Bacteria on Coralline Algae and Their Role As Sea Urchin Settlement Cues

Bacteria on coralline algae and their role as sea urchin settlement cues

Shaun Nielsen

A thesis in fulfilment of the requirements for the degree of Doctor of Philosophy

School of Biological, Earth and Environmental Sciences

Faculty of Science

2013

THE UNIVERSITY OF NEW SOUTH WALES - Thesis/Dissertation Sheet Surname or Family name:Nielsen

First name:Shaun Other name/s:

Abbreviation for degree as given in the University calendar: PhD

School:Biological, Earth and Environmental Sciences Faculty:Science Title:Bacteria on coralline algae and their role as sea urchin settlement cues

Abstract

Most benthic marine invertebrates have a biphasic life cycle, in which a planktonic larval stage alternates with a benthic ad ult stage. The transition between the larval and adult stage is typically guided by habitat-derived settlement cues and thus understanding the nature and distribution of settlement cues is a central theme in larval ecology. Both coralline algae and their epiphytic bacterial biofilms can be important settlement cues for marine invertebrate larvae, but the relationship between settlement and specific communities of bacteria is largely unknown. I investigated bacterial mediated settlement for larvae of the Australian sea urchins Heliocidaris erythrogramma and Holopneustes purpurascens and compared this to the community ecology of bacteria on coralline algae. I conducted a meta-analysis of putative larval cues from macroalgae to test the importance of coralline algae as settlement cues for invertebrate larvae generally and sea urchin larvae specifically. The meta-analysis revealed that coralline algae were the most inductive macroalgae for a variety of larval groups, but epiphytic bacteria only enhanced larval settlement for a few larval groups including sea urchins. Using larvae of H. erythrogramma and H. purpurascens, I next showed in larval settlement assays that bacteria on coralline algae enhanced settlement for both species but only larvae of H. erythrogramma responded to specific variation in the bacterial community composition. This specificity of response was then tested by isolating bacteria from the surface of coralline algae, and testing these against settlement by both sea urchins. One bacterium, Pseudoalteromonas luteoviolacea, was isolated from different species of corallines and induced larval settlement of both sea urchins, suggesting a common settlement cue across coralline algae. 16S rRNA tag sequencing surveys of the relative abundance of bacteria on these algae in the field indicated a high abundance of bacterial groups not examined in larval assays and an extremely low number of the genus Pseudoalteromonas, suggesting a very low abundance of P. luteoviolacea in natural communities. In summary, this thesis presents a systematic overview of larval settlement studies using meta -analysis, a statistical framework for correlating larval settlement and bacterial communities and provides an understanding of dominant bacterial members that have yet to be examined in larval settlement studies.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the Univ ersity of New South Wales or its agents the right to archiv e and to make av ailable my thesis or dis sertation in whole or in part in the Univ ersity libraries in all f orms of media, now or here af ter known, subject to the prov isions of the Copy right Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in f uture works (such as articles or books) all or part of this thesis or dissertation.

I also authorise Univ ersity Microf ilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only ).

…………………………………………………………… ……………………………………..……………… ………...…….…

Signature Witness Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ...... Date ......

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………......

Date ……………………………………………......

Table of contents

Abstract ...... I Chapter 1 ...... 1 1.1 The larval stage of benthic marine invertebrates ...... 1 1.2 Larval settlement cues ...... 2 1.3 Coralline algae as settlement cues ...... 4 1.4 Bacteria on coralline algae as settlement cues ...... 5 1.5 Understanding bacterial settlement cues ...... 6 1.6 Study organisms ...... 7 1.6.1 Sea urchins ...... 7 1.6.2 Coralline algae ...... 9 1.7 Thesis aims ...... 9 1.8 Chapter synopsis...... 10 Chapter 2 ...... 12 2.1 Introduction ...... 12 2.2 Methods ...... 15 2.2.1 Database searching ...... 15 2.2.2 Experimental criteria ...... 15 2.2.3 Meta-analysis predictor variables ...... 16 2.2.3 Statistical analysis ...... 17 2.3 Results ...... 19 2.3.1 Overall larval settlement across all substrata ...... 19 2.3.2 Effect of latitude ...... 19 2.3.3 Effect of larval time to competence ...... 19 2.3.4 Effect of experimental conditions ...... 19 2.3.5 Variation of larval settlement across common natural substrata ...... 20 2.3.6 Variation of larval settlement microbial biofilms, mixed vs. single species...... 21 2.3.7 Variation of larval settlement across macroalgal taxonomy...... 23 2.3.8 Variation of larval settlement to the Corallinales ...... 24 2.3.9 Variation of larval settlement among larval developmental type ...... 24 2.3.10 Variation in larval settlement across environmental regions ...... 24 2.3.11 Settlement in response to manipulated macroalgal biofilms ...... 25 2.4 Discussion ...... 41 Chapter 3 ...... 47 3.1 Introduction ...... 47 3.2 Methods ...... 50 3.2.1 Larval culturing ...... 50 3.2.2 Larval settlement in response to macroalgae ...... 50 3.2.3 Algal biofilm manipulations...... 51 3.2.4 Larval settlement in response to coralline algae treated with antibiotics ...... 52 3.2.5 Comparison of natural epiphytic bacterial communities on macroalgae and shell grit ...... 53 3.2.6 Correlation of larval settlement with bacterial community similarity and OTUs within communities ...... 54 3.2.7 Statistical analyses ...... 55 3.3 Results ...... 57 3.3.1 Larval settlement in response to coralline algae ...... 57 3.3.2 Verification of antibiotic effect on surface associated bacteria ...... 57 3.3.3 Larval settlement in response to algae treated with antibiotics ...... 58 3.3.4 Comparison of natural epiphytic bacterial communities ...... 58

3.3.5 Correlation of larval settlement percentage with bacterial community similarity and TRFs within communities ...... 58 3.4 Discussion ...... 71 3.5 Supplementary material...... 76 Chapter 4 ...... 79 4.1 Introduction ...... 79 4.2 Methods ...... 81 4.2.1 Isolation of Pseudoalteromonas luteoviolacea and other strains from coralline algae ...... 81 4.2.2 Sequencing of 16S rRNA genes from bacterial isolates ...... 81 4.2.3 The terminal restriction fragment of P. luteoviolacea ...... 82 4.2.4 Growth of biofilms for settlement experiments...... 82 4.2.5 Enumeration of bacteria in biofilms ...... 83 4.2.6 Culturing larvae ...... 83 4.2.7 Larval settlement assays ...... 83 4.2.8 Statistical Analysis ...... 84 4.3 Results ...... 85 4.3.1 Phylogenetic affiliation of bacterial isolates from coralline algae ...... 85 4.3.2 Settlement of Heliocidaris erythrogramma in response to Pseudoalteromonas luteoviolacea and other bacterial isolates from coralline algae ...... 85 4.3.3 Settlement of Heliocidaris erythrogramma in response to mixed species biofilms of Pseudoalteromonas luteoviolacea and other bacterial isolates from coralline algae ...... 86 4.3.4 Settlement of Holopneustes purpurascens in response to Pseudoalteromonas luteoviolacea and other bacterial isolates from different coralline algae ...... 86 4.4 Discussion ...... 93 4.5 Supplementary material...... 97 Chapter 5 ...... 98 5.1 Introduction ...... 98 5.2 Methods ...... 101 5.2.1 Sampling of coralline algae ...... 101 5.2.2 Analysis of bacterial communities ...... 101 5.2.3 DNA extraction...... 101 5.2.4 Terminal restriction fragment length polymorphism (tRFLP) ...... 102 5.2.5 Pyrosequencing of 16S rRNA genes ...... 102 5.2.6 Statistical analysis ...... 103 5.3 Results ...... 105 5.3.1 Variation in bacterial communities among four species of coralline algae ...... 105 5.3.2 Temporal and spatial variability of bacterial communities on coralline algae...... 106 5.3.3 Representation of settlement inducing isolates in sequence libraries ...... 107 5.4 Discussion ...... 116 5.5 Supplementary material...... 124 Chapter 6 ...... 128 6.1 Comparison of culture dependent and independent techniques for understanding bacterially mediated settlement of invertebrate larvae ...... 135 Literature cited ...... 137

Abstract

Most benthic marine invertebrates have a biphasic life cycle, in which a planktonic larval stage alternates with a benthic adult stage. The transition between the larval and adult stage is typically guided by habitat-derived settlement cues and thus understanding the nature and distribution of settlement cues is a central theme in larval ecology. Both coralline algae and their epiphytic bacterial biofilms can be important settlement cues for marine invertebrate larvae, but the relationship between settlement and specific communities of bacteria is largely unknown. I investigated bacterial mediated settlement for larvae of the Australian sea urchins Heliocidaris erythrogramma and Holopneustes purpurascens and compared this to the community ecology of bacteria on coralline algae. I conducted a meta-analysis of putative larval cues from macroalgae to test the importance of coralline algae as settlement cues for invertebrate larvae generally and sea urchin larvae specifically. The meta-analysis revealed that coralline algae were the most inductive macroalgae for a variety of larval groups, but epiphytic bacteria only enhanced larval settlement for a few larval groups including sea urchins. Using larvae of H. erythrogramma and H. purpurascens, I next showed in larval settlement assays that bacteria on coralline algae enhanced settlement for both species but only larvae of H. erythrogramma responded to specific variation in the bacterial community composition. This specificity of response was then tested by isolating bacteria from the surface of coralline algae, and testing these against settlement by both sea urchins. One bacterium, Pseudoalteromonas luteoviolacea, was isolated from different species of corallines and induced larval settlement of both sea urchins, suggesting a common settlement cue across coralline algae. 16S rRNA tag sequencing surveys of the relative abundance of bacteria on these algae in the field indicated a high abundance of bacterial groups not examined in larval assays and an extremely low number of the genus Pseudoalteromonas, suggesting a very low abundance of P. luteoviolacea in natural communities. In summary, this thesis presents a systematic overview of larval settlement studies using meta-analysis, a statistical framework for correlating larval settlement and bacterial communities and provides an understanding of dominant bacterial members that have yet to be examined in larval settlement studies.

Chapter 1

General introduction

1.1 The larval stage of benthic marine invertebrates

Most benthic marine invertebrate have a biphasic life cycle. The initial phase of the life cycle is characterised by planktonic larvae that are both morphologically and ecologically distinct from the adult (Thorson 1950). This early stage in benthic marine invertebrates is a critical period that has equally important consequences to an organism’s population dynamics as those experienced in the adult phase (Underwood and Fairweather 1989, Grosberg and Levitan 1992). Factors influencing the variation and timing of the ‘supply side’ of a species, and those associated with settlement and recruitment, have important implications on population dynamics and structure of marine assemblages (Keough 1983, Butler 1986, Minchinton and Scheibling 1991, Todd 1998, Barnay et al. 2003)

The supply of larvae can be influenced by factors associated with adult reproductive output (adult abundance and quantity of eggs or sperm), fertilisation success, the level of larval mortality, and the variety of oceanographic features (currents, tides, waves, wind) that disperse larvae (see Thorson 1950, Todd 1998). Clearly, the time spent in the water column, or dispersal period, has a large influence on the number of surviving offspring (Underwood and Fairweather 1989) by allowing such factors, especially mortality and ocean currents, to operate for longer periods of time.

The larval period varies greatly between species, lasting from minutes to months, and is generally dictated by the developmental mode (Strathmann 1985). Most larvae exhibit one of two main developmental or nutritional modes (Pawlik 1992) – planktotrophy or lecithotrophy. Planktotrophic larvae are characterised by feeding in the water column, longer developmental periods, and sophisticated morphologies Chapter 1 – General introduction

compared with lecithotrophic larvae, which do not feed but rely on egg yolk for nutrition, have short developmental periods and a much simpler body form (Thorson 1950, Strathmann 1985). Despite these differences, the development of both modes progress to a stage called ‘competence’ where larvae have matured enough (Pechenik and Gee 1993) to make the transition into the benthic mode of life and recruit into the adult population. This transition involves the descent of larvae from the water column to the substratum and the irreversible metamorphic transition into the adult body form. It must be noted that there are conflicting views on the terminology of this process in the literature, with some researchers defining ‘settlement’ (attachment to the substratum) and ‘metamorphosis’ as separate events, while others define the whole process as ‘settlement’. Hereon I define settlement according to the latter definition (the entire process) unless otherwise explicitly separated into attachment and metamorphic events.

The process of settlement is an important point in the life cycle of benthic marine invertebrates. Larvae must commence this process in a suitable habitat that will ensure subsequent survival and growth of healthy juveniles, leading to successful recruitment. This is especially important for completely sessile species as relocation is impossible after settlement has occurred. Also, it must be recognised that vulnerable early juveniles are often subject to high post-settlement mortality (Connell 1985, Hunt and Scheibling 1997). Ideally, some mechanism that guides larvae to suitable habitats must operate; otherwise random colonisation would occur, increasing the probability larvae end up in unsuitable habitats. Indeed, there is much observational and experimental evidence demonstrating that marine invertebrate larvae exhibit complex behaviour in response to environmental cues that presumably guide larvae to suitable habitats (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001).

1.2 Larval settlement cues

Larvae of many marine species respond to environmental cues of both physical and biochemical (chemical but derived from biological sources) nature. Responses to these 2

Chapter 1 – General introduction

types of stimuli are not likely to be mutually exclusive and highlight the complexity of interactions occurring during the settlement phase (Pawlik 1992). It has been suggested that different cues operate in a hierarchical fashion, for example, physical cues act in a manner that allows larvae to orientate themselves to substrata and chemical cues emanating from or associated with substrata initiate metamorphosis (Williams 1965, Pawlik 1992). Physical cues that influence larval behaviour include luminosity, gravity, temperature, salinity, hydrostatic pressure, water flow velocity and surface texture. Sponge larvae, for example, are known to be positively phototactic (swim toward light) during early periods of the larvae phase(a feature exploited to capture larvae in the field; Whalan et al. 2012) and it is thought that some hydroid species use this response to come in contact with the undersides of algal blades (Williams 1965). Despite the importance of physical cues (especially at larger spatial scales, 101-103 m and days-weeks, and which are key to delivering larvae closer to the shore), it appears biological and chemical characteristics of the substratum play a role as cues for settlement of marine invertebrate larvae to specific sites or habitats, either by attracting larvae to particular substrata or by inducing metamorphosis (Pawlik 1992, Hadfield and Paul 2001).

Many components of the marine environment induce the settlement and metamorphosis of invertebrate larvae (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001) and are generally ecologically consistent with the adult lifestyle, e.g. herbivorous species settling in response to algal food sources. These components include conspecifics (Pearce and Scheibling 1990b), prey species (Ritson-Williams et al. 2003), host organisms (Williamson et al. 2004), food sources (Pawlik 1989) and microbial biofilms (Thiyagarajan et al. 2006). These categories are not exclusive and the source of cues may act, for example, as both host and food (Williamson et al. 2004). Also, larval settlement of a species is not strictly in response to only one source of cue and settlement to different substrates is observed (Pearce and Scheibling 1991, Dworjanyn and Pirozzi 2008). However, certain sources of cue in the marine environment have received more attention than others as inducers of larval settlement for a number of species.

Chapter 1 – General introduction

1.3 Coralline algae as settlement cues

Coralline algae (Phylum Rhodophyta, Class Florideophyceae, Family Corallinales) are considered one of the most important sources of settlement cues for marine invertebrate larvae (Hadfield and Paul 2001, Roberts 2001, Steinberg and de Nys 2002). A taxonomically broad range of larval species are induced to settle in response to coralline algae including abalone (Daume et al. 1999b, Daume et al. 1999a, Roberts et al. 2004, Williams et al. 2008), corals (Heyward and Negri 1999, Negri et al. 2001), various gastropods (Hayakawa et al. 2007), sea urchins (Pearce and Scheibling 1990a, Kitamura et al. 1993, Huggett et al. 2006, Swanson et al. 2006, Dworjanyn and Pirozzi 2008, Swanson et al. 2012a), sea stars (Johnson and Sutton 1994), and sponges (Whalan et al. 2012). This highlights the importance corallines play in the early life history of marine invertebrates. However, the red algal family Corallinales encompasses many different species of coralline algae and indeed, many different species have been shown to induce the settlement of many different larvae. This phenomenon highlights a fundamental question for coralline algal settlement cues – is there a common cue from corallines involved in larval settlement in response to coralline algae? Or do different coralline algae harbour specific settlement cues?

The evidence to date indicates that larval settlement in response to coralline algae is a result of a variety of components of the alga including its associated epiphytes. The epiphytes are generally microbial biofilms that exist as aggregates of cells enmeshed in extracellular polymeric matrix that allow the biofilms to attach to the surface of the coralline algae. Thus settlement cues from coralline algae can originate from the algae (Morse and Morse 1984, Williams et al. 2009, Roberts et al. 2010a) but also by epiphytic bacterial biofilms or specific bacteria isolated from the surface of the algae (Johnson et al. 1991b, Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006, Tebben et al. 2011). In cases where cues have been associated with the algae per se, surface associated chemicals have been identified putative settlement cues (Morse and Morse 1984, Williams et al. 2009). Williams et al. (2009) suggested that the algal surface biomarker composition, characterised by mass-to-charge ratios, was involved in the variation of settlement to different coralline species by abalone. This suggested 4

Chapter 1 – General introduction

the involvement of integrated cues, with settlement linked to a complex of chemical cues rather than a single cue.

1.4 Bacteria on coralline algae as settlement cues

A number of studies have shown that the settlement response of invertebrate larvae to coralline algae is linked to bacteria on the surface of the algae. Various methods have been used in these studies, including treating algae with antibiotics(Johnson and Sutton 1994, Huggett et al. 2006, Swanson et al. 2006)or physically wiping algal surfaces (Huggett et al. 2006, Dworjanyn and Pirozzi 2008) to reduce the amount of epiphytic bacteria or by testing bacteria isolated from coralline algae (Negri et al. 2001, Huggett et al. 2006, Tebben et al. 2011) in settlement assays. While removal of bacteria from the surface of coralline algae can reveal the involvement of bacteria in the settlement response, it does not provide information of about the species of bacteria that may be involved.

In contrast, studies using isolated strains have revealed a number of bacterial species present on coralline algae that induce the settlement of invertebrate larvae. Larvae of sea urchin Heliocidaris erythrogramma were induced to settle by species of Shewanella, Vibrio, Photobacterium, Pseudomonas and Pseudoalteromonas bacteria (Swanson et al. 2006). Other species in the genus Pseudoalteromonas isolated from coralline algae have also been shown to induce the metamorphosis (but not settlement) of coral larvae (Negri et al. 2001, Tebben et al. 2011). Interestingly, Johnson et al. (1991b) was unable to show bacteria (including representatives of Vibrio and Alteromonas) isolated from coralline algae to induce larval settlement of the sea star, Acanthaster planci, unless the bacteria were attached to coralline algae (using experiments where bacteria were removed from coralline algae first before bacterial isolates were then allowed to colonise).

Culture-based approaches are biased towards a small fraction of the total bacterial species pool and thus limit our understanding of bacterially mediated settlement on

Chapter 1 – General introduction

coralline algae. Currently, little is known regarding the bacterial species composition on coralline algae besides those that are currently culturable. Different species of coralline algae may harbour similar bacterial communities and this might offer an explanation as to why different coralline algae can induce high levels of settlement. Clearly, better understanding of the composition and variability of bacteria on coralline algae will provide a stronger platform with which to make inferences of bacterial mediated larval settlement by coralline algae.

1.5 Understanding bacterial settlement cues

Despite biofilms being well known to induce larval settlement of many invertebrate species, the specific components of biofilms that induce larval settlement are still not well understood. General approaches to understanding the characteristics of biofilms that induce larval settlement have considered a variety of biofilm attributes (Thiyagarajan et al. 2006) such as cell densities, species composition, biomass and biofilm chemical attributes such as whole biofilm chemical signatures (Chung et al. 2010)or specific extracellular polymeric matrix components (Lam et al. 2005). From these, the hypothesis that the species composition of biofilms governs whether a larva will chose to accept or reject the biofilm for initiation of larval settlement has been suggested (Qian et al. 2003, Lau et al. 2005b, Thiyagarajan et al. 2006). However, it is unclear if the larval settlement in response to natural mixed species biofilms is due to a multispecies, emergent property of biofilms, if several inductive strains exist in biofilms, or if the induction of settlement is wholly or primarily due to one or only a few strongly inductive bacterial species.

Both culture-dependent and culture-independent techniques have been used to understand which species in biofilms induce larval settlement. For culture-dependent techniques, bacteria are isolated from the biofilm and tested in settlement assays with larvae (Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006). While this approach can be successful in identifying which individual bacteria can induce larval settlement, it is problematic if settlement is a result of a multispecies phenomenon or

Chapter 1 – General introduction

if the cultured strains are not representative of the bacterial community in situ. For culture-independent techniques, bacterial communities are characterised (generally by DNA sequencing or fingerprinting or other molecular techniques) and the variation among bacterial assemblages qualitatively compared to rates of larval settlement to these biofilms (Webster et al. 2004, Lau et al. 2005b). While these techniques have been used to examine whole bacterial communities in relation to larval settlement, there has been no unequivocal test to show that similar bacterial community compositions induce similar levels of larval settlement, which would suggest an emergent property of the biofilm induces larval settlement. In part this is due to the somewhat limited statistical evaluation used to date to compare bacterial community data to larval settlement; in general previous studies have not made use of a variety of (in particular) multivariate techniques available for the analysis of high diversity communities (Anderson and Willis 2003).Lastly, whether species identified to induce larval settlement through culture dependent and independent approaches are consistent is not clear but is critical in establishing the validity of conclusions obtained from both approaches.

1.6 Study organisms

1.6.1 Sea urchins

I used the larvae of Heliocidaris erythrogramma and Holopneustes purpurascens- two species of sea urchin common to rocky, shallow subtidal habitats in the temperate south-eastern (SE) coast of Australia in order to investigate coralline algal settlement cues. Both species produce lecithotrophic larvae that have short developmental periods, are relatively easy to culture in the laboratory, and respond to macroalgal and bacterial cues (Huggett et al. 2006, Swanson et al. 2006). Their reproductive periods are offset, allowing for near year round experimentations. These organisms provide good models with which to investigate coralline algal settlement cues.

H. erythrogramma (Family Toxopneustidae) is a bottom dwelling, regular sea urchin that is conspicuous on rock platforms from low to sub-tidal regions in SE Australia. It 7

Chapter 1 – General introduction

typically lives in depressions bored within the rock which offer substantial protection from physical disturbance (especially for aspiring PhD students trying to collect them). This species is considered an ecologically important component of temperate rocky reefs owing to its high density and ability to modify habitats through extensive grazing of seaweeds (Wright et al. , Ling et al. 2010). H. erythrogramma is a model species in developmental biology (see RA Raff, Indiana University) and much is known on the molecular biology of the lecithotrophic development of its larvae and its sister species with planktotrophic development H. turbuculata (Raff et al. 1999, Wilson et al. 2005). The reproductive season of H. erythrogramma is during the summer months (Williams and Anderson 1975) when shortly after, recruitment levels, especially on coralline algae, increase (Huggett et al. 2008).

H. purpurascens (Family Temnopleuridae) is a canopy dwelling, regular sea urchin, spending a majority of its life enmeshed in the fronds of seaweeds, primarily the red foliose alga Delisea pulchra and the kelp Ecklonia radiate (Steinberg 1995, Williamson et al. 2004). The reproductive period of H. purpurascens is predominately during winter as suggested by gonad indices (Williamson and Steinberg 2002), however, recruits are found year round (Swanson et al. 2006).

H. purpurascens is induced to settle in response to a range of macroalgae, especially coralline algae and the red alga Delisea pulchra (Swanson et al. 2006). Initial research into settlement cues of H. purpurascens has focused on those from D. pulchra (Williamson et al. 2004). Settlement has been linked to the production of histamine by this alga, which is now known as a strong, naturally occurring, chemical inducer of metamorphosis of H. purpurascens (Swanson et al. 2004). Histamine was identified through bio-guided settlement assays of chemical extracts and algal conditioned seawater of D. pulchra, and further shown to be present in field collected samples (Swanson et al. 2006). Indeed, smaller size classes (<30 mm) of H. purpurascens are mostly found enmeshed within D. pulchra (Williamson et al. 2004) which is consistent with its histamine production.

Despite this relationship with D. pulchra, comparable number of newly recruited urchins (<5mm) have also been found among coralline turfs by (Swanson et al. 2006).

Chapter 1 – General introduction

Coralline algae (Corallina officinalis or Amphiroa anceps) also induced high levels of larval settlement even though histamine has not been detected in these algae (Swanson et al. 2006). This led to the hypothesis that, unlike for histamine from D. pulchra, bacteria on the surface of coralline algae induce the settlement of H. purpurascens (Swanson et al. 2006) .

1.6.2 Coralline algae

Coralline algae (Rhodophyta) are a unique group of red algae characterised by the incorporation of calcium carbonate in the cell wall, resulting in hardened thalli (Johansen 1981). They are ubiquitous in many marine habitats across the globe from polar to tropical regions and at depths from intertidal zones to as deep as 270m (Harvey et al. 2005). Two growth forms occur among coralline species. Geniculate or articulated corallines are upright, branched forms consisting of calcified units connected by uncalcified joints called ‘genicula’. Non-geniculate or non-articulated coralline algae lack genicula and form crusts (‘encrusting’ species) on substrata. Taxonomy of coralline algae has undergone many changes in the recent decade (Harvey et al. 2005) and is still in a state of flux (Broom et al. 2008), although recent molecular studies have contributed to our understanding (Broom et al. 2008). Coralline algae play many important ecological roles besides acting as settlement cues for marine invertebrate larvae by providing habitat and food for a variety of fish and invertebrates (Steneck 1982, Garland et al. 1985, Littler et al. 1995), stabilising the substratum (Johansen 1981) and maintaining wave-resistant fronts in tropical coral reefs (Ady 1975).

1.7 Thesis aims

The major aims of my thesis are to:

1. Characterise previous studies on larval settlement through quantitative analysis. 9

Chapter 1 – General introduction

2. To determine the source of settlement cues from different coralline algae for the Australian sea urchins Holopneustes purpurascens and Heliocidaris erythrogramma. 3. Use quantitative statistical methods to understand correlation patterns of larval settlement with whole bacterial communities and individuals within them. 4. Compare larval settlement induction by bacteria isolated from different coralline algae. 5. Understand the taxonomic composition of bacterial communities on coralline algae.

1.8 Chapter synopsis

In Chapter 2 I conduct a quantitative literature review of larval settlement in response to common settlement substrata examined in laboratory based settlement assays (which make up the bulk of the experimental studies investigating settlement cues), using a meta-analysis (Glass 1976). While many reviews have been conducted on larval settlement cues (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001), the variation in larval settlement to natural substrata is often not discussed in detail and ‘tends to vanish’ when the work is summarised (Raimondi and Keough 1990), resulting in the notion that larvae either do or do not respond to particular substrata. Biological systems are inherently variable and a review which detailed this variation, and which also more formally characterised the literature on the response of invertebrate larvae to both corallines and biofilms was done. A particular focus in the analysis was the response of larvae to coralline algae and was used as a hypothesis driver for my thesis.

In Chapter 3, I investigated bacterially mediated larval settlement of sea urchins in response to coralline algae. I showed settlement cues on coralline algae were associated with epiphytic bacteria by use of manipulative experiments. I used culture independent techniques to test whether larval settlement of sea urchins is related to bacterial community similarity and bacterial operational taxonomic units (OTUs) within communities. In this chapter I employ multivariate statistical techniques not yet used

Chapter 1 – General introduction

in larval ecology to provide a much deeper level of analysis than currently shown in the field.

In Chapter 4, the aim is to further examine potentially inductive bacterial isolates from different coralline algae, including the presence of the highly inductive bacterial species Pseudoalteromonas luteoviolacea. P. luteoviolacea induces strong settlement of H. erythrogramma (Huggett et al. 2006) and has previously been isolated form the coralline alga A. anceps. I isolated P. luteoviolacea from other coralline algae and tested whether they can also induce the settlement of H. erythrogramma. Furthermore, I also tested whether P. luteoviolacea can induce settlement of the sympatric sea urchin species H. purpurascens and if P. luteoviolacea still induces larval settlement in mixed species biofilms.

The aim of Chapter 5 is to investigate the taxonomic identities of bacteria on coralline algae using 16S rRNA gene sequencing to deepen our understanding of the bacteria present on coralline algae. Furthermore, the experimental design allowed me to understand the temporal and spatial variability of bacterial communities among coralline algae. Since bacteria on the surface of coralline algae acted as settlement cues for the two species of sea urchin that I used in this study, it was clear that identifying bacteria on coralline algae and how they vary over space and time would contribute to our understanding of bacterial mediated settlement cues on coralline algae. The only bacterial species on coralline algae known to induce larval settlement were discovered using culture based methods (Johnson et al. 1991a, Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006, Tebben et al. 2011) and I investigated the presence of these species with the species identified using culture independent techniques to understand the ecological consistency of previous studies. The results of this chapter are discussed in the context of bacteria known to induce larval settlement, the emerging macroalgal microbiome concept and the discrepancy between culture dependent and independent methods.

In Chapter 6 I discuss the findings from each chapter in the context of the generality of coralline algae settlement cues, the difficulty in merging results obtained from culture dependent and independent approaches and avenues for future research.

Chapter 2

A meta-analysis of larval settlement to natural settlement cues

2.1 Introduction

The two-phase life history exhibited by many marine benthic invertebrates is critical for the establishment and persistence of populations in the marine environment (Strathmann 1974, Grosberg and Levitan 1992). The dispersive planktonic larval stage, which can last from hours to months and is subject to oceanographic currents that can carry larvae great distances, has given slow moving or completely sessile organisms the chance to extend their colonisation of home ranges or to inhabit new environments that would otherwise be largely inaccessible to the adult mode of life (Grantham et al. 2003). Indeed, the ‘supply side’ of an organism’s life history – the planktonic stage – has received increasing attention in the last 20 years as a key determinant in population dynamics of benthic marine organisms (Underwood and Fairweather 1989, Grosberg and Levitan 1992).

In order for planktonic larvae to transit into the benthic mode of life and recruit into the adult population, larvae must typically first come into contact with, and then respond to, specific environmental cues that set in motion the process of settlement and metamorphosis (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001, Steinberg et al. 2001).

Environmental cues that induce the settlement and metamorphosis of marine invertebrate larvae can be broadly categorised into physical and (bio-) chemical cues. While physical cues (e.g. water flow, gravity, luminosity and surface contour) are important and direct or orientate larvae to the substratum (Williams 1965), there is much experimental evidence suggesting that putative chemicals cues associated with the substratum are the probable cause of the physiological transition from the larva to Chapter 2 – A meta-analysis

the adult body form (Pawlik 1992, Hadfield and Paul 2001). In addition to chemical cues from natural substrates, there are also examples of artificial chemical inducers (i.e. likely not to be found naturally in the marine environment and which may structurally mimic true inducers) of larval settlement (Morse 1985, Pawlik 1990, Rodriguez et al. 1993, Zhao et al. 2003) highlighting the strong effect chemicals have on the induction of larval settlement. From an ecological perspective, however, it is natural chemical cues and their origin that are critical to our understanding of marine communities and the processes that underlie the patterns we observe (Hadfield and Paul 2001).

Various substrata in the marine environment act as settlement cues for invertebrate larvae. Primary examples include conspecifics (Pearce and Scheibling 1990b), macroalgae (Huggett et al. 2006, Williams et al. 2008), corals (Ritson-Williams et al. 2003), and microbial biofilms (Pearce and Scheibling 1991, Wieczorek and Todd 1998, Huang and Hadfield 2003, Webster et al. 2004). Generally, these substrata fall under ecologically meaningful categories such as host organisms (e.g. macroalgae for a canopy dwelling sea urchin; Swanson et al. 2004) or prey/food sources (e.g. corals for a carnivorious nudibranch; Ritson-Williams et al. 2003) and provide ecologically consistent explanations for the observed settlement response. It is clear, however, that variation in the settlement of larvae to each substrata or cue type exists even when adult ecology (e.g. diet preference) would suggest a very specific cue (Ritson- Williams et al. 2003). Understanding this variation is critical to our understanding of the ecology of benthic ecosystems and the predictive models associated with them (Raimondi and Keough 1990)

While various reviews on cues that induce the settlement and metamorphosis of marine invertebrate larvae (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001, Steinberg and de Nys 2002) have highlighted common substrata as settlement cues, the chemical (full or partial) identity of putative chemical cues from substrata and the ecological and evolutionary reasoning behind the observed responses to certain cues, the variation of larval settlement to natural settlement cues has not been evaluated to date. Most reviews are qualitative and narrative in form, and a primary

Chapter 2 – A meta-analysis

concern is how strong generalisations of larval settlement in response to common natural settlement substrata are, considering such summaries can neglect the variation present in the original studies (Raimondi and Keough 1990). Thus, our understanding of the importance of certain substrata as settlement cues could be misconstrued, particularly if there are many examples of larval settlement (that is not necessarily high) to a common substratum. This is particularly relevant for settlement in response to coralline algae, which are believed to be a well-known source of ‘strongly‘ inducing settlement cues for a variety of marine invertebrate larvae but for which the strength of the association have never been examined critically. In this thesis I specifically investigate coralline algae settlement cues but, here, first critically analyse general patterns of larval settlement in response to coralline algae.

Here, I conduct a quantitative meta-analysis of the settlement responses of marine invertebrate larvae to natural settlement cues. I compare larval settlement in response to commonly identified natural settlement substrata, and examine whether settlement is consistent across larval taxonomy, common larval type and larval developmental type. Macroalgae are a common source of natural settlement cues and I examine whether variation in larval settlement can be attributed to particular taxonomic groups, with a particular interest in the coralline algae that are often assumed to be highly inductive to a broad range of larvae. Microbial biofilms are another important source of natural settlement cues, and much research has been conducted on elucidating which microbial species in biofilms induce settlement (Wieczorek and Todd 1998). Subsequently, I compare the magnitude of larval settlement to microbial biofilms presented as natural mixed consortia or single species biofilms of bacteria or diatoms. Lastly, I evaluate the effect of experimental conditions of settlement assays on the settlement response of larvae.

Chapter 2 – A meta-analysis

2.2 Methods

2.2.1 Database searching

I selected relevant published studies of marine invertebrate larval settlement by searching the ISI Web of Science database using the following search criteria; (Larva* AND settlement OR metamorphosis) and supplemented these results with the search criteria; (settlement OR metamorphosis AND echinoderm* OR porifera OR cnidaria* OR mollusc* OR chordat* OR polychaet* OR annelid* OR bryozoa* OR barnacle). In addition, I included other relevant studies that were not found with my initial search criteria that I identified through referencing in published literature. The latter should not introduce an objective bias and it is unclear why such studies were not identified using the search criteria above.

2.2.2 Experimental criteria

I included studies that examined the settlement response of marine invertebrate larvae in laboratory based, single dish, no choice, static water settlement assays with substrata present within dishes or other similar experimental containers. I did not use data from ‘double dish’ or ‘choice’ assays as there is non-independence between treatments leading to correlations in the data which were difficult to integrate into the meta-analysis. I also did not examine experiments investigating combinations of treatments (e.g. diatom biofilms with conspecific mucus trails). Since there is evidence of bacterial mediated larval settlement in response to macroalgae, I included settlement experiments that compared settlement in response to macroalgae with manipulated epiphytic bacteria (physically wiped off or reduced with antibiotics) and associated unmanipulated controls.

I only included experiments that had a ‘filtered sea water’ (FSW) treatment or an equivalent negative control as these were required to standardise larval settlement responses across studies and experiments within the meta-analysis. The exception to this was when comparing bacterial mediated larval settlement in response to macroalgae where settlement to biofilm manipulated macroalgae was compared with

Chapter 2 – A meta-analysis

the corresponding unmanipulated controls. Settlement experiments analysed were conducted over various time periods ranging from 1 h to 72 h, however, a majority of experiments were conducted for 24 h periods. If settlement was examined at multiple time points then data from the nominal time point of 24 h was extracted. The great majority of experiments presented data as the percentage (%) settlement or metamorphosis. Means and measures of variance (converted to standard deviation) were extracted from plots using the distance measuring tool in Foxit Reader 4.3 (http://www.foxitsoftware.com) along with sample sizes.

2.2.3 Meta-analysis predictor variables

I categorised our dataset based on larval, substratum and experimental characteristics. Larval species were categorised by major larval groups (e.g. sea urchins, abalone, corals, polychaetes) and were classified to Phylum and Class level using currently accepted taxonomy (World register of marine species, WoRMS, www.marinespecies.org). I also recorded larval time to competence (days) and development type (planktotrophic or lecithotrophic). Substrate types were categorised into common natural substrates identified in the literature, including macroalgae, seagrass, microbial biofilms (present on abiotic surfaces), conspecifics, other animals and sterilised substrates. Algal and plant species were classified from Phylum to genus level using currently accepted taxonomy (AlgaeBase, www.algaebase.org). Microbial biofilms were further categorised according to whether they were present on natural (rock, coral rubble and shell) or artificial substrates (petri dish and glass slides) and whether they were presented as natural mixed consortia or single species bacteria or diatom biofilms. For the comparison of bacterial mediated settlement in response to macroalgae, data was categorised into biofilm present and biofilm removed. Experimental characteristics that were recorded included assay volume (mL),assay length (h), number of larvae per replicate, larval density per replicate, and substrate size as area either (mm2), volume (mm3), mass (g) (depending on what was given from a particular experiment). Lastly, I recorded the latitude of each experiment and classified experiments into tropical and temperate bioregions (no polar experiments were recorded).

Chapter 2 – A meta-analysis

2.2.3 Statistical analysis

I quantified the effect of difference settlement substrata on larval settlement response using the effect size Hedges d (Hedges et al. 1985), which is the standardised mean difference between treatment and control weighted by sample size. This effect size was chosen because the nature of the data from settlement assays is similar (see above) with means and variances on the same scale.

There were multiple studies from the same research groups, multiple experiments within studies and multiple treatments (sharing a common control) within experiments, highlighting the problem of different levels of non-independence of observations. If the same treatments were used in different experiments within the same study, I assumed these to be repeated observations of the same settlement response at a similar time by the same researcher and subsequently modelled this effect as random variation in a linear mixed model. I treated multiple treatments (e.g. different species of algae as settlement cues) sharing a common control as independent observations since the settlement response in FSW was generally marginal to zero. Ideally, having multiple random terms in the model would be preferable, but this was difficult to specify with the statistical package I used (see below) which was optimised for nested models and not fully developed.

I used linear mixed models (LMM) to estimate the variation in effect size attributed to the various predictor variable outlined above, including continuous and categorical factors. Missing combinations of levels in categorical factors (i.e. not all larval phyla were tested with every substrata type) prevented the creation of LMMs incorporating interactions; subsequently I created models within each level of a factor (e.g. substrata type) to contrast differences between levels (e.g. larval phyla) within the factors. Restricted maximum likelihood (REML) estimates of fixed effects and their 95% confidence intervals were obtained from the LMMs (using the “cell means” parameterisation (Pinheiro and Bates 2000) for categorical factors) and used to generate all figures. Significance of the variability of fixed effect terms in the LMMs was tested using analysis of variance (ANOVA). ANOVA of LMMs provides a test of significance for both the intercept (in the case of categorical factors, the estimate of

Chapter 2 – A meta-analysis

the mean response of all fixed factors across the population) and the variability among fixed factors(Pinheiro and Bates 2000). The significance between fixed effect estimates and each estimate from zero was made by observing whether 95% confidence intervals overlapped with each other or with zero, respectively. I used the R package LME4 (Bates 2010) to create and analyse all LMMs in this study. Estimates of R2 for LMMs with continuous predictors were derived using methods of Edwards et al. (Edwards et al. 2008)

Chapter 2 – A meta-analysis

2.3 Results

2.3.1 Overall larval settlement across all substrata

Laboratory based, single dish settlement assays conducted with marine invertebrate larvae exposed to natural settlement substrata have resulted, on average, in a positive settlement response across all substrata examined (REML estimate of Hedges d effect size with 95% confidence interval, d=6.15 +/- 1.07). Effect sizes ranged from -32 to 96 across all observations revealing cases of both inhibition and induction of larval settlement (Figure 1).

2.3.2 Effect of latitude

There was no strong latitude effect observed in the settlement response of larvae to all natural substrata. Larval settlement responses increased slightly with latitude although there was very little variance explained by the model (Figure 2, F1,332=5.82 P=0.0163, R2=0.017).

2.3.3 Effect of larval time to competence

The length of time (days) until larval competence was reached, and settlement assays begun, had no effect on the settlement response of larvae (Figure 3, F1,101=0.231 P=0.635, R2=0.0022).

2.3.4 Effect of experimental conditions

Overall, experimental conditions did not have any measureable effect on the settlement response of larvae. Neither the temperature at which settlement 2 experiments were conducted (Figure 4a, F1,109=1.26 P=0.263 , R =0.011), the assay 2 length (Figure 4b, F1,118=3.44 P=0.066 , R =0.028), nor the substrata size used (using mm2, which was the most widely reported measure of substrate size; Figure 4c, 2 F1,59=0.104, P=0.747 , R = 0.002) had an effect on the settlement response of larvae. A slight increase in the settlement response of larvae was observed as the density of 2 larvae increased in settlement assays (Figure 4d, F1,109=4.67, P=0.033 , R =0.041) but only a small proportion of variability was explained by the model. 19

Chapter 2 – A meta-analysis

2.3.5 Variation of larval settlement across common natural substrata

There was significant variation in the magnitude of larval settlement in response to the most common settlement substrata identified as settlement cues in the marine environment (Figure 5, F5,114=2.76, P=0.022). This was driven by the difference in settlement to macroalgae and sterile substrata (the only two groups with non- overlapping 95% confidence intervals). Overall, the greatest mean settlement was to seagrasses but this group was poorly represented in the analysis with respect to the types of larvae used (Echinoderms only, see below). Macroalgae and natural microbial Biofilms, which were much better represented in our analysis, both induced strong larval settlement that were of similar magnitudes and variation to each other. ‘Conspecifics’ and ‘Other Animals’ had lesser effects on larval settlement than the other groups tested, and there was a zero effect size of sterilised substrates on the settlement response of larvae.

Following these general comparisons, I then contrasted the settlement response of larvae at the level of larval phyla within each substrata type separately (Figure 6). There was significant variation in the settlement response of larval phyla to macroalgae (F4,190=2.61, P=0.037), likely driven by the negative response of chordates compared with other phyla. Echinoderms, molluscs and cnidarians showed the greatest settlement in response to macroalgae, with these responses similar each other (overlapping 95% confidence intervals). Limited evidence (n=4) suggested that the larvae of porifera do not respond consistently to macroalgae and chordates (n=3) responded negatively, although the wide confidence intervals in both cases suggested either poor model fits by these predictors or an effect of low observation numbers. There was significant variation in the settlement response of larval phyla to natural microbial biofilms (F5,68=3.01, P=0.017). Annelids and echinoderms responded the most strongly to natural microbial biofilms, while members of cnidaria, mollusca, proferia and arthropoda responded positively but not, significantly to biofilms (95% confidence intervals overlapping with zero). There was no variation in response to conspecific cues (F2,8=1.8,P=0.27), a measure of gregarious settlement, among larval phyla. However, wide 95% confidence intervals were observed suggesting a poor fit to

Chapter 2 – A meta-analysis

the model by these predictors, or an effect of low observation numbers. There was no difference in the settlement response of larvae to other (non-conspecific) animals

(F1,29=0.00266, P=0.96) among larval phyla, with settlement non-significant from zero for the two phyla represented. There was a significant settlement response of echinoderms, the only phyla represented, to seagrass (F1,8=7.26, P=0.027). There was significant variation in settlement response of larval phyla to sterilised substrata

(F1,16=42.0, P=<0.0001). Echinoderms showed a consistently low response to sterilised substrata, while cnidarians exhibited no settlement responsesterilised substrata.

I then contrasted the settlement response of larvae grouped by common name (e.g. sea urchin or abalone) within each substrata type separately (Figure 7). There was significant variation in the settlement response of common larval groups to macroalgae (F7,187=2.17, P=0.039), which was driven by the difference in settlement of sea urchins, abalone and coral versus ascidians (non-overlapping 95% confidence intervals).Sea urchins, snails (non-abalone), abalone and corals all responded significantly to macroalgae (95% confidence intervals not overlapping with zero) but sea stars, sand dollars, sponges and ascidians did not. There was significant variation in the settlement response of common larval groups to natural microbial biofilms

(F6,60=2.81, P=0.018). Polychaetes and sea urchins responded significantly to biofilm cues but oysters, barnacles, sponges, corals and abalone did not. There was significant variation in the settlement response of common larval groups to conspecific cues

(F2,5=4.4, P=0.08). Sand dollars and sea urchins responded significantly to conspecific cues but sponges did not. There was non-significant variation in the settlement response of common larval groups to other animals (F1,29=0.00266, P=0.96). Nudibranchs responded significantly to other animals but corals did not. There was significant variation in settlement response of larval phyla to sterilised substrata

(F1,16=42.0, P=<0.0001). Echinoderms showed a consistent low response to sterilised substrata while cnidarians exhibited no settlement response.

2.3.6 Variation of larval settlement microbial biofilms, mixed vs. single species

Natural mixed species microbial biofilms generally induce strong larval settlement. I contrasted the magnitude of larval settlement in response to natural biofilms with that

Chapter 2 – A meta-analysis

of single species biofilms of bacteria or diatoms (generally isolated from natural biofilms). Mixed natural biofilms induced a significantly greater larval settlement response than either single species biofilms of bacteria or diatoms (Figure 8,

F2,398=23.3, P=<0.0001; 95% confidence intervals not overlapping). However, there were significant levels of settlement to each of these biofilm types (95% confidence intervals not overlapping with zero), suggesting some level of larval settlement in response to all biofilm types.

I further contrasted the settlement response of larvae at the level of larval phyla within each biofilm type separately (Figure 8). There was significant variation in the settlement response of larvae to single species biofilms of bacteria (F4,209=13.5, P=<0.0001), which was likely driven by the difference in settlement response of annelids compared to cnidarians and arthropods (95% confidence intervals not overlapping). Annelids, echinoderms and cnidarians all significantly responded to single species biofilms with the settlement of annelids significantly greater than cnidarians. There was no effect of single species bacterial biofilms on settlement of molluscs and biofilms inhibited settlement of arthropods. There was significant variation in the settlement response of larval phyla to single species biofilms of diatoms (F2,107=8.8, P=0.0003). The settlement response of annelids and molluscs were similar and significantly greater than arthropods (based on 95% confidence intervals).

I further contrasted the settlement response of common larval groups within each biofilm type separately (Figure 9).There was significant variation in the settlement response of common larval groups to single species biofilms of bacteria (F6,54=2.8, P=0.0193), which was likely driven by the difference in settlement response of polychaetes, corals and barnacles (95% confidence intervals not overlapping). Polychaetes, corals and sea urchins significantly responded to single species biofilms but with settlement of polychaetes significantly greater than corals. There was no effect of single species bacterial biofilms on settlement of mussels or abalone and biofilms inhibited settlement of barnacles. There was significant variation in the settlement response of common larval groups to single species biofilms of diatoms

Chapter 2 – A meta-analysis

(F3,106=6.4, P=0.0005). The settlement response of polychaetes and abalone were similar and significantly greater than barnacles (based on 95% confidence intervals).

2.3.7 Variation of larval settlement across macroalgal taxonomy

I contrasted larval settlement in response to macroalgae at different algal taxonomic levels to determine if patterns at higher taxonomic levels were driven by particular taxa at lower levels (Figure 10). There was significant variation in the settlement response of larvae to macroalgae at the level of phylum (F2,48=4.138, P=0.022), driven by the difference in settlement to the Rhodophyta (red algae) and Ochrophyta (brown algae)(95% confidence intervals not overlapping). There was significant larval settlement in response to the Rhodophyta,which also had more than four times as many observations compared with the other two algal phyla. There was non-significant settlement in response to the Ochrophyta and the Chlorophyta (green algae) but there was, on average, a positive larval settlement response to these phyla. At the class level, patterns of larval settlement mirrored those observed at the phylum level as only a few classes were represented (F3,47=2.631, P=0.0609). There was significant variation in the settlement response of larvae to macroalgae at the level of order (F11,38=2.271, P= 0.0302). The Bonnemaisoniales and Corallinales were the only two algal orders to induce significant levels of larval settlement, withsettlement in response to Bonnemaisoniales significantly greater than Corallinales. Settlement in response to the remaining algal orders was, on average, positive, and at times large as significantly inducing algal orders, but wide 95% confidence intervals resulted in estimates that were non-significant from zero.

The Corallinales was the only algal order well represented at lower taxonomic levels, and thus I further contrasted larval settlement at lower taxonomic levels within the Corallinales only. There was no difference in the settlement response of larvae to the three coralline families represented in my analysis (F2,85=0.02019, P=0.818), with all inducing significant settlement. There were no significant differences in settlement in response to coralline genera (F10,72=0.5373, P=0.8581), likely due to the relatively wide confidence intervals. However, significant settlement in response to two articulated

Chapter 2 – A meta-analysis

genera, Amphiroa and Corallina, and two encrusting genera Phymatolithon and Hydrolithon was observed.

2.3.8 Variation of larval settlement to the Corallinales

I further contrasted the settlement response of larvae at the level of larval phyla among the well represented macroalgal order Corallinales (Figure 11). There was significant variation in the settlement of larval phyla to the Corallinales (F4,104=3.58, P=0.0089) which was driven by the difference in settlement of echinoderms, mollusc and cnidarians with chordates (95% confidence intervals not overlapping). There was significant settlement in response to the Corallinales by the echinoderms, mollusc and cnidarians, but there was no response by the porifera and settlement by the chordates was inhibited.

The significant settlement response of the echinoderms to the Corallinales was primarily driven by sea urchin settlement (Figure 12). The other echinoderms, sea stars and sand dollars, did not significantly respond to coralline algae. For the molluscs, abalone settled significantly in response to Corallinales but other snails did not despite exhibiting a similar mean level of settlement as abalone.

2.3.9 Variation of larval settlement among larval developmental type

There was no significant difference in the overall (across all natural settlement substrata tested) settlement response between planktotrophic and lecithotrophic larvae (Figure 13, F1,332=2.9, P=0.09), although there was a slightly greater settlement response by planktotrophic over lecithotrophic larvae. There was no significant difference between development types in response to macroalgae (F1,193=0.096,

P=0.76) or seagrass (F1,6=2.0, P=0.21) but there were significant differences in response to natural microbial biofilms (F1,74=16.3, P=0.0001), conspecifics (F1,11=12.1, P=0.0052), and sterile substrates (F1,16=42, P<0.0001).In each case where a significant difference was observed, planktotrophic larvae always exhibited a larger settlement response. Only lecithotrophic larvae have been used to examine settlement cues from other animals for which there was, on average, significant settlement (F1,32=20.6, P=0.0001).

2.3.10 Variation in larval settlement across environmental regions

Chapter 2 – A meta-analysis

There was significantly greater overall (across all natural settlement substrata tested) settlement of larvae in tropical regions compared with temperate regions (Figure 14,

F1,332=7.5, P=0.0065), however, settlement in both regions was significantly greater than zero. There was no difference in the settlement of larvae from tropical and temperate regions in response to macroalgae (F1,193=3.4, P=0.0066), natural microbial biofilms (F1,74=0.008, P=0.931), conspecifics (F1,11=0.23, P=0.638) or other animals

(F1,31=0.37, P=0.849). In each case, the settlement of larvae from tropical regions was significantly different from zero, and this was true for larvae from temperate regions for settlement in response to macroalgae and natural microbial biofilms. Larval settlement in response to seagrass was only examined in temperate regions and there was significant settlement (F1,8=7.26, P=0.0273). There was a significant difference in the settlement response of larvae between regions in response to sterile substrates

(F1,16=42, P<0.0001).Larvae from temperate regions showed a consistent low response to sterilised substrata while larvae from tropical regions exhibited no settlement response.

2.3.11 Settlement in response to manipulated macroalgal biofilms

Removing biofilms from the surface of macroalgae resulted in, on average, a reduction in larval settlement to manipulated macroalgae (Figure 15), however, there was insufficient evidence to suggest a significant reduction (F1,7=2.28, P=0.175). This trend was also observed when I contrasted manipulated algae from different phyla

(F1,7=0.369, P=0.563). Corallinales was again the only algal taxon that was well represented at lower taxonomic levels. There was no significant effect on larval settlement when biofilms on coralline algae were removed (F1,6=3.03, P=0.132) but, on average, there was a reduction in settlement to manipulated algae. I then contrasted the settlement of common larval groups (which is the same as larval phyla in this case due to the low number of observations) to coralline algae. There was no significant difference in the settlement between the common larval groups (F1,2=6.94, P=0.119), however, there was evidence to suggest that sea urchinsrespond to bacteria on coralline algae (95% confidence intervals not overlapping with zero).

Chapter 2 – A meta-analysis

Figure 1.The frequency histogram of the effect sizes (Hedges d) for larval settlement to natural substrata (not including settlement to single species biofilms of bacteria or diatoms). The dotted line represents the estimate of the mean effect size across all natural substrata by all larval species using a linear mixed model.

Chapter 2 – A meta-analysis

Figure 2. Variation in larval settlement with absolute latitude across all natural substrata examined. The regression slope was derived from linear mixed models of Hedges d against latitude.

Chapter 2 – A meta-analysis

Figure 3.Variation in larval settlement with time to larval competence (days) across all natural substrata examined. The regression slope was derived from linear mixed models of Hedges d against time to larval competence.

Chapter 2 – A meta-analysis

Figure 4. Variation in larval settlement with experimental variablesincluding a) assay temperature, b) assay length, c) substrate size and d) larval density within replicate assay dishes.The regression slopes were derived from linear mixed models of Hedges d with each of the assay variables.

Chapter 2 – A meta-analysis

Figure 5.Variation in larval settlement among major settlement substrata identified in the marine environment (except sterile substrata). Mean estimates and 95% confidence intervals are derived from linear mixed models of Hedges d against predictor variables. The dotted line at zero represents an effect of zero where there is no settlement response of larvae. Sample sizes for estimates are shown above the x- axis.

Chapter 2 – A meta-analysis

Figure 6.Variation in larval settlement among larval phyla within each major settlement substrata identified in the marine environment (except sterile substrata). Mean estimates and 95% confidence intervals are derived from linear mixed models created within each settlement substrata type separately.Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 7. Variation in larval settlement among common larval group within each major settlement substrata identified in the marine environment (except sterile substrata). Mean estimates and 95% confidence intervals are derived from linear mixed models created within each settlement substrata type separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 8. Variation in larval settlement among different biofilm types including natural mixed species biofilms (Mixed) and single species biofilms of previously isolated bacteria (Bacteria) or diatoms (Diatoms). Variation in larval settlement to different biofilm types is then contrasted among larval phyla. Mean estimates and 95% confidence intervals are derived from linear mixed models created within each biofilm type separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 9.Variation in larval settlement among common larval groups to different biofilm types. Mean estimates and 95% confidence intervals are derived from linear mixed models created within each biofilm type separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 10. Variation in larval settlement among different taxonomic levels of macroalgae. Open squares (□) represent taxonomic divisions within the Corallinales only. Mean estimates and 95% confidence intervals are derived from linear mixed models created within each taxonomic level separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 11. Variation in larval settlement among different larval phyla to the macroalgal order Corallinales.. Mean estimates and 95% confidence intervals are derived from linear mixed models. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 12. Variation in larval settlement among common larval groups to the macroalgal order Corallinales. Mean estimates and 95% confidence intervals are derived from linear mixed models. Sample sizes for estimates are shown above the x- axis.

Chapter 2 – A meta-analysis

Figure 13. Variation in larval settlement between larval developmental modes, and contrasted amongmajor settlement substrata identified in the marine environment (except sterile substrata). Mean estimates and 95% confidence intervals are derived from linear mixed models created within each substrata type separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 14. Variation in larval settlement between geographical zones, and contrasted amongmajor settlement substrata identified in the marine environment (except sterile substrata). Mean estimates and 95% confidence intervals are derived from linear mixed models created within each substrata type separately. Sample sizes for estimates are shown above the x-axis.

Chapter 2 – A meta-analysis

Figure 15. Effect of biofilm removal from macroalgae on larval settlement contrasted among all macroalgae, macroalgae phyla, the order Corallinales and common larval groups to the order Corallinales.Mean estimates and 95% confidence intervals are derived from linear mixed models created within contrast separately.Open squares (□) represent settlement to the macroalgal order Corallinales only.

Chapter 2 – A meta-analysis

2.4 Discussion

In this study, I provide the first broad attempt to quantify the magnitude of larval settlement in response to natural settlement cues by analysing more than 100 larval settlement assays constituting more than 800 settlement responses (multiple treatments were used in assays resulting in a greater ratio of responses to assays).Across all experiments there was, on average, a positive settlement response of invertebrate larvae to all natural substrata tested (not including single species biofilms of bacteria or diatoms) confirming that natural substrata in the marine environment play important roles in larval ecology by providing settlement cues. While many natural substrata in the marine environment have been identified to induce the settlement of invertebrate larvae (Pawlik 1992, Rodriguez et al. 1993, Hadfield and Paul 2001, Steinberg and de Nys 2002) a general quantitative analysis of the strength of variation in settlement to different substrata has not been previously investigated.

My results revealed significant variation in larval settlement to different types of natural substrata. Macroscopic substrata such as seagrass and macroalgae as well as microscopic organisms in biofilms on abiotic substrata induced strong larval settlement, while settlement cues from animals, including conspecifics and non- conspecifics (presumably prey species), had lesser impacts on larval settlement. These patterns were generally consistent across geographical zones and partially between larval developmental types (except for settlement to biofilms and conspecifics). Surprisingly, there was little influence overall of assay temperature, assay length, larval density or substrate size (area) on the settlement response of larvae, suggesting the conclusions reached here are robust with respect to variation in experimental methodology. Differences in larval settlement in response to natural substrata did, however, occur among larval phyla and common larval groups. Furthermore, variation in larval settlement was observed when finer details of natural substrata, such as macroalgae taxonomy, were investigated suggesting a strong role of phylogeny with regard to both larvae and substrata. This underlines the difficulty in generalising the relative importance of natural settlement cues for invertebrate larvae in the marine

Chapter 2 – A meta-analysis

environment. Instead, it appears more likely that the importance of different settlement cues will depend ultimately upon which larval species is under question.

It is not surprising that invertebrate larvae respond to macroalgae during larval settlement because macroalgae play important ecological roles in the marine environment by providing food and habitat for many invertebrates. I show evidence that settlement in response to macroalgae is strongly influenced by algal phylogeny, especially at the Order level, thus revealing that not all macroalgae function similarly in regards to larval settlement induction. Red algae (Rhodophyta) induced the greatest settlement response, but this was largely driven by the Order Corallinales where estimates and variation of larval settlement to Corallinales was almost identical to Rhodophyta. Despite this, the greatest settlement within Rhodophyta was observed to the Order Bonnemaisonailes. This was explained, however, mostly by several studies on two closely related urchin species (Holopneustes purpurascens and Holopneustes inflatus) which are induced to settle by Delisea pulchra (Swanson et al. 2012a), which limits broader level conclusions for Bonnemaisonailes.

In contrast, the Corallinales usually strongly induced settlement, consistent with the substantial literature on their activity as cues for larval settlement for a variety of marine invertebrates (Pearce and Scheibling 1990a, Daume et al. 1999b, Heyward and Negri 1999, Roberts et al. 2004, Huggett et al. 2006, Williams et al. 2008). This group of algae constituted more than 60% of the observations of larval settlement to macroalgae, and provides a significant proportion of the reason macroalgae in my study so strongly induce larval settlement. All families that were represented within the Corallinales induced significant settlement that was of a similar magnitude, but at the genus level I observed variation in the larval settlement to coralline algae. My analysis confirms coralline algae are important settlement cues for marine invertebrate larvae. This generality holds at the (Coralline) Family level but becomes difficult to predict at the Genus level. Interestingly, both geniculate (upright) and non- geniculate (encrusting) genera were represented as significant inducers of larval settlement, suggesting that both morphological forms are important settlement cues. Generally, it has been thought that the non-geniculate forms (commonly referred to as

Chapter 2 – A meta-analysis

crustose coralline algae, CCA) are more important to invertebrate larvae as settlement cues (Johnson et al. 1997, Harvey et al. 2005) than geniculate forms, but here I show comparable levels of larval settlement in response to both. It would be interesting to determine whether different species of coralline algae possess similar settlement cues or whether there are specific cues associated with each species that induce the settlement of invertebrate larvae.

Variation in the strength of coralline algae settlement cues was observed among larval taxonomy, indicating that not all larval species are strongly induced to settle in response to coralline algae. Larvae from the Echinodermata, Mollusca and Cnidaria were strongly induced to settle in response to coralline algae but this was not observed for larvae from Porifera or Chordata. Furthermore, when the taxonomic resolution of larvae was increased (here using ‘common larval group’) differences in the levels of larval settlement between groups from the same phylum to coralline algae were observed. This was especially prevalent in the Echinodermata where only sea urchin larvae responded strongly to coralline algae but not sea stars or sand dollars. This result further strengthens the idea that generalisations of important settlement cues in the marine environment are difficult to make and depend ultimately upon which larval species is under question.

The importance of microbial biofilms on surfaces in the marine environment as inducers of larval settlement has been recognised for almost eight decades now (Zobell and Allen 1935). Over this time a plethora of research has investigated the induction of larval settlement by microbial biofilms (see Wieczorek and Todd 1998, Hadfield and Paul 2001, Hadfield 2011). A central question from these studies is whether there is an emergent property of the biofilm that induces the settlement response, or whether it is simply a property of any one of the microorganisms within the biofilm. In an attempt to understand this, much research has focused on the latter (Johnson and Sutton 1994, Negri et al. 2001, Harder et al. 2002a, Lau et al. 2002, Huang and Hadfield 2003, Huggett et al. 2006, Tran and Hadfield 2011) due to the ease of isolation and culturing techniques. My analysis revealed that biofilms consisting of single species of either bacteria or diatoms from inductive biofilms generally do not

Chapter 2 – A meta-analysis

induce as large a settlement response compared natural mixed species biofilms. Of course, there may be one or few isolates that do, but this is largely overshadowed by the many that do not. This suggests that one, or few, species within these mixed natural biofilms are responsible for settlement induction.

However, this conclusion is complicated by recent experiments where mixing inductive with non-inductive bacteria resulted in a loss of settlement compared with the inductive single species biofilm (Tran and Hadfield 2011). It is thus unclear how single species biofilms distort the reality of what occurs naturally in the marine environment and this has to be considered in future experiments. For example, how likely would it be for larvae to be exposed to a monospecies biofilm (with unrealistic, high cells densities of only one species) in the marine environment?

Microbial biofilms are present on macroalgae and offer an explanation to the similarity in larval settlement to each of these groups in my analysis. Few experiments have been conducted examining the influence of epiphytic microbial biofilms on macroalgae towards larval settlement (e.g. Johnson et al. 1991b, Negri et al. 2001, Huggett et al. 2006, Swanson et al. 2006, Dworjanyn and Pirozzi 2008). Analysing these experiments in my meta-analysis revealed a reduction in larval settlement to macroalgae with removed biofilms, but across all experiments there was insufficient evidence to suggest a significant effect on larval settlement by removing biofilms from macroalgae.

Since it is likely that different larval species are induced to settle by different cues, a conclusion that is strongly drawn in this meta-analysis, it is not surprising that such a holistic analysis (combining settlement response of many different larvae) res ulted in a non-significant effect of macroalgal biofilms to larval settlement. Indeed, when I contrasted the settlement of different larval groups to coralline algae (the only well represented macroalgal group in this analysis) I found that larvae of sea urchins appeared to respond to bacteria on coralline algae while larvae of abalone did not. Thus it appears that bacteria on macroalgae are involved in the settlement response of some larval species but not others. More interesting, though, is that it seems different larval species respond to similar species of macroalgae (e.g. coralline algae) but by different mechanisms i.e. host algal related cues or epiphytic microbial biofilm cues.

Chapter 2 – A meta-analysis

The number of experiments examining bacterial mediated settlement on macroalgae was low and this suggests an avenue of focus for future research. In the case of bacterial mediated settlement on macroalgae, it might be possible to compare bacteria on different inductive algal species in order to determine which bacterial species are involved with the settlement response.

Much of what is known about larval settlement cues has been gained using laboratory based larval settlement assays, which constituted the experimental data the used in this analysis. I chose only experiments that were replicated and included controls measuring spontaneous settlement (negative control). Unfortunately, many experimental studies, especially earlier, pioneering studies before the 1980s, were rejected based on these criteria. (e.g. Cameron and Hinegardner 1974). Despite these failings, it is likely that such experiments still provide some level of information on the importance of different larval settlement cues albeit with a much lower confidence. Indeed, this problem has arisen in other areas of early larval ecological research. Oksanen (1991) cautioned whatever the apparent methodological weaknesses of the studies of pioneering ecologists, ‘It is a tremendous waste of time if we lose contact with their ideas and are forced to reinvent them’ (in Grosberg and Levitan 1992). Where replication was non-existent the variability in larval settlement responses will be remain unclear and understanding variability is paramount to any ecological experiment. This is especially important in larval settlement studies because variability in settlement responses can arise from both the larvae themselves but also from cues associated with the settlement substrates, which are known to vary (Swanson et al. 2006, Huggett et al. 2008). Where negative controls were not used, it is uncertain whether the absolute level of larval settlement in an experiment is indeed due to the cues examined, or some other physiological response by larvae resulting in increased spontaneous settlement, or simply if the larvae settle in the absence of cues. The latter, from observing our data set, is highly likely for barnacles, bryozoans, sponges and ascidians. I analysed the settlement response in negative controls in my data set and, interestingly, found that the estimated average settlement in filtered seawater (FSW) across all experiments was approximately 17%, further highlighting the need to understand larvae behaviours in the absence of any cue. 45

Chapter 2 – A meta-analysis

In this chapter, I for the first time investigated the relative importance of different settlement cues in a quantitative, meta-analytic approach. I demonstrated that larval settlement to natural substrates is wide spread but considerable variation occurs and is largely dependent on larval taxonomy, larval developmental type and finer substrate characteristics (e.g. macroalgal phylogeny). While greater overall settlement was observed for tropical regions compared with temperate regions, this observation seemed less important among substrates. My data confirms the importance of coralline algae as settlement cues for a variety of larval species but shows that larvae respond to coralline algae via different mechanisms (i.e. in response to host algae or associated epiphytic microbial biofilms). I further show the low likelihood in isolating bacterial or diatom species (i.e. most isolates will show little settlement induction) that induce as large a settlement response to natural mixed species biofilms which questions whether a few bacterial species do indeed induce settlement or whether larvae respond to an emergent property of a mixed species biofilm. Since bacteria on coralline algae are involved in the larval settlement of sea urchins, this system provides the opportunity to investigate biofilm related cues and further understand the generality of coralline algal cues for different species of larvae.

Chapter 3

The correlation between bacteria on coralline algae and the larval settlement of two Australian sea urchins

3.1 Introduction

Many benthic marine invertebrates have a two-stage life cycle that is distinguished by a dispersive planktonic larval stage and a sessile adult benthic stage. During larval development, larvae reach ‘competence’ and gain the ability to respond to environmental cues that induce larval settlement and metamorphosis, resulting in the benthic mode of life of marine invertebrates (Pawlik 1992).

Coralline red algae are an important source of cues for a range of marine invertebrates (Chapter 2), and are common components of benthic habitats worldwide. Coralline algae induce the settlement or metamorphosis of the larvae of sea urchins (Pearce and Scheibling 1990a, Huggett et al. 2006, Swanson et al. 2006), abalone (Daume et al. 1999b, Williams et al. 2008) and corals (Negri et al. 2001, Webster et al. 2004). Cues associated with coralline algae are produced not only by the algae itself (Morse and Morse 1984, Williams et al. 2009, Roberts et al. 2010b), but also by epiphytic bacteria present as biofilms on their surfaces(Johnson et al. 1991b, Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006, Tebben et al. 2011).

Microbial biofilms (whether associated with corallines or other submerged surfaces) can significantly affect settlement of marine invertebrate larvae, but the specific components of biofilms that induce settlement or metamorphosis are not well understood. Numerous approaches have been taken to address the nature of settlement inducers associated with biofilms. These include approaches based on understanding the chemistry of biofilms, for example comparing larval settlement with the chemical signatures of natural biofilm extracts (Chung et al. 2010) or specific Chapter 3 – Correlations between bacteria and larval settlement

biofilm components such as the biofilm extracellular polymeric matrix (Lam et al. 2005). Another approach relates the bacterial composition of biofilms to larval settlement either by: i) culture-dependent techniques, or ii) culture-independent techniques. For culture-dependent techniques, strains are isolated from the biofilm and tested in settlement assays with larvae (Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006). Culture-independent techniques characterise bacterial species composition in different biofilms (generally by DNA sequencing or other molecular techniques) and the variation among bacterial assemblages and qualitatively compared to rates of larval settlement to these biofilms (Webster et al. 2004, Lau et al. 2005b).

In recent years, the latter approach has been greatly facilitated by molecular techniques that characterize microbial communities primarily using the 16S rRNA gene as a taxonomic marker for microbial species differentiation and identification (Muyzer et al. 1993, Liu et al. 1997). Culture-independent techniques have shown in some instances that biofilms with different species composition consistently induce larval settlement at different rates (Qian et al. 2003, Lau et al. 2005b, Thiyagarajan et al. 2006), suggesting that larvae can distinguish between bacterial communities of different composition. However, it is unclear from previous culture-independent studies if the induction of larval settlement is due to a multispecies, emergent property of biofilms, if several inductive strains exist in biofilms, or if the induction of settlement is wholly or primarily due to one or only a few strongly inductive bacterial species. Distinguishing between these possibilities is critically important if the effects of biofilms on larval settlement are to be fully understood. The uncertainty among these alternatives is due at least in part to the somewhat limited statistical evaluation of bacterial community data that has been generated with respect to larval settlement, which in general has not made use of a variety of (in particular) multivariate techniques available for the analysis of high diversity communities (Anderson and Willis 2003).

The larvae of the Australian sea urchins Heliocidaris erythrogramma (Huggett et al. 2006)and Holopneustes purpurascens (Swanson et al. 2006) settle and metamorphose

Chapter 3 – Correlations between bacteria and larval settlement

in response to both coralline algal holobionts (the seaweeds and their associated microorganisms) and bacteria isolated from these algae. Using the larvae of these sea urchins, and statistical techniques previously not used in this context, I asked three questions: 1) do coralline algae (indeed, other co-occurring non-coralline algae also) induce similar settlement responses of larvae because of similar epiphytic bacterial communities; 2) can individual bacterial operational taxonomic units (OTUs) within bacterial communities on coralline algae be correlated with the settlement response of larvae, and; 3) are similar OTUs on coralline algae involved in larval settlement of different species of sea urchins?

Chapter 3 – Correlations between bacteria and larval settlement

3.2 Methods

3.2.1 Larval culturing

Sea urchin larvae were cultured using methods previously described (Williamson et al. 2004, Huggett et al. 2006, Swanson et al. 2006). Adult urchins (Holopneustes purpurascens and Heliocidaris erythrogramma) were collected from the shallow subtidal zone at Long Bay, Sydney (33°57′56″S 151°14′55″E) during reproductive periods (Williams and Anderson 1975, Williamson and Steinberg 2002). The release of gametes was induced by intraceolomic injection with 1-3 mL of 0.5 M potassium chloride (KCl, Ettensohn et al. 2004). Sperm and eggs were collected separately and pooled from at least three individuals. Eggs were washed with 2 L of 0.22 µm filtered, autoclaved seawater (SFSW). Cleaned eggs were fertilized with 100 µL diluted sperm solution in 1.8 L of SFSW for 10 min. Fertilization success was verified by the appearance of raised vitelline membranes in more than 90% of eggs. Fertilized eggs were washed twice to remove excess sperm and suspended in 1.8 L of SFSW to yield a concentration of approximately two eggs mL-1. Antibiotics (streptomycin sulphate 30 µg·mL-1, penicillin G 20 µg·mL-1) were added to larval cultures to prevent bacterial contamination. Larval cultures were reared at 19°C with slight aeration under a 12:12 h light:dark regime. The culture water was changed every day until larval competence was reached (6 d for H. purpurascens - (Williamson et al. 2004), and 4 d for H. erythrogramma (Williams and Anderson 1975).

3.2.2 Larval settlement in response to macroalgae

To examine the settlement response of larvae to coralline algae and co-occurring non- coralline macroalgae, a range of algal species were collected from sea urchin habitats. These included four species of articulated coralline algae (Amphiroa anceps, Corallina officinalis, Haliptilon roseum and Jania sp.- all identified with help by Dr Alan Miller of the Royal Botanical Gardens and Domain trust), two types of unidentified non- geniculate coralline algae (each type collected from similar habitat - from sea urchin barrens “CCABar” or the understory of Ecklonia radiate “CCAEck” and thus might not

Chapter 3 – Correlations between bacteria and larval settlement

be consistently the same species) and three other common co-occurring macroalgae (Ecklonia radiata, Delisea pulchra, and Sargassum vestitum).

For settlement assays, 30 mg pieces of individual thalli were carefully removed and randomly added to individual dishes in 6-well plates containing 4 mL of SFSW per well. SFSW without any substrate added served as a negative control. Ten replicates were used for each treatment. Ten larvae were used per replicate for H. erythrogramma and five larvae per replicate for H. purpurascens (the difference in numbers of larvae used between the two species was due to the greater fecundity and greater ease of collection of H. erthryogramma). Larval settlement assays were conducted at 19°C in a 12:12 h light:dark regime for 24 h, after which the proportion of metamorphosed larvae (within the entire dish) was counted. Metamorphosis was defined as the irreversible attachment of larvae together with the appearance of five tube feet and spines.

3.2.3 Algal biofilm manipulations

Manipulative experiments were employed to determine if larval settlement cues originated from epiphytic bacterial biofilms on macroalgae. For these the abundance and diversity of epiphytic bacteria was reduced (it is very difficult to completely eliminate biofilms) prior to testing in settlement assays following previously described methods developed for the same systems (Huggett et al. 2006). Algae were first soaked for 5 min in SFSW with 10% Betadine® (containing the antimicrobial providone- iodine) to control diatoms to disrupt bacterial cells. Algae were then rinsed twice in SFSW and soaked individually overnight in an antibiotic solution (SFSW containing 20 µg·mL-1 streptomycin sulphate, 10 µg·mL-1 penicillin G, 10 µg·mL-1 kanamycin) on an orbital shaker at 80 rpm. The following day, samples were rinsed with SFSW before being used in experiments. Procedural controls for antibiotic treatment consisted of the same handling process, but SFSW replacing the antimicrobial treatments. Freshly collected field samples served as unmanipulated controls.

To verify whether the antibiotic treatments were successful in reducing biofilms, the abundance and diversity of bacteria on two species of coralline algae (A. anceps and C. officinalis) were examined after treatment with antibiotics and compared with 51

Chapter 3 – Correlations between bacteria and larval settlement

unmanipulated and procedural controls (abundance measurements were determined using two extra treatments: 1) Betadine-only application and 2) an antibiotics-only soak without previous Betadine application). The abundance of epiphytic bacteria was visualised using LIVE/DEAD BacLight bacterial viability stain (Invitrogen) following the manufacturer’s instructions. Twenty µL of each stain was dripped (uniformly coating the surface) onto the surface of samples before visualisation using a fluorescent microscope (Olympus BX50). Only the fluorescence of live bacteria was visualised due to interference from prominent algal auto-fluorescence generated in the same excitation spectrum (617 nm) as that used for stained dead bacteria. Fortunately, the propidium iodide (“dead” stain) causes a reduction in the SYTO 9 (“live” stain) fluorescence when both dyes are present (LIVE/DEAD product manual) such that the effect of SYTO 9 staining all bacteria (live and dead) was mitigated. Live bacterial abundance was thus measured as green (517 nm) fluorescence. Five fields of view were taken per replicate at 200x magnification with five replicates per treatment. The percentage of green fluorescence per field of view was calculated using ImageJ image analysis software (http://rsb.info.nih.gov/ij/).

The effect of antibiotics on surface bacterial community composition was analysed with the culture-independent bacterial profiling method of terminal restriction fragment length polymorphism (TRFLP; Liu et al. 1997, see below for method). Each treatment (n = 5) was compared with the unmanipulated control, procedural control and antibiotic treatments.

3.2.4 Larval settlement in response to coralline algae treated with antibiotics

A subset of the macroalgae that induced settlement (A. anceps, C. officinalis, H. roseum and Jania sp., D. pulchra, S. vestitum) and biofilm coated shell grit (a major constituent of the benthos where the urchins and algae were collected) were examined for bacterial mediated settlement induction using manipulative antibiotic treatments (see above) before examining settlement in assays. D. pulchra releases a known settlement inducer for both urchins, histamine (Swanson et al. 2004, Swanson et al. 2012b) and was included here as an additional “procedural control” to show that 52

Chapter 3 – Correlations between bacteria and larval settlement

antibiotic treatments do not affect larval response to these algae other than by modifying the algae-associated bacterial community. Unmanipulated and procedural control treatments were also included in settlement assays and there were five to six replicates per treatment. Experimental conditions and the counting regime of the settlement assay followed protocol outlined above.

3.2.5 Comparison of natural epiphytic bacterial communities on macroalgae and shell grit

Since changes in bacterial community composition were associated with a reduction in settlement responses, I compared the natural epiphytic bacterial communities on macroalgal and shell grit surfaces. Bacterial communities were characterised by TRFLP. Total DNA was extracted by adding 30 mg of algae to 1 mL extraction buffer (100 mM Tris-HCl pH 8, 20 mM EDTA pH 8, 2.5 M sodium acetate, 8 mg cetyl trimethylammonium bromide (CTAB), 300 µL chloroform:isoamyl alcohol 24:1 (C:IAA), 200 µL 0.1 mm silica beads and 1 mg PVPP in 2 mL cryotubes. Samples were processed at 5.5 m·s-1 for 30 s using a FastPrep FP120 tissue homogeniser. Samples were centrifuged at 13000g for 1 min and the supernatant added to an equal volume of C:IAA and inverted multiple times. Samples were centrifuged at 13000g for 5 min and the supernatant was transferred to a 1.5 mL Eppendorf tube and centrifuged again at 16 000 g at 4°C for 30 min. DNA in the supernatant was precipitated with 0.1 volume (of the supernatant) 3M sodium acetate(pH 5.2) and 1 volume (of the supernatant)of isopropanol at -20°C for 24 h. DNA was pelleted by centrifugation at 16 000g at 4°C for 30 min and the supernatant discarded. The DNA pellet was washed twice with ice-cold 70% ethanol, reconstituted in 30 µL TE buffer. DNA extracts werevisualised on a 0.8 % agarose gel in sodium borate (SB) buffer containing the nucleic acid stain gelRED (biotin). DNA extracts were kept at -20 °C until further use.

A polymerase chain reaction (PCR) of 16S rRNA genes of bacterial communities was carried out in 25 µL reactions containing Econotaq 2X master mix (Lucigen) and the eubacterial primers 27F (AGAGTTTGATCCTGGCTCAG) and 519R (GWATTACCGCGGCKGCTG). The 27F primer was labelled at the 5’ end with 6-carboxy

Chapter 3 – Correlations between bacteria and larval settlement

fluorescein (FAM) dye. The thermocycling conditions were; initial denaturation at 94°C for 3 min, 25 cycles of 94°C for 30 s (denaturation), 55°C for 30 s (primer annealing), 72°C for 45 sec (extension) and a final extension at 72°C for 5 min. PCR products were visualised on a 1.2% agarose gel in sodium borate (SB) buffer containing the nucleic acid stain gelRED (biotin). On confirmation of successful amplification, PCR products were purified using silica spin columns (Zymo Research) following the manufacturers’ instructions. Two hundred ng of purified PCR products were digested with 5 U of RsaI (chosen on availability of a 4 bp recognition site enzyme, providing the highest probability of finding a restriction site, at time of initial experiments – sufficient detail was observed in electropherograms and thus this enzyme was used throughout this thesis in order for comparison of results among experiments if required) where for 3 h at 37°C. Digested products were desalted using silica spin columns (Zymo Research). Five ng of purified PCR product were then analysed on an ABI3730 capillary sequencer in genotyping mode with the size standard LIZ600 (Ramacoitti Centre, UNSW).

Terminal restriction fragments (TRFs) sizes were determined after electrophoresis by comparison with internal size standards using Peak Scanner v1.0 (AB systems). All TRFs above a threshold of one fluorescent unit were included such that determination of ‘true’ peaks could be statistically filtered from background noise (Abdo et al. 2006). TRFs <30 and >550 bp were excluded from the analysis. For a given sample, peak areas were standardised by total peak area to give the relative abundance of each TRF. TRFs were aligned using the T-Align algorithm (Smith et al. 2005) with a 0.5 bp threshold and the resulting data matrix was used for multivariate statistical analysis.

3.2.6 Correlation of larval settlement with bacterial community similarity and OTUs within communities

Settlement assays were conducted to assess the relationship between larval settlement and bacteria in communities on coralline algae, where epiphytic bacterial communities were investigated directly from samples used in settlement assays. The two most common coralline algae in urchin habitats, A. anceps and C. officinalis, were used for this experiment. Two larval settlement assays were conducted, separately for

Chapter 3 – Correlations between bacteria and larval settlement

each urchin species, with substrate mass, experimental conditions and counting regime as outlined above. In both assays six replicates were used for A. anceps and 12 for C. officinalis. At the completion of the larval settlement assays, sea urchins were removed from algae and algal samples were immediately frozen for subsequent extraction of the bacterial community DNA and TRFLP genotyping.

3.2.7 Statistical analyses

Analysis of variance (ANOVA) was conducted on larval settlement experiment data, which were checked for homogeneity of variance (Levene’s test) and transformed if necessary (Underwood 1997). Post-hoc tests were conducted using Tukey’s HSD.

For multivariate community data, Bray-Curtis similarity coefficients were calculated between every pair of samples. The resulting similarity matrix was visualised using non-metric multidimensional scaling (nMDS). Permutational multivariate analysis of variance (PERMANOVA, Anderson 2001) was used to test for differences between Bray-Curtis similarity coefficients of bacterial communities. Homogeneity of Bray- Curtis similarities were analysed using PERMDISP (Anderson 2006) to complement the interpretation of PERMANOVA results.

Mantel tests (Mantel 1967) were used to investigate the correlation between larval settlement and bacterial community similarity on coralline algae, as this intrinsic method to compare bacterial communities results in a pairwise distance matrix, which cannot be used with conventional univariate correlation methods. A Mantel test calculates the correlation between two matrices and the generation of a pairwise distance matrix of settlement responses is not trivial. I created pairwise matrices of larval settlement using Euclidian distance and matrices of community similarity using Bray-curtis coefficients. Both Pearson’s product moment (measure of a linear relationship) and Spearman rank (measure of a monotonic relationship) coefficients were calculated to investigate the degree and shape of the correlation between larval settlement and bacterial community similarity. For both coefficients, a correlation coefficient value close to one indicates that similar bacterial community composition

Chapter 3 – Correlations between bacteria and larval settlement

result in the similar range of larval settlement. Conversely, a correlation coefficient close to zero indicates similar settlement responses in dissimilar communities.

Canonical correlation analysis of principle coordinates (CAP) was applied to more specifically identify TRFs (i.e. bacterial OTUs) within the community data matrix that correlated with the percentage of larval settlement(Anderson and Willis 2003). This constrained ordination method approximates an axis through the multivariate community data cloud that maximises the relationship with a predictor variable, in this case, the percentage of larval settlement. This method also provides another approach, compared with Mantel tests (see above), for examining the relationship between larval settlement and bacterial community similarity but uses reduced dimensionality and rotation of the multivariate data cloud (unlike Mantel tests that are a direct correlative approach). However, provided a CAP correlation exists, individual TRFs within communities can then be identified with relative abundances that correlate with this CAP axis and are likely then to also correlate with larval settlement. We chose a threshold canonical correlation coefficient (ρ) of >0.5 before individuals within the communities were further investigated. We defined TRFs of interest that had absolute Pearson’s correlations with the CAP axis of > 0.3 and were observed in > 80% of the samples. A high percentage of observations (> 80%) were required because ambiguously high correlations resulting from many zero abundant TRFs and few abundant TRFs occurred in some samples. I used univariate linear regression to further investigate TRFs of interest.

Chapter 3 – Correlations between bacteria and larval settlement

3.3 Results

3.3.1 Larval settlement in response to coralline algae

Larvae of Heliocidaris erythrogrammaand Holopneustes purpurascens settled to varying degrees in response to all macroalgae tested andFSW elicited no or very little settlement(Figure 1). The amount of settlement varied significantly across the different algal species for both urchin species (Table 1). Post-hoc analyses indicated a complex grouping of settlement responses (Figure 1), however, larval settlement in response to articulated coralline algae, the red alga Delisea pulchra and the brown alga Sargassum vestitum was generally high for both urchin species. Larval settlement in response to crustose coralline algae was high for H. erythrogramma but moderate for H. purpurascens, while larval settlement in response to the kelp Ecklonia radiata was low for both urchin species (Figure 1).

3.3.2 Verification of antibiotic effect on surface associated bacteria

The abundance and community composition of bacteria on the surface of Amphiroa anceps and Corallina officinalis were significantly affected by the application of antibiotics and related treatments. The abundance of bacteria was reduced by up to 50% in the antibiotic treatment used in settlement assays (Betadine® + antibiotics), and in procedural controls the abundance of bacteria almost doubled (Figure 2, 3). ANOVA confirmed these patterns were statistically similar for bacteria on both coralline algae (Table 2). The composition of epiphytic bacterial communities on both coralline algae also changed in response to antibiotic treatments used in settlement assays (Figure 4). Here, bacterial communities in antibiotic treatments did not cluster with communities in either unmanipulated or procedural control treatments when visualised using nMDS ordination. The bacterial communities on coralline algae under the latter two treatments clustered in the nMDS ordination plot (Figure 4), suggesting that community compositions were stable despite differences in abundance (Figure 2). PERMANOVA confirmed these changes in community composition in response

Chapter 3 – Correlations between bacteria and larval settlement

toexperimental treatments were statistically similar for bacteria on both coralline algae (Table 2b).

3.3.3 Larval settlement in response to algae treated with antibiotics

The larvae of H. erythrogramma and H. purpurascens settled significantly less on all algae treated with antibiotics, except for D. pulchra (Figure 5, Table 3). No observable deleterious effects on the macroalgae treated with antibiotics were observed. Since D. pulchra releases a known settlement inducer for both urchins, histamine (Swanson et al. 2004, Swanson et al. 2012a), these observations act as an additional “procedural control”, in that they are consistent with the hypothesis that the antibiotic treatments do not affect larval response to these algae other than by modifying the algae- associated bacterial community. The additional procedural control of soaking algae in FSW without antibiotics had no effect on larval settlement, except for H. purpurascens in response to C. officinalis and Haliptilon roseum (which was not observed in repeat experiments, data not shown).

3.3.4 Comparison of natural epiphytic bacterial communities

Comparisons of bacterial communitiesamong different macroalgae and shell grit using tRFLP analysis revealed significant differences in the bacterial communities among the substrata (Figure 6, Table 4).Analysis of Bray-Curtis similarities followed by post-hoc analysis revealed all substrates contained distinct community compositions from each other. Communities from replicates of each species generally clustered together, with the exception of S. vestitum, which showed considerable variation in community structure among replicates compared to the other algal species (Figure 6; PERMDISP analysis, data not shown).

3.3.5 Correlation of larval settlement percentage with bacterial community similarity and TRFs within communities

Chapter 3 – Correlations between bacteria and larval settlement

The correlation between larval settlement and bacterial community similarity differed between the two urchin species. There was a correlation between the settlement response of H. erythrogramma to communities from A. anceps and C. officinalis (Pearson’s r = 0.386, p < 0.001; Spearman’s ρ = 0.414, p<0.001) but this was not observed for H. purpurascens (Pearson’s r = -0.04, p = 0.418; Spearman’s ρ = -0.02, p = 0.627). Despite the lack of a direct correlation between larval settlement and bacterial community similarity for H. purpurascens, the Canonical analysis of principle coordinate (CAP) analysisindicated the possibility of an axis through the multivariate community data cloud that better correlated with larval settlement (i.e. using a constrained ordination approach).

For the CAP analysis of H. erythrogramma settlement, the first four PCO axes explained 64% of the variability of Bray-Curtis similarities in the original dissimilarity matrix and provided the best number (m) of reduced axes to be included in the canonical analysis . For H. purpurascens, the first eight PCO axes explained 82% of the variability of Bray- Curtis similarities in the original dissimilarity matrix and provided the best number (m) of reduced axes to be included in the canonical analysis. The resulting canonical analyses based on these axes produced a highly significant squared canonical correlation of δ = 0.733 (P = 0.0001 using 9999 permutations, Figure 9a) for H. erythrogramma and a significant squared canonical correlation of δ = 0.460 (P = 0.027 using 9999 permutations, Figure 9b) for H purpurascens. Thus, samples were positioned on a new CAP axis that significantly correlated with the percentage of larval settlement and highlights that pure variation in bacterial community similarity does not necessarily correlate with variation in larval settlement (at least for H. purpurascens). Positioning the samples along the CAP axis allowed for the identification of TRFs of interest, those likely to be involved in the settlement response, by observing which TRFs within communities correlated with the new CAP axis.

The number, identity and the strengths of the relationships of TRFs with settlement differed with respect to each urchin species. CAP analysis revealed that 18 (~26%) TRFs had a relationship (both positive and negative) with larval settlement of H.

Chapter 3 – Correlations between bacteria and larval settlement

erythrogramma, whereas eight (~10%) TRFs had a relationship (negative only) with larval settlement of H. purpurascens. Ad hoc univariate linear regression was carried out on the relative abundance of these TRFs and the percentages of larval settlement of the urchin species. Three positively and four negatively correlated TRFs (~10% of total tRFs) had significant linear relationships with the settlement of H. erythrogramma, while only one TRF (~1% of total) had significant linear relationship with the settlement of H. purpurascens (Table 5). Interestingly, there were stronger negative relationships between the relative abundance of tRFs and the settlement of H. erythrogramma.

Chapter 3 – Correlations between bacteria and larval settlement

Figure 1.Mean (+/- SE, n=10) larval settlement of a) Heliocidaris erythrogramma and b) Holopneustes purpurascens after 24 hours to different macroalgae and FSW negative control. Bars sharing the same letters denote statistically similar settlement response (Tukeys HSD).

Chapter 3 – Correlations between bacteria and larval settlement

Figure 2.Mean (+/- SE, n=5) abundance of live bacteria on coralline algae measured by percent green fluorescence of stained live bacteria using LIVE/DEAD BacLight bacterial viability stain. Bars denote statistically similar groups (Tukeys HSD).

Chapter 3 – Correlations between bacteria and larval settlement

Figure 3. Live stained bacteria using LIVE/DEAD BacLight bacterial viability stain on Amphiroa anceps a) unmanipulated control at 100x magnification (scale bar 500µm), b) unmanipulated control at 200x magnification (scale bar 200 µm) c) antibiotic treated at 200x magnification (scale bar 100 µm). And Cat.

Chapter 3 – Correlations between bacteria and larval settlement

Figure 4. nMDS of Bray-Curtis similarities of bacterial communities, analysed using tRFLP, on a) Amphiroa ancepsand b)Corallina officinalisfrom fresh (unmanipulated), FSW (procedural control) and AB (antibiotic treated) treatments.

Chapter 3 – Correlations between bacteria and larval settlement

Figure 5.Mean (+/- SE, n=5) larval settlement of a) Heliocidaris erythrogramma and b) Holopneustes purpurascens after 24 hours to fresh (unmanipulated), FSW (procedural control) and AB (antibiotic treated) macroalgae and shell grit with corresponding FSW negative controls. Tukeys tests were conducted within species only due a significant ANOVA interaction term. Bars with overlapping lines denote statistically similar settlement response within a species (Tukeys HSD).

Chapter 3 – Correlations between bacteria and larval settlement

Figure 6.nMDS of Bray-Curtis similarities of bacterial communities, analysed using tRFLP,on different macroalgae and shell grit from sea urchin habitats.

Chapter 3 – Correlations between bacteria and larval settlement

a. Heliocidaris erythrogramma 100 • • •• • •• •• 80 . -••• 60 ..• ·-• 40 ...... • Corallina officinalis 20 ...... &. Amphiroa anceps ...... '#.- • ..... 0 -c::: Q) -0.3 -0.2 -0.1 0 0.1 0.2 0.3 E Q) b. Holopneustes purpurasens t: 100 • • • ...... enQ) 80 ...... 60 •• • 40

20 •

0 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 CAP axis 1

Figure 7. Scatterplot of CAP axis scores of bacterial communities on coralline algae and the associated larval settlement response of a) Heliocidaris erythrogramma and b) Holopneustes purpurascens in response to the bacterial communities after 24 hours.

Chapter 3 – Correlations between bacteria and larval settlement

a) Heliocidaris erythrogramma settlement Source DF MS F P Species 9 0.9118 10.49 <0.001 Error 90 0.0869 b) Holopneustes purpurascens settlement Source DF MS F P Species 9 1.4065 12.33 <0.001 Error 90 0.1141 Table 1.Analysis of larval settlement of a) Heliocidaris erythrogramma and b) Holopneustes purpurascens after 24 hours in response to macroalgae.

a) Abundance of live bacteria Source DF MS F P Species 1 0.01947 0.58 0.45 Treatment 4 0.34957 10.43 <0.001 Species x Treatment 4 0.02934 0.88 0.487 Error 40 0.0335

b) Community composition of bacteria Source DF MS F P Species 1 26141 19.93 <0.001 Treatment 2 2637.4 2.01 0.0083 Species x Treatment 2 2524.4 1.92 0.013 Error 24 1312 Table 2.Analysis of the a) abundance of bacteria and the b) composition of bacteria on coralline algae between fresh (unmanipulated), FSW (procedural control) and AB (antibiotic treated) treatments.

Chapter 3 – Correlations between bacteria and larval settlement

a) Heliocidaris erythrogramma settlement Source DF MS F P Species 7 3.3614 43.39 <0.001 Treatment 2 7.2095 93.05 <0.001 Species x Treatment 14 0.5533 7.14 <0.001 Error 120 0.0775 b) Holopneustes purpurascens settlement Source DF MS F P Species 6 2.3238 24.64 <0.001 Treatment 2 3.7054 39.29 <0.001 Species x Treatment 12 0.5422 5.75 <0.001 Error 84 0.0943 Table 3.Analysis of larval settlement of a) Heliocidaris erythrogramma and b) Holopneustes purpurascens after 24 hours in response tofresh (unmanipulated), FSW (procedural control) and AB (antibiotic treated) macroalgae and shell grit.

Source DF MS F P Species 6 6540.1 8.206 0.0001 Error 31 796.99 Table 4.Analysis of the variation in bacterial communities found naturally on macroalgae and shell grit.

Chapter 3 – Correlations between bacteria and larval settlement

a) Heliocidaris erythrogramma 2 tRF Relationship F1,34 P R 437 Negative 35.01 <0.001 0.507 109 Negative 23.97 <0.001 0.413 80 Negative 14.45 0.001 0.298 113 Positive 8.6 0.006 0.202 102 Positive 7.04 0.012 0.172 438 Negative 6.26 0.017 0.155 487 Positive 5.64 0.023 0.142 469 Positive 3.5 0.07 0.093 516 Negative 2.89 0.098 0.078 115 Positive 2.11 0.155 0.058 525 Negative 1.98 0.169 0.055 312 Positive 1.37 0.251 0.039 125 Negative 1.21 0.279 0.034 527 Negative 0.79 0.38 0.023 111 Positive 0.65 0.427 0.019 173 Positive 0.03 0.869 0.001 58 Negative 0 0.988 0 411 Positive 0.01 0.915 0

b) Holopneustes purpurascens 2 tRF Relationship F1,34 P R 105 Negative 7.75 0.009 0.186 107 Negative 3.9 0.057 0.103 78 Negative 1.45 0.237 0.041 472 Negative 0.96 0.334 0.027 435 Negative 0.92 0.344 0.026 171 Negative 0.07 0.787 0.002 67 Negative 0 0.954 0 430 Negative 0 0.973 0 Table 5.Regression analysis of tRFs identified to correlatewith the larval settlement of a) H. erythrogrammaand b)H. purpurascens using canonical correlation analysis of principle coordinates (CAP).

Chapter 3 – Correlations between bacteria and larval settlement

3.4 Discussion

Previous studies probing the role of bacteria in the mediation of marine invertebrate larval settlement have mainly relied on bacterial culture-dependent techniques. Numerous bacterial isolates have been screened as mono- or multispecies bacterial biofilms in larval settlement assays (Unabia and Hadfield 1999, Negri et al. 2001, Lau et al. 2002, Huang and Hadfield 2003, Huggett et al. 2006), but among the high number of bacterial strains tested only remarkably few induced high levels of larval settlement (Tebben et al. 2011, Huang et al. 2012), whereas the majority had moderate or no effect on larval settlement (although see Huggett et al. 2006 for an exception). Among the few bacterial strains shown to induce larval settlement, the inductive capacity varied both between species of a single genus and between strains of single species, suggesting that small genetic changes can result in disparate inductive capacities (Huang et al. 2012). Given the limited spectrum of bacteria accessible through bacterial culture-dependent analyses it remains speculative if the induction of invertebrate larval settlement is indeed mainly due to only few “inductive” bacteria, or if other properties of complex microbial biofilms (Webb et al. 2003), such as the community composition of bacteria may provide integrated cues that induce larval settlement.

Culture-independent approaches provide an alternative to better describe the effect of the natural microbial communities in biofilms on invertebrate larvae settlement choice. Studies using these approaches have revealed that invertebrate larvae can differentiate between characteristics of a biofilm such as age (Keough and Raimondi 1995, Wieczorek et al. 1995), bacterial abundance (Huang and Hadfield 2003), chemical biofilm properties (Kirchman et al. 1982, Khandeparker et al. 2003), bacterial metabolites (Tebben et al. 2011), and bacterial community composition (Qian et al. 2003, Webster et al. 2004, Lau et al. 2005b, Thiyagarajan et al. 2006). These findings support the notion of highly developed receptor systems in invertebrate larvae. However, using this molecular approach it remains unclear whether, emergent properties of the bacterial community in microbial biofilms are correlated with

Chapter 3 – Correlations between bacteria and larval settlement

variation in larval settlement or whether the presence of a subset of certain bacteria within complex biofilms is driving differential larval settlement.

Both the sea urchins studied here, Heliocidaris erythrogramma and Holopneustes purpurascens, settle and metamorphose in response to a range of phylogenetically different coralline algae (Figure 1) as well as bacteria isolated from these algae (Huggett et al. 2006, Swanson et al. 2006). However, different species of urchin larvae revealed different specificities for some of the coralline algae investigated here. While H. purpurascens settled at high rates in response to Amphiroa anceps and Corallina officinalis, H. erythrogramma settled significantly less in response to A. anceps than to C. officinalis(Figure 7). Strong settlement induction by different coralline algae, albeit with inherent variation, has also been observed with other urchins, such as Strongylocentrotus droebachiensis (Pearce and Scheibling 1990b), and together with my results, these observations suggest that coralline algae and their epiphytic bacterial assemblages may provide common settlement cues to sea urchins.

Larval settlement by these two urchins was significantly decreased when the epiphytic bacterial abundance and community composition on a range of coralline algae was manipulated by antibiotic treatment of algae (Figures 2 – 5), indicating that epiphytic bacteria were indeed the main drivers of larval settlement. This was in contrast to host derived settlement cues emanating from D. pulchra (Swanson et al. 2004, Swanson et al. 2012b) which were not affected by the antibiotic treatment used, and which also acted as an additional procedural control to reveal little effect of antibiotics on host algae. However, it must be noted that there was no direct measurement of algal health in response to antibiotic treatment.

One hypothesis generated from this result is that bacterial communities on coralline algae may be similar and provide an integrated cue that larvae respond to. In this chapter, I examined the bacterial communities on coralline algae and co-occurring non-coralline algae using the molecular profiling technique of TRFLP for microbial communities (Liu et al. 1997) which suggested host-specific bacterial communities were associated with each algal species (Figure 6). Qualitatively, these differences were not consistent with the settlement of larvae to these bacterial communities.

Chapter 3 – Correlations between bacteria and larval settlement

However, in order to make a stronger argument relating bacterial community similarity and larval settlement response more rigorous tools were required.

I then tested if the induction of larval settlement by two temperate coralline algae, A. anceps and C. officinalis, correlated with the community composition of epiphytic bacteria on these algae. To determine if the variation in bacterial community composition on different coralline algae was reflected in a similar variation of the magnitude of larval settlement, Mantel tests (Mantel 1967) were applied to correlate the distance matrices of larval settlement and bacterial community similarity, an approach that has yet to be seen in the larval settlement literature. This involved directly examining bacterial communities larvae were exposed to in settlement assays. However, it is possible that the larval footprint may affect the community profile but I believe this not likely due the short assay time (24 h) and little carry-over of microbes from larval cultures (larvae were cultured in SFSW with antibiotics that was changed daily). The percentage of larval settlement of both urchin species correlated differently with respect to bacterial community similarities on coralline algae. Larval settlement of H. erythrogramma correlated to some degree with bacterial community similarities within and between the coralline algae A. anceps and C. officinalis, whereas no correlation was observed for the other urchin species, H. purpurascens (Figure 8). Despite larvae of both urchin species settling in response to bacteria on coralline algae, it appeared that larvae of each urchin species had a different preference and specificity of bacterial community compositions on coralline algae (Figure 9). A similar discrepancy in larval settlement patterns of tube-worm and barnacle larvae to the same bacterial biofilm community has been previously observed (Lau et al. 2005b). Along with our results, these findings highlight the complexity of marine invertebrate larvae interactions with, and responses to, the composition of microbial biofilms.

Variation of urchin larval settlement to coralline algae carrying different bacterial communities was tested with respect to a few, inductive operational taxonomic units (OTUs). A constrained ordination technique was applied to determine which OTUs within bacterial communities correlated with larval settlement, a process that was followed by regression analysis of the OTUs. Larval settlement of H. erythrogramma

Chapter 3 – Correlations between bacteria and larval settlement

correlated not only with the epiphytic bacterial community similarity on coralline algae, but also with the relative abundance of a few but distinct bacterial OTUs. In contrast, settlement of H. purpurascens only correlated with one OTU. These results seemingly rejected the hypothesis that similar bacterial species on coralline algae induced larval settlement of congeneric urchin species. However, it has been hypothesised that H. purpurascens larvae respond to bacterially produced histamine on the surface of coralline algae (Swanson et al. 2006). A promising alternative to 16S rRNA gene methods may be to focus on histamine producing genes in the bacterial communities (Björnsdóttir-Butler et al. 2010), given that the threshold concentrations of histamine may not be adequately reflected by simply correlating the presence of bacteria in the biofilm community, and that the abundance of histamine producing bacteria may be overshadowed by the presence of other bacterial species.

Importantly, but also unexpectedly, larval settlement of H. erythrogramma not only correlated with ‘inductive’ but also with ‘inhibitive’ OTUs in bacterial communities, and indeed, there were as many negatively correlated OTUs as positive ones (Figure 10, 11). While it is known from previous studies – utilising bacterial culture-dependent methods - that both settlement-inducing (Unabia and Hadfield 1999, Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006) and settlement-inhibiting(Holmstrom et al. 1998, Harder et al. 2004, Rao et al. 2007) bacterial species exist in bacterial biofilms, no culture-independent study to date has unequivocally demonstrated that positive and negative larval settlement patterns are correlated with the presence or absence of inductive or inhibitive bacteria in complex biofilm communities (see Webster et al. 2004, Lau et al. 2005b, Huggett et al. 2006).

In this study, I demonstrated for the first time that larval settlement in response to a bacterial community can be correlated with the relative abundance of ‘inductive’ and ‘inhibitive’ OTUs within these communities. While our experiments only revealed correlative associations between larval settlement and ‘inductive’ and ‘inhibitive’ OTUs within these communities, it still remains speculative as to whether the observed OTUs are indeed inductive or inhibitive to larval settlement. Manipulative experiments are required to address this uncertainty, e.g. by targeting the abundance of either group of

Chapter 3 – Correlations between bacteria and larval settlement

bacteria, yet still preserving the natural community composition. An ideal experimental strategy would merge culture-independent and dependent techniques, using the former to identify bacterial species that correlate with larval settlement, and the latter to target these strains by culturing and verify their effect in laboratory larval settlement assays.

In addition, other molecular techniques (e.g. qPCR, FISH) may be used to quantify such bacterial species in the field and compare their densities to those that induce larval settlement in the laboratory. This approach would address the critical question of larval settlement studies, that is, do laboratory observations of larval settlement represent ecological realistic responses? Because the culture-independent molecular method of TRFLP employed in this study does not provide phylogenetic information about bacterial OTUs of interest, we were limited in adopting this approach. An approach to address the identity of these tRFs could involve generating 16S clone libraries, followed by sequencing of the clones to obtain phylogenetic information. These sequences could then be digested using the same restriction enzyme either empirically or in silico to obtain the tRF associated with the sequence and hence link taxonomy with tRF. This approach, however, is time consuming and cumbersome by generating only a small number of sequences at a high cost. Since there is limited information on the taxonomy of bacteria on coralline algae, more recent technologies such as high-throughput sequencing would provide a greater detail of taxonomic information to better understand which bacteria might be involved in settlement induction of sea urchin larvae.

Chapter 3 – Correlations between bacteria and larval settlement

3.5 Supplementary material

Supplementary figure 1. Example electrophoresis gels of a) DNA extracts from algal biofilm manipulation - Lane 1 = DNA ladder (GeneRuler 100 bpLanes 2-6 = Corallina Fresh, Lanes 7-11 = Corallina AB, b) PCR products from algal biofilm manipulation - Lane 1 = DNA ladder (GeneRuler 100 bp), Lane 2 = Positive control, Lane 3 = Negative control, Lanes 4-13 = Corallina Fresh, Lanes 7-11 = Corallina AB, c) DNA extracts from different macroalgae – Lane 1 = DNA ladder (GeneRuler 1kb), Lanes 2-7 = Corallina, Lanes 8-12 = Amphiroa, d) PCR products from different macroalgae – Lane 1 = DNA ladder (GeneRuler 1kb), Lanes 2-7 = Corallina, Lanes 8-12 = Amphiroa.

Chapter 3 – Correlations between bacteria and larval settlement

Supplementary figure 2. Example electropherograms of terminal restriction fragments from bacterial communities on Corallina officinalis in the antibiotic experiment a) Fresh - unmanipulated control, b) FSW - procedural control, and c) AB - antibiotic treated.

Chapter 3 – Correlations between bacteria and larval settlement

Supplementary figure 1. Example electropherograms of terminal restriction fragments from bacterial communities on different macroaglae and an abiotic surface including a) Amphiroa anceps, b) Haliptilon rosem, c) Jania sp., d) Corallina officinalis, e) Delisea pulchra, f) Sand/Shell grit and f) Sargassum vestitum.

Chapter 4

Induction of settlement of sea urchin larvae by Pseudoalteromonas luteoviolacea and other bacteria isolated from three species of coralline algae

4.1 Introduction

Microbial biofilms are ubiquitous in the marine environment, including on living organisms (Davey and O'toole 2000). Many marine invertebrate larvae respond – positively or negatively - to microbial biofilms when making the transition from a planktonic to a benthic mode of life (Johnson et al. 1997, Hadfield 2011). This widespread observation has led many researchers to investigate settlement cues associated with biofilms (Unabia and Hadfield 1999, Harder et al. 2002b, Thiyagarajan et al. 2006, Chung et al. 2010) including studies of which microbial taxa within biofilms induce settlement of larvae of benthic marine invertebrates. (Wieczorek and Todd 1998, Harder et al. 2002a, Dobretsov and Qian 2004, Webster et al. 2004)

Biofilms on the surface of coralline algae have received increasing attention as inducers of larval settlement of a number of invertebrate species (Johnson and Sutton 1994, Huggett et al. 2006, Swanson et al. 2006, Dworjanyn and Pirozzi 2008). In Chapter 3, I demonstrated that the settlement of larvae of two species of Australian sea urchin was strongly affected by bacteria on the surface of coralline algae and using molecular tools to investigate these bacteria, revealed that different bacteria might be involved in the settlement induction of each sea urchin. The identities of bacteria however were not determined, due to methodological constraints of the DNA fingerprinting method (terminal restriction fragment length polymorphism or tRFLP) used. Chapter 4 – Larval settlement in response to isolates

An alternative approach to molecular characterization for understanding which bacterial species on coralline algae induce larval settlement is to isolate and culture potentially inductive strains and test them directly against larval settlement. Such studies are common (Unabia and Hadfield 1999, Negri et al. 2001, Huang and Hadfield 2003) and one particularly relevant example is the settlement of larvae of the Australian sea urchin Heliocidaris erythrogramma, which was induced to settle by a number of bacterial strains isolated from coralline algae (Huggett et al. 2006). Of these different bacteria, Pseudoalteromonas luteoviolacea induced the greatest level of settlement. This is consistent with other recent observations on the importance of P. luteoviolacea as a settlement inducing bacterium (Huang and Hadfield 2003, Hadfield 2011, Tran and Hadfield 2011).

P. luteoviolacea is particularly amenable for culture based studies because it produces a suite of secondary metabolites including violacein (Jiang et al. 2000, Vynne et al. 2011), which gives colonies of this species a characteristic dark purple to black colour. Using this as a selection tool for identifying this species (which can then be confirmed by sequencing of the 16S rRNA gene), it is possible to screen isolated bacteria from different species of coralline algae for the presence of P. luteoviolacea with the goal of identifying whether variation in induction of settlement across different corallines can be attributed to the abundance of P. luteoviolacea.

In this chapter, I isolated bacterial strains from three species of corallines and tested their abilities when in biofilms to induce settlement of larvae of the sea urchins H. erythrogramma and, H. purpurascens. Moreover, because in previous studies of single isolates as inducers of larval settlement with H. erythrogramma (Huggett et al. 2006), the density of the bacteria was not measured, which may confound cross strain comparisons, I further characterized densities of isolated strains in settlement experiments. My results indicated that the major inductive bacteria cultured from all three coralline hosts were P. Luteoviolacea strains, although not all P. luteoviolacea strains were inductive. Given P. luteoviolacea occurs in multi-species biofilms in the field, I also investigated whether P. luteoviolacea is inductive in multi-species species biofilms created with other isolates.

Chapter 4 – Larval settlement in response to isolates

4.2 Methods

4.2.1 Isolation of Pseudoalteromonas luteoviolacea and other strains from coralline algae

Three species of coralline algae, Corallina officinalis, Amphiroa anceps and Haliptilon roseum, were collected in triplicate from Long Bay, Sydney (33°57′56″S 151°14′55″E). Algae were briefly rinsed in sterile filtered sea water (FSW) before adding each sample to individual falcon tubes containing 5 µL of FSW. Samples were vortexed vigorously for 20 mins to remove bacteria cells from the surface of the algae. Bacterial suspensions were serially diluted to 10-5 and 20 µL of each dilution plated out on to marine agar 2216 (Difco). Dark purple colonies, characteristic of the bacterial species Pseudoalteromonas luteoviolacea, were identified and isolated first, followed by random isolation of other bacterial colonies based on distinguishing colony characteristics such as colony colour, form and margin.

4.2.2 Sequencing of 16S rRNA genes from bacterial isolates

DNA from bacterial isolates was liberated from cells by heating (95 °C) a small amount of colony biomass (transferred using the tip of a sterile toothpick) in 50ul TE (10mM Tris, 1mM EDTA) buffer for 3 min. An aliquot (1 µL) of this extract was added to a 24 µL PCR solution containing 12.5 µL Econotaq 2X master mix, 1 µL each of 10 uM 27F (AGAGTTTGATCMTGGCTCAG) and 1492R (GYTACCTTGTTACGACTT) primers, and 11.5 µL water. PCR conditions were 94 °C for 3 min and then 25 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s, and finally 72 °C for 5 min. PCR amplicons were verified by gel electrophoresis, purified using silica spin columns (Zymo research, following manufactures instructions) and diluted to 10 ng·µL-1.

PCR amplicons were sequenced bi-directionally using the Applied Biosystems BigDye Terminator (v3.1 4337455) sequencing kit. The same primers as in the PCR were used in the sequencing reaction. An aliquot of purified of PCR amplicon (10 ng, 1 µL) was added to a sequencing solution consisting of 0.5 µL BigDye, 0.75 µL 5x Buffer, 0.5 µL 10 uM primer and 7.25 µL water. Thermocycling conditions were 96°C for 1 min and then

Chapter 4 – Larval settlement in response to isolates

25 cycles of 96 °C for 10 s, 50 °C for 5s, 60 °C for 4 min. BigDye amplified products were purified using ethanol precipitation, air-dried and then submitted to the Ramaciotti Centre, UNSW. Amplicons were sequenced on an ABI3730 sequencer.

Forward and reverse sequences were trimmed of poor base calls (

4.2.3 The terminal restriction fragment of P. luteoviolacea

The terminal restriction fragment of P. luteoviolacea was generated to enable a comparison of results with Chapter 3. The tRFLP protocol, described in Chapter 3, was carried out on the DNA of one P. luteoviolacea isolate from Haliptilon roseum.

4.2.4 Growth of biofilms for settlement experiments

Single species biofilms were grown on glass microscope coverslips following methods described in Huggett et al (2006), with slight modifications. Sterile coverslips were added to 35 mm petri dishes followed by 3 mL of Marine broth inoculated with a bacterial isolate. Cultures were left overnight at 25 °C, allowing biofilms to form on coverslips. The following day, culture media were removed from petri dishes and washed twice with FSW to remove loosely attached bacterial cells. A final aliquot of 4 mL of FSW was added to each dish before either 1) larvae (see below) were introduced for settlement assays or 2) 4 % formaldehyde was added to fix cells for enumeration.

Mixed species biofilms were grown on the bottom of 12-well culture plates by mixing cultures of different isolates. Isolates other than P. luteoviolacea were chosen only based upon the colour of colonies formed on marine agar, such that inoculation densities of each isolate in the mixture could be subsequently measured by spread plating. These strains were identified via 16S rRNA gene sequencing (see above). Overnight cultures of Pseudoalteromonas luteoviolacea (purple colony colour), two strains identified as Pseudoalteromonas rubra (red colony colour) and Vibrio pomeryi (light brown colony colour), and an unidentified isolate (light pink colony colour) were 82

Chapter 4 – Larval settlement in response to isolates

washed twice in 5 mL SFSW and reconstituted in 5 mL of SFSW. Mixtures were made by adding 100 µL of each culture to 9.5 mL SFSW and included one mixture of all 5 isolates and one mixture of 4 isolates excluding P. luteoviolacea. 3 mL of mixtures were then added to wells and cells allowed to attach for 6 hours. After 6 hours, loosely attached cells were rinsed off with two washes of 2 mL SFSW. 3 mL of SFSW was added and larvae introduced for settlement assays.

4.2.5 Enumeration of bacteria in biofilms

The densities of bacteria in single species biofilms (mixed species were not measured) to which larvae were exposed, was measured in a subset of replicates in order to account for the variation in larval settlement due to variation in abundance of one strain, as well as to differences among strains. Biofilms were fixed in 4 % formaldehyde and left at 4 °C overnight. Residual formaldehyde was rinsed from fixed samples by washing twice with 4 mL of FSW. 15-30 µL of DAPI solution (300nM DAPI, FSW, 10% v/v Glycerol) was applied to each coverslip to stain cells for 15 mins in the dark. Residual DAPI was rinsed free by washing twice with 100 µL FSW. Stained bacteria were visualised by epifluorescence microscopy using an Olympus BX50 microscope with UV excitation. 5 – 10 random fields of view were photographed per replicate glass slide at 600x magnification and the number of cells counted using the image analysis software ImageJ (Rasband http://imagej.nih.gov/ij/). Cell density was converted to cells per mm2 for statistical analysis.

4.2.6 Culturing larvae

Culturing of larvae of H. erythrogramma and H. purpurascens was performed as described in Chapter 3.

4.2.7 Larval settlement assays

Larval settlement assays were conducted in individual 35mm petri dishes containing 4 mL of SFSW with a bacterial biofilm presented on a microscope coverslip. Assays investigating mixed species biofilms which were conducted in 12 well tissue culture plates with 3 mL SFSW with a bacteria biofilm presented on the bottom of the dish. SFSW without any substrate added served as a negative control. Coralline algae 83

Chapter 4 – Larval settlement in response to isolates

(Corallina officinalis or Amphiroa anceps) served as a positive control. Three replicates were used for each treatment. Ten larvae were used per replicate for H. erythrogramma and five larvae per replicate for H. purpurascens. Larval settlement assays were conducted at 19°C for 24 h, after which the proportion of metamorphosed larvae was counted. Metamorphosis was defined as the irreversible attachment of larvae together with the appearance of five tube feet and spines.

Settlement of H. purpurascens in response to bacterial isolates from coralline algae was conducted using Pseudoalteromonas luteoviolacea (A316) and a random subset of other bacterial isolates from the culture library generated previously by (Huggett et al. 2006). Larval settlement of H. erythrogramma to single and mixed species biofilms was investigated using bacteria isolated from coralline algae in this study.

4.2.8 Statistical Analysis

Variation in settlement and bacterial densities were analysed by analysis of variance (ANOVA). Data were checked for homogeneity of variance (Levene’s test) and transformed if necessary(Underwood 1997). If transformed data did not conform to the assumptions of ANOVA, Kruskal-Wallis non-parametric one way ANOVA was used.

The relationship between settlement and bacterial density was analysed using linear regression and ANOVA. Since bacterial densities were measured using a subset of samples that larvae were not exposed to, we compared the average density to the average settlement response across treatments (isolates).

Chapter 4 – Larval settlement in response to isolates

4.3 Results

4.3.1 Phylogenetic affiliation of bacterial isolates from coralline algae

Phylogenetic analysis of bacterial isolates from the coralline algae Corallina officinalis, Amphiroa anceps and Haliptilon roseum using near full length 16S rRNA gene sequences revealed a relatively low diversity of strains, with isolates restricted to the genera Pseudoalteromonas, Shewanella, Vibrio, Bacillus, Cellulophaga and Alteromonas (Table 1). Sequencing of all dark purple colonies from each of the three coralline algal species indicated that all these isolates were affiliated with Pseudoalteromonas luteoviolacea, confirming that the pigmentation of this species is characteristic among the culturable fraction of bacteria from these coralline algae (Table 1).

The terminal restriction fragment of P. luteoviolacea, using the same protocol as in Chapter 3, was 517bp (“tRF 517”, Supplementary figure 1)

4.3.2 Settlement of Heliocidaris erythrogramma in response to Pseudoalteromonas luteoviolacea and other bacterial isolates from coralline algae

Pseudoalteromonas luteoviolacea strains induced the highest levels of larval settlement compared with other bacterial isolates from coralline algae across both of the two experiments conducted (Figure 1a, experiment 1 - black bars, Kruskal-Wallis,

H24 = 88.77, P < 0.001; experiment 2 - white bars, Kruskal-Wallis H24 = 83.52, P < 0.001). At least one P. luteoviolacea isolate from all of the three coralline species induced strong larval settlement of Heliocidaris erythrogramma (Figure 1a). However, not all P. luteoviolacea isolates induced larval settlement of H. erythrogramma, with larvae not settling at all in response to some isolates. Overall, the settlement response of larvae to P. luteoviolacea isolates ranged from 0 to 90%.

The range of bacterial densities larvae were exposed to was between 0.89 – 35.71 x 3 2 10 cells/mm (Figure 1b, ANOVA - experiment 1: F18,38=22.37, P < 0.001; Experiment 2:

F19,40 = 41.33, P < 0.001), indicating cells were present in all biofilms grown for assays. 85

Chapter 4 – Larval settlement in response to isolates

Variation in settlement to the different strains of P. luteoviolacea was not due to differences in density among the biofilms; there was no relationship between the average larval settlement and the average cell densities of different P. luteoviolacea 2 biofilms (Figure 2, F1,8 = 1.66, P = 0.238, R = 0.192). There was also no relationship overall between bacterial density of the different strains and settlement (Figure 3, F1,36 = 4.87, P = 0.0.034, R2 = 0.116).

4.3.3 Settlement of Heliocidaris erythrogramma in response to mixed species biofilms of Pseudoalteromonas luteoviolacea and other bacterial isolates from coralline algae

No larval settlement of H. erythrogramma was observed in response to mixed species biofilms containing P. luteoviolacea and each of four other bacterial isolates (Figure 4). Settlement did occur in the positive control and when P. luteoviolacea was presented as a single species biofilm, indicating that the larvae were competent (Figure 4). Inoculation densities of each bacterial isolate within the mixture were very similar to each other, ranging from 1-5 x 105 cells / ml, but the densities of the bacteria in the resultant biofilm were not measured.

4.3.4 Settlement of Holopneustes purpurascens in response to Pseudoalteromonas luteoviolacea and other bacterial isolates from different coralline algae

Larval settlement of H. purpurascens varied significantly in response to isolates from coralline algae, (Figure 5a, Kruskal-Wallis, H33 = 95.36, P < 0.001). Pseudoalteromonas luteoviolacea (isolate A316, Huggett et al. 2006) induced consistently high larval settlement (>90%) of H. purpurascens (Figure 5a). This was consistent when the experiment was repeated with a subset of the isolates and a different batch of larvae

(Figure 5b, F14,74 = 45.24, P < 0.001).

Chapter 4 – Larval settlement in response to isolates

Figure 1.a) Mean (+/- SE, n=3) larval settlement of Heliocidaris erythrogramma aafter 24 hours in response to bacteria isolated from different coralline algae in two replicate experiments (black and white bars). b) Mean (+/- SE, n=3) cell densities in biofilms cultured in parallel with larval settlement biofilms. Isolates from the coralline algae Amphiroa anceps, Corallina officinalis, Haliptilon roseum are represented by A, C, and H, respectively. See Table 1 taxonomies of isolates. Isolate numbers underlined are Pseudoalteromonas luteoviolacea (Table 1).

Chapter 4 – Larval settlement in response to isolates

Figure 2. Mean (n=3) cell densities of Pseudoalteromonas luteoviolacea biofilms (measured in a subset of assay dishes cultured in parallel with biofilms tested against larvae) vs. mean (n=3) larval settlement of Heliocidaris erythrogramma from two experiments. There is no significant relationship between the parameters (F1,8=1.66, P=0.238, R2 = 0.192).

Chapter 4 – Larval settlement in response to isolates

Figure 3.Mean (n=3) cell densities of bacterial biofilms of isolates from coralline algae (measured in a subset of assay dishes cultured in parallel with biofilms tested against larvae) vs. mean (n=3) larval settlement of Heliocidaris erythrogramma from two experiments. There is no significant relationship between the parameters (F1,36=4.87, P=0.0.034, R2 = 0.116).

Chapter 4 – Larval settlement in response to isolates

Figure 4. Mean (+/- SE, n=3) larval settlement of Heliocidaris erythrogramma after 24 hours in response to the coralline algae Corallina officinalis, single species biofilms of Pseudoalteromonas luteoviolacea (P. luteo) and a mixed species biofilm containing P. luteoviolacea and four other bacterial isolates (MSB + P. luteo). FSW = sterile filtered sea water.

Chapter 4 – Larval settlement in response to isolates

Figure 5. Mean (+/- SE, n=3) larval settlement of Holopneustes purpurascens after 24 hours in response to bacteria previously isolated from coralline algae(Huggett et al. 2006) in two experiments (a and b). A316 = Pseudoalteromonas luteoviolacea. See Huggett et al. (2006) for taxonomy of other isolates.

Chapter 4 – Larval settlement in response to isolates

Table 1.Taxonomic affiliation of bacteria isolated from different coralline algae using 16S rRNA gene sequences. Isolates from the coralline algae Amphiroa anceps, Corallina officinalis, Haliptilon roseum are represented by A, C, and H, respectively. * indicates dark purple colonies on marine agar 2216.

Isolate Basepairs S_ab score RDP closest match A1 1330 0.974 Pseudoalteromonas rubra; C312; DQ005900 A2 1318 0.986 Shewanella pacifica; KMM 3605; AY366086 A3 1325 0.981 Pseudoalteromonas sp. DG1135; AY258115 A4 1323 0.99 Pseudoalteromonas sp. AS-43; AJ391204 A5 1338 0.981 Vibrio pomeroyi (T); type strain: LMG 20537; AJ491290 A6 n.d A7 n.d A8* 1333 0.962 Pseudoalteromonas luteoviolacea (T); NCIMB 1893T; X82144 C1* 1330 0.95 Pseudoalteromonas luteoviolacea (T); NCIMB 1893T; X82144 C2 1334 0.999 Pseudoalteromonas sp. AS-43; AJ391204 C3* 1323 0.972 Pseudoalteromonas luteoviolacea (T); NCIMB 1893T; X82144 C4 1341 0.992 Bacillus oceanisediminis (T); H2; GQ292772 C5 1331 0.99 Shewanella sp. MJ5323; DQ531951 C6 1333 0.98 Shewanella waksmanii (T); KMM 3823; AY170366 C7 1325 0.99 Cellulophaga lytica; MBIC1544; AB032509 C8 n.d C9 1340 0.999 Alteromonas macleodii; MS907; AJ414399 C10* 1331 0.969 Pseudoalteromonas luteoviolacea (T); NCIMB 1893T; X82144 H1* 1323 0.977 Pseudoalteromonas luteoviolacea (T), A136 H2 1332 0.959 Vibrio neptunius (T); LMG 20536; AJ316171 H3 1318 0.976 Pseudoalteromonas rubra; C312; DQ005900 H4 1357 0.914 Alteromonas macleodii; MS907; AJ414399 H5 n.d H6* 1324 0.971 Pseudoalteromonas luteoviolacea (T); NCIMB 1893T; X82144

Chapter 4 – Larval settlement in response to isolates

4.4 Discussion

Coralline algae are an important source of settlement cues for invertebrate larvae (Daume et al. 1999b, Roberts et al. 2004, Huggett et al. 2006, Williams et al. 2008) and at least some of this induction of larvae can be attributed to bacteria on the surface of these algae. This includes induction of settlement for larvae of sea urchins (Chapter 3, Huggett et al. 2006, Dworjanyn and Pirozzi 2008). Understanding which bacterial species in complex communities induce settlement is important to our understanding of both settlement cues from coralline algae and bacterially mediated larval settlement. In this chapter, I examined the activity of bacterial strains isolated from coralline algae, including that of Pseudoalteromonas luteoviolacea, a previously recognized strong inducer of larval settlement (Huang and Hadfield 2003, Huggett et al. 2006). I isolated P. luteoviolacea from each of three species of coralline algae suggesting a widespread distribution of this bacterial species on coralline algae. At least one representative strain of P. luteoviolacea from each of the three coralline species induced the settlement of larvae of H. erythrogramma. However, not all P. luteoviolacea strains induced larval settlement. Induction of settlement by other bacterial strains was low. Variation in settlement was not explained by variation in cell densities, either for P. luteoviolacea, or across a diversity of bacterial strains. When P. luteoviolacea was presented to larvae in mixed species biofilm no settlement occurred, raising the question as to whether P. luteoviolacea is inductive in natural, highly diverse mixed species biofilms. However, this last conclusion is limited by unknown total or species specific cell densities in mixed species biofilms (but see Tran and Hadfield 2011, who arrived a similar conclusions with measured cell densities )

Pseudoalteromonas luteoviolacea A316, previously isolated from one of the coralline algae used in this study (Huggett et al. 2006), also induced larval settlement of H. purpurascens, a sympatric sea urchin species of H. erythrogramma. This is the first time that one bacterial species has been shown to induce the larval settlement of two larval species sharing the same habitat, which is intriguing given the very different lifestyles of these two urchins. These results also further highlight the common observation of P. luteoviolacea as a settlement inducing bacterium, now shown to 93

Chapter 4 – Larval settlement in response to isolates

induce settlement of sea urchins (Huggett et al. 2006), polychaetes (Huang and Hadfield 2003) and corals (Tran and Hadfield 2011).

Not all P. luteoviolacea strains in this chapter induced a strong settlement response. Strain to strain variation in the larval settlement inducing abilities of P. luteoviolacea has been previously observed(Hadfield 2011, Huang 2012). This has been attributed to genetic variation between strains, with inductive strains possessing ‘inductive’ genes that are not present in non-induction strains (Huang 2012); this may explain the patterns observed in this chapter. Furthermore, there was some variation in settlement to the same isolates, both P. luteoviolacea and others, across experiments (Figure 1) that cannot be explained by genetic variation. Considering similar cell densities were observed in these cases, it is unclear why variation in settlement across experiments occurred. This could be linked with unknown factors or possible differences in expression of genes.

Pseudoalteromonas luteoviolacea is becoming a model bacterial species for larval settlement (Hadfield 2011, Huang 2012), microbial chemical ecology (Matz et al. 2004) and studies of bacterial natural products (Jiang et al. 2000, Vynne 2011, Vynne et al. 2011). In culture, this species produces a suite of bioactive compounds (Jiang et al. 2000, Vynne et al. 2011), which is consistent among strains isolated globally (Vynne 2011), and might explain why many interactions with other marine species occurs. P. luteoviolacea has been observed to interact antagonistically with other marine bacteria by producing antibacterial compounds (Radjasa et al. 2009). The dark purple colour of the colonies is a result of violacein production, a compound known to deter nano-flagellates (Matz et al. 2004). Interactions also extend to other higher organisms, with P. luteoviolacea inducing the settlement of the polychaete Hydroides elegans (Huang and Hadfield 2003), the coral Pocillopora damicornis (Tran and Hadfield 2011), and two species of sea urchins, Heliocidaris erythrogramma (Huggett et al. 2006, this study) and Holopneustes purpurascens (this study). The factors that cause P. luteoviolacea to induce the settlement of invertebrate larvae are much less clear but are currently under investigation by different research groups. One approach is to extract the chemistry (Tebben, unpublished data) another is to understand the genes

Chapter 4 – Larval settlement in response to isolates

involved (Huang 2012). Interestingly, the latter has shown that genes involved in cell adhesion and bacterial secretion systems could be linked with larval settlement (Huang 2012).

While P. luteoviolacea is a fascinating bacterium biologically, and increasingly a focus of larval settlement studies, some of the results here (also see Chapter 5) raise questions about the extent to which this bacterium induces settlement in the field. In this chapter, cell densities of P. luteoviolacea at 5 x 103 cells·mm-2 induced a high rates of larval settlement (~80%) of Heliocidaris erythrogramma. Huggett et al. (2008) showed that the Pseudoalteromonas genus constitutes approximately 0.1 % (102cells·mm-2) of the natural bacterial communities on two species of coralline algae. Considering that many other Pseudoalteromonas species have also been isolated from these coralline algae (Negri et al. 2001, Huggett et al. 2006), the density of P. luteoviolacea may represent a very small fraction of the total bacterial community present on coralline algae (≤102 cells·mm-2). The densities of P. luteoviolacea in biofilms generated in this chapter were above this threshold, so it is unclear whether induction of settlement occurs at ≤102 cells/mm2. Densities of < 5 x 103cells·mm-2 induced a low settlement response (~10%) of the polychaete worm Hydroides elegans suggesting densities of ≤102 cells·mm-2 might be below the threshold required for P. luteoviolacea to induce larval settlement.

This concern about whether culture based experiments test ecologically relevant densities of putatively inductive bacteria was also highlighted here by characterizing the tRF of P. luteoviolacea. This fragment did not appear to constitute any of the tRFs associated with the larval settlement of the sea urchin used in Chapter 3. This reveals a discrepancy between results obtained using two contrasting techniques (culture dependent and independent techniques) to understand bacterially mediated larval settlement. This is a critical point to consider since various studies in the literature use either culture dependent (Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006) and independent techniques (Webster et al. 2004, Chung et al. 2010) to try an elucidate cues associated with bacterial biofilms. More generally, it is critical to understand the abundance of P. luteoviolacea in natural bacterial communities on

Chapter 4 – Larval settlement in response to isolates

coralline algae and compare those densities to those used in settlement assays in future larval settlement studies.

In my meta-analysis (Chapter 2), I identified a low effect size of single species bacterial biofilms on the induction of larval settlement, and this was likely driven by many experiments resulting in few inductive bacteria and many non-inductive (Negri et al. 2001, Huang and Hadfield 2003, Roberts et al. 2010b). The results of this chapter reflected this pattern as few bacterial isolates induced any appreciable settlement except for P. luteoviolacea. In Chapter 3, I demonstrated that few bacterial members within complex communities induce the settlement of sea urchin larvae. This chapter confirms that observation, with only P. luteoviolacea inducing larval settlement among a variety of isolates from different coralline algae. The results presented in this chapter thus reiterate the observation that only few bacterial members with complex communities may be responsible for larval settlement (Negri et al. 2001, Huang and Hadfield 2003). However, given the somewhat contrasting results from Chapter 3 and 4, which specific bacteria in the field induce settlement is not yet clear.

Chapter 4 – Larval settlement in response to isolates

4.5 Supplementary material

Supplementary figure 1. Electropherogram showing the terminal restriction fragment length (labelled peak at 517 bp) of Pseudoalteromonas luteoviolacea isolated from the coralline alga Haliptalon rosem.

Chapter 5

The community ecology and taxonomic identities of bacteria on coralline algae and their relationship with larval settlement

5.1 Introduction

Coastal rocky shores in temperate marine environments are typically dominated by macroalgae. These ecologically important primary producers provide both habitat and food for a variety of marine organisms and hence are important ecosystem engineers for temperate coastal waters. Moreover, the surface of macroalgae is host to a rich diversity of micro-organisms, especially bacteria (Burke et al. 2011b, Lachnit et al. 2011). These surface-dwelling microbes are involved in a suite of ecological interactions – with other microbes (Weinberger et al. 1997, Longford 2008), with their host alga (Steinberg and de Nys 2002, Goecke et al. 2010) and with other higher organisms (Johnson and Sutton 1994, Huggett et al. 2006).

Bacteria on the surface of coralline algae play an important role in the mediation of settlement and metamorphosis of marine invertebrate larvae, which are critical periods in the life histories of benthic marine invertebrates. Reducing the abundance and diversity of these epiphytic bacteria on coralline algae can result in the loss of settlement induction for marine invertebrate larvae (Huggett et al. 2006, Swanson et al. 2006). Further, individual bacterial isolates from coralline algae can induce settlement (Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006) and metamorphosis (Negri et al. 2001) of larvae. Thus, there is considerable evidence that bacteria on coralline algae – but not the algae itself per se – play an important role in settlement of invertebrate larvae.

However, with culture-based approaches, as used in the latter it is unclear whether the bacterial species identified as settlement inducers are representative of the Chapter 5 – Community ecology of bacteria on coralline algae

ecologically relevant bacteria that mediate settlement under natural conditions, given that only a small proportion of bacterial diversity is culturable – the ‘great plate anomaly’ (Amann et al. 1995). Previous culture-based approaches have shown that bacteria belonging to the genera Pseudoalteromonas, Vibrio and Shewanella could induce sea urchin larval settlement and that they constituted a large proportion of culture libraries of bacteria from coralline algae (Huggett et al. 2006). Molecular biological approaches (specifically, fluorescent in situ hybridisation, FISH) have verified that these cultured, inductive strains are present in natural communities of bacteria on coralline algae (Huggett et al. 2008), but have also shown that these bacteria constituted only a small fraction of the total community. While it is possible that such low densities of bacteria in mono-species biofilms can affect settlement of higher organisms (Huang and Hadfield 2003, Rao et al. 2007), recent evidence suggests that double-species biofilms containing one inductive stain and one non-inductive strain do not confer the same properties as the inductive mono-species biofilm (Tran and Hadfield 2011). Thus, it is not clear how laboratory results relate to the relative abundance of bacteria in the field, and how those bacteria influence settlement of larvae in situ. Clearly, a better understanding of the bacteria on coralline algae using culture-free methods is needed and a first attempt should simply entail a description of the bacterial species present.

In Chapter 3, I described bacterial communities on coralline algae using DNA fingerprinting, a technique that overcomes culture bias, and then compared bacterial community similarities to larval settlement rates. I observed host-specific bacterial communities on coralline algae, despite relatively high rates of larval settlement in response to these communities. Furthermore, larval settlement showed both positive and negative correlation with OTUs within communities. These initial results lacked the phylogenetic information associated with the OTUs that would enable a direct comparison with species previously identified as settlement inducers from culture based approaches (Huggett et al. 2006). Thus the need to identify the bacteria present on coralline algae became apparent. To achieve this, a more in-depth analysis of the bacterial communities on these algae was undertaken in order to understand a) the relative abundances of potentially inductive strains, and b) the match or mismatch 99

Chapter 5 – Community ecology of bacteria on coralline algae

between laboratory and field results. In addition to this, a better understanding of the temporal and spatial variability of bacterial communities on coralline algae is needed to make inferences about bacterial mediated larval settlement on coralline algae. For example, if coralline algae harbour characteristic bacterial communities that are stable through space and time then larvae may have evolved recognition of this characteris tic biological signature. Alternatively, if a large variation in bacterial communities exists, then it is possible that functional traits of bacteria could be involved in the mediation of larval settlement (Huang 2012).

In this study, I analysed the bacterial community composition of four co-occurring coralline algae that induce larval settlement of sea urchins, including the temporal and spatial variability of communities on the two most dominant species, using a DNA fingerprinting (terminal restriction length polymorphism) and a next generation sequencing (454 pyrosequencing) approach. The aims of this chapter were ask how variation in communities on coralline algae reflected previously described variation in settlement (Chapter 3) and to investigate the species identities of bacteria on coralline algae and compare those with cultured isolates (Huggett et al. 2006). Further, I – for the first time - describe the tempo-spatial variability of bacterial communities on coralline algae in the context of larval settlement and general community ecology

(characterisation of microbial communities on closely related hosts).

100

Chapter 5 – Community ecology of bacteria on coralline algae

5.2 Methods

5.2.1 Sampling of coralline algae

Two separate investigations of bacteria on coralline algae were conducted in this study;

The first involved the comparison of epiphytic bacterial communities among four different coralline algae; Corallina officinalis, Amphiroa anceps, Haliptilon roseum) and Jania sp. all collected from Long Bay, Sydney, Australia during March 2010.

The second involved the investigation of the spatial and temporal variability of epiphytic bacterial communities of the two most dominant articulated coralline algae, C. officinalis and A. anceps, in the Sydney region (personal observation). These two species were collected from Clovelly Bay (Site 1), Long Bay (Site 2) and Bare Island (Site 3) during mid-spring (October 2010, Time 1) and late summer (March 2011, Time 2).

Individual samples were randomly taken from areas at least 2 m apart and at 2-3 m depth. Four-five replicate thalli of each species were collected into individual plastic zip lock bags and transferred to the laboratory. 30-50 mg samples were then rinsed briefly in autoclaved filtered (0.2 μm) seawater and frozen until DNA extraction.

5.2.2 Analysis of bacterial communities

A combination of two molecular-based techniques targeting the 16S rRNA gene of bacteria was used to analyse bacterial communities on coralline algae. Terminal restriction fragment length polymorphism (tRFLP), a DNA fingerprinting method, was used to understand broad patterns of bacterial community composition and increase the sampling power of experiments (due to a relatively cheaper cost per sample) at the expense of providing no phylogenetic information. Identification of bacterial OTUs was achieved by sequencing of the community 16S rRNA genes using 454 pyrosequencing.

5.2.3 DNA extraction

101

Chapter 5 – Community ecology of bacteria on coralline algae

Total DNA from the surface of coralline algal samples was extracted using a bead beating method as described in Chapter 3.

5.2.4 Terminal restriction fragment length polymorphism (tRFLP) tRFLP analysis method was described in Chapter 3. All samples collected from both investigations described above were analysed using tRFLP.

5.2.5 Pyrosequencing of 16S rRNA genes

A subset of samples from both investigations described above was used for sequencing of community 16S rRNA genes. These included all samples from the first investigation (four species of coralline algae from March 2010) and a random subset from the spatial-temporal study that resulted in a fully orthogonal design (three replicates from two sites, Long Bay and Bare Island, and from both time points and from both coralline algae).

454 Pyrosequencing of 16S rRNA amplicons was conducted by Research and Testing Laboratories (Houston, TX, USA) using FLX chemistry (Roche). The primer set used was the same as in the tRFLP analysis (27F and 519R). The resulting raw sequence reads were analysed using the sequence analysis pipeline Mothur (Schloss et al. 2009). Sequences were initially quality-filtered using the pyronoise algorithm (Quince et al. 2009) followed by removing sequences that were too short (< 180 bp), contained ambiguous bass calls or had > six homopolymers. Quality-filtered sequences were aligned using the Silva 16S reference alignment (Pruesse et al. 2007), screened to include only overlapping regions of sequence, pre-clustered (diffs = 2) and chimera checked (uchime). Sequences were classified twice using the Greengenes and the RDP7 training sets in order to compare taxonomic identities, however, the Greengenes taxonomy was primarily used as it has been suggested to provided better results (Werner et al. 2011). Classified algal chloroplasts were removed (taxon=Chloroplast). Sequences were then clustered into OTUs at 97% similarity with consensus taxonomy. A multivariate data matrix comprising the number of all OTUs across all samples was generated for statistical analysis (below). The number of sequences per OTU was standardised by the total number of sequences within a sample to account for

102

Chapter 5 – Community ecology of bacteria on coralline algae

differences in sequencing depth between samples. This was used as a measure of relative abundance of different OTUs.

5.2.6 Statistical analysis

Multivariate community data generated from both tRFLP and 454pyrosequencing were analysed similarly, however, for the pyrosequencing dataset two taxonomic levels, phylum and OTU, were analysed to provide data at different taxonomic resolutions. The Bray-Curtis similarity coefficient was calculated between every pair of samples and the resulting matrix was visualised using non-metric multidimensional scaling (nMDS). Permutational multivariate analysis of variance (PERMANOVA) was used to test for differences between treatments in terms of Bray-Curtis similarities. Where the number of unique permutations for significance testing was below 35, Monte Carlo simulations were used (denoted by superscript “MC” in PERMANOVA tables). Homogeneity of Bray-Curtis similarities was examined using PERMDISP and the outcome used to accompany interpretation of PERMANOVA. For the analysis of spatial and temporal variability in bacterial communities, the time and site factors were random and the algal host factor fixed.

Similarity of percentages (SIMPER) was conducted to identify the OTUs contributing the most to the average Bray-Curtis similarity of communities on each coralline species. Such OTUs are considered to be consistently abundant within communities (Clarke and Warwick 2001). The average relative abundance of these OTUs was visualised graphically. For the comparison of bacterial communities among the four species of coralline algae, the cut-off for the SIMPER analysis was 50% (i.e. only OTUs that contributed the first 50% of the cumulative average Bray-Curtis similarity for communities on each coralline species). SIMPER analysis was conducted for the factor “host algae” only for the analysis of spatial and temporal variability in bacterial communities, using a cut-off of 75% of the cumulative average Bray-Curtis similarity, to determine which OTUs were consistently abundant on each coralline algal species through time and space. Using different cut-off values simply reflected the fact that more OTUs were observed in the first investigation (among the four species of coralline algae) compared with the second (two species of coralline algae among space

103

Chapter 5 – Community ecology of bacteria on coralline algae

and time) resulting in a larger tail of rarer species in the former dataset reducing the overall contribution each OTU made to the community similarity.

104

Chapter 5 – Community ecology of bacteria on coralline algae

5.3 Results

5.3.1 Variation in bacterial communities among four species of coralline algae tRFLP analysis of bacterial community 16S rRNA genes revealed clear host-specific bacterial communities on the four species of coralline algae. There was clear clustering of bacterial communities from replicate samples of each algal species (Figure 1b) with similar variability among replicates (PERMDISP - F3,12=2.8, P=0.221). The composition of bacteria within communities among the four species of coralline algae, compared using the Bray-Curtis similarity coefficient, was significantly different (PERMANOVA -

F3,12=6.53, P=0.001, 997 permutations). Pairwise comparisons between bacterial communities on each coralline species were significantly different (P < 0.05) confirming that bacterial community compositions were host-specific. Pyrosequencing analysis of bacterial communities using OTUs reflected results from tRFLP (Supplementary 1).

Approximately 100,000 bacterial 16S rRNA tag sequences were obtained from the four species of coralline algae. After sequence processing, approximately 30,000 sequences were attributed to 6,500 OTUs at 97% similarity.

There was a high diversity at the phylum level with > 99% of OTUs classified among 50 different phyla including 24 formerly described phyla and 24 candidate phyla. Almost 90% of sequences affiliated with the four major phyla Proteobacteria, Bacteriodetes, Actinobacteria and Cyanobacteria (Figure 2). The composition of bacterial phyla within communities among the four species of coralline algae, measured using Bray-Curtis similarities, was significantly different (PERMANOVA – F3,12= 6.02, P < 0.0001, Permutations 9903). Proteobacteria, Bacteriodetes, Actinobacteria, Cyanobacteria and Planctomycetes constituted > 90% of the average similarities within, and between, communities on coralline algae (SIMPER). Thus, changes in the relative abundance of these four major phyla were associated with the differences detected in the composition of the bacterial phyla within communities among the four species of coralline algae. Despite these compositional differences, it appeared that consistent bacterial phyla were present on all coralline algae.

105

Chapter 5 – Community ecology of bacteria on coralline algae

There was only a small number of OTUs that contributed largely to the host specific bacterial communities on each coralline species (SIMPER, Figure 3, Table 1). Bacterial communities on Amphiroa were largely dominated by OTUs from the genus Granulosicoccus and the families Flavobacteriaceae, SC3-41 (Acidimicrobiales) and JdFBGBact (Acidimicrobiales). Corallina had large relative abundances of OTUs from the families SC3-41 (Acidimicrobiales), JdFBGBact (Acidimicrobiales) and Rhodobacteraceae as well as from the genus Granulosicoccus. The most abundant OTUs in communities on Haliptilon were from the phyla Cyanobacteria and representatives from the families SC3-41 (Acidimicrobiales), Xenococcaceae and the genus Granulosicoccus. Bacterial communities on Jania were mostly dominated by OTUs from different genera associated with the family Rhodobacteracea.

5.3.2 Temporal and spatial variability of bacterial communities on coralline algae tRFLP analysis of bacterial community 16S rRNA genes found a strong host-specific relationship for bacterial communities on the coralline algae Amphiroa and Corallina irrespective of sampling time or sampling site (Figure 4). Additional to host specificity, bacterial communities on both coralline species differed between sampling times, while patterns between sampling sites were difficult to visually assess (Figure 4). Bacterial communities from Amphiroa, regardless of time or site, appeared to be more variable than those from Corallina. There was significant heterogeneity in bacterial communities between the two algal species (measured using Bray-Curtis similarity,

PERMDISP, F11,36= 4.4 P = 0.012), complementing visual interpretation of bacterial community variability. Pyrosequencing analysis of bacterial communities using OTUs reflected results from tRFLP (Supplementary 2).

Host algae, sampling time and sampling site all influenced bacterial composition on coralline algae (three-way PERMANOVA found a significant three-way interaction of algae, sampling time and sampling site on bacterial community similarity, F2,36=3.30, P <0.0001, 9904 permutations). However, the contribution strength and interaction of these factors on bacterial composition differed. The largest component of variation was attributed to the ‘Algae’ factor (28.8 Bray-Curtis units) followed by the three-way interaction (15.7 Bray-Curtis units), the ‘Time’ x ‘Site’ interaction (13.0 Bray-Curtis

106

Chapter 5 – Community ecology of bacteria on coralline algae

units) and the ‘Time’ factor (11.0 Bray-Curtis units). Clearly, algal hosts played the largest role in structuring the bacterial composition than factors associated with time or space.

Approximately 150,000 bacterial 16S rRNA tag sequences were obtained from the two coralline algae from two sites and two sampling times. After sequence processing approximately 50 000 sequences were attributed to 3600 OTUs at 97% similarity.

Only a few OTUs were consistently abundant in bacterial communities on each of the corallines across space and time (Figure 5, Table 2). Bacterial communities on Amphiroa had consistently abundant OTUs from the genus Granulosicoccus and the families SC3-41 (Acidimicrobiales), Flavobacteriaceae, and Rhodobacteracease. Corallina had large relative abundances of OTUs from the families SC3-41 (Acidimicrobiales), Rhodobacteraceae, Pseudanabaenaceae, Flavobacteriaceae as well as from the genus Granulosicoccus.

5.3.3 Representation of settlement inducing isolates in sequence libraries

Representatives of the genera Pseudoalteromonas, Vibrio and Shewanella, which were consistently isolated from coralline algae(Chapter 4, Huggett et al. 2006) and induced the settlement of invertebrate larvae were extremely underrepresented in all sequence libraries generated in this study, constituting < 0.001% of the total number of sequence counts. These results corroborated with previous estimates that these genera constitute < 1% of the natural bacterial community on coralline algae (Huggett et al. 2008). Clearly, cultured representatives do not reflect the true bacterial species pool present on coralline algae and likely introduce bias in our current knowledge of bacterial species on coralline algae that induce larval settlement.

107

Chapter 5 – Community ecology of bacteria on coralline algae

Figure 1. a) Group-average dendrogram and b) non-parametric multi-dimensional scaling (nMDS) plots comparing the similarities between bacterial communities from four species of coralline algae, analysed by tRFLP. Bacterial communities were compared using the Bray-Curtis similarity coefficient.

108

Chapter 5 – Community ecology of bacteria on coralline algae

Figure 2.Relative abundance of bacterial phyla in communities on four species of coralline algae. Data is generated using 454 pyrosequencing of bacterial communities 16S rRNA genes.

109

Chapter 5 – Community ecology of bacteria on coralline algae

Figure 3. The relative abundances of OTUs (97% sequence similarity) which were consistently abundant in bacterial communities on each of the four coralline algal species. OTUs were identified by the average contribution to the within-algal-species Bray-Curtis similarity using SIMPER. The cumulative contribution of each OTU to the first 50% of the within-species Bray-Curtis similarity is given on the right axis.

110

Chapter 5 – Community ecology of bacteria on coralline algae

Figure 4. a) Group-average dendrogram and b) non-parametric multi-dimensional scaling (nMDS) plots comparing the similarities between bacterial communities from two species of coralline algae at different sites and times, analysed by tRFLP. Bacterial communities were compared using the Bray-Curtis similarity coefficient.

111

Chapter 5 – Community ecology of bacteria on coralline algae

Figure 5. The relative abundances of OTUs (97% sequence similarity) which were consistently abundant, in both space and time, in bacterial communities on each of the two coralline algal species. OTUs were identified by the average contribution to the within-algal-species Bray-Curtis similarity using SIMPER. The cumulative contribution of each OTU to the first 50% of the within-species Bray-Curtis similarity is given on the right axis.

112

Chapter 5 – Community ecology of bacteria on coralline algae

Table 1. Taxonomic identities of OTUs (97% sequence similarity) consistently abundant of each of the four coralline algal species shown in Figure 3.

OTU Phylum Class Order Family Genus OTU 18 Actinobacteria Actinobacteria Acidimicrobiales Acidimicrobiaceae Ilumatobacter OTU 22 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Algibacter OTU 29 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Algibacter OTU 43 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Croceitalea OTU 33 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Flagellimonas OTU 6 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Maribacter OTU 27 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Pibocella OTU 125 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Winogradskyella OTU 30 Proteobacteria Alphaproteobacteria Rhizobiales Hyphomicrobiaceae Cucumibacter OTU 144 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Ahrensia OTU 10 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Amaricoccus OTU 53 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Jannaschia OTU 67 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Jannaschia OTU 123 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Jannaschia OTU 93 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litoreibacter OTU 95 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litoreibacter OTU 100 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litoreibacter OTU 310 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litoreibacter OTU 89 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litorimicrobium OTU 244 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Litorimicrobium OTU 82 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Loktanella OTU 61 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Pseudoruegeria OTU 12 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Ruegeria OTU 119 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Silicibacter OTU 127 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Silicibacter OTU 80 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Thalassobius OTU 34 Proteobacteria Alphaproteobacteria Rhodospirillales Rhodospirillaceae Caenispirillum OTU 63 Proteobacteria Alphaproteobacteria Sphingomonadales Erythrobacteraceae Erythrobacter OTU 74 Proteobacteria Alphaproteobacteria Sphingomonadales Sphingomonadaceae Sandaracinobacter OTU 5 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 8 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 9 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 15 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 24 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 35 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 38 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 48 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 58 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus

113

Chapter 5 – Community ecology of bacteria on coralline algae

OTU Phylum Class Order Family Genus OTU 44 Proteobacteria Gammaproteobacteria Chromatiales Halothiobacillaceae Thioalkalibacter OTU 78 Proteobacteria Gammaproteobacteria incertae sedis Porticoccus unclassified OTU 99 Proteobacteria Gammaproteobacteria incertae sedis Porticoccus unclassified OTU 26 Proteobacteria Gammaproteobacteria incertae sedis Sedimenticola unclassified OTU 25 Proteobacteria Gammaproteobacteria Methylococcales Methylococcaceae Methylococcus OTU 60 Proteobacteria Gammaproteobacteria Methylococcales Methylococcaceae Methylococcus OTU 20 Proteobacteria Gammaproteobacteria Oceanospirillales Oceanospirillaceae Neptuniibacter OTU 138 Proteobacteria Gammaproteobacteria Oceanospirillales Oceanospirillaceae Neptuniibacter OTU 40 Proteobacteria Gammaproteobacteria Oceanospirillales Oceanospirillaceae Oleibacter OTU 70 Proteobacteria Gammaproteobacteria Oceanospirillales Oceanospirillaceae Oleibacter OTU 71 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae Leucothrix OTU 79 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae Leucothrix OTU 114 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae Leucothrix Table 1 continued.

114

Chapter 5 – Community ecology of bacteria on coralline algae

Table 2. Taxonomic identities of OTUs (97% sequence similarity) consistently abundant of each of the four coralline algal species shown in Figure 3.

OTU Phylum Class Order Family Genus OTU 2 Actinobacteria Actinobacteria Acidimicrobiales SC3-41 unclassified OTU 64 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Aquimarina OTU 2978 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae Aquimarina OTU 154 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae unclassified OTU 146 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae unclassified OTU 40 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae unclassified OTU 1240 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae unclassified OTU 2986 Bacteroidetes Flavobacteria Flavobacteriales Flavobacteriaceae unclassified OTU 275 Bacteroidetes Sphingobacteria Sphingobacteria Sphingobacteria unclassified OTU 1096 Cyanobacteria Oscillatoriophycideae Chroococcales Xenococcaceae unclassified OTU 224 Cyanobacteria Synechococcophycideae Pseudanabaenales Pseudanabaenaceae Halomicronema OTU 217 Cyanobacteria Synechococcophycideae Pseudanabaenales Pseudanabaenaceae Leptolyngbya OTU 313 Cyanobacteria Synechococcophycideae Pseudanabaenales Pseudanabaenaceae Leptolyngbya OTU 526 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae Sulfitobacter OTU 3 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae unclassified OTU 43 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae unclassified OTU 241 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae unclassified OTU 66 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae unclassified OTU 222 Proteobacteria Alphaproteobacteria Rhodobacterales Rhodobacteraceae unclassified OTU 540 Proteobacteria Alphaproteobacteria Sphingomonadales Erythrobacteraceae Porphyrobacter OTU 63 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 37 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 7 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 62 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 663 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 127 Proteobacteria Gammaproteobacteria Chromatiales Granulosicoccaceae Granulosicoccus OTU 5 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae Leucothrix OTU 252 Proteobacteria Gammaproteobacteria Thiotrichales Thiotrichaceae Thiothrix OTU 143 Proteobacteria Gammaproteobacteria unclassified unclassified unclassified

115

Chapter 5 – Community ecology of bacteria on coralline algae

5.4 Discussion

Bacteria on the surface of coralline algae play important ecological roles in the marine environment. They mediate the critical transition between planktonic and benthic modes of life for a variety of marine invertebrates (Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006). Identifying those bacteria associated with this transition is important to understanding coralline algal settlement cues. Previous investigations (Johnson et al. 1991b, Huggett et al. 2006, Huggett et al. 2008) have been limited by culture dependent techniques that are highly biased towards a small fraction of the true diversity of bacteria (Amann et al. 1995). In Chapter 3, I attempted to overcome culture bias by using a DNA fingerprinting approach to investigate bacterial communities on coralline algae and further correlated the relative abundance of bacterial OTUs generated from this approach with larval settlement. Despite using a molecular approach and identifying relationships between larval settlement and specific OTUs, I encountered the problem of identifying the taxonomy of these important OTUs. Studies on the phylogenetic make-up of bacterial communities on coralline algal are scarce and limited to those using culture-based approaches (Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006). In order to gain a better understanding of the diversity and variability of epiphytic bacterial communities on coralline algae, and the identities of the bacteria constituting thes e communities, a combination of two molecular-based bacterial community profiling techniques, tRFLP and pyrosequencing, were used, providing information on broad patterns of community composition and bacterial identities, respectively.

The combination of both techniques identified host-specific bacterial communities on coralline algae, however, together with meaningful spatial and temporal variability. Importantly, a large number of sequences were obtained that allowed me to identify many different bacterial OTUs and provided a comprehensive, taxonomy-based comparison of coralline algae communities. Investigation of a sequencing library for bacterial genera on coralline algae known to induce settlement (Huggett et al. 2006),

116

Chapter 5 – Community ecology of bacteria on coralline algae

found its isolates, as well as other cultured representatives, to be extremely underrepresented. While there may be issues with short DNA sequence read lengths with the technologies (454 sequencing platform) used in this study limiting phylogenetic resolution, discrepancies occurred at higher taxonomic levels that are successfully resolved with such read lengths. Importantly, this highlights the fact that much less is known about the bacterial species that are consistently abundant on coralline algae and the role these species have in larval settlement induction. These findings question the applicability of culture-based approaches for sound ecological experiments. This is especially true in larval settlement studies where a majority of bacteria known to induce settlement have been identified using culture dependent methods (Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006).

One hypothesis I raise from the data presented here is that the bacterial species constituting the consistently abundant fraction of the epiphytic community (the core microbiome, see below) on coralline algae are associated with the mediation of larval settlement. In order to test this idea, future studies should aim to target these groups of bacteria by either 1) correlating abundances in natural communities with larval settlement and/or 2) isolating these from natural communities using targeted culture techniques and testing them in single species settlement assays. The former approach is aided by use of the 16s rRNA gene sequences generated in this study. Here, probes can be designed that target specific groups of bacteria, and absolute abundances can be measured in natural populations (e.g. with FISH) to which larval settlement rates have also been recorded. Outcomes from these experiments will shed light on the role of these uncultured bacteria in the mediation of larval settlement and address problems associated with culture based approaches.

Interestingly, the consistently abundant bacteria observed on coralline algae in this study are also observed on other macroalgae species (see below). Provided these bacteria are associated with settlement induction, it would suggest that significant rates of larval settlement should also be observed on other macroalgal species. Indeed, we observed

117

Chapter 5 – Community ecology of bacteria on coralline algae

high rates of sea urchin larval settlement to other microalgae as well as coralline algae (Chapter 3). For example, the brown alga Sargassum induced similar rates of larval settlement as coralline algae including the loss of inductive cues when the abundance and diversity of bacteria were affected with antibiotics. This result provides evidence to support our hypothesis.

Whether there is any interaction between the surface of coralline algae and bacteria was highlighted in larval settlement experiments with the crown-of- thorn sea star, Acanthaster planci (Johnson and Sutton 1994). Bacteria isolated from coralline algae would not induce the settlement and metamorphosis of larvae unless present on the algal surface (via re-inoculation experiments of antibiotic treated coralline algae), demonstrating the interaction, presumably chemically mediated, between coralline algae and bacteria. These results highlight the complex interactions between coralline algae, bacteria and invertebrate larvae.

Much of what is known about bacterial mediated larval settlement on coralline algae has been gathered using culture based techniques and hence is largely biased to those groups of bacteria easily cultured. For example, a broad range of bacterial isolates from the coralline algae Amphiroa anceps and Corallina officinalis induced the settlement and metamorphosis of larvae of the sea urchin Heliocidaris erythrogramma (Huggett et al. 2006). The bacterial isolates were dominated by representatives of the genera Pseudoalteromonas, Vibrio and Shewanella. But these groups were extremely underrepresented in our sequence libraries, constituting less than 0.5% of the relative abundance of bacteria in communities. Not only were they underrepresented among different coralline algae, they were also underrepresented over spatial and temporal scales. This suggests that the sampling scheme outlined in this chapter was robust. It enabled effective identification of the possible variability of genera that might confound results should sampling have only occurred at one time point and one spatial point.

While knowledge that these bacterial genera constitute less than 1% of natural communities on coralline algae was previously available, it was hypothesised that such 118

Chapter 5 – Community ecology of bacteria on coralline algae

low densities could still induce settlement (Huggett et al. 2008). Other settlement experiments with single-species bacterial biofilms have shown that low densities of bacteria can still lead to interactions with settling larvae/higher organisms (Rao et al. 2007). Recent evidence, however, has highlighted the artefacts associated with such single species settlement assays. When an inductive bacterial isolate was allowed to form a biofilm in the presence of a non-inductive isolate, the settlement response of larvae was significantly lower than in the single species treatment (Tran and Hadfield 2011). While low densities of a single bacterial species may lead to an interaction with a higher organism, this effect can be overshadowed by the presence of other bacteria, which is an expected scenario in natural communities. This highlights a problem in extrapolating results from these experiments to understanding what realistically happens in the marine environment in addition to whether the bacterial species used are ecologically relevant.

The data presented here show host-specific bacterial communities on coralline algae. As the algal samples were subject to similar bacterial coloniser pools (when from the same site, or site/time combination) it is possible that coralline algae regulate or influence their associated epiphytic bacterial community. Indeed, this has been proposed by Johnson et al. (1991) in their work on coralline algae suggesting that they may provide unique nutritional environments for particular bacterial species. These results support the concept of host-specific bacterial communities on a variety of species of temperate macroalgae (Lachnit et al. 2009, Nylund et al. 2010, Lachnit et al. 2011, Sneed and Pohnert 2011). As such, macroalgae from different lineages could be expected to have distinctive surface chemistries leading to the selection of host-specific bacterial communities. Indeed, the similarity of epiphytic bacterial communities appears to be better correlated with closely related macroalgae than distantly related ones (Lachnit et al. 2009, Sneed and Pohnert 2011). The coralline algae used in this study are all in the Corallinaceae family. For these algae it appears that bacterial community composition correlates well at this lower level of phylogeny. However, not all macaroalgae show such strong host specificity of their bacterial communities. In contrast to these patterns, the green seaweed Ulva australis exhibited significant within-species variation in epiphytic bacterial communities (Tujula et 119

Chapter 5 – Community ecology of bacteria on coralline algae

al. 2009, Burke et al. 2011b), highlighting the difficulties in making broad statements about host-specificity of surface associated bacterial communities on macroalgae.

Whether host-specific bacterial communities on coralline algae were stable in space and time was further investigated. There was spatial and temporal variability observed in bacterial communities on coralline algae, however, the algal hosts played a much larger role structuring bacterial communities. These patterns have been observed in bacterial communities on other macroalgae (Lachnit et al. 2009, Nylund et al. 2010, Sneed and Pohnert 2011). While sampling sites used in this study were generally influenced by similar environmental conditions, the cause of the resulting variability in communities between sites and time within each individual coralline algal species remains unclear. Temporal and spatial variation in the pool of colonising bacteria might contribute to the final community. Alternatively, redundancy in functional traits required for a successful epiphytic community may account for the variability in community structure (Burke et al. 2011a), and a meta-genomic approach would assist in elucidating the variability in space and time observed here.

While functional redundancy might explain spatial and temporal variability, it does not account for the consistent differences in bacterial communities among species of coralline algae. Here, the effect of algal host chemistry on surface interactions with bacteria needs to be considered (Steinberg and de Nys 2002, Goecke et al. 2010). Secondary metabolites released by some red algae have a greater influence in shaping host-associated bacterial communities than other factors, such as spatial effects (Nylund et al. 2010). There is a gap in our understanding of the surface chemistry of such systems and how it relates to associated bacterial communities, including those used in this present study. Further effort is needed to understand the intricacies of these relationships and how host chemistry might drive differences or similarities in community structure.

Proteobacteria, Bacteriodetes, Actinobacteria and Cyanobacteria were the most abundant bacterial phyla on coralline algae found in this study, dominating bacterial communities in every sample. These four phyla are well represented in bacterial communities on other 120

Chapter 5 – Community ecology of bacteria on coralline algae

macroalgae (Longford et al. 2007, Burke et al. 2011b, Lachnit et al. 2011, Webster et al. 2013). Despite the qualitative dominance of these phyla, there was strong evidence to suggest quantitative phylum level differences exist in communities between corallines. These results are unlike previous studies of epiphytic bacterial communities on living surfaces where phylum-level similarity has been observed between hosts (Longford et al. 2007) or between surface and planktonic communities (Burke et al. 2011b).

However, these contrasting results may be attributed to differences in statistical tests used between these studies. In our case, we compared the relative abundance of OTUs within phyla using Bray-Curtis similarity, which was not conducted previously. The relevance of Phylum level comparisons remains unclear as there was greater diversity at lower higher taxonomic levels, which may not have biological meaning at the high lower taxonomic levels (i.e. Phylum).

At the lowest taxonomic level, OTUs clustered at 97% sequence similarity (hereafter OTU), there was a large diversity of bacteria but the most abundant belonged to few taxonomic groups. Interestingly, OTUs belonging to these phylogenetic groups were also observed to be the consistently abundant in communities on corallines that were sampled at different locations and times. It is likely then that these bacteria constitute a significant proportion of the core micro-biome of coralline algae. They are abundant on different species of coralline algae and consistent in both time and space. These groups of bacteria have also been observed in other algal-associated bacterial communities (Longford et al. 2007, Staufenberger et al. 2008, Burke et al. 2011b, Lachnit et al. 2011) preventing the conclusion that they are coralline specific bacteria. They might, however, represent part of the largely unexplored macroalgal microbiome. For example, Granulosicoccus was found to be a major constituent of bacterial communities on the brown algae Fucus vesiculosus (Lachnit et al. 2011) and has few cultured representatives from other marine macrophytes (Kurilenko et al. 2010, Bengtsson et al. 2011). In addition, Rhodobacteraceace and Flavobacteriaceae constituted greater than 50 % of sequences from the green alga Ulva australis (Burke et al. 2011b). Understanding why there is this association between these

121

Chapter 5 – Community ecology of bacteria on coralline algae

groups of bacteria and macroalgae is important for future research as it is speculated to be linked to host health and function. Changes in the algal micro biome can lead to a loss of host health, the development of disease (Campbell et al. 2011) and changes to interactions of other organisms – including larval settlement (Webster et al. 2013). Indeed, the study of micro-biomes in other systems has gained much attention recently due to the recognition of the significant role microbes play in health of higher organisms. For example, the human microbiome project aims to understand the influence of microbes to human normal physiology and predispositions to disease (Turnbaugh et al. 2007).

While a shift is occurring towards using molecular techniques to understand the ecology and interactions of bacteria on macroalgae, many studies continue to use broad spectrum culturing methods (Goecke et al. 2010). It is becoming clear that these isolated bacteria are probably not indicative of those functioning in natural communities, questioning what we have learnt with these methods in the last couple of decades. I have shown here that a large proportion of the dominant bacteria in communities on coralline algae have never been tested as single species biofilms in larval settlement assays. This is critical toward a better understanding of bacterial induced larval settlement.

However, there is still the issue that the molecular techniques used here still obtain their own biases such as capturing the more dominant populations or only providing limited resolution of taxonomic information through short DNA sequence read lengths (i.e. 454 sequencing platform). Minor populations of bacteria could still play a role in mediating larval settlement but are overlooked. Also, there may also be issues associated with the primer sets used (e.g. targeting Bacteria) and the importance of other potentially important members of the community such as the Archaea and nano-eukroyotes might also be overlooked here. However, targeted culture-based approaches, such as genome targeted cultivation (Hugenholtz 2002), address the problem of merging culture dependent and independent approaches by first understanding which species and functional genes are present in the environment (i.e. through molecular techniques) and

122

Chapter 5 – Community ecology of bacteria on coralline algae

then designing cultivation techniques based on the outcomes. Following this approach in larval settlement studies will address the problem of identifying the bacteria that induce settlement without isolating and testing individually in settlement assays. Settlement assays involving single species biofilms are critical to confirming the inductive capacity of bacterial species and thus complementary to sequencing and ‘omics approaches. Such merging of these approaches is suggested for future studies.

123

Chapter 5 – Community ecology of bacteria on coralline algae

5.5 Supplementary material

S1. Variation in bacterial communities among four species of coralline algae using a sequencing approach

Pyrosequencing analysis (using 97% OTU taxonomic level) of bacterial communities on coralline algae results mirrored those from the tRFLP analysis revealing clear host- specific bacterial communities on the four species of coralline algae (Supplementary Figure 1). There was homogeneity of variance in Bray-Curtis similarities among coralline algae (PERMDISP F3,12=1.4 P=0.49).The composition of bacteria within communities among the four species of coralline algae, compared using the Bray-

Curtis similarity coefficient, was significantly different (PERMANOVA - F3,12= 3.95 P <0.001, 9848 permutations). Pairwise comparisons between bacterial communities on each coralline species were significantly different (P < 0.05) confirming that bacterial community compositions were host-specific.

Supplementary figure 2. a) Group-average dendrogram and b) non-parametric multidimensional scaling (nMDS) plots comparing the similarities between bacterial communities from four species of coralline, analysed by 454 pyrosequencing. Bacterial communities were compared using the Bray-Curtis similarity coefficient.

S2. Temporal and spatial variability of bacterial communities on coralline algae using a sequencing approach

124

Chapter 5 – Community ecology of bacteria on coralline algae

Pyrosequencing analysis (using 97% OTU taxonomic level) of bacterial communities on coralline algae results mirrored those from the tRFLP analysis, finding strong host- specific relationship for bacterial communities on the coralline algae Amphiroa and Corallina irrespective of sampling time or sampling site (Supplementary figure 1). Additional to host specificity, bacterial communities on both coralline species differed between sampling times, while patterns between sampling sites were difficult to visually assess (Figure 4). Bacterial communities from Amphiroa, regardless of time or site, appeared to be more variable than those from Corallina. There was significant heterogeneity in bacterial communities between the two algal species (measured using Bray-Curtis similarity, PERMDISP, F7,16= 4.04 P = 0.136), complementing visual interpretation of bacterial community variability.

Host algae, sampling time and sampling site all influenced bacterial composition on coralline algae (three-way PERMANOVA found a significant three-way interaction of algae, sampling time and sampling site on bacterial community similarity, F1,16=1.53 P=0.032, 998 permutations). However, the contribution strength and interaction of these factors on bacterial composition differed. The largest component of variation was attributed to the ‘Algae’ factor (30.5 Bray-Curtis units) followed by the three-way interaction (19.8 Bray-Curtis units), the ‘Time’ x ‘Site’ interaction (16.0 Bray-Curtis units) and the ‘Time’ factor (11.8 Bray-Curtis units). Clearly, algal hosts played the largest role in structuring the bacterial composition than factors associated with time or space.

Supplementary figure 3.a) Group-average dendrogram and b) non-parametric multidimensional scaling (nMDS) plots comparing the similarities between bacterial 125

Chapter 5 – Community ecology of bacteria on coralline algae

communities from two species of coralline algae at different sites and times, analysed by 454 pyrosequencing. Bacterial communities were compared using the Bray-Curtis similarity coefficient.

Supplementary figure 6. Example electrophoresis gel of a) DNA extractions used for tRFLP and pyrosequencing - Lane 1 = Low Range DNA ladder, Lane 2 = 1500 bp 16S PCR product, Lanes 3-6 = Corallina Time 1 Site 1, Lanes 7-10 = Corallina Time 1 Site 2, Lanes 11:13 = Corallina Time 1 Site 3, b) PCR products from DNA extracts for tRFLP. Lane 1 = Low Range DNA ladder, Lane 2 = 16S positive control, Lane 3 = Negative control, Lanes 4-7 = Corallina Time 1 Site 1, Lanes 8-11 =Corallina Time 1 Site 2, Lanes 12-15 = Corallina Time 1 Site 3.

126

Chapter 5 – Community ecology of bacteria on coralline algae

Supplementary figure 8.Example electropherograms of terminal restriction fragments from bacterial communities on Corallina officinalis (a-f) and Amphiroa anceps (g-l) from different times and sites.

127

Chapter 6

General discussion

The settlement of larvae is a critical stage in the life history of most benthic marine invertebrates. Larvae must encounter and settle on suitable substrata within a habitat that ensures subsequent survival and growth, thus allowing for successful recruitment. Settlement cues associated with substrata in favourable habitats guide larvae through this process (Pawlik 1992, Hadfield and Paul 2001). There is an abundant literature showing that coralline algae are a major source of such settlement cues for a remarkable wide range of marine invertebrates (Pearce and Scheibling 1990a, Johnson and Sutton 1994, Daume et al. 1999b, Daume et al. 1999a, Heyward and Negri 1999, Hadfield and Paul 2001, Negri et al. 2001, Roberts 2001, Steinberg and de Nys 2002, Roberts et al. 2004, Huggett et al. 2006, Swanson et al. 2006, Dworjanyn and Pirozzi 2008, Williams et al. 2008, Whalan et al. 2012) and the major focus of this thesis was to further develop our understanding of larval settlement cues from coralline algae.

I initially investigated larval settlement in response to coralline algae and associated epiphytic biofilms by a meta-analysis of literature studies (Chapter 2). It is often cited that coralline algae are broadly important settlement cues but this idea has never been rigorously assessed using a quantitative approach. I demonstrated that there is greater settlement of marine invertebrate larvae in response to coralline algae compared with other settlement substrata, and this was especially true for larvae of corals, abalone and sea urchins. Previous studies have shown that settlement cues from coralline algae may be of host origin or associated with epiphytic microbial biofilms and the meta- analysis further addressed this phenomenon, revealing that larvae of abalone and sea urchins generally responded most strongly to host cues vs. bacterial cues, respectively.

Following on from the results of the meta-analysis for the importance of biofilms as settlement inducers for sea urchins, I then experimentally examined bacterially Chapter 6 – General discussion

mediated settlement of sea urchin larvae in response to coralline algae, using both molecular and culture based methods, to understand whether similar cues exist on a number of different corallines and if these cues can induce larvae of different species of urchins. I demonstrated that bacteria on the surface of different coralline algae were important for inducing larval settlement of two species of sea urchin by treating algae with antibiotics and showing a significant decrease in larval settlement. Both molecular and culture based approaches indicated that settlement was due to a relatively small subset of bacteria taxa (Chapter 3 and 4). Indeed, in culture based approaches, only one species, Pseudoalteromonas luteoviolacea, represented by several strains from different coralline algae, consistently induced settlement (Chapter 4). However, when I compared the specific inductive species identified by the two methods, I found there was poor agreement, with little evidence for P. luteoviolacea inducing larval settlement from the molecular approach.

I further examined the taxonomic identities of bacteria on coralline algae through space and time using higher resolution and DNA sequencing providing molecular tools (pyrosequencing), and found almost no representatives of bacteria which had been previously identified (Chapter 4, Huggett et al. 2006) as inducers of larval settlement. While limited taxonomic information at low taxonomic levels (e.g. genus), associated with the short DNA sequence reads, was obtained here, this discrepancy still occurred at higher taxonomic levels (e.g. class and family). Thus while my results indicate specificity in the identity of inducing bacteria, the specific identity of these bacteria is still unresolved, as is the ecological relevance of many culture based settlement experiments using individual strains of bacteria.

The meta-analytic approach used in Chapter 2 proved very useful for understanding general patterns of larval settlement in response to natural substrata in the marine environment, and more specifically for investigating the assumption of wide spread larval settlement in response to coralline algae. While previous reviews on larval settlement of marine invertebrates have highlighted the importance of coralline algae as settlement cues for larvae (Hadfield and Paul 2001, Roberts 2001, Steinberg and de Nys 2002), the strength of this association had not previously been rigorously,

129

Chapter 6 – General discussion

quantitatively examined. Combining results among many studies provides a much greater potential to understand general patterns, and as a consequence meta-analysis has become increasingly used in the ecological literature (Brett 1997, Gurevitch et al. 2001, Johnston and Roberts 2009, Moles et al. 2011, Poore et al. 2012). My analysis showed that coralline algae were an important source of settlement cues for some marine invertebrate larvae (Chapter 2). In particular, larvae of sea urchins, coral and abalone were strongly induced to settle by coralline algae, with larvae of other taxa exhibiting only very weak settlement in response to coralline algae, or even inhibition. Overall, I concluded that coralline algae are important to the early life stages of some marine invertebrates, likely providing food and habitat for post-settlement juveniles. In addition to establishing the general importance of coralline algae settlement cues, the results in Chapter 2 also provided a comprehensive investigation of broad settlement patterns of invertebrate larvae in response to natural substrata.

In Chapter 2 I further used this meta-analytic approach to examine the question of the importance of biofilms on coralline algae as inducers of larval settlement. Experimental studies of settlement cues from coralline algae have revealed that cues are produced not only by the algae (Morse and Morse 1984, Williams et al. 2009, Roberts et al. 2010a) but also by bacteria associated with the surface of corallines (Johnson et al. 1991b, Johnson and Sutton 1994, Negri et al. 2001, Huggett et al. 2006, Tebben et al. 2011) and I again looked for overall patterns across these studies. I compared host alga and bacterial mediated settlement of marine invertebrate larvae in response to coralline algae. Interestingly, abalone larvae generally responded to host cues, while sea urchin larvae respond to surface associated microbial biofilms. Thus while different settlement cues from coralline algae induce different larval groups, ultimately larvae are recruiting to a similar substrata - coralline algal surfaces.

In Chapter 3, I investigated the link between bacteria on coralline algae and larval settlement using molecular based tools that provide a much more in depth analysis of bacterial communities than do culture based approaches. Various characteristics of bacterial biofilms have been considered in linking biofilms with larval settlement including cell densities, biomass and species composition (Qian et al. 2003, Lau et al.

130

Chapter 6 – General discussion

2005a, Thiyagarajan et al. 2006). Species composition of biofilms is thought to play an important role in whether larvae choose to accept, or reject, the substrata as a site for settlement(Qian et al. 2003, Lau et al. 2005a, Thiyagarajan et al. 2006) and I focused on this hypothesis initially when investigating bacterially mediated larval settlement using molecular tools.

I found statistically distinct bacterial communities on different coralline algae, and then used a suite of multivariate statistical tools, not yet used in the context of bacterially mediated larval settlement, to provide a deeper level of analysis of the link between bacterial community structure and larval settlement. My first approach involved directly comparing the similarities in bacterial community compositions with larval response to those communities, using Mantel tests. Here, I found that one species of sea urchin, H. erythrogramma, appeared to respond to changes in bacterial community composition on coralline algae but the other urchin species, H. purpurascens, did not. This result was interesting because it appeared that different larval species were responding to bacteria on coralline algae (shown by a change in settlement when biofilms were manipulated with antibiotics – Chapter 4) but in different ways. This further increased the complexity of coralline algal settlement cues revealed by Chapter 2; e.g., abalone settle in response to host algal cues, while sea urchins settle in response to bacteria on coralline algae (Chapter 2, 4) - but different sea urchin species respond to bacteria in different ways.

The second multivariate statistical approach I used involved a constrained ordination technique (Canonical correlation analysis of principle coordinates, CAP - Anderson and Willis 2003) in which an axis through the multivariate community data cloud is determined that maximises the relationship with a predictor variable, in this case, larval settlement. These results of this analytical technique reflected the first, but importantly, with CAP I could use the position of samples along the new axis generated to weigh the importance of the relative abundance of the many bacterial members within communities to larval settlement, and thus determine whether specific bacterial taxa in the community correlated with larval settlement.

131

Chapter 6 – General discussion

The number, identity and the strengths of the relationships of bacterial members (as characterised by terminal restriction fragments or tRFs) with larval settlement differed between the two urchin species. This demonstrated that different bacteria on coralline algae induced settlement of different larval species. Only a very small proportion of bacterial members had a significant relationship with larval settlement. Unexpectedly, larval settlement of H. erythrogramma not only correlated with ‘inductive’ but also with ‘inhibitive’ bacteria, with the latter showing stronger correlations. Both settlement-inducing (Unabia and Hadfield 1999, Negri et al. 2001, Huang and Hadfield 2003, Huggett et al. 2006) and settlement-inhibiting (Holmstrom et al. 1998, Harder et al. 2004, Rao et al. 2007) bacterial species have been isolated from communities but no culture-independent study to date has previously demonstrated that positive and negative larval settlement patterns are correlated with the relative abundance of inductive or inhibitive bacteria in natural, complex biofilm communities.

Based on the Chapter 3, and following on from related work by Huggett et al. (2006) on settlement of H. erythrogramma, I investigated settlement of larvae of both species of urchins in response to strains of bacteria isolated form coralline algae. I was particularly interested in whether Pseudoalteromonas luteoviolacea induced larval settlement from molecular approaches (Chapter 3) as this species was the most inductive bacterial isolate from coralline algae identified previously (Huggett et al. 2006), and in the role bacterial density plays in indicating variation in larval settlement. My interest in bacterial density was both in the context of whether differences in induction were due to differences in density as opposed to differences among bacterial species, but also ultimately whether densities which induced settlement were ecologically realistic – e.g., comparable to those one might find in the field.

I demonstrated the consistent presence of inductive P. luteoviolacea strains across three species of coralline algae, with at least one strain of P. luteoviolacea cultured from each coralline species inducing larval settlement of H. erythrogramma. Bacterial densities of P. luteoviolacea did not correlate with the amount of settlement induction by the different P. luteoviolacea strains. P. luteoviolacea also induced larval settlement of the sympatric sea urchin H.purpurascens revealing for the first time a bacterial

132

Chapter 6 – General discussion

species which induced settlement of two larval species from the same habitat. However, the potential importance of P. luteoviolacea was constrained by the observation that it does not induce settlement mixed species biofilms (Chapter 3, Tran and Hadfield 2011). Indeed, the tRFs linked with larval settlement from Chapter 3 did not include the tRF indicative of P. luteoviolacea (Chapter 4). Thus there was a significant discrepancy in the specific inductive bacteria identified between culture dependent and culture independent methods.

P.luteoviolacea induces the settlement various larval species including sea urchins (Chapter 4, Huggett et al. 2006), a coral (Tran and Hadfield 2011) and a polychaete(Huang and Hadfield 2003) highlighting the importance of this bacterial species in larval settlement. However, previous results were all based on culturable bacteria and it is important for future studies to consider the ecological context and evidence for this bacterial species to prevent conclusion being based upon artefacts, either through culture bias and/or laboratory based experiments.

In Chapter 5, I used a community wide 16S rRNA gene sequencing approach to better understand the bacterial species on coralline algae and further allow me to investigate the abundances of potentially inductive strains and the match or mismatch between laboratory (bacterial isolation) and field results (natural bacterial communities). The sequencing results revealed a large diversity of bacteria that were not identified in previous culture dependent studies of bacteria on coralline algae (Johnson et al. 1991a, Huggett et al. 2006).Representatives of the genera Pseudoalteromonas, Vibrio and Shewanella, which were consistently isolated from coralline algae and induced the settlement of invertebrate larvae (Huggett et al. 2006) were extremely underrepresented in all sequence libraries generated, constituting < 0.001% of the total number of sequence counts. Clearly, culture dependent approaches are problematic because they do not represent the true diversity and relative abundances of bacteria in these communities and as a consequence and may lead to ecologically inconsistent or misleading conclusions.

The most common bacteria (in terms of relative abundance) present on coralline algae belonged to the genus Granulosicoccus, and the families Acidimicrobiales,

133

Chapter 6 – General discussion

Flavobacteriaceae, and Rhodobacteraceae (Chapter 5). These bacterial groups were consistently abundant on all coralline algal species examined across the temporal and spatial extent of my sampling (Chapter 5). Given that it would make ecological sense for larvae to be responding to signals in the community which are consistently present, it would be expected that such bacteria would more likely be involved in larval settlement than transient species. However, the involvement of these bacteria in larval settlement is unclear because few representatives of these strains have been cultured, preventing the examination of these bacteria in settlement assays(although few exist but were isolated outside of larval settlement studies e.g. Granulosicoccus, Kurilenko et al. 2010). One at least partial solution to this would be to design specific probes that target these bacteria, allowing correlations to be generated between the abundance of these taxa and larval settlement e.g. conducting settlement assays and using fluorescent in situ hybridisation to investigate the abundance of these bacteria in assays. Importantly, the 16S rRNA gene sequences that were generated in Chapter 5 provide a starting point for such future studies.

It has been hypothesised that the specificity of larval settlement in response to coralline algae is a result of coralline algae supporting characteristic bacteria (Johnson et al. 1991a). However, many of the core groups of bacteria that I identified on coralline algae have also been observed on other species of macroalgae. For example, Granulosicoccus was a major constituent of bacterial communities on the brown algae Fucus vesiculosus(Lachnit et al. 2011) and has appeared on other marine macrophytes (Kurilenko et al. 2010, Bengtsson et al. 2011). In addition, Rhodobacteraceace and Flavobacteriaceae constituted a large proportion of bacteria from the green alga Ulva australis (Burke et al. 2011b). Therefore, it appears that many of the bacteria that have been identified occur on a broad range of to macroalgae. Larvae in this study did settle at high rates to other macroalgae (e.g., Sargassum – Chapter 3), indicating that settlement in response to algal associated bacteria is not restricted to corallines, and a closer examination of the bacterial communities of these other seaweeds, in the context of larval settlement, is merited.

134

Chapter 6 – General discussion

6.1 Comparison of culture dependent and independent techniques for understanding bacterially mediated settlement of invertebrate larvae

In this thesis, I used both culture dependent and independent techniques to investigate bacterial mediated larval settlement in response to coralline algae. There were both consistencies and discrepancies between the results I obtained from the two methods.

Some broad conclusions, such as support for the idea that only a few bacterial species within complex communities are linked to larval settlement, were obtained using both approaches. Using culture independent methods to characterise bacterial communities on coralline algae, I found only a few bacterial members within these communities correlated with the larval settlement of H. erythrogramma (Chapter 3). Similarly, only a few bacterial isolates from coralline algae induced the larval settlement of this urchin species (Chapter 4).

Identifying which bacterial species in particular were involved in larval settlement was much harder to reconcile between the two approaches. Pseudoalteromonas luteoviolacea was isolated from three species of coralline algae and induced larval settlement of both species of sea urchins. However, this bacterial species was not found on coralline algae using two molecular techniques (tRFLP and pyrosequencing) to characterise bacterial communities (Chapter 3 and 5). In fact, the majority of bacterial isolates that constituted a previously generated culture library of bacteria on coralline algae (Huggett et al. 2006) were not identified using the sequencing approach in chapter 5.The identification of which bacteria, in natural communities, induce larval settlement is a key goal in studies of larval settlement but relies on ecologically representative experiments that might not be addressed using culture dependent studies.

Huggett et al. (2008) were the first to attempt to quantify the abundances of inductive bacterial isolates in natural communities, providing the first stepping stone to linking culture dependent and independent approaches. However, the approach taken there still suffered from culture bias, as probes were designed from bacterial isolates, which I 135

Chapter 6 – General discussion

have shown in this thesis to represent a significant problem. An alternative approach is to design probes for the abundant bacterial taxa I have identified in this thesis and link these with larval settlement.

Isolating bacteria and testing them in settlement assays is an important experimental method, despite the problems with culture bias. The issue of culture bias (as a result of using general purpose media e.g. marine agar) can be addressed using targeted culturing techniques, which first involve identifying bacterial species of interest using molecular tools and then developing specific selective media to isolate these species (Tyson et al. 2005). This approach has not been used in larval settlement studies.

Understanding the ecological relevance of Pseudoalteromonas luteoviolacea is another avenue of future research that involves the linking of culture dependent and independent approaches. Investigation into P. luteoviolacea settlement cues has gained much greater attention following the recognition of its ability to induce the settlement of various larval species (Hadfield 2011). However, whether the identification of this species is a culturing artefact is still unclear until the presence and abundance of this species in natural communities is further investigated. A species specific probe designed for this species is the obvious approach to answering this question.

136

Literature cited

Abdo, Z., U. M. E. Schuette, S. J. Bent, C. J. Williams, L. J. Forney, and P. Joyce. 2006. Statistical methods for characterizing diversity of microbial communities by analysis of terminal restriction fragment length polymorphisms of 16S rRNA genes. Environmental Microbiology 8:929-938. Ady, W. H. 1975. The algal ridges and coral reefs of St Croix and their structure and Holocene development. Atoll Research Bullet in:1-67. Amann, R. I., W. Ludwig, and K. H. Schleifer. 1995. Phylogenetic identification and in- situ detection of individual microbial-cells without cultivation. Microbiological Reviews 59:143-169. Anderson, M. J. 2001. A new method for non-parametric multivariate analysis of variance. Austral Ecology 26:32-46. Anderson, M. J. 2006. Distance-based tests for homogeneity of multivariate dispersions. Biometrics 62:245-253. Anderson, M. J. and T. J. Willis. 2003. Canonical analysis of principle coordinates: a useful method of contrained ordination for ecology. Ecology 84:511-525. Barnay, A., C. Ellien, F. Gentil, and E. Thiébaut. 2003. A model study on variations in larval supply: are populations of the polychaete Owenia fusiformis in the English Channel open or closed? Helgoland Marine Research 56:229-237. Bates, D. M. 2010. lme4: Mixed-effects modeling with R. URL http://lme4. r-forge. r- project. org/book. Bengtsson, M., K. Sjøtun, J. Storesund, and L. Øvreås. 2011. Utilization of kelp-derived carbon sources by kelp surface-associated bacteria. Aquatic Microbial Ecology 62:191-199. Björnsdóttir-Butler, K., G. E. Bolton, L.-A. Jaykus, P. D. McClellan-Green, and D. P. Green. 2010. Development of molecular-based methods for determination of high histamine producing bacteria in fish. International Journal of Food Microbiology 139:161-167. Brett, M. T. 1997. Meta-analysis in ecology. Bulletin of the ecological society of America 78:92-94. Broom, J. E. S., D. R. Hart, T. J. Farr, W. A. Nelson, K. F. Neill, A. S. Harvey, and W. J. Woelkerling. 2008. Utility of psbA and nSSU for phylogenetic reconstruction in the Corallinales based on New Zealand taxa. Molecular Phylogenetics and Evolution 46:958-973. Burke, C., P. Steinberg, D. Rusch, S. Kjelleberg, and T. Thomas. 2011a. Bacterial community assembly based on functional genes rather than species. Proceedings of the National Academy of Sciences 108:14288-14293. Burke, C., T. Thomas, M. Lewis, P. Steinberg, and S. Kjelleberg. 2011b. Composition, uniqueness and variability of the epiphytic bacterial community of the green alga Ulva australis. Isme Journal 5:590-600. Butler, A. J. 1986. Recruitment of sessile invertebrates at five sites in Gulf St. Vincent, South Australia. Journal of Experimental Marine Biology and Ecology 97:13-36. Cameron, R. A. and R. T. Hinegardner. 1974. Initiation of metamorphosis in laboratory cultured sea urchins. The Biological Bulletin 146:335-342. Literature cited

Campbell, A. H., T. Harder, S. Nielsen, S. Kjelleberg, and P. D. Steinberg. 2011. Climate change and disease: bleaching of a chemically defended seaweed. Global Change Biology 17:2958-2970. Chung, H. C., O. O. Lee, Y. L. Huang, S. Y. Mok, R. Kolter, and P. Y. Qian. 2010. Bacterial community succession and chemical profiles of subtidal biofilms in relation to larval settlement of the polychaete Hydroides elegans. Isme Journal 4:817-828. Clarke, K. and R. Warwick. 2001. Change in marine communities: an approach to statistical analysis and interpretation. 2 edition. PRIMER-E, Plymouth. Connell, J. H. 1985. The consequences of variation in initial settlement vs. post- settlement mortality in rocky intertidal communities. Journal of Experimental Marine Biology and Ecology 93:11-45. Daume, S., S. Brand-Gardner, and W. J. Woelkerling. 1999a. Preferential settlement of abalone larvae: diatom films vs. non-geniculate coralline red algae. Aquaculture 3-4:243-254. Daume, S., S. Brand-Gardner, and W. J. Woelkerling. 1999b. Settlement of abalone larvae (Haliotis laevigata Donovan) in response to non-geniculate coralline red algae (Corallinales, Rhodophyta). Davey, M. E. and G. A. O'toole. 2000. Microbial biofilms: from ecology to molecular genetics. Microbiology and Molecular Biology Reviews 64:847-867. Dobretsov, S. and P. Y. Qian. 2004. The role of epibotic bacteria from the surface of the soft coral Dendronephthya sp in the inhibition of larval settlement. Journal of Experimental Marine Biology and Ecology 299:35-50. Dworjanyn, S. A. and I. Pirozzi. 2008. Induction of settlement in the sea urchin Tripneustes gratilla by macroalgae, biofilms and conspecifics: A role for bacteria? Aquaculture 274:268-274. Edwards, L. J., K. E. Muller, R. D. Wolfinger, B. F. Qaqish, and O. Schabenberger. 2008. An R2 statistic for fixed effects in the linear mixed model. Statistics in medicine 27:6137-6157. Ettensohn, C. A., G. A. Wray, and G. M. Wessel. 2004. Development of sea urchins, ascidians, and other invertebrate deuterostomes: experimental approaches. Academic Press. Garland, C. D., S. L. Cooke, J. F. Grant, and T. A. McMeekin. 1985. Ingestion of the bacteria on and the cuticle of crustose (non-articulated) coralline algae by post- larval and juvenile abalone Haliotis ruber Leach) from Tasmanian waters. Journal of Experimental Marine Biology and Ecology 91:137-149. Glass, G. V. 1976. Primary, secondary, and meta-analysis of research. Educational researcher:3-8. Goecke, F., A. Labes, J. Wiese, and J. F. Imhoff. 2010. Chemical interactions between marine macroalgae and bacteria. Marine Ecology-Progress Series 409:267-299. Grantham, B., G. L. Eckert, and A. Shaunk. 2003. Disperal potential of marine invertebrates in diverse habitats. Ecological Applications 13:S108-S116. Grosberg, R. K. and D. R. Levitan. 1992. For adults only? Supply-side ecology and the history of larval biology. Trends in Ecology and Evolution 7:130 - 133. Gurevitch, J., P. S. Curtis, and M. H. Jones. 2001. Meta-analysis in ecology. Advances in Ecological Research 32:199-247.

138

Literature cited

Hadfield, M. G. 2011. Biofilms and marine invertebrate larvae: What bacteria produce that larvae use to choose settlement sites. Annual review of marine science 3:453-470. Hadfield, M. G. and V. J. Paul. 2001. Natural chemical cues for settlement and metamorphosis of marine invertebrate larvae.in J. B. McClintock and W. Baker, editors. Marine chemical echology. CRC press, New York. Harder, T., S. Dobretsov, and P. Y. Qian. 2004. Waterborne polar macromolecules act as algal antifoulants in the seaweed Ulva reticulata. Marine Ecology-Progress Series 274:133-141. Harder, T., C. Lam, and P. Y. Qian. 2002a. Induction of larval settlement in the polychaete Hydroides elegans by marine biofilms: an investigation of monospecific diatom films as settlement cues. Marine Ecology-Progress Series 229:105-112. Harder, T., S. C. K. Lau, H. U. Dahms, and P. Y. Qian. 2002b. Isolation of bacterial metabolites as natural inducers for larval settlement in the marine polychaete Hydroides elegans (Haswell). Journal of Chemical Ecology 28:2029-2043. Harvey, A., T. Farr, K. Neill, W. Woelkerling, and W. A. Nelson. 2005. Coralline algae of central New Zealand: an identification guide to common "crustose" species. NIWA, Wellington, New Zealand. Hayakawa, J., T. Kawamura, T. Horii, and Y. Watanabe. 2007. Settlement of larval top shell Turbo (Batillus) cornutus in response to several marine algae. Fisheries Science 73:371-377. Hedges, L. V., I. Olkin, M. Statistiker, I. Olkin, and I. Olkin. 1985. Statistical methods for meta-analysis. Academic Press New York. Heyward, A. J. and A. P. Negri. 1999. Natural inducers of coral larval metamorphosis. Holmstrom, C., S. James, B. A. Neilan, D. C. White, and S. Kjelleberg. 1998. Pseudoalteromonas tunicata sp. nov., a bacterium that produces antifouling agents. International Journal of Systematic Bacteriology 48:1205-1212. Huang, S. Y. and M. G. Hadfield. 2003. Composition and density of bacterial biofilms determine larval settlement of the polychaete Hydroides elegans. Marine Ecology-Progress Series 260:161-172. Huang, X. and A. Madan. 1999. CAP3: A DNA sequence assembly program. Genome research 9:868-877. Huang, Y. 2012. An analysis of genes in the marine bacterium Pseudoalteromonas luteoviolacea underlying its capacity to induce larval setttlement in the polychaete Hydroides elegans. Trial 37:00. Huang, Y., S. Callahan, and M. G. Hadfield. 2012. Recruitment in the sea: bacterial genes required for inducing larval settlement in a polychaete worm. Scientific Reports 2. Hugenholtz, P. 2002. Exploring prokaryotic diversity in the genomic era. Genome Biology 3:reviews0003.0001 - reviews0003.0008. Huggett, M. J., G. R. Crocetti, S. Kjelleberg, and P. D. Steinberg. 2008. Recruitment of the sea urchin Heliocidaris erythrogramma and the distribution and abundance of inducing bacteria in the field. Aquatic Microbial Ecology 53:161-171. Huggett, M. J., J. E. Williamson, R. de Nys, S. Kjelleberg, and P. D. Steinberg. 2006. Larval settlement of the common Australian sea urchin Heliocidaris

139

Literature cited

erythrogramma in response to bacteria from the surface of coralline algae. Oecologia 149:604-619. Hunt, H. and R. Scheibling. 1997. Role of early post-settlement mortality in recruitment of benthic marine invertebrates. Marine Ecology Progress Series 155:269-301. Jiang, Z., K. G. Boyd, A. Mearns-Spragg, D. R. Adams, P. C. Wright, and J. G. Burgess. 2000. Two diketopiperazines and one halogenated phenol from cultures of the marine bacterium, Pseudoalteromonas luteoviolacea. Natural Product Letters 14:435-440. Johansen, H. W. 1981. Coralline algae, a first synthesis. CRC Press, Boca Raton, Florida. Johnson, C. R., D. G. Muir, D. S. Nichols, and B. M. Degnan. 1997. Bacterial induction of settlement and metamorphosis in marine invertebrates. Proceedings of the 8th International Coral Reef Symposium 2. Johnson, C. R., D. G. Muir, and A. L. Reysenbach. 1991a. Characteristic Bacteria Associated with Surfaces of Coralline Algae - a Hypothesis for Bacterial Induction of Marine Invertebrate Larvae. Marine Ecology-Progress Series 74:281-294. Johnson, C. R. and D. C. Sutton. 1994. Bacteria on the Surface of Crustose Coralline Algae Induce Metamorphosis of the Crown-of-Thorns Starfish Acanthaster Planci. Marine Biology 120:305-310. Johnson, C. R., D. C. Sutton, R. R. Olson, and R. Giddins. 1991b. Settlement of Crown- of-Thorns Starfish - Role of Bacteria on Surfaces of Coralline Algae and a Hypothesis for Deep-Water Recruitment. Marine Ecology-Progress Series 71:143-162. Johnston, E. L. and D. A. Roberts. 2009. Contaminants reduce the richness and evenness of marine communities: A review and meta-analysis. Environmental Pollution 157:1745-1752. Keough, M. J. 1983. Patterns of recruitment of sessile invertebrates in two subtidal habitats. Journal of Experimental Marine Biology and Ecology 66:213-245. Keough, M. J. and P. T. Raimondi. 1995. Responses of settling invertebrate larvae to bio-organic films - effects of different types of films. Journal of Experimental Marine Biology and Ecology 185:235-253. Khandeparker, L., A. C. Anil, and S. Raghukumar. 2003. Barnacle larval destination: piloting possibilities by bacteria and lectin interaction. Journal of Experimental Marine Biology and Ecology 289:1-13. Kirchman, D., S. Graham, D. Reish, and R. Mitchell. 1982. Lectin may mediate in the settlement and metamorphosis of Janua (dexiospira) brasiliensis grube (Polychaeta, Spirorbidae). Marine Biology Letters 3:131-142. Kitamura, H., S. Kitahara, and H. B. Koh. 1993. The induction of larval settlement and metamorphosis of 2 sea urchins, Pseduocentrotus depressus and Anthocidaris crassispina, by free fatty-acids extracted from the coralline red alga Corallina pilulifera. Marine Biology 115:387-392. Kurilenko, V. V., R. Christen, N. V. Zhukova, N. I. Kalinovskaya, V. V. Mikhailov, R. J. Crawford, and E. P. Ivanova. 2010. Granulosicoccus coccoides sp. nov., isolated from leaves of seagrass (Zostera marina). International Journal of Systematic and Evolutionary Microbiology 60:972-976.

140

Literature cited

Lachnit, T., M. Blumel, J. F. Imhoff, and M. Wahl. 2009. Specific epibacterial communities on macroalgae: phylogeny matters more than habitat. Aquatic Biology 5:181-186. Lachnit, T., D. Meske, M. Wahl, T. Harder, and R. Schmitz. 2011. Epibacterial community patterns on marine macroalgae are host-specific but temporally variable. Environmental Microbiology 13:655-665. Lam, C., T. Harder, and P. Y. Qian. 2005. Induction of larval settlement in the polychaete Hydroides elegans by extracellular polymers of benthic diatoms. Marine Ecology-Progress Series 286:145-154. Lau, S. C. K., K. K. W. Mak, F. Chen, and P. Y. Qian. 2002. Bioactivity of bacterial strains isolated from marine biofilms in Hong Kong waters for the induction of larval settlement in the marine polychaete Hydroides elegans. Marine ecology. Progress series 226:301-310. Lau, S. C. K., V. Thiyagarajan, S. C. K. Cheung, and P.-Y. Qian. 2005a. Roles of bacterial community composition in biofilms as a mediator for larval settlement of three marine invertebrates. Aquatic Microbial Ecology 38:41-51. Lau, S. C. K., V. Thiyagarajan, S. C. K. Cheung, and P. Y. Qian. 2005b. Roles of bacterial community composition in biofilms as a mediator for larval settlement of three marine invertebrates. Aquatic Microbial Ecology 38:41-51. Ling, S., S. Ibbott, and J. Sanderson. 2010. Recovery of canopy-forming macroalgae following removal of the enigmatic grazing sea urchin< i> Heliocidaris erythrogramma. Journal of Experimental Marine Biology and Ecology 395:135-146. Littler, M. M., D. S. Littler, and P. R. Taylor. 1995. Selective herbivore increases biomass of its prey: a chiton-coralline reef-building association. Ecology:1666-1681. Liu, W. T., T. L. Marsh, H. Cheng, and L. J. Forney. 1997. Characterization of microbial diversity by determining terminal restriction fragment length polymorphisms of genes encoding 16S rRNA. Applied and Environmental Microbiology 63:4516- 4522. Longford, S. 2008. The ecology of epiphytic bacteria on the marine red alga Delisea pulchra. PhD Thesis. University of New South Wales, Sydney. Longford, S. R., N. A. Tujula, G. R. Crocetti, A. J. Holmes, C. Holmstrom, S. Kjelleberg, P. D. Steinberg, and M. W. Taylor. 2007. Comparisons of diversity of bacterial communities associated with three sessile marine eukaryotes. Aquatic Microbial Ecology 48:217-229. Mantel, N. 1967. The Detection of Disease Clustering and a Generalized Regression Approach. Cancer Research 27:209-220. Matz, C., P. Deines, J. Boenigk, H. Arndt, L. Eberl, S. Kjelleberg, and K. Jürgens. 2004. Impact of Violacein-Producing Bacteria on Survival and Feeding of Bacterivorous Nanoflagellates. Applied and Environmental Microbiology 70:1593-1599. Minchinton, T. E. and R. E. Scheibling. 1991. The Influence of Larval Supply and Settlement on the Population Structure of Barnacles. Ecology 72:1867-1879. Moles, A. T., S. P. Bonser, A. G. Poore, I. R. Wallis, and W. J. Foley. 2011. Assessing the evidence for latitudinal gradients in plant defence and herbivory. Functional Ecology 25:380-388.

141

Literature cited

Morse, A. N. C. and D. E. Morse. 1984. Recruitment and metamorphosis of Haliotis larvae induced by molecules uniquely available at the surfaces of crustose red algae. Journal of Experimental Marine Biology and Ecology 75:191-215. Morse, D. E. 1985. Neurotransmitter-mimetic inducers of larval settlement and metamorphosis. Bulletin of Marine Science 37:697-706. Muyzer, G., E. C. de Waal, and A. G. Uitterlinden. 1993. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Applied and Environmental Microbiology 59:695-700. Negri, A. P., N. S. Webster, R. T. Hill, and A. J. Heyward. 2001. Metamorphosis of broadcast spawning corals in response to bacteria isolated from crustose algae. Nylund, G. M., F. Persson, M. Lindegarth, G. Cervin, M. Hermansson, and H. Pavia. 2010. The red alga Bonnemaisonia asparagoides regulates epiphytic bacterial abundance and community composition by chemical defence. FEMS Microbiology Ecology 71:84-93. Oksanen, L. 1991. Trophic levels and trophic dynamics: A consensus emerging? Trends in Ecology & Evolution 6:58-60. Pawlik, J. R. 1989. Larvae of the sea hare Aplysia californica settle and metamorphose on an assortment of macroalgal species. Marine ecology progress series. Oldendorf 51:195-199. Pawlik, J. R. 1990. Natural and artificial induction of metamorphosis of Phragmatopoma lapidosa californica (Polychaeta: Sabellariidae), with a critical look at the effects of bioactive compounds on marine invertebrate larvae. Bulletin of Marine Science 46:512-536. Pawlik, J. R. 1992. Chemical ecology of the settlement of benthic marine invertebrates. Oceanography and Marine Biology 30:273-335. Pearce, C. M. and R. E. Scheibling. 1990a. Induction of metamorphosis of larvae of the green sea-urchin, Strongylocentrotus-Droebachiensis, by coralline red algae. Biological Bulletin 179:304-311. Pearce, C. M. and R. E. Scheibling. 1990b. Induction of settlment and metamorphosis in the sand dollar Echinarachnius parma - evidence of an adult associated factor. Marine Biology 107:363-369. Pearce, C. M. and R. E. Scheibling. 1991. Effect of macroalgae, microbial films and conspecifics on the induction of metamorphosis of the green sea-urchin Strongylocentrotus droebachiensis. Journal of Experimental Marine Biology and Ecology 147:147-162. Pechenik, J. A. and C. C. Gee. 1993. Onset of metamorphic competence in larvae of the gastropod Crepidula fomicata (L.), judged by a natural and an artificial cue. Journal of Experimental Marine Biology and Ecology 167:59-72. Pinheiro, J. and D. Bates. 2000. Mixed Effects Models in S and S-Plus. Springer. Poore, A. G., A. H. Campbell, R. A. Coleman, G. J. Edgar, V. Jormalainen, P. L. Reynolds, E. E. Sotka, J. J. Stachowicz, R. B. Taylor, and M. A. Vanderklift. 2012. Global patterns in the impact of marine herbivores on benthic primary producers. Ecology letters 15:912-922. Pruesse, E., C. Quast, K. Knittel, B. M. Fuchs, W. Ludwig, J. Peplies, and F. O. Glöckner. 2007. SILVA: a comprehensive online resource for quality checked and aligned

142

Literature cited

ribosomal RNA sequence data compatible with ARB. Nucleic Acids Research 35:7188-7196. Qian, P.-Y., V. Thiyagarajan, S. C. K. Lau, and S. C. K. Cheung. 2003. Relationship between bacterial community profile in biofilm and attachment of the acorn barnacle Balanus amphitrite. Aquatic Microbial Ecology 33:225-237. Quince, C., A. Lanzen, T. P. Curtis, R. J. Davenport, N. Hall, I. M. Head, L. F. Read, and W. T. Sloan. 2009. Accurate determination of microbial diversity from 454 pyrosequencing data. Nature methods 6:639-641. Radjasa, O. K., T. Martens, H. P. Grossart, A. Sabdono, M. Simon, and T. Bachtiar. 2009. Antibacterial property of a coral-associated bacterium Pseudoalteromonas luteoviolacea against shrimp pathogenic Vibrio harveyi (In Vitro Study). HAYATI Journal of Biosciences 12:77. Raff, E. C., E. M. Popodi, B. J. Sly, F. R. Turner, J. T. Villinski, and R. A. Raff. 1999. A novel ontogenetic pathway in hybrid embryos between species with different modes of development. Development 126:1937-1945. Raimondi, P. T. and M. J. Keough. 1990. Behavioural variability in marine larvae. Australian Journal of Ecology 15:427-437. Rao, D., J. S. Webb, C. Holmström, R. Case, A. Low, P. Steinberg, and S. Kjelleberg. 2007. Low densities of epiphytic bacteria from the marine alga Ulva australis inhibit settlement of fouling organisms. Applied and Environmental Microbiology 73:7844-7852. Rasband, W. S. 1997. ImageJ. Page http://imagej.nih.gov/ij/. U. S. National Institutes of Health, Bethesda, Maryland, USA. Ritson-Williams, R., S. Shjegstad, and V. Paul. 2003. Host specificity of four corallivorous Phestilla nudibranchs (Gastropoda: Opisthobranchia). Marine Ecology Progress Series 255:207-218. Roberts, R. 2001. A review of settlement cues for larval abalone. Journal of Shellfish Research 20:571-586. Roberts, R. D., M. F. Barker, and P. Mladeno. 2010a. Is settlement of Haliotis iris larvae on coralline algae triggered by the its surface biofilm? Roberts, R. D., M. F. Barker, and P. Mladenov. 2010b. Is settlement of Haliotis iris larvae on coralline algae triggered by the alga or its surface biofilm? Journal of Shellfish Research 29:671-678. Roberts, R. D., H. F. Kaspar, and R. J. Barker. 2004. Settlement of abalone (Haliotis iris) larvae in response to five species of coralline algae. Rodriguez, S., F. Ojeda, and N. Inestrosa. 1993. Settlement of benthic marine invertebrates Marine Ecology-Progress Series 97:193-207. Schloss, P. D., S. L. Westcott, T. Ryabin, J. R. Hall, M. Hartmann, E. B. Hollister, R. A. Lesniewski, B. B. Oakley, D. H. Parks, C. J. Robinson, J. W. Sahl, B. Stres, G. G. Thallinger, D. J. V. Horn, and C. F. Weber. 2009. Introducing mothur: open- source, platform-independent, community-supported software for describing and comparing microbial communities. Applied and environmental microbiology 75:7537-7541. Smith, C. J., B. S. Danilowicz, A. K. Clear, F. J. Costello, B. Wilson, and W. G. Meijer. 2005. T-Align, a web-based tool for comparison of multiple terminal restriction fragment length polymorphism profiles. FEMS Microbiology Ecology 54:375- 380. . 143

Literature cited

Sneed, J. M. and G. Pohnert. 2011. The green alga Dicytosphaeria ocellata and its organic extracts alter natural bacterial biofilm communities. Biofouling 27:347- 356. Staufenberger, T., V. Thiel, J. Wiese, and J. F. Imhoff. 2008. Phylogenetic analysis of bacteria associated with Laminaria saccharina. FEMS Microbiology Ecology 64:65-77. Steinberg, P. D. 1995. Interaction between the canopy dwelling echinoid Holopneustes purpurescens and its host kelp. Marine Ecology Progress Series 127:169-181. Steinberg, P. D. and R. de Nys. 2002. Chemical mediation of colonization of seaweed surfaces. Journal of Phycology 38:621-629. Steinberg, P. D., R. de Nys, and S. Kjelleberg. 2001. Chemical mediation of surface colonization. Pages 335-387. CRC Press, Boca Raton, FL. Steneck, R. S. 1982. A limpet-coralline alga association: adaptations and defenses between a selective herbivore and its prey. Ecology:507-522. Strathmann, R. 1974. Spread of sibling larvae of sedentary marine invertebrates. American Naturalist 108:29-44. Strathmann, R. R. 1985. Feeding and non-feeding larval development and life-history evolution in marine invertebrates. Annual Review of Ecology and Systematics 16:339-361. Swanson, R. L., M. Byrne, T. A. A. Prowse, B. Mos, S. A. Dworjanyn, and P. D. Steinberg. 2012a. Dissolved histamine: a potential habitat marker promoting settlement and metamorphosis in sea urchin larvae. Marine Biology 159:915-925. Swanson, R. L., M. Byrne, T. A. A. Prowse, B. Mos, S. A. Dworjanyn, and P. D. Steinberg. 2012b. Dissolved histamine: a potential habitat marker promoting settlement and metamorphosis in sea urchin larvae. Marine Biology:1-11. Swanson, R. L., R. de Nys, M. J. Huggett, J. K. Green, and P. D. Steinberg. 2006. In situ quantification of a natural settlement cue and recruitment of the Australian sea urchin Holopneustes purpurascens. Marine Ecology-Progress Series 314:1-14. Swanson, R. L., J. E. Williamson, R. De Nys, N. Kumar, M. P. Bucknall, and P. D. Steinberg. 2004. Induction of settlement of larvae of the sea urchin Holopneustes purpurascens by histamine from a host alga. The Biological Bulletin 206:161-172. Tebben, J., D. M. Tapiolas, C. A. Motti, D. Abrego, A. P. Negri, L. L. Blackall, P. D. Steinberg, and T. Harder. 2011. Induction of larval metamorphosis of the coral Acropora millepora by tetrabromopyrrole isolated from a Pseudoalteromonas Bacterium. PloS one 6: e19082. Thiyagarajan, V., S. C. K. Lau, S. C. K. Cheung, and P.-Y. Qian. 2006. Cypris habitat selection facilitated by microbial films influences the vertical distribution of subtidal barnacle Balanus trigonus. Microbial ecology 51:431-440. Thorson, G. 1950. Reproductive and larval ecology of marine bottom invertebrates. Biological Reviews 25:1-45. Todd, C. 1998. Larval supply and recruitment of benthic invertebrates: do larvae always disperse as much as we believe? Hydrobiologia 375-376:1-21. Tran, C. and M. G. Hadfield. 2011. Larvae of Pocillopora damicornis (Anthozoa) settle and metamorphose in response to surface-biofilm bacteria. Marine Ecology Progress Series 433:85-96.

144

Literature cited

Tujula, N. A., G. R. Crocetti, C. Burke, T. Thomas, C. Holmstrom, and S. Kjelleberg. 2009. Variability and abundance of the epiphytic bacterial community associated with a green marine Ulvacean alga. Isme Journal 4:301-311. Turnbaugh, P. J., R. E. Ley, M. Hamady, C. M. Fraser-Liggett, R. Knight, and J. I. Gordon. 2007. The human microbiome project. Nature 449:804-810. Tyson, G. W., I. Lo, B. J. Baker, E. E. Allen, P. Hugenholtz, and J. F. Banfield. 2005. Genome-directed isolation of the key nitrogen fixer Leptospirillum ferrodiazotrophum sp. nov. from an acidophilic microbial community. Appl Environ Microbiol 71:6319-6324. Unabia, C. R. C. and M. G. Hadfield. 1999. Role of bacteria in larval settlement and metamorphosis of the polychaete Hydroides elegans. Marine Biology 133:55- 64. Underwood, A. J. 1997. Experiments in ecology: Their logical design and interpretation using analysis of variance. Cambridge University Press, Cambridge, UK. Underwood, A. J. and P. G. Fairweather. 1989. Supply-side ecology and benthic marine assemblages. Trends in Ecology and Evolution 4:16-20. Vynne, N. G. 2011. Bioactivity and phylogeny of the marine bacterial genus Pseudoalteromonas. Technical University of Denmark, Denmark. Vynne, N. G., M. Månsson, K. F. Nielsen, and L. Gram. 2011. Bioactivity, chemical profiling, and 16S rRNA-based phylogeny of Pseudoalteromonas strains collected on a global research cruise. Marine Biotechnology 13:1062-1073. Webb, J. S., M. Givskov, and S. Kjelleberg. 2003. Bacterial biofilms: prokaryotic adventures in multicellularity. Current Opinion in Microbiology 6:578-585. Webster, N. S., L. D. Smith, A. J. Heyward, J. E. M. Watts, R. I. Webb, L. L. Blackall, and A. P. Negri. 2004. Metamorphosis of a scleractinian coral in response to microbial bioﬁlms. Applied and Environmental Microbiology 70. Webster, N. S., S. Uthicke, E. S. Botté, F. Flores, and A. P. Negri. 2013. Ocean acidification reduces induction of coral settlement by crustose coralline algae. Global Change Biology 19:303-315. Weinberger, F., H.-G. Hoppe, and M. Friedlander. 1997. Bacterial induction and inhibition of a fast mecrotic response in Gracilaria conferta (Rhodophyta). Journal of Applied Phycology 9:277-285. Werner, J. J., O. Koren, P. Hugenholtz, T. Z. DeSantis, W. A. Walters, J. G. Caporaso, L. T. Angenent, R. Knight, and R. E. Ley. 2011. Impact of training sets on classification of high-throughput bacterial 16s rRNA gene surveys. The ISME journal 6:94- 103. Whalan, S., N. S. Webster, and A. P. Negri. 2012. Crustose coralline algae and a cnidarian neuropeptide trigger larval settlement in two coral reef sponges. PloS one 7:e30386. Wieczorek, S. K., A. S. Clare, and C. D. Todd. 1995. Inhibitory and facilitatory effects of microbial films on settlement of Balanus amphitrite larvae. Marine Ecology- Progress Series 119:221-228. Wieczorek, S. K. and C. D. Todd. 1998. Inhibition and facilitation of settlement of epifaunal marine invertebrate larvae by microbial biofilm cues. Biofouling 12:81-118. Williams, D. and D. Anderson. 1975. The reproductive system, embryonic development, larval development and metamorphosis of the sea urchin 145

Literature cited

Heliocidaris erythrogramma (Val.) (Echinoidea : Echinometridae). Australian Journal of Zoology 23:371-403. Williams, E. A., A. Craigie, A. Yeates, and S. M. Degnan. 2008. Articulated coralline algae of the genus Amphiroa are highly effective natural inducers of settlement in the tropical abalone Haliotis asinina. Biological Bulletin 215:98-107. Williams, E. A., S. Cummins, and S. M. Degnan. 2009. Settlement specifics: Effective induction of abalone settlement and metamorphosis corresponds to biomolecular composition of natural cues. Communicative & integrative biology 2:347-349. Williams, G. B. 1965. Observations on the behaviour of the planulae larvae of Clava squamata. Journal of the Marine Biological Association of the United Kingdom 45:257-273. Williamson, J. E., D. G. Carson, R. De Nys, and P. D. Steinberg. 2004. Demographic consequences of an ontogenetic shift by a sea urchin in response to host plant chemistry. Ecology 85:1355-1371. Williamson, J. E. and P. D. Steinberg. 2002. Reproductive cycle of the sea urchin Holopneustes purpurascens (Temnopleuridae : Echinodermata). Marine Biology 140:519-532. Wilson, K. A., M. E. Andrews, and R. A. Raff. 2005. Dissociation of expression patterns of homeodomain transcription factors in the evolution of developmental mode in the sea urchins Heliocidaris tuberculata and H. erythrogramma. Evol Dev 7:401-415. Wright, J. T., S. A. Dworjanyn, C. N. Rogers, P. D. Steinberg, J. E. Williamson, and A. G. B. Poore. 2005. Density-dependent sea urchin grazing : differential removal of species, changes in community composition and alternative community states. Marine Ecology Progress Series 298:143-156. Zhao, B., S. Zhang, and P.-Y. Qian. 2003. Larval settlement of the silver- or goldlip pearl oyster Pinctada maxima (Jameson) in response to natural biofilms and chemical cues. Aquaculture 220:883-901. Zobell, C. E. and E. C. Allen. 1935. The significance of marine bacteria in the fouling of submerged surfaces. Journal of bacteriology 29:239.

146