81.

117

Khan, M.S.R., Whittington, C.M., Thompson, M.B. & Byrne, M. (2019b) Arrangement and size variation of intra-gonadal offspring in a viviparous asterinid sea star. Zoosymposia, 15, 71–82. Kishimoto, T., Usui, N. & Kanatani, H. (1984) Breakdown of starfish ovarian follicle induced by maturation-promoting factor. Developmental Biology, 101, 28–34. Komatsu, M., Kano, Y.T. & Oguro, C. (1990) Development of a true ovoviviparous sea star, Asterina pseudoexigua pacifica Hayashi. Biological Bulletin, 179, 254–263. Landschoff, J. & Griffiths, C.L. (2015) Three-dimensional visualisation of brooding behaviour in two distantly related brittle stars from South African waters. African Journal of Marine Science, 37, 533–541. Lardies, M.A. & Fernandez, M. (2002) Effect of oxygen availability in determining clutch size in Acanthina monodon. Marine Ecology Progress Series, 239, 139–146. Lawrence, J.M. & Herrera, J. (2000) Stress and deviant reproduction in echinoderms. Zoological Studies, 39, 151–171. Lesoway, M.P., Abouheif, E. & Collin, R. (2014) The development of viable and nutritive embryos in the direct developing gastropod Crepidula navicella. International Journal of Developmental Biology, 58, 601–611. Levin, L.F. & Bridges, T.S. (1995) Pattern and diversity in reproduction and development. Ecology of marine invertebrate larvae (ed. L.R. McEdward), pp. 1–48. CRC Press, Boca Raton/London/New York/Washington, DC. Lim, J.N., Senior, A.M. & Nakagawa, S. (2014) Heterogeneity in individual quality and reproductive trade‐offs within species. Evolution, 68, 2306–2318. Liversage, K. & Byrne, M. (2018) A note on life-history traits and conservation concerns for viviparous Australian sea stars (Parvulastra parvivipara and P. vivipara). Research Ideas and Outcomes, 4, e29766. Mah, C.L. (2006) A new species of Xyloplax (Echinodermata: Asteroidea: Concentricycloidea) from the northeast Pacific: comparative morphology and a reassessment of phylogeny. Invertebrate Biology, 125, 136–153. Manahan, D.T. (1990) Adaptations by invertebrate larvae for nutrient acquisition from seawater. American Zoologist, 30, 147–160. Manahan, D.T., Davis, J.P. & Stephens, G.C. (1983) Bacteria-free sea urchin larvae: selective uptake of neutral amino acids from seawater. Science, 220, 204–206. Marshall, D.J., Bolton, T.F. & Keough, M.J. (2003) Offspring size affects the post‐ metamorphic performance of a colonial marine invertebrate. Ecology, 84, 3131–3137. 118

Marshall, D.J., Bonduriansky, R. & Bussiere, L.F. (2008) Offspring size variation within broods as a bet-hedging strategy in unpredictable environments. Ecology, 89, 2506– 2517. Marshall, D.J., Heppell, S.S., Munch, S.B. & Warner, R.R. (2010) The relationship between maternal phenotype and offspring quality: do older mothers really produce the best offspring? Ecology, 91, 2862–2873. Marshall, D.J. & Keough, M.J. (2006) Complex life cycles and offspring provisioning in marine invertebrates. Integrative and Comparative Biology, 46, 643–651. Marshall, D.J. & Keough, M.J. (2008) The evolutionary ecology of offspring size in marine invertebrates. Advances in Marine Biology (ed. D.W. Sims), pp. 1–60. Elsevier Academic Press Inc, San Diego. Marshall, D.J., Krug, P.J., Kupriyanova, E.K., Byrne, M. & Emlet, R.B. (2012) The biogeography of marine invertebrate life histories. Annual Review of Ecology, Evolution, and Systematics (ed. D.J. Futuyma), pp. 97–114. Annual Reviews, Palo Alto. Materia, C.J., Monagle, J.F. & O’Loughlin, P.M. (1991) Seasonal coelomic brooding in southern Australian cucumariids (Echinodermata, Holothurioidea). Biology of Echinodermata, 301–307. McArdle, B.H. (1988) The structural relationship - regression in biology. Canadian Journal of Zoology, 66, 2329–2339. McClary, D.J. & Mladenov, P.V. (1990) Brooding biology of the sea star Pteraster militaris (OF Müller): energetic and histological evidence for nutrient translocation to brooded juveniles. Journal of Experimental Marine Biology and Ecology, 142, 183–199. McEdward, L.R. & Janies, D.A. (1993) Life cycle evolution in asteroids: what is a larva? Biological Bulletin, 184, 255–268. McEdward, L.R. & Janies, D.A. (1997) Relationships among development, ecology, and morphology in the evolution of echinoderm larvae and life cycles. Biological Journal of the Linnean Society, 60, 381–400. McEdward, L.R. & Miner, B.G. (2001) Larval and life-cycle patterns in echinoderms. Canadian Journal of Zoology, 79, 1125–1170. Menge, B.A. (1975) Brood or broadcast - adaptive significance of different reproductive strategies in two intertidal sea stars Leptasterias hexactis and Pisaster ochraceus. Marine Biology, 31, 87–100.

119

Mercier, A., Sun, Z., Parrish, C.C. & Hamel, J.F. (2016) Remarkable shifts in offspring provisioning during gestation in a live-bearing cnidarian. PLoS One, 11, e0154051. Messing, C.G. (1984) Brooding and paedomorphosis in the deep-water feather star Comatilia iridometriformis (Echinodermata: Crinoidea). Marine Biology, 80, 83–91. Mileikovsky, S. (1971) Types of larval development in marine bottom invertebrates, their distribution and ecological significance: a re-evaluation. Marine Biology, 10, 193– 213. Miloslavich, P. & Dufresne, L. (1994) Development and effect of female size on egg and juvenile production in the neogastropod Buccinum cyaneum from the Saguenay Fjord. Canadian Journal of Fisheries and Aquatic Sciences, 51, 2866–2872. Moran, A.L. & Emlet, R.B. (2001) Offspring size and performance in variable environments: field studies on a marine snail. Ecology, 82, 1597–1612. Morrison, K.R., Ngo, V., Cardullo, R.A. & Reznick, D.N. (2017) How fish eggs are preadapted for the evolution of matrotrophy. Proceedings of the Royal Society B: Biological Sciences, 284. Mortensen, T. (1920) On hermaphroditism in viviparous ophiurids. Acta Zoologica, 1, 1–19. Mousseau, T.A. & Fox, C.W. (1998) The adaptive significance of maternal effects. Trends in Ecology & Evolution, 13, 403–407. Musick, J.A. & Ellis, J.K. (2005) Reproductive evolution of Chondrichthyans. Reproductive biology and phylogeny of Chondrichthyes: Sharks, Batoids and Chimaeras (ed. W.C. Hamlett), pp. 45–79. CRC Press, USA. Nasution, S., Roberts, D., Farnsworth, K., Parker, G.A. & Elwood, R.W. (2010) Maternal effects on offspring size and packaging constraints in the whelk. Journal of Zoology, 281, 112–117. Naylor, J., Taylor, E. & Bennett, D. (1999) Oxygen uptake of developing eggs of Cancer pagurus (Crustacea: Decapoda: Cancridae) and consequent behaviour of the ovigerous females. Journal of the Marine Biological Association of the United Kingdom, 79, 305–315. Okazaki, K. (1956) Skeleton formation of sea urchin larvae. I. Effect of Ca concentration of the medium. The Biological Bulletin, 110, 320–333. Olivera-Tlahuel, C., Ossip-Klein, A.G., Espinosa-Perez, H.S. & Zuniga-Vega, J.J. (2015) Have superfetation and matrotrophy facilitated the evolution of larger offspring in poeciliid fishes? Biological Journal of the Linnean Society, 116, 787–804.

120

Orlando, L., Pereyra, I., Silveira, S., Paesch, L., Oddone, M.C. & Norbis, W. (2015) Determination of limited histotrophy as the reproductive mode in Mustelus schmitti Springer, 1939 (Chondrichthyes: Triakidae): analysis of intrauterine growth of embryos. Neotropical Ichthyology, 13, 699–706. Ostrovsky, A.N. (2013a) Evolution of sexual reproduction in marine invertebrates: example of gymnolaemate bryozoans. Springer Verlag: Dordrecht Heidelberg, NewYork, London. Ostrovsky, A.N. (2013b) From incipient to substantial: evolution of placentotrophy in a phylum of aquatic colonial invertebrates. Evolution, 67, 1368–1382. Ostrovsky, A.N., Gordon, D.P. & Lidgard, S. (2009) Independent evolution of matrotrophy in the major classes of Bryozoa: transitions among reproductive patterns and their ecological background. Marine Ecology Progress Series, 378, 113–124. Ostrovsky, A.N., Lidgard, S., Gordon, D.P., Schwaha, T., Genikhovich, G. & Ereskovsky, A.V. (2015) Matrotrophy and placentation in invertebrates: a new paradigm. Biological Reviews of the Cambridge Philosophical Society, 91, 673–711. Oyarzun, F.X. & Brante, A. (2014) The effects of nurse eggs and sibling interactions on the larval development of the poecilogonous annelid Boccardia proboscidea (Spionidae). Invertebrate Biology, 133, 340–353. Parker, G.A. & Begon, M. (1986) Optimal egg size and clutch size - effects of environment and maternal phenotype. American Naturalist, 128, 573–592. Parker, G.A., Royle, N.J. & Hartley, I.R. (2002) Intrafamilial conflict and parental investment: a synthesis. Philosophical Transactions of the Royal Society B: Biological Sciences, 357, 295–307. Pearse, J. (1965) Reproductive periodicities in several contrasting populations of Odontaster validus Koehler, a common Antarctic asteroid. Biology of the Antarctic seas II, 5, 39– 85. Pearse, J.S., Eernisse, D.J., Pearse, V.B. & Beauchamp, K.A. (1986) Photoperiodic regulation of gametogenesis in sea stars, with evidence for an annual calendar independent of fixed daylength. American Zoologist, 26, 417–431. Pearse, J.S. & Mcclintock, J.B. (1990) A comparison of reproduction by the brooding spatangoid echinoids Abatus shackletoni and A. nimrodi in McMurdo Sound, Antarctica. Invertebrate Reproduction & Development, 17, 181–191. Peters, R.H. (1986) The ecological implications of body size. Cambridge University Press.

121

Pfohl, R.J. (1975) Alkaline phosphatase of sea urchin embryos: chromatographic and electrophoretic characterization. Developmental Biology, 44, 333–345. Pollux, B.J.A., Pires, M.N., Banet, A.I. & Reznick, D.N. (2009) Evolution of placentas in the fish family Poeciliidae: an empirical study of macroevolution. Annual Review of Ecology Evolution and Systematics, pp. 271–289. Annual Reviews, Palo Alto. Pollux, B.J.A. & Reznick, D.N. (2011) Matrotrophy limits a female's ability to adaptively adjust offspring size and fecundity in fluctuating environments. Functional Ecology, 25, 747–756. Poulin, É. & Féral, J.P. (1996) Why are there so many species of brooding Antarctic echinoids? Evolution, 50, 820–830. Prestedge, G.K. (1998) The distribution and biology of Patiriella vivipara (Echinodermata: Asteroidea: Asterinidae) a sea star endemic to southeast Tasmania. Records of the Australian Museum, 50, 161–170. Prowse, T.A. & Byrne, M. (2012) Evolution of yolk protein genes in the Echinodermata. Evolution & Development, 14, 139–151. Prowse, T.A.A., Sewell, M.A. & Byrne, M. (2008) Fuels for development: evolution of maternal provisioning in asterinid sea stars. Marine Biology, 153, 337–349. Puritz, J.B., Keever, C.C., Addison, J.A., Byrne, M., Hart, M.W., Grosberg, R.K. & Toonen, R.J. (2012) Extraordinarily rapid life-history divergence between Cryptasterina sea star species. Proceedings of the Royal Society B: Biological Sciences, 279, 3914– 3922. Raff, R.A. & Byrne, M. (2006) The active evolutionary lives of echinoderm larvae. Heredity (Edinb), 97, 244–252. Reznick, D.N., Mateos, M. & Springer, M.S. (2002) Independent origins and rapid evolution of the placenta in the fish genus Poeciliopsis. Science, 298, 1018–1020. Riesch, R., Plath, M., Schlupp, I. & Marsh-Matthews, E. (2010) Matrotrophy in the cave molly: an unexpected provisioning strategy in an extreme environment. Evolutionary Ecology, 24, 789–801. Rivest, B.R. (1983) Development and the influence of nurse egg allotment on hatching size in Searlesia dira (Reeve, 1846) (Prosobranchia, Buccinidae). Journal of Experimental Marine Biology and Ecology, 69, 217–241. Roediger, L.M. (2011) Population and reproductive ecology of the direct-developing sea stars Parvulastra parvivipara and Cryptasterina hystera. PhD Thesis, Flinders University, Adelaide. 122

Roediger, L.M. & Bolton, T.F. (2008) Abundance and distribution of South Australia's endemic sea star, Parvulastra parvivipara (Asteroidea: Asterinidae). Marine and Freshwater Research, 59, 205–213. Rollinson, N. & Rowe, L. (2016) The positive correlation between maternal size and offspring size: fitting pieces of a life‐history puzzle. Biological Reviews, 91, 1134– 1148. Rowe, F.W., Baker, A.N. & Clark, H.E. (1988) The morphology, development and taxonomic status of Xyloplax Baker, Rowe and Clark (1986)(Echinodermata: Concentricycloidea), with the description of a new species. Proceedings of the Royal Society of London B: Biological Sciences, 233, 431–459. Royle, N.J., Smiseth, P.T. & Kölliker, M. (2012) The evolution of parental care. pp. 376. Oxford University Press. Sakai, S. & Harada, Y. (2001) Why do large mothers produce large offspring? theory and a test. American Naturalist, 157, 348–359. Schinner, G.O. & McClintock, J.B. (1993) Structural characteristics of marsupial brood pouches of the Antarctic sea urchins Abatus nimrodi and Abatus shackletoni (Echinoidea: Spatangoida). Journal of Morphology, 216, 79–93. Schoener, A. (1972) Fecundity and possible mode of development of some deep‐sea ophiuroids 1. Limnology and Oceanography, 17, 193–199. Schoenmakers, H., Colenbrander, P., Peute, J. & Van Oordt, P. (1981) Anatomy of the ovaries of the starfish Asterias rubens (Echinodermata). Cell and Tissue Research, 217, 577–597. Schrader, M. & Travis, J. (2009) Do embryos influence maternal investment? evaluating maternal‐fetal coadaptation and the potential for parent‐offspring conflict in a placental fish. Evolution, 63, 2805–2815. Schrader, M. & Travis, J. (2012) Variation in offspring size with birth order in placental fish: a role for asymmetric sibling competition? Evolution, 66, 272–279. Scrimshaw, N.S. (1944) Superfetation in poeciliid fishes. Copeia, 1944, 180–183. Sewell, M. & Chia, F.-S. (1994) Reproduction of the intraovarian brooding apodid Leptosynapta clarki (Echinodermata: Holothuroidea) in British Columbia. Marine Biology, 121, 285–300.

123

Sewell, M.A. (1993) Reproductive cycle, sex change, and mortality during brooding of a viviparous sea cucumber Leptosynapta clarki. PhD Thesis, University of Alberta, Edmonton. Sewell, M.A. (1994) Small size, brooding, and protandry in the apodid sea cucumber Leptosynapta clarki. Biological Bulletin, 187, 112–123. Sewell, M.A., Koss, R.O.N., Turner, A. & Chia, F.-S. (2006) Evidence for matrotrophy in the viviparous sea cucumber Leptosynapta clarki: a role for the genital haemal sinus? Invertebrate Reproduction and Development, 49, 225–236. Shine, R. (1978) Propagule size and parental care: the “safe harbor” hypothesis. Journal of Theoretical Biology, 75, 417–424. Shine, R. (1983) Reptilian viviparity in cold climates: testing the assumptions of an evolutionary hypothesis. Oecologia, 57, 397–405. Shirai, H., Yoshimoto, Y. & Kanatani, H. (1981) Mechanism of starfish spawning. IV. Tension generation in the ovarian wall by 1-methyladenine at the time of spawning. Biological Bulletin, 161, 172–179. Smiley, S. (1988) The dynamics of oogenesis and the annual ovarian cycle of Stichopus californicus (Echinodermata: Holothuroidea). Biological Bulletin, 175, 79–93. Smiley, S. (1990) A review of echinoderm oogenesis. Journal of Electron Microscopy Technique, 16, 93–114. Smith, C. & Reay, P. (1991) Cannibalism in teleost fish. Reviews in Fish Biology and Fisheries, 1, 41–64. Smith, C.C. & Fretwell, S.D. (1974) The optimal balance between size and number of offspring. American Naturalist, 108, 499–506. Spight, T.M. (1975) Factors extending gastropod embryonic development and their selective cost. Oecologia, 21, 1–16. Steer, M.A., Moltschaniwskyj, M.A., Nichols, D.S. & Miller, M. (2004) The role of temperature and maternal ration in embryo survival: using the dumpling squid Euprymna tasmanica as a model. Journal of Experimental Marine Biology and Ecology, 307, 73–89. Stewart, J.R. (1989) Facultative placentotrophy and the evolution of squamate placentation: quality of eggs and neonates in Virginia striatula. The American Naturalist, 133, 111– 137. Strathmann, R.R. (1978) The evolution and loss of feeding larval stages of marine invertebrates. Evolution, 32, 894–906. 124

Strathmann, R.R. (1985) Feeding and nonfeeding larval development and life-history evolution in marine invertebrates. Annual Review of Ecology and Systematics, 16, 339–361. Strathmann, R.R. (1995) Peculiar constraints on life histories imposed by protective or nutritive devices for embryos. American Zoologist, 35, 426–433. Strathmann, R.R. & Chaffee, C. (1984) Constraints on egg masses. 2. Effect of spacing, size, and number of eggs on ventilation of masses of embryos in jelly, adherent groups, or thin-walled capsules. Journal of Experimental Marine Biology and Ecology, 84, 85– 93. Strathmann, R.R. & Strathmann, M.F. (1982) The relationship between adult size and brooding in marine invertebrates. American Naturalist, 119, 91–101. Strathmann, R.R. & Strathmann, M.F. (1995) Oxygen-supply and limits on aggregation of embryos. Journal of the Marine Biological Association of the United Kingdom, 75, 413–428. Strathmann, R.R., Strathmann, M.F. & Emson, R.H. (1984) Does limited brood capacity link adult size, brooding, and simultaneous hermaphroditism? a test with the starfish Asterina phylactica. American Naturalist, 123, 796–818. Sun, Z., Hamel, J.F. & Mercier, A. (2012) Marked shifts in offspring size elicited by frequent fusion among siblings in an internally brooding marine invertebrate. American Naturalist, 180, E151–160. Swift, D.G., Dunning, L.T., Igea, J., Brooks, E.J., Jones, C.S., Noble, L.R., Ciezarek, A., Humble, E. & Savolainen, V. (2016) Evidence of positive selection associated with placental loss in tiger sharks. BMC Evolutionary Biology, 16, 126. Thompson, M.B. & Speake, B.K. (2006) A review of the evolution of viviparity in lizards: structure, function and physiology of the placenta. Journal of Comparative Physiology B Biochemical Systemic and Environmental Physiology, 176, 179–189. Thompson, M.B., Stewart, J.R. & Speake, B.K. (2000) Comparison of nutrient transport across the placenta of lizards differing in placental complexity. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 127, 469– 479. Travis, J., Farr, J.A., Henrich, S. & Cheong, R.T. (1987) Testing theories of clutch overlap with the reproductive ecology of Heterandria formosa. Ecology, 68, 611–623. Trexler, J.C. (1997) Resource availability and plasticity in offspring provisioning: embryo nourishment in sailfin mollies. Ecology, 78, 1370–1381. 125

Trexler, J.C. & DeAngelis, D.L. (2003) Resource allocation in offspring provisioning: an evaluation of the conditions favoring the evolution of matrotrophy. American Naturalist, 162, 574–585. Trivers, R.L. (1974) Parent-offspring conflict. American Zoologist, 14, 249–264. Trumbo, S.T. (1996) Parental care in invertebrates. Advances in the Study of Behavior, pp. 3– 51. Elsevier. Turner, R.L. & Dearborn, J.H. (1979) Organic and inorganic composition of post- metamorphic growth stages of Ophionotus hexactis (EA Smith)(Echinodermata: Ophiuroidea) during intraovarian incubation. Journal of Experimental Marine Biology and Ecology, 36, 41–51. Valentinsson, D. (2002) Reproductive cycle and maternal effects on offspring size and number in the neogastropod Buccinum undatum (L.). Marine Biology, 140, 1139– 1147. Van Dyke, J.U., Brandley, M.C. & Thompson, M.B. (2014) The evolution of viviparity: molecular and genomic data from squamate reptiles advance understanding of live birth in amniotes. Reproduction, 147, R15–26. Vance, R.R. (1973a) More on reproductive strategies in marine benthic invertebrates. American Naturalist, 107, 353–361. Vance, R.R. (1973b) On reproductive strategies in marine benthic invertebrates. American Naturalist, 107, 339–352. Vidavsky, N., Addadi, S., Mahamid, J., Shimoni, E., Ben-Ezra, D., Shpigel, M., Weiner, S. & Addadi, L. (2014) Initial stages of calcium uptake and mineral deposition in sea urchin embryos. Proceedings of the National Academy of Sciences, 111, 39–44. Vidavsky, N., Addadi, S., Schertel, A., Ben-Ezra, D., Shpigel, M., Addadi, L. & Weiner, S. (2016) Calcium transport into the cells of the sea urchin larva in relation to spicule formation. Proceedings of the National Academy of Sciences, 113, 12637–12642. Villinski, J.T., Villinski, J.C., Byrne, M. & Raff, R.A. (2002) Convergent maternal provisioning and life‐history evolution in echinoderms. Evolution, 56, 1764–1775. Voogt, P., Broertjes, J. & Oudejans, R. (1985) Vitellogenesis in sea star: physiological and metabolic implications. Comparative Biochemistry and Physiology Part A: Physiology, 80, 141–147. Walker, C. & Lesser, M. (1989) Nutrition and development of brooded embryos in the brittlestar Amphipholis squamata: do endosymbiotic bacteria play a role? Marine Biology, 103, 519–530. 126

Walker, C.W. (1979) Ultrastructure of the somatic portion of the gonads in asteroids, with emphasis on flagellated‐collar cells and nutrient transport. Journal of Morphology, 162, 127–161. Walker, C.W. (1980) Spermatogenic columns, somatic cells, and the microenvironment of germinal cells in the testes of asteroids. Journal of Morphology, 166, 81–107. Webb, J.K., Shine, R. & Christian, K.A. (2006) The adaptive signigicance of reptilian viviparity in the tropics: testing the maternal manipulation hypothesis. Evolution, 60, 115–122. Whitaker, M. (2008) Calcium signalling in early embryos. Philosophical Transactions of the Royal Society B: Biological Sciences, 363, 1401–1418. Whittington, C.M., Griffith, O.W., Qi, W., Thompson, M.B. & Wilson, A.B. (2015) Seahorse brood pouch transcriptome reveals common genes associated with vertebrate pregnancy. Molecular Biology and Evolution, 32, 3114–3131. Wourms, J.P. (1981) Viviparity - the maternal-fetal relationship in fishes. American Zoologist, 21, 473–515. Wourms, J.P., Grove, B.D. & Lombardi, J. (1988) The maternal-embryonic relationship in viviparous fishes. Fish Physiology, 11, 1–134. Wourms, J.P. & Lombardi, J. (1992) Reflections on the evolution of piscine viviparity. American Zoologist, 32, 276–293. Wray, G. (1995) Evolution of larvae and developmental modes. Ecology of Marine Invertebrate Larvae (ed. L.R. McEdward), pp. 413–447. CRC Press, Boca Raton, London, New York, Washington DC. Wray, G.A. (1996) Parallel evolution of nonfeeding larvae in echinoids. Society of Systematic Biologists. Wright, S.H. & Ahearn, G.A. (1997) Nutrient absorption in invertebrates. Handbook of Physiology, 2, 1137–1205. Wu, Y.-Q., Qu, Y.-F., Wang, X.-J., Gao, J.-F. & Ji, X. (2017) Does the oviparity-viviparity transition alter the partitioning of yolk in embryonic snakes? BMC Evolutionary Biology, 17, 235. Zazueta-Novoa, V., Onorato, T.M., Reyes, G., Oulhen, N. & Wessel, G.M. (2016) Complexity of yolk proteins and their dynamics in the sea star Patiria miniata. Biological Bulletin, 230, 209–219.

127

APPENDIX

128

Zoosymposia 15: 071–082 (2019) ISSN 1178-9905 (print edition) http://www.mapress.com/j/zs/ ZOOSYMPOSIA Copyright © 2019 · Magnolia Press ISSN 1178-9913 (online edition) http://dx.doi.org/10.11646/zoosymposia.15.1.8 http://zoobank.org/urn:lsid:zoobank.org:pub:69CC2980-53BF-4313-A456-FC695B6CBD9F

Arrangement and size variation of intra-gonadal offspring in a viviparous asterinid sea star

M. S. R. KHAN1, C. M. WHITTINGTON1, 2, M. B. THOMPSON1 & M. BYRNE1, 3 1The University of Sydney, School of Life and Environmental Sciences (A08), Sydney, New South Wales, Australia. 2The University of Sydney, Sydney School of Veterinary Sciences, New South Wales, Australia. 3The University of Sydney, School of Medical Sciences (F13), New South Wales, Australia. Corresponding author, M. S. R. Khan: e-mail: [email protected]

Abstract

Sibling competition and developmental asynchrony may greatly influence the arrangement and size of offspring of marine invertebrates that care for their young. In Parvulastra parvivipara, an asterinid sea star that incubates its young in the gonads, sibling cannibalism supports post-metamorphic development. Offspring size varies within (coefficient of variation, CV = 22.6 %) and among (CV = 17.7%) the gonads. Confocal microscopy was used to visualize early embryos and oocytes, and revealed the presence of several developmental stages within individual gonads. The eggs were a mean diameter of 84 µm. The observation of a gastrula at 86 µm smaller than the largest egg observed (134 µm) suggests that terminal egg size varies. The appearance of early embryos surrounded by somatic cells suggests that they may receive nutrients through histotrophy. Sibling competition intensifies once the digestive tract is functional in the tiny juveniles which then start to consume siblings. The arrangement of the offspring in the gonads was observed using micro-computed tomography. The juveniles were oriented with their oral surface facing each other, presumably as a defensive strategy to protect themselves from being eaten. Periodic release of offspring in single or several cohorts indicates continual reproduction. Released and retained juveniles varied in size. It is not known what initiates birth but it may be mediated by sibling competition. Larger adults had a greater allocation to female reproductive output than smaller adults.

Key words: adelphophagy, brooding, viviparity, lecithotrophy, matrotrophy, micro-CT

Introduction

Diverse modes of parental care have evolved independently in marine invertebrate lineages (McClary & Mladenov 1990; Byrne 1991; Trumbo 1996; Ostrovsky et al. 2015). Some species care for their offspring internally (gonads, bursae, gastric pouches), whereas others have their young on or under the body surface (Strathmann & Strathmann 1982; Trumbo 1996; Gillespie & McClintock 2007; Larson 2017). Internal care of offspring is considered to have evolved to provide a ‘safe harbour’ for offspring to enhance their survival by providing protection from predators, UV radiation and pathogens (Shine 1978; Trumbo 1996; Gillespie & McClintock 2007). Incubation chambers also serve important functions such as supplying dissolved oxygen and nutrients to offspring and removing excretory materials (Strathmann & Chaffee 1984; Strathmann 1995; Gillespie & McClintock 2007). As there are limitations in the space and resources that can be provided to offspring (Bernardo 1996; Kamel et al. 2010A), marine invertebrates that care for their young adopt diverse strategies to arrange or package their offspring to maximize reproductive output and offspring fitness. For example, early developmental stages in the sea urchin, Amphipneustes lorioli, are partitioned in the deeper recesses of the brood pouches to reduce overcrowding, an adaptation to limited space (Galley et al. 2005). In the clam, Transennella tantilla, the smallest embryos are positioned dorsally adjacent to the oviduct and, as

Accepted by M. Komatsu: 13 Feb. 2019; published: 21 Oct. 2019 71 they develop, progressively move towards the ventral surface of the gill that facilitates continuous reproduction and hatching (Kabat 1985). The maternal provisioning strategy may play a role in determining the arrangement and size of the offspring in the incubation space (Carrasco & Phillips 2014). For species that care for their young internally, offspring interact with the parent and may compete with siblings for limited resources; these interactions may influence size variation at birth (Schrader & Travis 2009; Kamel et al. 2010B). Parent-offspring conflict and sibling competition are more intense in species that provide extra-embryonic nutrition, a mode of parental care called matrotrophy (Frick 1998; Schrader & Travis 2009; Mercier et al. 2016; Ostrovsky et al. 2015; Kamel & Williams 2017). Extreme sibling competition may result in cannibalism where the progeny ingest eggs (oophagy) or siblings (embryophagy/adelphophagy) (Lesoway et al. 2014; Ostrovsky et al. 2015). Sibling cannibalism greatly increases size variation of offspring in some gastropods, sharks and sea stars (Rivest 1983; Smith & Reay 1991; Byrne 1996; Collin & Spangler 2012). Thus, adelphophagic matrotrophic species provide an opportunity to assess how sibling competition mediates offspring size variation and how the young are arranged in the incubation space. Non-invasive imaging techniques such as micro-computed tomography (µ-CT) has been used to observe offspring inside the brood chamber non-destructively in situ in ophiuroids (Landschoff & Griffiths 2015). The asterinid sea star, Parvulastra parvivipara, exhibits a highly derived mode of matrotrophic incubation, sibling cannibalism, and a high level of offspring size variation (Byrne 1996; Khan et al. 2019). This species care for its young within the gonads to an advanced juvenile stage that varies in size from 0.5–2.5 mm in diameter (Byrne 1996; Khan et al. 2019). Adelphophagy provides significant nourishment for the juveniles, and gonad histology shows that cannibalism may mediate the position of juveniles within the gonads (Byrne 1996). In P. p a r v i v i p ar a , the size differentiation of offspring starts post-metamorphosis due to sibling cannibalism and to asynchronous gamete maturation and fertilization (Byrne 1996; Byrne & Cerra 1996). The extent to which the offspring size varies among and within gonads has not been characterised. Parvulastra parvivipara is a simultaneously hermaphroditic self-fertile species and has young in the gonad year-round with a range of developmental stages from embryo to juvenile (Byrne 1996: Khan et al. 2019). The observation that the parents die after releasing all of the juveniles (Keough & Dartnall 1978) requires confirmation. There are several features of the reproductive biology of this species that are yet to be determined including the minimum and maximum size at birth, and whether parents release all of the juveniles at the same time or in cohorts. It is also not known if adults release larger juveniles over a certain size range and retain the smaller juveniles for a longer period. These questions are addressed here.

Materials and methods

Sample collection Parvulastra parvivipara were collected from the Smooth Pool (32ʹ54ʹʹS, 134ʹ04ʹʹE), Eyre Peninsula, South Australia (Ministerial Exemption from the Ministry of Agriculture, Fisheries and Forestry, Government of South Australia- ME9902902) and transported to the University of Sydney. Each P. pa r v i v i p a r a (n = 54) was kept in a separate plastic beaker that was checked daily for newly released juveniles. Released juveniles were photographed using a camera connected to a dissection microscope (Olympus SZX10) to measure juvenile diameter using the software Micropublisher 3.0. The juveniles were weighed using a microbalance (Mettler H35AR) to 0.0001 g and then preserved in 70% ethanol in a separate Eppendorf tube for each individual. If the parents did not release offspring, the juveniles were collected from the dissected gonads of the parent. Then the juveniles were photographed, counted, measured, weighed, and preserved in 70% ethanol. Before dissection, the arm radius (R) of the parent was measured from the center of the mouth to the tip of one arm and wet weight was measured. A total of 500 gonads were dissected from 50 P. parv iv ipar a and 491 progeny (juveniles and embryos) were counted and measured to compare the size of the released (n = 197) and retained (n = 294) juveniles. The coefficient of variation (CV) in offspring size within gonads was determined for 64 gonads from 22 adults. Among gonad CV within individual stars was also determined. The CV in offspring size was calculated from the gonads with more than one developing young using the following formula: CV = Standard deviation/mean × 100. The relationship between within gonad CV and number of offspring was analysed using linear regression to determine whether the number of offspring influences CV in offspring size within a gonad. The relationship between the total number of offspring and adult size was analysed using linear

72 · Zoosymposia 15 © 2019 Magnolia Press KHAN ET AL. regression. The length of the gonads was measured from the base of the gonad to the farthest tip (or longest length) of the gonad from the photographs taken using Micropublisher 3.0. The relationship between the number and size of gonads that contained offspring per adult and adult size was analysed using linear regression. It was noted if gonads were predominantly testes. The relationship between the number and size of the male gonads (which only had sperm) and the parent size was also analysed. These data were used to determine relative investment in the male (testes) and female (brooding) function using linear regression. Prior to analyses, data were tested for normality with Shapiro-Wilk test and for homogeneity of variance with Levene’s test. All linear regressions were analysed using IBM SPSS statistics 24 and significance was assigned at the 5% level.

Micro-computed tomography and confocal microscopy One P. p ar v i v i p a r a was preserved in 70% ethanol. Micro-CT scans of this individual was performed using Skyscan 1172 with isotropic voxel resolution 2.94 µm. The specimen was firmly positioned within an Eppendorf tube using polystyrene foam and then scanned. A scan of 11 hours at 23 kV and 200 µA resulted in 1170 images that visualized the skeleton. The images were analysed using the software Avizo 9.5 version. Manual “draw tool” and “magic wand” were applied to each image stack to isolate each of the juveniles. A different color was applied to each juvenile within a single gonad to visualize them separately. The position and orientation of the juveniles were visualized by applying three-dimensional (3-D) reconstruction and by making the parent body virtually transparent. For confocal microscopy, P. p a r v i v i pa r a were fixed in 2.5% glutaraldehyde in 1 µm filtered seawater (FSW) for 24 hours and then stored in 70% ethanol. Individual sea stars were then dissected, and each gonad was isolated and dehydrated in an ethanol series (70–95 %). The gonads were cleared with benzyl alcohol and benzyl benzoate, 2:1 or 1:1 (v/v) (Sigma-Aldrich Co.) and mounted on a glass slide. The glutaraldehyde auto- fluorescence was used as a signal for analysis using a Leica SPE-2 Confocal Laser Microscope or ZEISS LSM 800 plus Spectral Confocal Microscope. The samples were excited at 488 nm wavelength with an argon laser contrasting separate gain and offset values for each sample. Images were taken from the different focal planes at 1024×1024 pixel array with a frame average that varied from 4 to 12 based on sample response to excitation. A series of images were stacked for viewing 3D view along the Z axis (Z-stack). The diameter of egg and embryo were measured using confocal microscope online assistance.

Results

Biology of viviparity in Parvulastra parvivipara Parvulastra parvivipara with young in the gonads were a range of sizes (R = 1.8–5.2 mm, 0.0094–0.1234 g). The gonads are attached by the gonoduct to the aboral body wall and the juveniles emerge through the gonopore (Fig. 1a). The gonads varied in size and shape (Fig. 1b–c). Predominantly male gonads were usually tubular and branched and had occasional oocytes (Fig. 1c). Gonads that contained juveniles were oval to round. In gonads with one juvenile, the gonad shape reflected the juvenile form (Fig 1c). Only two of the 50 adults had juveniles in all ten of their gonads. The gonads ranged in size from 120–2,630 µm (x̅ = 941, SD = 624, n = 49) due to the wide size range of juveniles. Predominantly male gonads were small (x̅ = 422, SD = 18, Range = 147–940 µm, n = 56 gonads from 27 adults) with a mean of 2.11 testes per individual (SD = 0.96, 2 Range = 0–4, n = 28). The number of male gonads was not related to adult size (r = 0.079, F1,26 = 2.242, P > 2 0.05), but the size of these gonads was larger in the bigger adults (r = 0.388, F1,25 = 16.156, P < 0.05). The 2 number of gonads that contained juveniles was positively related to adult size (r = 0.466, F1,38 = 33.107, P < 0.05). Offspring within gonads ranged from developing gastrulae, brachiolaria larvae, and metamorphic juvenile to fully developed juveniles (Fig 1f). Late brachiolaria ranged from 146–247 µm length (x̅ = 197 µm, SD = 28, n = 6) and pre-metamorphic juveniles ranged from 243–326 µm diameter (x̅ = 277 µm, SD = 35, n = 5). There was a marked variation in the size of the juvenile that were released in the laboratory (x̅ = 1311 µm diameter, SD = 474 µm, Range = 473–3,013 µm, n = 197) (Fig. 2). The juveniles that remained in the gonad also varied in size (x̅ = 860 µm diameter, SD = 408, Range = 146–2,514 µm, n = 294) (Fig. 2). Parvulastra parvivipara often retained juveniles larger or equal to the largest released juvenile (Fig. 2). Among the 54

ARRANGEMENT AND SIZE OF VIVIPAROUS ASTERINID SEA STAR Zoosymposia 15 © 2019 Magnolia Press · 73 adults monitored, 47 released juveniles at least once, 12 released juveniles more than once (2–5 different time points), six released all of their juveniles in a single event, and seven of them did not release juveniles, even though large juveniles (x̅ = 960, SD = 47 µm, Range = 219–2,514 µm diameter, n = 115) were present in the gonads (Fig. 2). The largest number of offspring (50 juveniles ranging from 418 to 1,280 µm diameter) were isolated from an adult R = 4.7 mm. The smallest adult had only one juvenile (1,397 µm diameter), whereas the largest adult (R = 5.2 mm) had 31 juveniles, of which 16 emerged through the gonopores at five different time points over 14 days, while the remaining 15 juveniles were dissected from the gonads.

FIGURE 1. Parvulastra parvivipara: a) five juveniles emerging through the gonopores (white arrows); b) ten gonads with juveniles (white arrows); c) asynchronous development with juveniles (white arrows) at variable in size and predominantly male gonads (black arrows); d) a juvenile (white arrow) crawling out after rupturing its parent’s body; e) a juvenile (arrow) eating adjacent torn (black arrow) pyloric caeca (P), inset is the larger view (scale 0.5 mm); f) pre-metamorphic (black arrow), newly metamorphic (upper top) and fully developed (white arrow) juveniles from a gonad. Scale: a and c = 4.0 mm, b = 2 mm, d = 1.5 mm, e = 3.0 mm and f = 0.5 mm.

Four adults died after releasing all of their young. These were small adults (R = 2.5–3.2 mm). In two other adults, the body wall was ruptured by the emerging juveniles but the parents were still alive (Fig 1d, e). In one adult (R = 1.8 mm), the juvenile (1,398 µm diameter) was 37 % of the parent size. In the second adult (R = 4.2 mm), the juvenile (1,443 µm) was around 18 % of the parent diameter. One juvenile was observed eating the

74 · Zoosymposia 15 © 2019 Magnolia Press KHAN ET AL. 2 pyloric caeca of the parent (Fig. 1e). Total brood weight was significantly related to adult size (r = 0.367, F1,52 2 = 30.101, P < 0.05) and weight (r = 0.414, F1,46 = 32.616, P < 0.05).

FIGURE 2. Size of released and retained juveniles in Parvulastra parvivipara in the laboratory from 54 parents having different arm radius (µm). The size of the retained juveniles was determined after dissecting the parent.

The CV in offspring size within an individual gonad was 24.7 % (SD = 24.18, Range = 0.14–112.57, n = 64 gonads from 22 adults). Among gonad CV within an adult was 19.5 % (SD = 14.56, Range = 1.91–39.01, n = 22 adults 64 gonads). The largest number of juveniles in a single gonad was nine.

Confocal Microscopy The gonads are sac-like structures with outer and inner epithelial layers (Fig. 3a). Developing eggs are attached to the germinal epithelium by a basal stalk-like structure and were surrounded by a layer of follicle cells (Fig. 3b). Egg development is continuous. The mean egg size was 84 µm (SD = 21, Range = 54–134 µm, n = 43) and the largest oocyte observed was 134 µm diameter. The gonads contained a range of developmental stages including gastrula, brachiolaria larva and juveniles (Fig. 3a–i). The eggs and early embryos may remain surrounded by somatic cells (Fig. 3a–b). Juveniles remain orally apposed to each other (Fig. 3e). Amorphous material was observed adjacent to brachiolaria and metamorphic juvenile within the gonadal lumen which may be haemal fluid (Fig. 3i). The mean diameter of the gastrula was 91 µm (SD = 4, n = 3), early brachiolaria was 112 µm (SD = 6, n = 4), late brachiolaria was 145 µm (SD = 13, n = 3), and pre-metamorphic juvenile was 218 µm (SD = 4, n = 3). The presence of eggs and offspring at different stages of development indicates asynchronous reproduction (Fig. 3a–i). Among 63 gonads observed from 13 adults, four were predominantly male with no oocytes (Fig. 3f) and offspring, 32 contained juveniles, five with embryos only, four with embryos and juveniles, and 18 had developing oocytes. All of the gonads except four male gonads (Fig. 3f) had oocytes of variable stages and sizes. The smallest early embryo encountered was 86 µm diameter (early gastrula). Gonads with developing oocytes were tubular or dome-shaped and become oval or round with increasing size and number of juveniles. The profile of gonads that had only one large juvenile reflected the shape of the juvenile (Fig. 3c). Juveniles may extrude their stomach into the lumen (Fig. 3i).

ARRANGEMENT AND SIZE OF VIVIPAROUS ASTERINID SEA STAR Zoosymposia 15 © 2019 Magnolia Press · 75 FIGURE 3. Confocal microscopy of Parvulastra parvivipara gonads: a) a gonad containing a gastrula surrounded by somatic cells in the lumen adjacent to an egg, outer and inner gonad wall (arrows); b) eggs surrounded by follicle cells (arrow); c) oral face of a large juvenile with two pairs of tube feet in each arm contrasting in size with a small metamorphosing juvenile (arrow); d) a 3-D view of a gonad containing two juveniles and eggs; e) a 3-D view of two juveniles with their oral faces (arrow) opposite to each other; f) sperm in a predominantly male gonad; g) gonad containing eggs, early gastrula (arrow), gastrula, brachiolaria, late brachiolaria, and metamorphic juvenile and juvenile; h) flocculent material and somatic cells adjacent to a gastrula; i) juvenile with stomach extruded (arrow), amorphous material (AM) next to the late brachiolaria and around egg that may be haemal fluid. E, egg; G, gastrula; B, brachiolaria; LB, late brachiolaria; MJ, metamorphic juvenile; J, Juvenile; L, lumen; SC, somatic cells; M, mouth; S, sperm; T, tube feet. Scale: a, b, f and i = 50 µm; c, e and g = 200 µm, d = 100 µm, h = 20 µm.

Offspring arrangement using µ-CT The 3D tomographic video of P. parv iv ipar a shows 13 juveniles (392–1,620 µm diameter) in the gonads (Fig. 4a–b). Four gonads each contained one juvenile, two gonads each contained two juveniles, one gonad had five juveniles and three gonads had no juvenile (Fig. 4a–b). In gonads with two juveniles these were facing orally opposite to each other (Fig. 4c). In the gonad that had five juveniles, the largest juvenile was 926 µm diameter and the smallest juvenile was 721 µm diameter. All five juveniles were oriented with their oral face opposite to each other and they had their oral face towards the center of the gonad (Fig. 4d). The position and orientation of juvenile had no specific pattern in the gonads that had a single juvenile. The largest juvenile was more than four times larger than the smallest solitary juvenile.

Discussion

Parvulastra parvivipara exhibits a unique spatial arrangement of its young in the gonads that is influenced by maternal provisioning strategy and sibling cannibalism. The juveniles usually position their oral surface

76 · Zoosymposia 15 © 2019 Magnolia Press KHAN ET AL. opposing one another, an orientation that may defend them from being preyed upon by siblings (Byrne 1996). Juveniles that achieve an advantageous position and face the aboral surface of another juvenile may successfully cannibalize their sibling (Byrne 1996). Large juveniles appear to be relatively stationary in the gonads potentially to prevent rupture of the gonad wall. Juveniles often had their stomachs extruded, indicative of predation and perhaps using gonadal fluid and eggs as sources of nutrient. The 3-D reconstruction of P. p a r v i v i pa r a offspring add a new dimension in observing brooding in marine invertebrates. There is one previous three-dimensional reconstruction (µ-CT) of the offspring in the brood chamber of echinoderms, those of the ophiuroids, Ophioderma wahlbergii and Amphipholis squamata (Landschoff & Griffiths 2015). In these ophiuroids, the juveniles had their oral surface pressed against the bursal sinus, which provides nutrients to developing young (Byrne 1991; Hendler & Tran 2001). In P. parvivipara, amorphous material in the gonad may be haemal fluid and a potential source of nutrients for the juveniles. For this species, however, adelphophagy is the main source of nutrients that supports for substantial post-metamorphic growth.

FIGURE 4. Micro-Computed tomography of Parvulastra parvivipara: a) 3-D image showing the position of 13 juveniles (coloured differently) within the gonads (A–J), gonad B, I and J had no juveniles; b) juveniles (coloured differently) with parent body removed virtually, separate numbers indicates juveniles from separate gonads (A–H); c) two juveniles orally apposed to each other in a gonad; d) position of five juveniles (coloured differently) within a gonad oriented with their oral surfaces opposed. Scale: a and b = 1.0 mm, b and c = 0.5 mm.

Studies that relate sibling competition to the arrangement of offspring in the incubation chamber in marine invertebrates are rare. This unusual mode of maternal provisioning is also reported in the holothuroid Leptosynapta clarki and the sea star Pteraster militaris, and P. v i v i pa r a , (McClary & Mladenov 1990; Byrne & Cerra 1996; Sewell et al. 2006). Sibling cannibalism is common in marine gastropods and polychaetes (Cable & Tinsley 1991; Cubillos et al. 2007; Carrasco & Phillips 2014). Sibling cannibalism has evolved to support embryonic nutrition and generation of large offspring size in sharks and mollusks (Wourms 1981; Smith & Reay 1991; Strathmann 1995; Cubillos et al. 2007). For example, in sand tiger sharks, Carcharias taurus, 9–10 embryos begin development in each oviduct, but only one gigantic offspring survives (Wourms

ARRANGEMENT AND SIZE OF VIVIPAROUS ASTERINID SEA STAR Zoosymposia 15 © 2019 Magnolia Press · 77 1981). Similarly, several mollusks place hundreds of embryos within a single egg case to support few large hatchlings that grow through cannibalism (Strathmann & Strathmann 1995; Carrasco & Phillips 2014). As the juveniles of P. p ar v i v i p a r a develop from a very small secondarily reduced egg (84 µm diameter), intragonadal sibling cannibalism may have evolved in parallel to a decrease in egg size to support post-metamorphic nutrition and producing large offspring size (Byrne 1996; Byrne & Cerra 1996). Developing embryos and juveniles may also get nutrition from surrounding somatic cells, gonadal fluid and haemal fluid. These extra- embryonic sources remain to be investigated. Sibling cannibalism may be an adaptation to reduce the number of embryos competing for oxygen, as in the mollusk Acanthina monodon, where adelphophagy increases with increasing oxygen demand (Lardies & Fernandez 2002). It is not known whether sibling cannibalism in P. p a r v i v i p a r a decreases embryonic development time as it does in several muricid gastropods (e.g. Crepipatella fecunda) (Spight 1975; Cubillos et al. 2007). Adelphophagy should be favoured if structural and energetic constraints limit maximum egg size (Strathmann 1995). This may be the case for P. p a r v i v i p a r a , which is the smallest known sea star. The offspring produced by P. pa rv i v i par a are much larger than the juveniles released by the closely related asterinid brooders that have large eggs and lecithotrophic development (Byrne & Cerra 1996). Sibling cannibalism in P. pa r v i v i p a r a may be an adaptation to produce large offspring size (>1000 µm) to enhance their success in the harsh high intertidal environment occupied by this species (Byrne 1996; Trumbo 1996; Gillespie & McClintock 2007). The largest juveniles produced by P. par v i v i pa r a are nearly reproductively mature (Byrne 1996). Large offspring may be selected for early reproductive maturity, high survival and stress tolerance (Rivest 1983; Moran & Emlet 2001; Olivera-Tlahuel et al. 2015). As seen here for P. p ar v i v i p a r a, the presence of several development stages in the gonad space also occurs for the young in the bursa of the ophiuroid, Amphiura carchara (Hendler & Tran 2001), bivalves Transennella tantilla and T. confusa (Kabat 1985; Russell et al. 1992), sponge Rhopaloeides odorabile (Whalan et al. 2007), and annelid Diopatra marocensis (Arias et al. 2013). Asynchronous development of several ontogenic stages or superfetation commonly occurs in live-bearing fishes (Clinidae, Zenarchopteridae, Poeciliidae) and was thought to be restricted to fishes (Wourms 1981; Frick 1998). Frick (1998) first used the term ‘superfetation’ in the marine invertebrate literature to describe viviparity in the holothuroid Synaptula hydriformis that cares for multiple cohorts of embryos in the perivisceral coelom. In P. parvivipara, The presence of an early gastrula at 86 µm diameter, which is smaller than the largest egg observed (134 µm diameter), suggests that the terminal egg size may vary. Superfetation is thought to be evolutionarily linked with matrotrophy in many live-bearing poeciliid fishes (Scrimshaw 1944; Pollux et al. 2014; Olivera-Tlahuel et al. 2015) and is advantageous with respect to energetics and fecundity, because a parent can produce small eggs with the outcome of producing large juveniles freeing the adult from fixed initial cost to produce a large egg (Olivera-Tlahuel et al. 2015). This strategy results in many small early-stage embryos and few large-late stage embryos, avoids the constraints limiting the number of offspring due to maternal size, and allows production of large hatchlings (Scrimshaw 1944; Wourms 1981; Trexler 1997; Olivera-Tlahuel et al. 2015). Thus, asynchronous viviparity may also be evolutionarily linked to matrotrophy (sibling cannibalism) in P. p ar v i v i p a r a where producing a small egg (80 µm) freed from fixed energy investment at the onset of embryo development, supported continuous reproduction and large offspring to maximize energetic efficiency (Travis et al. 1987). Continuous production of offspring may be a mechanism to reduce overcrowding and utilize limited space available to accommodate offspring in the small-bodied sea star P. pa r v i v i pa r a, similar to the sea urchin A. lorioli (Galley et al. 2005). According to the energetic hypothesis, this tiny sea star may not have sufficient energy to produce more and larger eggs to invest all of its resources at a time (Chia 1974). However, larger adults of P. pa r v i v i p r a are more capable of allocating a greater amount of energy at a time for female functioning or offspring care than smaller adults. Parvulastra parvivipara exhibits substantial among and within gonad variation in offspring size. Size variation of offspring among broods often reflect a bet-hedging strategy influenced by the environment on maternal provisioning or availability of the resources (Bernardo 1996; Moran & Emlet 2001; Marshall & Keough 2008). Selection favours parents investing optimum resources for all offspring (Vance 1973; Smith & Fretwell 1974), but individual offspring will gain a fitness benefit by securing a greater share of parental resources (Trivers 1974; Kamel & Williams 2017). Thus, offspring will compete with their siblings and the parent. In cases where outcrossing occurs, individual offspring are more closely related to themselves than to siblings or the parent (Parker et al. 2002; Kamel & Williams 2017) and this may promote competition. In

78 · Zoosymposia 15 © 2019 Magnolia Press KHAN ET AL. contrast, in P. parv iv ipar a and other viviparous asteroids, the parent and offspring have the same genotype due to self-fertilization (Keever et al. 2013). As all juveniles are genetically identical to each other, sibling relatedness or mating system has a little role in sibling competition and size variation. In species with asynchronous viviparity, early hatched larger embryos/juveniles out compete less- developed siblings (asymmetric competition) due to their physical superiority or greater competitive ability (Olivera-Tlahuel et al. 2015). Hence, they will be larger at birth, as occurs in some live-bearing fish (Schrader & Travis 2012). In P. p a r v i v i p a r a, asynchronous vivparity initiates an asymmetric competition among early developing embryos, and dynamic sibling cannibalism reduces parental control on offspring size (Pollux & Reznick 2011) and intensifies the size variation of offspring at the post-metamorphic stage. Within brood variation in offspring size in P. par v iv ip ar a may also be an adaptation to an unpredictable intertidal environmental (bet-hedging) (Marshall & Keough 2008), or some offspring may simply obtain slightly more resources (randomly) in the early development stages (Rivest 1983; Moran & Emlet 2001). In the laboratory, P. parvivipara released all of its juveniles in a single cohort or at several time points. Parvulastra parvivipara may retain juveniles in the gonads which are larger than those released. This species does not normally die after birth of offspring, counter to early reports (Keough & Dartnall 1978). This species releases offspring at a diverse size range (0.4–3.0 mm diameter) and retains juveniles up to 2.5 mm in diameter. Parvulastra parvivipara releases juveniles through the aboral gonopore associated with softening of the surrounding collagenous tissue (Byrne 1996), but is not known if this process is controlled by the parent or if it is mediated by a signal from the offspring. As the adults retain juveniles that are larger than the released juveniles, the parent may have little control over juvenile release or influence on offspring size at birth. Some small juveniles may vacate the gonad to avoid being preyed on by the larger siblings (Byrne 1996) or may emerge as a group with other siblings. Large single juveniles may remain in the gonad to a very advance stage and occasionally may tear the gonad and body wall of the parent. In extreme cases, they may cause major damage to the parent, possibly when gonad has reached its maximum capacity. This resulted in occasional adult mortality and was most commonly seen in the smallest adults.

Acknowledgements

The authors acknowledge the Bosch Institute Advance Microscopy and the Australian Centre for Microscopy & Microanalysis at the University of Sydney and assistance of the staff of these facilities. The authors are grateful to Liz McTaggart (Department of Environment, Water and Natural Resources, South Australian Government) for assistance in sample collection. Thanks also to the Ministry of Agriculture, Fisheries and Forestry, Government of South Australia for the Ministerial Exemption (ME9902902) to collect samples.

Funding

This research was funded by the University of Sydney.

Conflict of Interest

The authors declare that they have no conflicts of interest.

References

Arias, A., Richter, A., Anadon, N. & Paxton, H. (2013) Evidence of simultaneous hermaphroditism in the brooding Diopatra marocensis (Annelida: Onuphidae) from northern Spain. Journal of the Marine Biological Association of the United Kingdom, 93 (6), 1533–1542. https://doi.org/10.1017/S002531541300012X Bernardo, J. (1996) The particular maternal effect of propagule size, especially egg size: Patterns, models, quality of evidence

ARRANGEMENT AND SIZE OF VIVIPAROUS ASTERINID SEA STAR Zoosymposia 15 © 2019 Magnolia Press · 79 and interpretations. American Zoologist, 36 (2), 216–236. https://doi.org/10.1093/icb/36.2.216 Byrne, M. (1991) Reproduction, development and population biology of the Caribbean ophiuroid Ophionereis olivacea, a protandric hermaphrodite that broods its young. Marine Biology, 111 (3), 387–399. https://doi.org/10.1007/BF01319411 Byrne, M. (1996) Viviparity and intragonadal cannibalism in the diminutive sea stars Patiriella vivipara and P parvivipara (family Asterinidae). Marine Biology, 125 (3), 551–567. Byrne, M. & Cerra, A. (1996) Evolution of intragonadal development in the diminutive asterinid sea stars Patiriella vivipara and P. parvivipara with an overview of development in the Asterinidae. Biological Bulletin, 191 (1), 17–26. https://doi.org/10.2307/1543057 Cable, J. & Tinsley, R.C. (1991) Intra-uterine larval development of the polystomatid monogeneans, Pseudodiplorchis americanus and Neodiplorchis scaphiopodis. Parasitology, 103 Pt 2 (02), 253–266. https://doi.org/10.1017/S0031182000059539 Carrasco, S.A. & Phillips, N.E. (2014) Encapsulation and development of three New Zealand neogastropods with contrasting embryo packaging and maternal provisioning. New Zealand Journal of Zoology, 41 (3), 171–186. https://doi.org/10.1080/03014223.2014.921205 Chia, F. (1974) Classification and adaptive significance of developmental patterns in marine invertebrates. Thalassia Jugoslavica, 10, 121–130. Collin, R. & Spangler, A. (2012) Impacts of adelphophagic development on variation in offspring size, duration of development, and temperature-mediated plasticity. Biological Bulletin, 223 (3), 268–277. https://doi.org/10.1086/BBLv223n3p268 Cubillos, V.M., Chaparro, O.R., Montiel, Y.A. & Veliz, D. (2007) Unusual source of food: impact of dead siblings on encapsulated embryo development of Crepipatella fecunda (Gastropoda: Calyptraeidae). Marine and Freshwater Research, 58 (12), 1152–1161. https://doi.org/10.1071/MF07094 Frick, J.E. (1998) Evidence of matrotrophy in the viviparous holothuroid echinoderm Synaptula hydriformis. Invertebrate Biology, 117 (2), 169–179. https://doi.org/10.2307/3226968 Galley, E.A., Tyler, P.A., Clarke, A. & Smith, C.R. (2005) Reproductive biology and biochemical composition of the brooding echinoid Amphipneustes lorioli on the Antarctic continental shelf. Marine Biology, 148 (1), 59–71. https://doi.org/10.1007/s00227-005-0069-3 Gillespie, J.M. & McClintock, J.B. (2007) Brooding in echinoderms: how can modern experimental techniques add to our historical perspective? Journal of Experimental Marine Biology and Ecology, 342 (2), 191–201. https://doi.org/10.1016/j.jembe.2006.10.055 Hendler, G. & Tran, L.U. (2001) Reproductive biology of a deep-sea brittle star Amphiura carchara (Echinodermata : Ophiuroidea). Marine Biology, 138 (1), 113–123. https://doi.org/10.1007/s002270000446 Kabat, A.R. (1985) The allometry of brooding in Transennella tantilla (Gould) (Mollusca, Bivalvia). Journal of Experimental Marine Biology and Ecology, 91 (3), 271–279. https://doi.org/10.1016/0022-0981(85)90181-9 Kamel, S.J., Grosberg, R.K. & Marshall, D.J. (2010A) Family conflicts in the sea. Trends in Ecology & Evolution, 25 (8), 442– 449. https://doi.org/10.1016/j.tree.2010.05.008 Kamel, S.J., Oyarzun, F.X. & Grosberg, R.K. (2010B) Reproductive biology, family conflict, and size of offspring in marine invertebrates. Integrative and Comparative Biology, 50 (4), 619–629. https://doi.org/10.1093/icb/icq104 Kamel, S.J. & Williams, P.D. (2017) Resource exploitation and relatedness: implications for offspring size variation within broods. Oikos, 126 (8), 1219–1226. https://doi.org/10.1111/oik.04034 Keever, C.C., Puritz, J.B., Addison, J.A., Byrne, M., Grosberg, R.K., Toonen, R.J. & Hart, M.W. (2013) Shallow gene pools in the high intertidal: extreme loss of genetic diversity in viviparous sea stars (Parvulastra). Biological Letters, 9 (5), 20130551. https://doi.org/10.1098/rsbl.2013.0551 Keough, M.J. & Dartnall, A.J. (1978) A new species of viviparous asterinid asteroid from Eyre Peninsula South Australia. Records of the South Australian Museum (Adelaide), 17 (28), 406–416. Khan, M.S.R., Whittington, C.M., Thompson, M.B. & Byrne, M. (2019) Intragonadal incubation of progeny in three viviparous

80 · Zoosymposia 15 © 2019 Magnolia Press KHAN ET AL. asterinid sea stars that differ in offspring provisioning, lecithotrophy vs matrotrophy. Marine Biology, 166 (6), 81. https://doi.org/10.1007/s00227-019-3507-3 Landschoff, J. & Griffiths, C.L. (2015) Three-dimensional visualisation of brooding behaviour in two distantly related brittle stars from South African waters. African Journal of Marine Science, 37 (4), 533–541. https://doi.org/10.2989/1814232X.2015.1095801 Lardies, M.A. & Fernandez, M. (2002) Effect of oxygen availability in determining clutch size in Acanthina monodon. Marine Ecology Progress Series, 239, 139–146. https://doi.org/10.3354/meps239139 Larson, P. (2017) Brooding sea anemones (Cnidaria: Anthozoa: Actiniaria): paragons of diversity in mode, morphology, and maternity. Invertebrate Biology, 136 (1), 92–112. https://doi.org/10.1111/ivb.12159 Lesoway, M.P., Abouheif, E. & Collin, R. (2014) The development of viable and nutritive embryos in the direct developing gastropod Crepidula navicella. International Journal of Developmental Biology, 58 (6–8), 601–611. https://doi.org/10.1387/ijdb.140136rc Marshall, D.J. & Keough, M.J. (2008) The evolutionary ecology of offspring size in marine invertebrates. In: Sims, D.W. (Ed.), Advances in Marine Biology (Vol. 53). Elsevier Academic Press Inc., San Diego, pp. 1–60. https://doi.org/10.1016/S0065-2881(07)53001-4 McClary, D.J. & Mladenov, P.V. (1990) Brooding biology of the sea star Pteraster militaris (OF Müller): energetic and histological evidence for nutrient translocation to brooded juveniles. Journal of Experimental Marine Biology and Ecology, 142 (3), 183–199. https://doi.org/10.1016/0022-0981(90)90090-Y Mercier, A., Sun, Z., Parrish, C.C. & Hamel, J.F. (2016) Remarkable shifts in offspring provisioning during gestation in a live- bearing cnidarian. PLoS ONE, 11 (4), e0154051. https://doi.org/10.1371/journal.pone.0154051 Moran, A.L. & Emlet, R.B. (2001) Offspring size and performance in variable environments: field studies on a marine snail. Ecology, 82 (6), 1597–1612. https://doi.org/10.2307/2679803 Olivera-Tlahuel, C., Ossip-Klein, A.G., Espinosa-Perez, H.S. & Zuniga-Vega, J.J. (2015) Have superfetation and matrotrophy facilitated the evolution of larger offspring in poeciliid fishes? Biological Journal of the Linnean Society, 116 (4), 787– 804. https://doi.org/10.1111/bij.12662 Ostrovsky, A.N., Lidgard, S., Gordon, D.P., Schwaha, T., Genikhovich, G. & Ereskovsky, A.V. (2015) Matrotrophy and placentation in invertebrates: a new paradigm. Biological reviews of the Cambridge Philosophical Society, 91 (3), 673– 711. https://doi.org/10.1111/brv.12189 Parker, G.A., Royle, N.J. & Hartley, I.R. (2002) Intrafamilial conflict and parental investment: a synthesis. Philosophical Transactions of the Royal Society of London B Biological Sciences, 357 (1419), 295–307. https://doi.org/10.1098/rstb.2001.0950 Pollux, B.J.A., Meredith, R.W., Springer, M.S., Garland, T. & Reznick, D.N. (2014) The evolution of the placenta drives a shift in sexual selection in livebearing fish. Nature, 513 (7517), 233–236. https://doi.org/10.1038/nature13451 Pollux, B.J.A. & Reznick, D.N. (2011) Matrotrophy limits a female's ability to adaptively adjust offspring size and fecundity in fluctuating environments. Functional Ecology, 25 (4), 747–756. https://doi.org/10.1111/j.1365-2435.2011.01831.x Rivest, B.R. (1983) Development and the influence of nurse egg allotment on hatching size in Searlesia dira (Reeve, 1846) (Prosobranchia, Buccinidae). Journal of Experimental Marine Biology and Ecology, 69 (3), 217–241. https://doi.org/10.1016/0022-0981(83)90071-0 Russell, M.P., Huelsenbeck, J.P. & Lindberg, D.R. (1992) Experimental taphonomy of embryo preservation in a cenozoic brooding bivalve. Lethaia, 25 (4), 353–359. https://doi.org/10.1111/j.1502-3931.1992.tb01637.x Schrader, M. & Travis, J. (2009) Do embryos influence maternal investment? evaluating maternal?fetal coadaptation and the potential for parent?offspring conflict in a placental fish. Evolution, 63 (11), 2805–2815. Schrader, M. & Travis, J. (2012) Variation in offspring size with birth order in placental fish: a role for asymmetric sibling competition? Evolution, 66 (1), 272–279. https://doi.org/10.1111/j.1558-5646.2011.01422.x Scrimshaw, N.S. (1944) Superfetation in poeciliid fishes. Copeia, 1944 (3), 180–183.

ARRANGEMENT AND SIZE OF VIVIPAROUS ASTERINID SEA STAR Zoosymposia 15 © 2019 Magnolia Press · 81 https://doi.org/10.2307/1437814 Sewell, M.A., Koss, R.O.N., Turner, A. & Chia, F.S. (2006) Evidence for matrotrophy in the viviparous sea cucumber Leptosynapta clarki: a role for the genital haemal sinus? Invertebrate Reproduction and Development, 49 (4), 225–236. https://doi.org/10.1080/07924259.2006.9652212 Shine, R. (1978) Propagule size and parental care: the “safe harbor” hypothesis. Journal of Theoretical Biology, 75 (4), 417– 424. https://doi.org/10.1016/0022-5193(78)90353-3 Smith, C. & Reay, P. (1991) Cannibalism in teleost fish. Reviews in Fish Biology and Fisheries, 1 (1), 41–64. https://doi.org/10.1007/BF00042661 Smith, C.C. & Fretwell, S.D. (1974) The optimal balance between size and number of offspring. American Naturalist, 108 (962), 499–506. https://doi.org/10.1086/282929 Spight, T.M. (1975) Factors extending gastropod embryonic development and their selective cost. Oecologia, 21 (1), 1–16. https://doi.org/10.1007/BF00345889 Strathmann, R.R. (1995) Peculiar constraints on life histories imposed by protective or nutritive devices for embryos. American Zoologist, 35 (4), 426–433. https://doi.org/10.1093/icb/35.4.426 Strathmann, R.R. & Chaffee, C. (1984) Constraints on egg masses. 2. effect of spacing, size, and number of eggs on ventilation of masses of embryos in jelly, adherent groups, or thin-walled capsules. Journal of Experimental Marine Biology and Ecology, 84 (1), 85–93. https://doi.org/10.1016/0022-0981(84)90232-6 Strathmann, R.R. & Strathmann, M.F. (1982) The relationship between adult size and brooding in marine invertebrates. American Naturalist, 119 (1), 91–101. https://doi.org/10.1086/283892 Strathmann, R.R. & Strathmann, M.F. (1995) Oxygen-supply and limits on aggregation of embryos. Journal of the Marine Biological Association of the United Kingdom, 75 (2), 413–428. https://doi.org/10.1017/S0025315400018270 Travis, J., Farr, J.A., Henrich, S. & Cheong, R.T. (1987) Testing theories of clutch overlap with the reproductive ecology of Heterandria formosa. Ecology, 68 (3), 611–623. https://doi.org/10.2307/1938466 Trexler, J.C. (1997) Resource availability and plasticity in offspring provisioning: embryo nourishment in sailfin mollies. Ecology, 78 (5), 1370–1381. https://doi.org/10.1890/0012-9658(1997)078[1370:RAAPIO]2.0.CO;2 Trivers, R.L. (1974) Parent-offspring conflict. American Zoologist, 14 (1), 249–264. https://doi.org/10.1093/icb/14.1.249 Trumbo, S.T. (1996) Parental care in invertebrates. In Advances in the Study of Behavior (Vol. 25). Elsevier, Amsterdam, pp. 3– 51. https://doi.org/10.1016/S0065-3454(08)60329-0 Vance, R.R. (1973) More on reproductive strategies in marine benthic invertebrates. American Naturalist, 107 (955), 353–361. https://doi.org/10.1086/282839 Whalan, S., Battershill, C. & de Nys, R. (2007) Sexual reproduction of the brooding sponge Rhopaloeides odorabile. Coral Reefs, 26 (3), 655–663. https://doi.org/10.1007/s00338-007-0236-8 Wourms, J.P. (1981) Viviparity - the maternal-fetal relationship in fishes. American Zoologist, 21 (2), 473–515. https://doi.org/10.1093/icb/21.2.473