THE ECOLOGY OF CHEMICAL DEFENCE

IN A FILAMENTOUS MARINE RED ALGA

NICHOLAS A. PAUL

A thesis submitted to the University of New South Wales for the degree of Doctor of Philosophy

July 2006 i

ACKNOWLEDGEMENTS··························································································································· iii

CHAPTER 1. GENERAL INTRODUCTION ...... 1 NATURAL PRODUCTS CHEMISTRY OF MACROALGAE ...... 2 THE RED ALGAE ...... 2 ECOLOGICAL ROLES OF SECONDARY METABOLITES ...... 3 i) Chemical mediation of interactions with herbivores ...... 3 ii) Chemical mediation of interactions with fouling organisms ...... 4 ULTRASTRUCTURE OF SPECIALISED CELLS AND STRUCTURES...... 5 RESOURCE ALLOCATION TO SECONDARY METABOLITE PRODUCTION...... 6 THESIS AIMS ...... 7 CHAPTER 2. CHEMICAL DEFENCE AGAINST BACTERIA IN ASPARAGOPSIS ARMATA: LINKING STRUCTURE WITH FUNCTION...... 9 INTRODUCTION ...... 9 MATERIAL AND METHODS...... 11 Study organism...... 11 Screening of algal extracts against known microorganisms ...... 12 Chemical analysis...... 13 Quantification of metabolite release ...... 15 Screening of metabolites against known microorganisms ...... 16 Effect of removing bromine on algal structure and chemistry ...... 16 Bacterial densities on algae cultured in bromide (+) and bromide (-) media...... 17 Antibacterial assays using bromide (+) and bromide (-) algae ...... 18 Statistical analysis...... 19 RESULTS...... 20 Antibiotic tests of extracts ...... 20 Natural product chemistry...... 20 Release of major metabolites ...... 21 Antibiotic testing of major metabolites...... 22 Effect of bromine on cell structure and chemistry...... 23 Gland cells and related structures ...... 23 Epiphytic bacterial densities on bromide (+) and bromide (-) algae...... 24 Antibacterial assays with bromide (+) and bromide (-) algae...... 24 DISCUSSION ...... 37 Conclusions ...... 41 CHAPTER 3. ULTRASTRUCTURE OF THE GLAND CELLS OF THE RED ALGA ASPARAGOPSIS ARMATA (BONNEMAISONIACEAE)...... 43 INTRODUCTION ...... 43 MATERIALS AND METHODS...... 45 Study organism...... 45 Light Microscopy...... 46 Manipulation of gland cells...... 46 Epifluorescence microscopy...... 47 TEM of cryofixed and freeze-substituted material ...... 47 RESULTS...... 48 Light microscopy ...... 48 Bromine manipulation...... 49 Epifluorescence microscopy...... 50 Transmission electron microscopy...... 50 DISCUSSION ...... 59 ii

CHAPTER 4. SEAWEED-HERBIVORE INTERACTIONS AT SMALL SCALES: A DIRECT TEST OF FEEDING DETERRENCE USING FILAMENTOUS ALGAE ...... 64 INTRODUCTION ...... 64 MATERIALS AND METHODS ...... 67 Feeding assay with an amphipod ...... 67 Collection and culturing of Asparagopsis armata ...... 68 Effects of bromide (+) and bromide (-) artificial media on algae...... 68 Herbivores...... 69 Feeding assays - Whole plants ...... 70 Feeding assays - Artificial diets ...... 72 Statistical analysis...... 75 RESULTS...... 76 Feeding assays with the amphipod Hyale nigra...... 76 Effects of bromide (+) and bromide (-) artificial media on algal thalli...... 76 Whole-plant feeding assays...... 77 Artificial-diet feeding assays ...... 78 DISCUSSION ...... 88 CHAPTER 5. TRADE-OFFS BETWEEN GROWTH AND CHEMICAL DEFENCE IN THE RED ALGA ASPARAGOPSIS ARMATA ...... 95 INTRODUCTION ...... 95 METHODS...... 98 Study organism...... 98 Chemical analyses...... 99 Growth and chemistry across a light gradient...... 99 Correlations between growth and metabolite production in the laboratory...... 100 Use of gland cell size to inform allocation to defence...... 101 Resource allocation at the cellular scale in the laboratory ...... 102 Comparison of laboratory-cultured and field-sampled algae...... 103 Cellular correlates between defence and growth in the field...... 104 Statistical analysis...... 105 RESULTS...... 106 Growth and chemistry across a light gradient...... 106 Covariance between growth and metabolite production in the laboratory...... 106 Variation in cellular allocation to defence and growth at two light levels ...... 107 Comparison of field-sampled and laboratory-cultured algae...... 108 Allocation of resources along the growth axis in field samples...... 109 DISCUSSION ...... 121 Trade-offs between growth and defence in Asparagopsis armata ...... 121 Cellular correlates of allocation to chemical defence...... 123 Application of plant defence models to A. armata ...... 125 Marine versus terrestrial systems...... 126 Conclusions ...... 128 CHAPTER 6. GENERAL DISCUSSION...... 130 CHEMICAL DEFENCES OF MARINE ALGAE...... 131 Large versus small macroalgae ...... 132 STRUCTURE/FUNCTION RELATIONSHIPS ...... 134 RESOURCE ALLOCATION AND COSTS OF ALGAL CHEMICAL DEFENCE...... 135 ASPARAGOPSIS ARMATA – A MODEL ORGANISM FOR MARINE CHEMICAL ECOLOGY? ...... 137 REFERENCES...... 141

APPENDIX ONE Paul NA, Cole L, de Nys R, Steinberg PD (2006) J. Phycol 42:637-645 ·············· 154 iii

Acknowledgements

First and foremost, thank you to my two supervisors Prof. Peter Steinberg and Prof. Rocky de Nys for the opportunity to do this Ph. D. and for the mentoring that it entailed. Peter, thanks for the wisdom and teachings and all the support. I will be hard pressed to find another boss whose reaction is to burst into song at any opportune time – and I will miss it! Rocky, you have undoubtedly created a supervisorial relationship that should be the envy of many; meetings just won’t be the same in the future when conducted in an office and not on fibreglass sticks in the surf.

I owe much gratitude to members (or ex-) of the Steinberg emporium and School of BEES at UNSW for their expert help. Special mention should go to Neda Shakibaee, Peter Schupp, Louise Cole, Daniel Davies, Richard Taylor, Joe Zuccarello, Carola Holmström, Tim Charlton, Keyne Monro, Adriana Verges and Sharon Longford for assistance with the diverse methods and techniques that I have tackled over the years and the help with getting this thesis out. To Bec and Megan, well, it was a great journey that we all had together and I will cherish it always.

Thanks to the gang of people who have made my life at work that much more enjoyable – whether it be sampling or something more of a “Royal” decree. Thanks for the good times (there had to be some excuses for taking so long!).

A big thank you to my family. Mum, Dad and Kristie you have all been there with unblinking support over the years. Then, of course, my beautiful Caroline. Thank you so much for your love. And while you may never love seaweed as much as I do (does it bother you that I have amorous feelings towards algae?), I hope that it has been worth it for you too – well, maybe I just shouldn’t ask that question!

Next! CHAPTER 1

General Introduction

Marine macroalgae are a polyphyletic group comprised of species from three separate divisions – Chlorophyta, Phaeophyta and Rhodophyta (Clayton and King 1990, Lee 1999). These organisms are typically found in multispecies assemblages on benthic reefs, from the shoreline to depths greater than 200 m (Lee 1999). Macroalgae are important primary producers and are central to the ecology of coastal marine systems (Mann 1973, Lüning 1981, Paine 1990). They are consumed by various types of herbivores, both large and small (Underwood 1980, Lubchenco and Gaines 1981, Choat 1982, Hay and Fenical 1988, Brawley 1992). Macroalgae also compete for the limited space available on benthic reefs (Carpenter 1990, Paine 1990), whilst bacteria and other fouling organisms vie for a position on the algae themselves (Wahl 1989, Schmitt et al. 1995, Steinberg and de Nys 2002, Kubanek et al. 2003). For many macroalgae, their interactions with other organisms can be mediated by producing bioactive natural products (McClintock and Baker 2001). Chapter 1 General Introduction 2

Natural Products Chemistry of Macroalgae

Numerous novel and interesting compounds have been isolated from marine algae (reviewed by Fenical 1982, Faulkner 1996, Paul and Puglisi 2004). Many of the ecological interactions involving macroalgae are mediated by compounds that are not fashioned by primary metabolism. Such compounds are known as secondary metabolites and have a range of structures – including terpenes, acetogenins, alkaloids and phenolics – with various molecular weights (Fenical 1982). While there are some basic structural similarities between marine and terrestrial secondary metabolites, some notable distinctions result from the availability of different elements to these organisms (Fenical 1982). For instance, halogenated compounds are widespread in marine organisms but are comparatively rare in terrestrial systems (Fenical 1982, Gribble 1994). Halogenation is a particularly common feature of the secondary metabolites from red macroalgae, which produce a variety of compounds that are unique to this division (Fenical 1975, Faulkner 1996).

The Red Algae

The largest group of macroalgae are the red algae (Rhodophyta), containing upwards of 5000 species (Cole and Sheath 1990, Huisman 2000). The red algal division is recognised for its complex life histories and varied growth forms (Cole and Sheath 1990). For example, some red algae have a triphasic life history, consisting of free- living gametophytes (haploid), sporophytes (diploid) and a third stage, carposporophyte (diploid), that is parasitic on the gametophyte (Feldmann and Feldmann 1939). The different life history stages may have heteromorphic morphologies that can be important to algal ecology, in particular, their interactions with herbivores (Searles 1980). The first example of a heteromorphic life history in macroalgae was identified in a red alga, Asparagopsis armata Harvey (Bonnemaisoniaceae), by Feldmann and Feldmann Chapter 1 General Introduction 3

(1939). Despite that red algae typically have simple body plans (Coomans and Homersand 1990), this group has a diverse array of specialised cells (Murray and Dixon 1992). Furthermore, natural products from red algae have demonstrated activity against a wide range of organisms, from bacteria to fish (McConnell and Fenical 1979, Hay et al. 1987b, de Nys et al. 1995, Wright et al. 2004, Nylund et al. 2005). Taken together, these features make red algae particularly interesting for studies of the ecology of chemical defences.

Ecological Roles of Secondary Metabolites

Secondary metabolites from marine algae can act as consumer deterrents (Hay and Fenical 1988), antifoulants (Wahl et al. 1989, de Nys et al. 1995), antibiotics (McConnell and Fenical 1979, Hellio et al. 2001), sperm attractants (Maier and Muller 1986) and in allelopathy (de Nys et al. 1991). Research to date has predominantly focused on roles of secondary metabolites in herbivore deterrence (Hay and Fenical 1988, Paul 1992, Hay 1996). However, there has been renewed interest in their role as defence against colonisation by fouling organisms (Steinberg and de Nys 2002).

i) Chemical mediation of interactions with herbivores

All algae are susceptible to consumers (Littler and Littler 1980, Steneck and Dethier 1994). These herbivores are diverse and include large grazers, such as fish and urchins (Choat 1982, Paul and Hay 1986, Hay et al. 1987b), a variety of molluscs (Underwood 1980, Pennings 1990), and small, plant-dwelling herbivores (mesograzers), such as amphipods and gastropods (Brawley and Adey 1981, Brawley 1992, Duffy and Hay 2000). In order to reduce the impact of these herbivores, algae may tolerate consumption through fast growth or they may escape herbivory by occupying spatial or temporal niches (Littler and Littler 1980, Lubchenco and Gaines 1981). Algae can also exhibit Chapter 1 General Introduction 4

resistance traits, such as structural (Paul and Hay 1986, Hay et al. 1988b), nutritional (Duffy and Paul 2000, Cruz-Rivera and Hay 2003) or chemical defences (Hay and Fenical 1988, and see Paul 1992, McClintock and Baker 2001). Most work on chemical resistance by macroalgae has been on large, apparent species (Steinberg 1984, Hay et al. 1988a, Van Alstyne et al. 2001, Wright et al. 2004, Toth et al. 2005). These algae produce a broad range of chemical defences, which can be complemented by physical defences (Paul and Hay 1986, Hay et al. 1994). To date, grouping of macroalgal susceptibility to herbivores by functional form predicts that smaller, less complex algae will have tolerance traits (i.e. growth-based) rather than resistance characteristics (Littler and Littler 1980, Steneck and Dethier 1994). However, as other predictions from the functional form paradigm may not hold (Padilla and Allen 2000), the possibility that small individuals also produce chemical defences requires investigation.

ii) Chemical mediation of interactions with fouling organisms

Although biofouling is a more subtle process than herbivory, its effects may at times be far more detrimental. Fouling by both macro- and micro-organisms can lead to physical damage through increased drag (Wahl 1989), or reduce photosynthesis by limiting light attenuation (Sand-Jensen 1977). Fouling can also make the host organism more susceptible to collateral damage resulting from the consumption of fouling organisms (Wahl et al. 1997). Furthermore, a large variety of pathogenic bacteria can cause disease (Littler & Littler 1995, Correa & Sanchez 1996, Kushmaro et al. 1997, Potin et al. 2002). Chemical mediation of the interactions with bacteria and other fouling organisms can provide fitness benefits to the producing organisms (Wahl 1989, Steinberg and de Nys 2002). Many marine algae contain natural products that are active as antibiotics (Sieburth and Conover 1965, Hornsey and Hide 1974, McConnell and Fenical 1979, Chapter 1 General Introduction 5

Hellio et al. 2001). However, it is no longer appropriate to demonstrate the presence of active compounds within (i.e. extracted from) an organism when testing surface- mediated interactions of natural products (such as bacterial antifouling). Only a handful of studies have adequately related the localisation of known natural products to surface- mediated functions in algae (Schmitt et al. 1995, de Nys et al. 1998, Kubanek et al. 2003). The localisation of secondary metabolites in marine algae remains a major issue for determining the ecological roles of these compounds.

Ultrastructure of Specialised Cells and Structures

As secondary metabolites can be cytotoxic (or bioactive) they are typically localised in specialised structures to avoid autotoxicity in the producing organism (Fenical 1982). The location of sequestered compounds and the ultrastructure of the specialised cells and structures in which they are found will ultimately dictate the potential roles for secondary metabolites. However, despite that specialised cells and subcellular structures are common features in macroalgae (Kylin 1927, Young 1977, Womersley 1996 & 1998) comparatively little is known of their ultrastructure or for what purposes they have evolved (Murray and Dixon 1992). Ultrastructural investigation of the specialised structures in algae may inform the ecological function of their stored components (Young and West 1979, Young et al. 1980). Some of the functions attributed to these structures or, more specifically, their contents, include light collection, nutrition, excretion and defence (Young and West 1979). However, these roles have rarely been empirically tested. Interestingly, the specialised cells of red algae have long been thought to contain compounds that are herbivore deterrents (Kylin 1927). The various specialised structures and high diversity of natural products present in macroalgae may relate to their broad defensive requirements against multiple natural enemies. However, the links between the types of Chapter 1 General Introduction 6

compounds found in the specialised structures and their ecological roles in most marine algae are yet to be determined (but see Ragan 1976, Young and West 1979, Young et al. 1980, Dworjanyn et al. 1999, Schoenwaelder 2002).

Resource Allocation to Secondary Metabolite Production

Several different models attempt to explain the diversity, abundance and concentration of secondary metabolites found in terrestrial plants (Feeny 1976, Rhoades 1979, Herms and Mattson 1992, Stamp 2003). The evolution of plant chemical defences is typically examined in the context of resource allocation, as it is assumed that it is costly for an organism to produce defence compounds (Rhoades 1979, Simms 1992, Zangerl and Bazzaz 1992, Bergelson and Purrington 1996). This cost exists as the resources allocated to chemical defences cannot be used in primary processes, and this will manifest as a reduction in fitness for those plants with higher investment in defence in the absence of any natural enemies. Terrestrial plant defence models have also been applied to marine macroalgae. For instance, within-plant variation in resource allocation with respect to optimal defence (Rhoades 1979) has been used to explain the distribution of allelochemicals in macroalgae with differentiated tissues (Steinberg 1984, Hay et al. 1988b, Pfister 1992, Cronin and Hay 1996a,b, Van Alstyne et al. 1999, Pavia et al. 2002). Others have examined the changes in allocation patterns in response to variation in the available resources (Yates and Peckol 1993, Steinberg 1995, Cronin and Hay 1996a & 1996b, Puglisi and Paul 1997). A limited number of studies have tested whether it is costly for marine algae to produce secondary metabolites (Pavia et al. 1999, Dworjanyn et al. in press). While the natural products chemistry of many marine algae is well established (Fenical 1982, Faulkner 1996, MarinLit 2004, Paul and Puglisi 2004), comparatively Chapter 1 General Introduction 7

little is known of the factors that generate variation in secondary metabolite production in marine algae and how this relates to the evolution of algal chemical defences. Furthermore, most of the existing research on chemical defence and resource allocation in marine systems has been on brown algae (Steinberg 1984, Cronin and Hay 1996a, Pavia et al. 1999, Van Alstyne et al. 1999, Jormalainen and Honkanen 2004; but see Puglisi and Paul 1997, Dworjanyn et al. in press). Many red algae with established chemical defences remain to be examined in the context of the plant defence models.

Thesis Aims

In this thesis, I address key ecological questions in marine chemical ecology with the red alga Asparagopsis armata Harvey (Bonnemaisoniaceae) as a model. The natural product chemistry of the genus Asparagopsis has been described in detail by McConnell and Fenical (1977). It is comprised of numerous halogenated metabolites, including halomethanes, haloacids and haloacetones. However, little is known of (1) the ultrastructure of the storage structures for these halogenated metabolites, (2) the ecological roles that the metabolites have in defence, and (3) the sources of variation in levels of the major halogenated metabolites, in particular, differences in metabolite production among genotypes or in response to resource variation. Chapters 2 and 3 examine the relationship between the storage of halogenated metabolites in A. armata and their ecological function. Localisation of metabolites is particularly important to defence functions that require the specific delivery of metabolites, for example, secondary metabolites that function as antifoulants must be delivered to the algal-water interface. In Chapter 2, I develop a quantitative method for the analysis of the major halogenated metabolites in A. armata. This allows for the comparison of metabolite levels between the two free-living life history stages (i.e. tetrasporophytes and Chapter 1 General Introduction 8

gametophytes) and also between laboratory-cultured algae and those from the field. I subsequently test the antibacterial roles of the halogenated metabolites in a novel assay by manipulating the production of these metabolites in cultured algae. This provided a means to examine the relationship between cellular structures and the potential release of metabolites from the specialised cells in this alga. In Chapter 3, I further examine the relationship between storage structures and ecological functions of secondary metabolites in the tetrasporophyte of A. armata. The ultrastructure of the specialised cells involved in the production of halogenated secondary metabolites are investigated in detail with a range of techniques, including light, epifluorescence and transmission electron microscopy. In Chapter 4, I test whether filamentous red algae can produce metabolites that function in chemical defence against herbivores. I specifically test the efficacy of the halogenated metabolites from the tetrasporophyte of A. armata against herbivores. This is done by manipulating the compounds in the alga and offering these manipulated individuals to herbivores in feeding assays. Furthermore, a range of smaller herbivores (mesograzers) are examined, as these mesograzers make choices at a scale corresponding to the small size of these algae. Chapter 5 tests resource allocation to chemical defence in A. armata. Variation in halogenated secondary metabolite production across light levels, and across and within individuals is examined using chemical and cellular analyses. I further test whether the production of these chemicals incurs a cost to this alga. Some aspects of the biology of A. armata (and macroalgae in general) that may limit the application of the terrestrial plant defence models are also investigated. Chapter 6 provides a general discussion for the thesis. CHAPTER 2

Chemical defence against bacteria in Asparagopsis armata: linking structure with function*

Introduction

Marine pathogenic bacteria can have substantial impacts on algae (Littler and Littler 1995, Correa and Sanchez 1996, Kushmaro et al. 1997, Potin et al. 2002). Furthermore, epiphytic bacteria can indirectly lead to adverse effects, such as drag or reduced photosynthesis, by enhancing the settlement of macrofoulers (Wahl 1989). Yet despite the constant threat of bacterial epibiosis, many algae remain largely free of disease and heavy fouling (Correa and Sanchez 1996, Potin et al. 2002). Such observations have encouraged widespread screening of algae for novel antimicrobial compounds (McConnell and Fenical 1979). However, algae-derived antimicrobials have generally been tested by applying whole plants, crude extracts, or pure compounds to plated microorganisms, most of which have limited ecological relevance (Hornsey and Hide 1974, Al-Ogily and Knight-Jones 1977, Hellio et al. 2001). While these antimicrobial assays provide insight into potential functions for algal metabolites, their roles in mediating interactions with epiphytic bacteria are often unknown.

* This chapter is published in Marine Ecology Progress Series (2006) 306:87-101 Chapter 2 Chemical defence against bacteria 10

The role of algal metabolites as inhibitors of surface colonisation (antifoulants) by macro- and microorganisms in a more ecological context has received renewed attention (reviewed by Steinberg and de Nys 2002). In part, this increased focus is due to investigations of the localisation and delivery mechanisms of biologically-active secondary metabolites (Schmitt et al. 1995, de Nys et al. 1998, Dworjanyn et al. 1999). This is a crucial issue for surface active inhibitors, and the presentation of metabolites for chemical defence will ultimately depend on the nature of the structures in which these metabolites are produced and stored. However, few studies have examined the ecological roles of algal secondary metabolites in microbial antifouling at relevant surface concentrations (Kubanek et al. 2003). In general, the production of biologically-active metabolites is inherently linked to an ability to partition compounds into specialised storage structures in order to avoid autotoxicity (McKey 1979). Storage structures of marine algae have long been proposed to contain materials that have ecological function (Kylin 1927). These specialised structures can be single cells, or components thereof, and have a varied ultrastructure. They include gland cells (or vesicle cells), in which the vesicle occupies practically the entire cellular space (Young and West 1979, Dworjanyn et al. 1999), and cells with refractile inclusions, such as corp en cerise (Young et al. 1980) and physodes (Schoenwaelder 2002). Most specialised storage cells are located at, or close to, the surface cell layer; however, not all algae have the means to deliver natural products to the surface (Young and West 1979). Delivery is an essential criterion for surface mediated interactions such as the inhibition of fouling by bacteria. The release of a metabolite to the algal surface and its demonstrated efficacy against sympatric bacteria represent the two major issues in distinguishing a natural antifoulant from other potentially bioactive metabolites. I investigated the localisation of secondary metabolites in the red alga Asparagopsis armata and their role as natural inhibitors of the growth of marine bacteria. A. armata is a member of the Bonnemaisoniaceae, a family with well characterised chemistry (Fenical 1975). Over 100 halogenated compounds are found in Chapter 2 Chemical defence against bacteria 11

the genus Asparagopsis alone, including haloforms, haloacids and haloketones (McConnell and Fenical 1977, Woolard et al. 1979). The natural products of A. armata have been proposed to act in bacterial antifouling (Hornsey and Hide 1974, McConnell and Fenical 1979), although ecological tests against relevant bacteria have not been performed. While it is generally accepted that these halogenated metabolites are localised in specialised gland cells, or ‘vesicle’ cells, (Wolk 1968, Fenical 1975, McConnell and Fenical 1977), the delivery and release of metabolites to the surface has not been determined. This needs to be demonstrated in order to confirm an ecological role as natural antifoulants. In this Chapter, I examined how the natural products chemistry and morphology of Asparagopsis armata related to its antibacterial properties. More specifically I asked: (1) Do extracts of this alga inhibit bacterial growth? (2) What are the major metabolites in the 2 life-history stages of A. armata? (3) Where are these metabolites stored and are they released from the alga? (4) Do the metabolites of A. armata inhibit epiphytic bacteria? I also describe a novel test of the ecological roles for algal secondary metabolites by manipulating the production of these compounds in A. armata.

Material and Methods

Study organism

Asparagopsis armata Harvey has a heteromorphic life-history, alternating between a filamentous tetrasporophyte (diploid) and an erect, plumose gametophyte (haploid). The tetrasporophyte is often referred to in the literature as the Falkenbergia-stage of A. armata, having once been described as a separate species. Both tetrasporophyte and gametophyte of A. armata were collected from the shallow subtidal at Bare Island (33° 59' 38' S, 151° 14' 00' E), Sydney, Australia. At this site, the tetrasporophyte is found throughout the year, while the gametophyte is found only in late winter to summer. Chapter 2 Chemical defence against bacteria 12

Screening of algal extracts against known microorganisms

Tetrasporophytes and gametophytes of Asparagopsis armata were collected and transported back to the laboratory. Freeze-dried material was either extracted exhaustively in methanol (MeOH) or in dichloromethane (DCM). These extracts were made up at 100 μg μl–1 in diethyl-ether (DEE). MeOH and DCM extracts of A. armata were tested against 6 strains of bacteria; 2 marine (Vibrio harveyii, V. alginolyticus) and 4 biomedical strains (Pseudomonas aeruginosa, Staphylococcus aureus, S. epidermis and Escherichia coli). V. harveyii and V. alginolyticus were cultured in 37.4 g l–1 Marine Broth (DIFCOTM), while E. coli, S. aureus, S. epidermis and P. aeruginosa were cultured in 13 g l–1 Nutrient Broth (OXOID). Broths were inoculated with plated strains of microorganisms and shake-incubated for 20 h at 30°C. Broth aliquots of 150 μl were spread over 3 replicate agar plates (1.5 % agar-agar in the respective growth media) for each bacterial strain. Assays were performed by stab-inoculating 100 μg of MeOH or DCM extract dissolved in DEE into the agar surface. Whole pieces of tetrasporophyte were also tested against Chromobacterium violaceum (CV026). C. violaceum produces the pigment violacein, and mutants such as this can be used in assays of the regulation of bacterial signalling (Manefield et al. 1999). Growth plates were prepared with a synthetic acylated homoserine lactone (oxo- hexanoyl-homoserine lactone at 0.5 ng ml–1 of media), which regulates the production of violacein. While the signalling component was not assessed in this study, the C. violaceum assay was used as the absence of pigment is a clear means of determining inhibition of growth. Algae (~50 mg fresh weight [FW] per piece) were placed on top of LB-10 (10 g l–1 NaCl, 5 g l–1 yeast extract, 10 g l–1 tryptone) agar plates. LB-10 agar (0.6 % agar) inoculated with C. violaceum was then poured until slightly covering the algal tissue. In all antibacterial assays, the size (cm2) of the activity zone surrounding the inoculation point of each treatment was measured. For the DCM extracts, each replicate bacterial strain (n = 3) had a single inoculate from both tetrasporophyte and Chapter 2 Chemical defence against bacteria 13

gametophyte in a split-plot design. A comparison of the size of the activity zone against the 6 bacterial strains was made using a partially nested ANOVA, with Strain and Life Stage as fixed factors, and Plate nested in Strain. A direct comparison was not made between the MeOH and DCM extracts.

Chemical analysis

Quantitative analysis of the major halogenated metabolites of Asparagopsis armata was performed by gas chromatography-mass spectrometry (GC-MS). Ion fragments characteristic of the major compounds were: dibromochloromethane (CHBr2Cl)

(molecular ion m/e 208 and M+ - 129); bromoform (CHBr3) (molecular ion cluster at m/e 250, 252, 254, 256 [1:2:2:1] and M+ - 173); bromochloroacetic acid (methyl ester) (molecular ion m/e 188 and M+ - 129); dibromoacetic acid (DBA) (methyl ester) (molecular ion cluster m/e 230, 232, 234 [1:2:1] and M+ - 173); and 3,3 dibromoacrylic acid (methyl ester) (molecular ion cluster m/e 242, 244, 246 [1:2:1] and M+ - 213). Initial comparisons of extraction techniques (including drying method and solvent type) demonstrated that methanol extraction of freeze-dried material consistently yielded the highest amounts of the major halogenated metabolites. No qualitative differences in the above compounds were found when extracting from freeze-dried versus wet material. The presence of methyl esters (of the acids) was an artefact of extraction in MeOH, as EtOH extraction yielded ethyl esters of the same acids. This was determined by fragmentation patterns of the mass spectra and increases in retention times. Consequently, all haloacids were quantified using methyl esters as standards. There was no evidence of non-esterified haloacids in the samples. To quantify the major metabolites, algae were freeze-dried, weighed and extracted in MeOH with 10 μg ml–1 naphthalene as an internal standard. The extraction volume was approximately 30 μl MeOH mg(dry weight [DW])–1. MeOH was added to the tissue in the GC-MS inserts, sealed and then sonicated for approximately 15 min. Samples were left at –30°C for at least 72 h to ensure esterification before transfer by glass syringe into a clean GC-MS vial. MeOH extracts were directly quantified using Chapter 2 Chemical defence against bacteria 14

GC-MS. Gas chromatography was performed using a Hewlett Packard (HP) 5890 series II gas chromatograph and a polyethylene-glycol-coated phase on a polyimide-coated fused silica capillary column (Sol-gel wax, 30 m, 0.25 mm ID). All injections (2 Pl) were performed in the splitless mode (1.5 min) with an inlet pressure of 8 psi. The inlet liner (4 mm i.d., 78.5u 6.3 mm o.d., single taper-SGE part no. 092018) was replaced after 50 samples. The injection port was held at 250qC and the GC-MS interface at

300qC. The GC was held at 40qC for 1 min, ramped at 16qC min–1 to 250qC, then held at this temperature for 2 min. Helium was used as the carrier gas. Mass spectrometry was performed on a HP 5972 Mass Selective Detector (MSD). Ions characteristic of the internal standard (naphthalene) and target compounds were monitored in the selected ion monitoring (SIM) mode. Target compound standards were run at the start of each sample set and at regular intervals within sample sets. Target compound concentration in each sample was calculated from the peak areas ratio of target compound over the internal standard. This ratio was then converted to concentration by reference to standard curves. Masses of target compounds (bromoform, dibromoacetic acid, dibromochloromethane and bromochloroacetic acid) are expressed as mass compound per unit algal dry weight. Commercial standards used were bromoform and dibromochloromethane (Sigma-Aldrich), dibromoacetic acid (methyl ester) and bromochloroacetic acid (methyl ester) (Chem Service Inc.). This quantification method was used to determine the levels of the major metabolites in both the tetrasporophyte and gametophyte of Asparagopsis armata. Comparisons were also made between field-sampled and laboratory-cultured tetrasporophytes. Laboratory-culture conditions were at 19°C under 20 μmol photons m–2 s–1 with 36 W daylight fluorescent lighting with a 16:8 h light: dark cycle. For field samples, approximately 6 individual tetrasporophytes and 6 partial samples of the morphologically-larger gametophyte were collected at 4 times when the 2 life-history stages co-exist in the same habitat. Levels of bromoform in field samples were analysed using a 2-factor mixed model ANOVA with Life Stage (N = 2) as a fixed factor, and Sample Time (N = 4) as a random factor. Levels of DBA in field samples were analysed Chapter 2 Chemical defence against bacteria 15

using a 2-factor mixed model ANOVA with Life Stage (N = 2) as a fixed factor, and Sample Time (N = 3) as a random factor. Comparisons between the tetrasporophyte and gametophyte were made for each metabolite using 2-factor mixed model ANOVAs, with Life Stage as a fixed factor and Sample Time as a random factor. Levels of the bromoform and DBA from laboratory-cultured tetrasporophytes and field-sampled tetrasporophytes were analysed individually with 1-factor ANOVA.

Quantification of metabolite release

To determine whether metabolites were released from Asparagopsis armata, the concentration of bromoform and DBA was quantified in culture medium using the method of Cancho et al. (1999). This method of extraction allowed these metabolites to be extracted and esterification of the acids, simultaneously. Individuals of tetrasporophytes of A. armata (N = 9, ~4 mg[DW] ind.–1 ) were cultured in 25 ml of sterile seawater in 70 ml sealed dishes at 19°C under 14 μmol photons m–2 s–1 with a 16:8 h light:dark cycle for 7 d, after which 20 ml of culture media was transferred to

40 ml screw-cap vials. Three ml of concentrated H2SO4, 12 g of anhydrous Na2SO4, 3 g of Cu(II)SO4 and 2 ml of methyl-tertiary-butyl ether (MTBE) (with naphthalene 50 μg ml–1 as internal standard) were added, and the vials were shaken for 2 min and left to stand for 5 min to allow for liquid-liquid partitioning. A 1 ml sample of the MTBE layer was transferred into 2 ml of MeOH/H2SO4 (10%) and placed in a water bath at 50°C for 1 h. This mixture was cooled to 4°C for 10 min, 5 ml of

Cu(II)SO4 (10%)/Na2SO4 (5%) added and subsequently shaken for 2 min. The MTBE layer was removed and 2 μl injected into the gas chromatograph. Whole thalli from each replicate container were frozen, freeze-dried and extracted in MeOH. GC-MS analysis on the extracted medium and algae were run in SIM mode as previously described. Standards for bromoform and DBA were made in MTBE (with naphthalene at 50 μg ml– 1 as internal standard). Chapter 2 Chemical defence against bacteria 16

Screening of metabolites against known microorganisms

Bromoform and DBA were tested against the same 6 strains of bacteria used in the assay of Asparagopsis armata extracts: Vibrio harveyii, V. alginolyticus, Pseudomonas aeruginosa, Staphylococcus aureus, S. epidermis and Escherichia coli. Bacteria were cultured as described previously, and both treatments added to each replicate plate as a split-plot design. Of each compound, 5 μmol were stab-inoculated onto the agar plate. As bromoform is liquid at room temperature, it was applied directly to the agar plate. DBA was made up in DEE at 1 μmol μl–1, and 5 μl applied. DEE controls (5 μl) were applied adjacent to the DBA. The analysis of the size of activity zones for each bacterial strain was made using a partially-nested ANOVA (with each Plate nested within Strain) and Compound (bromoform or DBA) as the fixed factor. Penicillin-G was tested as a positive control against Escherichia coli. Penicillin-G (5 μmol) was applied directly to the bacterial lawn, as above. The size of the inhibition zone was compared to 5 μmol inoculates of bromoform and DBA.

Effect of removing bromine on algal structure and chemistry

I investigated the importance of bromine in the formation of vesicles in Asparagopsis armata gland cells through the manipulation of bromine in the culture medium. This method uses half strength PES-enriched artificial culture medium (Provasoli 1968) with

–1 the addition of 60 μg l NaIO3. This enabled me to make culture media with and without bromide ions, bromide (+) and bromide (-), respectively, which has been successful in culturing Delisea pulchra (Dworjanyn et al. 1999). For bromide (+) medium, NaBr was added to give bromide ions at a concentration of 64 mg g–1. An equimolar amount of NaCl was added to bromide (-) medium to maintain salinity. Apical cuttings of the tetrasporophyte of A. armata were taken from field specimens and Chapter 2 Chemical defence against bacteria 17

cultured in these media for 4 wk at 19°C under 20 μmol photons m–2 s–1 with a 16:8 h light: dark cycle. The effects of the removal of bromine from the culture medium on algae were measured using light microscopy and GC-MS. As the major vesicle of the gland cell is highly refractive, light microscopy was used to examine the development of vesicles in algae cultured in bromide (+) and bromide (-) media. Tetrasporophytes and gametophytes grown in the 2 different media were viewed and photographed using a Leica DM LB microscope fitted with a Leica DC100 digital camera. Small clippings or whole pieces of algae were mounted (unstained) in seawater and viewed with bright field optics (40u and 100u lenses) to examine the position of the gland cells and identify potential structures involved in surface release of metabolites. The effects of bromine on the natural product chemistry of the algae (both halogenated and non-halogenated) were examined using GC-MS (as described above).

Bacterial densities on algae cultured in bromide (+) and bromide (-) media.

Bromide (-) plants do not produce halogenated metabolites (see ‘Results’). This outcome was used to test the antibacterial activity of Asparagopsis armata by counting bacterial numbers on the surface of bromide (+) and bromide (-) algae. If the brominated metabolites are active against epiphytic bacteria, then densities should be higher on the surface of those plants grown in bromide (-) media. A. armata collected for culturing was not cleansed of bacteria, as this alga grows poorly under axenic conditions (N. Paul unpublished data). Consequently, epiphytic bacteria from the field were present in all cultures. Surface bacteria on tetrasporophytes (from clippings) and gametophytes (settled from spores) of A. armata grown in bromide (+) and bromide (-) media for 4 wk were stained with the green fluorescent nucleic acid stain Syto 9 ® (Molecular

Probes) for 3 min. Bacteria were viewed under a 100u lens with a Leica DM LB microscope fitted with a 100 W Hg fluorescence lamp and a GFP (green fluorescent Chapter 2 Chemical defence against bacteria 18

protein)-specific filter. Total bacterial densities were determined for 600 μm2 areas at apical and mature regions of tetrasporophytes and gametophytes, 100 and 700 μm from the apical tip, respectively. The latter was chosen as the ‘mature’ region as this was close to the maximum size of the algal cells. For the gametophyte, lateral branches (determinate branches 700 μm from the tip) were also measured. In both experiments, separate individuals were used for each region, and replicate measurements (N = 3 for gametophytes and N = 6 for tetrasporophytes) were made on different branches within an individual. Bacterial density was analysed by 3- factor mixed model ANOVA with Media (bromide [+] versus bromide [-], including a ‘seawater control’ as an additional treatment for the tetrasporophyte) and Region (apex versus mature, including ‘lateral branch’ as an additional treatment for the gametophyte) as fixed factors, and the random factor Individual (N = 3) nested in each Media by Region combination. In order to control for possible unknown effects on the epiphytic bacterial community caused by the removal of bromine, surface bacterial densities on the filamentous red alga Bostrychia moritziana (Sonder ex Kuetzing) J. Agardh, an alga with no known halogenated metabolites (as determined by GC-MS analysis of the MeOH extract), were measured for individuals grown with and without bromine. Bacterial densities were analysed by 3-factor mixed model ANOVA with Media (bromide [+] and bromide [-]) and Region (apex and mature) as fixed factors, and Individual (N = 3) nested in Media by Region. Six replicate measurements (i.e. branches) were taken for each individual in Media by Region.

Antibacterial assays using bromide (+) and bromide (-) algae

If metabolites from Asparagopsis armata are active against bacteria, then bacteria that are common on metabolite-free algae may be more susceptible to these metabolites than bacteria common on metabolite-producing algae. To test this, bacteria were sampled from bromide (+) and bromide (-) algae and assayed against bromoform and DBA. Chapter 2 Chemical defence against bacteria 19

Epiphytic strains were grown with Marine broth/agar. Bacterial colonies were obtained either by streaking algae over agar or from a dilution series (after algae were vortexed in sterile seawater for 5 min). Six bacterial isolates unique to bromide (+) and 6 from bromide (-) algal cultures were selected based on differences in colony colour and morphology (bromide [+] 1–6, and, bromide [-] 1–6). These bacteria were not characterised beyond their origin (either bromide [+] or bromide [-] algae). Isolates were added to 5 ml of Marine Broth and shaken at 33°C for 24 h. Aliquots (150 μl) of culture were spread evenly over Marine Agar. Both compounds (5 μmol) were stab-inoculated into bacterial growth plates. Three replicate plates of each bacterial isolate were inoculated with both bromoform and DBA in a split-plot design. Bromoform is liquid at room temperature and was stabbed into the surface of the agar plate. DBA was dissolved in DEE and 5 μl of solution and then stabbed into the plate. A DEE control (5 μl) was inoculated adjacent to DBA treatment. Plates were incubated at 30°C for 24 h at which time activity zones surrounding the inoculation stab point were calculated. Activity zones are reported as the surface area (cm2) of growth limitation, as indicated by a front in the resulting bacterial lawn. The size of the activity zones was analysed as a split-plot, partially-nested ANOVA (with each Plate nested within Media, and Compound as fixed factor). Culture-based assays were also used to determine the effects of Asparagopsis armata cultured in bromide (+) and bromide (-) media on a laboratory bacterial strain Chromobacterium violaceum. C. violaceum was grown with LB-10 agar. For C. violaceum assays, bromide (+) and bromide (-) tetrasporophytes were immersed in agar inoculated with C. violaceum and incubated for 24 h at 30°C. Zones of activity were determined by the absence of pigment from the growth medium.

Statistical analysis

The analyses of the experimental designs outlined in each of the above sub-sections were performed using Systat 10 (SPSS). Where appropriate, transformations of the data (detailed in ‘Results’) were made to satisfy the ANOVA assumptions of normality and Chapter 2 Chemical defence against bacteria 20

heterogeneity of variance (Quinn and Keough 2002). Post -hoc comparisons were made using Tukey’s HSD test for multiple comparisons where required.

Results

Antibiotic tests of extracts

Both MeOH and DCM extracts (Fig. 2.1A) were active against all 6 strains of bacteria. The solvent control (DEE) had low levels of activity against all bacterial strains, and the size of this zone of activity was subtracted from the total activity zone of the extract treatments for each replicate plate. For the DCM extracts, there was no difference in efficacy between the tetrasporophyte and gametophyte (Table 2.1, ANOVA: F1,13 = 1.68, P = 0.217). Differences in susceptibility between the different strains were also found

(ANOVA: F5,13 = 3.71, P = 0.026, Fig. 2.1A); Escherichia coli and Pseudomonas aeruginosa were the most susceptible to the DCM extract, while Vibrio spp. were the least susceptible (Fig. 2.1A). Whole pieces of the tetrasporophyte of Asparagopsis armata inhibited growth of Chromobacterium violaceum (Fig. 2.1B).

Natural product chemistry

Mass spectral fragmentation patterns of the major peaks from the methanol extracts of Asparagopsis armata were used in the structural assignment of bromoform, DBA (methyl ester), dibromochloromethane, bromochloroacetic acid (methyl ester) and dibromoacrylic acid (methyl ester). These structures were confirmed using commercial standards, except for dibromoacrylic acid (methyl ester). The mass spectrum of dibromoacrylic acid (methyl ester) matched the spectrum previously described for methyl 3,3 dibromoacrylate (Woolard et al. 1979). Chapter 2 Chemical defence against bacteria 21

Bromoform was the dominant metabolite in Asparagopsis armata, with mean levels across sampling times of 1.45 % DW (±0.12 SE) for the tetrasporophyte (ranging from 0.58 to 3.11 % DW) and 1.67 % DW (±0.16 SE) for the gametophyte (ranging from 0.66 to 4.35 % DW). The amount of bromoform varied depending on both the life stage and sampling time (Table 2.2, ANOVA: F3,50 = 7.15, P < 0.001). Dibromochloromethane was present in the tetrasporophyte (mean of 0.02 % DW [±0.002 SE]) and the gametophyte (mean of 0.03 % DW [±0.003 SE]). The amount of the methyl esters of the haloacetic acids (DBA and bromochloroacetic acid) were converted to weights (molar equivalent) of the corresponding acids. Levels of DBA were higher in the tetrasporophyte than in the gametophyte (Fig. 2.2, and Table 2.2, ANOVA: F1,34 = 67.46, P = 0.015), and overall levels varied between 0.12 to 2.6 % DW (mean 0.74 % DW [±0.16 SE]) for the tetrasporophyte, and 0.02 to 1.1 % DW for the gametophyte (mean 0.25 % DW [±0.06 SE]). Bromochloroacetic acid was found in the tetrasporophyte (mean of 0.14 % DW [±0.02 SE]) and the gametophyte (mean of 0.08 % DW [±0.01 SE]) (Fig. 2.2). Dibromoacrylic acid (methyl ester) was not quantified due to the absence of a standard. Amounts and ratios of the major halogenated metabolites were the same for field- collected and laboratory-cultured tetrasporophytes (Bromoform: F3,22 = 2.79, P = 0.065;

DBA: F3,22 = 0.43, P = 0.716).

Release of major metabolites

The detection of bromoform and the methyl ester of DBA in culture medium using the MTBE extraction demonstrate that these same compounds are also released from the tetrasporophyte. The levels of these compounds in the medium indicated a release rate of 1110 ng g(DW)–1 h–1 (±393 SE) for bromoform, and 539 ng g(DW)–1 h–1 (±166 SE) for DBA. Background levels of bromoform (but not DBA) were detected in seawater controls and the mean value of this subtracted from each bromoform measurement. The average internal levels in the test algae were 1.57 % DW (±0.19 SE) Chapter 2 Chemical defence against bacteria 22

for bromoform, and 0.79 % DW (±0.05 SE) for DBA. The total amounts of each compound released over a 1 wk period are small relative to the amounts retained in the tissue (released: internal, bromoform = 1:77, and DBA = 1:83; assuming no new growth of the alga in this time). The 3 other major halogenated metabolites in Asparagopsis armata (dibromochloromethane, bromochloroacetic acid and dibromoacrylic acid) were also detected in some of the culture media replicates, but were not quantitatively analysed. The data for the release rates of bromoform and DBA were highly variable. There were no significant correlations between the internal concentration of either metabolite and the amount released; bromoform (Pearson’s coefficient = 0.234, P = 0.545) and DBA (Pearson’s coefficient = 0.314, P = 0.412). There was also no correlation between the amounts of bromoform and DBA released (Pearson’s coefficient = 0.370, P = 0.328), despite a strong positive relationship between the internal levels of these compounds (Pearson’s coefficient = 0.955, P < 0.001). While the individual release rates were variable, the ratio of the average level of release of the metabolites (bromoform: DBA = 1.93) is close to that of the ratio of the averages of the internal level in the algae (bromoform: DBA = 1.88).

Antibiotic testing of major metabolites

Both bromoform and dibromoacetic acid (5 μmol of each) showed activity against the 6 bacterial strains assayed. Different strains were more susceptible to one compound than the other, as shown by a significant interaction between strain and compound

(Table 2.1, ANOVA: F1,13 = 6.62, P = 0.003). The size of activity zones surrounding the inoculation point was much larger for DBA than for bromoform against all bacteria

(Fig. 2.3, and Table 2.1, ANOVA: F1,13 = 247.28, P < 0.001), although the potential differences in solubility of each compound in agar limits direct comparison. The mean size of the Escherichia coli inhibition zone surrounding Penicillin-G was 6.17 cm2 (±0.25 SE), which was a larger effect than either bromoform (0.019 cm2) Chapter 2 Chemical defence against bacteria 23

or DBA (1.05 cm2), though direct comparisons are constrained by differences in solubility.

Effect of bromine on cell structure and chemistry

Gland cells of Asparagopsis armata occupy space within other cells of the alga (Fig. 2.4A (arrow heads) and Fig. 2.5A–E). The removal of bromine from the culture medium resulted in the loss of refractile vesicles in both the tetrasporophyte (Fig. 2.4B) and gametophyte (data not shown). The loss of vesicles corresponded to the absence of halogenated metabolites (Fig. 2.4D). Algae cultured with bromine continued to produce the major metabolites at comparable levels to seawater cultures (bromide [+]; e.g. bromoform: 0.44 DW [±0.14 SE]; DBA: 0.08 % [±0.06 SE]), although these values were in the lower range of production compared to both seawater cultures and environmental samples. Removal of bromide from the medium did not interfere with the production of non-halogenated compounds (Compounds 1–5, Fig. 2.4B). Growth of algae in artificial seawater without bromine appears to have minimal effect on cells other than the loss of the vesicle from gland cells.

Gland cells and related structures

The gland cell is located inside the parent cell (Fig. 2.5). Thus, an additional mechanism is required to transfer material from the gland cell to the surface for metabolites to be released. Such a link is maintained via a structure which connects the gland cell to the opposite (outer) wall of the parent (pericentral) cell in tetrasporophytes (Fig. 2.5A) and the parent (epidermal) cell of the gametophyte (Fig. 2.5B). These structures are extremely distinct in the lateral branches of the gametophyte, and material may be associated with them (Fig. 2.5B, arrow). In the sequence of images rising through focal planes from 2 gland cells (arrow heads) in the gametophyte (Fig. 2.5C–E), the structure appears to be tube-like in nature. Chapter 2 Chemical defence against bacteria 24

Epiphytic bacterial densities on bromide (+) and bromide (-) algae

Densities of bacteria were significantly higher on both tetrasporophytes and gametophytes of Asparagopsis armata which lacked brominated secondary metabolites (bromide [-]) than on algae that produced metabolites (bromide [+]) (Table 2.3, tetrasporophyte: F2,12 = 39.14, P < 0.001; gametophyte: F1,12 = 33.19, P < 0.001, Fig. 2.6A&B). For the tetrasporophyte, bromide (-) algae had significantly higher densities than both other treatments (Tukey’s test: (P = 0.05) bromide (-) > bromide (+) = seawater, Fig. 2.6A). This represents a 14-fold (apex) to 20-fold (mature) difference in mean bacterial densities (Br [-]:Br [+]). For the gametophyte of A. armata, the mean difference ranged from 3:2 to 3:1 between the 3 regions (apical cells, mature cells and lateral branch cells). There was no difference in bacterial density on the control alga Bostrychia moritziana in bromide (+) and bromide (-) media (Table 2.4, F1,8 = 0.51, P = 0.496, Fig. 2.6C). For both the tetrasporophyte of Asparagopsis armata and control alga

Bostrychia moritziana, bacterial densities were higher (F1,12 = 11.59, P = 0.005; F1,8 = 30.28, P < 0.001, respectively) on older (mature) parts of the algal filament than on younger (apex) parts (Fig. 2.6, Tables 4 and 5). There was no difference in bacterial densities between the regions of the gametophyte of A. armata (Table 2.3). For both tetrasporophytes and gametophytes of A. armata, there was no effect of ‘Individual’ (Table 2.3), indicating variance within each Media by Region group was minimal.

Antibacterial assays with bromide (+) and bromide (-) algae

Bacterial isolates from the surface of algae that no longer produced halogenated metabolites (bromide [-]) were more susceptible to bromoform and DBA than bacteria isolated from metabolite-producing algae (Fig. 2.7A&B, and Table 2.5, F1,34 = 12.48, Chapter 2 Chemical defence against bacteria 25

P = 0.001). DBA was active against all 12 strains assayed. Bromoform was active against 7 of the 12 strains. Of the 5 strains that were not affected by bromoform, 4 were isolates from bromide (+)-cultured algae. There was a highly significant effect of compound (F1,34 = 374.67, P < 0.001); however, direct comparisons are constrained by the potential difference in solubility in the agar. DEE controls had a negligible effect on bacterial growth. There appeared to be some variation in the susceptibility of individual isolates to the different compounds (Fig. 2.7A&B); however, this was not formally analysed. Direct tests of algae with and without halogenated secondary metabolites (bromide [+] and bromide [-], respectively) were consistent with an antibacterial function for the halogenated metabolites of Asparagopsis armata. When bromide (+) and bromide (-) algae were infused in cultures of Chromobacterium violaceum, bromide (+) algae produced highly visible areas of antibacterial activity (Fig. 2.7C&D). Chapter 2 Chemical defence against bacteria 26

A. a a 0.20

0.15 ab ) 2

0.10 bc bc Activity zone (cm Activity 0.05 c

0.00 i a s is s yii s m . col cu ve ino ureu yti r E a l a . ider S no ep lgi V. h . aerug S. .a P V

Bacteria B.

Figure 2.1. (A) Areas of antibacterial activity (mean ± SE, N = 3) of the dichloromethane extracts of tetrasporophytes and gametophytes (combined) against known bacteria. Bacteria sharing the same letter do not differ at P = 0.05 (Tukey’s HSD). (B) Chromobacterium violaceum (arrow head) is growth-inhibited (arrow) around the filamentous tetrasporophyte of Asparagopsis armata. Scale bar = 5 mm. Chapter 2 Chemical defence against bacteria 27

2.0

Tetrasporophyte: Culture

Tetrasporophyte: Field 1.5 Gametophyte: Field

1.0

0.5 Mean internal level (% dry weight)

0.0

CHBr3 DBA BCA CHBr2Cl

Metabolite

Figure 2.2. Internal level of the major halogenated metabolites (mean ± SE, N = 10 to 20 for field, N = 6 for culture) in field-sampled tetrasporophytes and gametophytes, and laboratory-cultured tetrasporophytes of A. armata (CHBr3: bromoform; DBA: dibromoacetic acid; BCA: bromochloroacetic acid; CHBr2Cl: dibromochloromethane). Chapter 2 Chemical defence against bacteria 28

2.0 Bromoform DBA

1.5 ) 2

1.0 Activity zone (cm

0.5

0.0 ii s y coli sa . o reus rmis ve E in e lyticu g .au o ar ru S pid in e .e g V.h S l P.a V.a Bacteria

Figure 2.3. Areas of antibacterial activity (mean ± SE, N = 3) of pure compounds (bromoform and dibromoacetic acid) from A. armata against different bacterial strains. There was a significant interaction between bacteria and compound (see Table 2.1 for ANOVA results). Chapter 2 Chemical defence against bacteria 29

A. B.

pc pc pc pc

Bromide (+) Bromide (-) C. B D. Ion Abundance Ion Ion Abundance Ion

IS 1 IS 1 D E2 345 2 345 A C

5101520 5101520 Retention time (min.) Retention time (min.)

Figure 2.4. Effects of bromine manipulation on the morphology (A, B) and chemistry (C, D) of the tetrasporophyte of A. armata. (A) Vesicles (arrow heads) are prominent in the gland cells of bromide (+) algae, which are located inside the pericentral cell (pc) as shown in this section of filament. (B) These vesicles are no longer formed in algae grown without bromide (scale bars = 20 μm). (C) Gas chromatograph of the MeOH extract of the tetrasporophyte of A. armata showing major halogenated peaks (A: dibromochloromethane; B: bromoform; C: bromochloroacetic acid [methyl ester]; D: dibromoacetic acid [methyl ester]; E: 3,3 dibromoacrylic acid [methyl ester]; IS: internal standard [naphthalene]). (D) Gas chromatograph showing that brominated metabolites (A–E) are no longer present in bromide (-) algae. Peaks 1–5 represent non-halogenated compounds (IS: Internal Standard) and these are found in both bromide (+) and bromide (-) algae (C, D). Chapter 2 Chemical defence against bacteria 30

A C

pc pc

D

pc pc

B

E

pc

Figure 2.5. Light micrographs of the tetrasporophyte (A) and gametophyte (B–E) of A. armata. (A) Gland cells (arrow) within pericentral (parent) cells (pc) of the tetrasporophyte and connective structure (arrow head). (B) Lateral branch of the gametophyte with a gland cell within the parent cell (pc) showing a structure connecting the gland cell to the outer cell wall. Material may be seen on this structure (B, arrow). (C–E) Images of these structures (arrows) originating at 2 gland cells (C, arrow heads) in the parent (epidermal) cells (pc) of the gametophyte as seen moving up through 2 focal planes towards the algal surface. The structure appears tubular in nature (D, arrows). Scale bars: (A, B) = 10 μm; (C–E) = 20 μm. Chapter 2 Chemical defence against bacteria 31

A. 7

) b

2 2 6 Seawater Bromide (+) 5 Bromide (-) ( # / 100 μm ( # / 4

3

2 b1

1 a2 a2 a1 a1 Bacterial Density

0 APEX MATURE REGION

B. 20

18 ) 2

m 16 ** P 14

12 *

10

8

6

4

Bacterial Density ( # / 100 Density Bacterial 2

0 APEX MATURE LATERAL

REGION

Figure 2.6. Mean surface density of bacteria (± SE) on the tetrasporophyte (A) and gametophyte (B) of A. armata and the control alga Bostrychia moritziana (C) grown in bromide (+) and bromide (-) media (tetrasporophyte of A. armata has Seawater as a third media treatment). (A) Bars sharing the same letter in each region are not significantly different (Tukey’s multiple comparison). (B, C) Significant (*) or non-significant (NS) differences between bromide (+) and bromide (-) bacterial densities are displayed for each region. Standard errors (SE) were calculated from the averages of individuals within each subgroup (Tetrasporophyte A. armata: N = 6; Gametophyte of A. armata, N = 3). Chapter 2 Chemical defence against bacteria 32

C.

20

18 ) 2 Bromide (+) 16 Bromide (-) 14 NS ( # / 100 μm ( # / 12

10

8

6 Bacterial Density 4 NS 2

0 APEX MATURE Region

Figure 2.6. continued. Mean surface density of bacteria (± SE) on the control alga Bostrychia moritziana (N = 3). Chapter 2 Chemical defence against bacteria 33

A. 0.30

0.25 ) 2 0.20

0.15

0.10 Activity zone ( cm

0.05

0.00 Br(-) 1 Br(-) 2 Br(-) 3 Br(-) 4 Br(-) 5 Br(-) 6 Br(+) 1 Br(+) 2 Br(+) 3 Br(+) 4 Br(+) 5 Br(+) 6

Bacterial Isolate B. 3.0

2.5 ) 2 2.0

1.5

Activity ZoneActivity (cm 1.0

0.5

0.0 Br(-) 1 Br(-) 2 Br(-) 3 Br(-) 4 Br(-) 5 Br(-) 6 Br(-) Br(+) 1 Br(+) 2 Br(+) 3 Br(+) 4 Br(+) 5 Br(+) 6 Br(+) Bacterial Isolate C. D.

Figure 2.7. (A, B) Areas of antibacterial activity (mean ± 1 SE, N = 3) of bromoform (A) and dibromoacetic acid (B) against bacterial isolates from the surface of algae with (Br[+]) and without (Br[-]) halogenated metabolites. There was a significant effect of Media (P = 0.001, Table 2.5), indicating that bacteria from Br [-] algae were less tolerant of the brominated metabolites. (C, D) Chromobacterium violaceum assays with bacterial inhibition zone (arrow) around bromide (+) algae (C), and the corresponding assay with bromide (-) algae (D). Scale bars = 5 mm. Chapter 2 Chemical defence against bacteria 34

Table 2.1. Results of 3-factor partially-nested ANOVAs testing the antibacterial effect of the (DCM) extracts (Life Stage: tetrasporophyte vs. gametophyte) and pure compounds (Compound: bromoform vs dibromoacetic acid) on 6 known bacterial strains. Significant values at P < 0.05 are in bold.

Source df MS F P

DCM extracts Strain 5 0.019 3.71 0.026 Plate (Strain) 13 0.005 3.74 0.012 Life Stage 1 0.002 1.68 0.217 Strain u Life Stage 5 0.001 0.79 0.576 Error 13 0.001

Pure compounds Strain 5 0.201 7.58 0.002 Plate (Strain) 13 0.026 0.80 0.656 Compound 1 8.220 247.28 <0.001 Strain u Compound 5 0.220 6.62 0.003 Error 13 0.059

Table 2.2. Results of 2-factor mixed-model ANOVAs testing the effect of Life Stage (tetrasporophyte vs. gametophyte) and sample time on the internal levels of bromoform and dibromoacetic acid (DBA). All data were log (x +1)-transformed. Significant values at P < 0.05 are in bold.

Source df MS F P

Bromoform Life Stage 1 0.002 0.01 0.926 Sample Time 3 0.578 17.04 <0.001 Life Stage u Sample Time 3 0.243 7.15 <0.001 Error 50 0.034

DBA Life Stage 1 0.836 67.46 0.015 Sample Time 2 0.059 0.73 0.489 Life Stage u Sample Time 2 0.012 0.15 0.858 Error 34 0.081 Chapter 2 Chemical defence against bacteria 35

Table 2.3. Results of 3-factor mixed-model ANOVA testing the effect of Media (Br[+] vs. Br[-]) and Region (Apex vs. Mature) on bacterial densities on the surface of the tetrasporophyte and gametophyte of A. armata. All data were log (x + 1)-transformed. Significant values at P < 0.05 are in bold. Tukey’s multiple comparisons were used for post-hoc analyses. Tetrasporophyte (Media), Sea = Br(+) < Br(-).

Source df MS F P

Tetrasporophyte Media 2 10.09 39.14 <0.001 Region 1 2.98 11.59 0.005 Media u Region 2 0.87 3.38 0.069 Individual (Media u Region) 12 0.26 0.26 0.259 Error 90 0.21

Gametophyte Media 1 15.90 33.19 <0.001 Region 2 1.05 2.18 0.155 Media u Region 2 0.30 0.62 0.552 Individual (Media u Region) 12 0.48 1.54 0.153 Error 36 0.31

Table 2.4. Results of 3-factor mixed model ANOVA testing the effect of Media (Br[+] vs Br[-]) and Region (Apex vs. Mature) on the bacterial densities on the surface of B. moritziana. Data were log (x + 1)-transformed. Significant values at P < 0.05 are in bold.

Source df MS F P

Media 1 0.98 0.51 0.496 Region 1 46.37 25.95 0.001 Media u Region 1 0.01 0.006 0.941 Individual (Media u Region) 8 1.79 4.05 0.001 Error 61 0.44 Chapter 2 Chemical defence against bacteria 36

Table 2.5. Results for 3-factor partially-nested ANOVA testing the antibacterial effect of Compound (bromoform vs dibromoacetic acid) against bacteria isolated from A. armata grown in different Media (Br[+] vs Br[-]). Data were log (x + 1)-transformed. Significant values at P < 0.05 are in bold.

Source df MS F P

Media 1 0.256 12.48 0.001 Dish (Media) 34 0.021 0.64 0.904 Compound 1 12.086 374.67 <0.001 Media u Compound 1 0.044 1.35 0.253 Error 34 0.032 Chapter 2 Chemical defence against bacteria 37

Discussion

Asparagopsis armata produces compounds that inhibit a range of bacteria in standard antibacterial plate assays. These compounds are stored in and released from specialised gland cells. The gland cell is internal to the parent cell, but maintains a physical connection with the outer cell wall. This connection appears to be a means for the release of metabolites to the algal surface. Support for this assertion is the observation that the major halogenated metabolites quantified in A. armata were also detected in the surrounding culture medium. An ecological role for the natural products of A. armata as surface-active antimicrobial agents was demonstrated by manipulating metabolite production in the alga by omitting bromide from the medium. Epiphytic bacterial densities were significantly lower on algae that produced halogenated secondary metabolites, and bacteria isolated from the surface of these algae were more tolerant to the major compounds than bacteria from metabolite-free algae. Combined with traditional assays of natural products against bacteria, including those isolated from the algal surface, these results demonstrate that the metabolites from A. armata could be important in limiting growth of bacteria on the surface of this alga. Traditionally, marine secondary metabolites have primarily been seen as defensive compounds against consumers (Hay et al. 1987b, Hay and Fenical 1988). However, recent reviews have highlighted important roles for secondary metabolites as inhibitors of surface colonisation (Potin et al. 2002, Steinberg and de Nys 2002). More specifically, algal natural products can mediate interactions with surface microbes (Maximilien et al. 1998), which may include specific pathogens (Weinberger and Friedlander 2000, Bouarab et al. 2001, Kubanek et al. 2003). Prior to these recent tests, antimicrobial activity of marine algae had generally been determined using terrestrial microbes, principally those bacteria of interest to humans (Sieburth and Conover 1965, Hornsey and Hide 1974, Al-Ogily and Knight-Jones 1977, McConnell and Fenical 1979). I have demonstrated that extracts of Asparagopsis armata are active in laboratory Chapter 2 Chemical defence against bacteria 38

tests against Escherichia coli, Staphylococcus spp., Pseudomonas aeruginosa and 2 marine Vibrio spp. Previous investigations of A. armata using whole pieces (Hornsey and Hide 1974) and isolated compounds (McConnell and Fenical 1979) also found that the natural products from this alga can inhibit bacterial growth. In the present study, the major natural products in Asparagopsis armata were small halocarbons (e.g. bromoform) and short-chained haloacids (e.g. DBA). No iodinated acids were detected, despite being the major acid products of Asparagopsis taxiformis (Woolard et al. 1979). Haloacetones and halobutanones, previously described for A. armata (McConnell and Fenical 1977), were not found. The total levels of the 4 major halogenated metabolites ranged between 2 and 6 % of algal DW. To date, the major metabolites had not been quantified in the genus Asparagopsis, and this range represents a high level of compounds compared to another member of the Bonnemaisoniaceae (Delisea pulchra has 0.2 to 2 % DW; de Nys et al. 1996). Levels of the major metabolite (bromoform) varied with sample time and life-history stage of A. armata (filamentous tetrasporophyte or plumose gametophyte). The levels of DBA, the second most abundant metabolite, were consistently higher in the tetrasporophyte than in the gametophyte, with little temporal variation. Importantly, the levels of the major metabolites in the tetrasporophyte of A. armata in seawater culture were consistent with field-sampled tetrasporophytes. This supports the use of laboratory manipulations of A. armata to demonstrate the release of metabolites from the alga into the surrounding media and to test their ecological roles. A key aspect to studies of the function of algal metabolites is the localisation of active metabolites in the alga (de Nys et al. 1998, Dworjanyn et al. 1999). Such active metabolites can be autotoxic, and the producing organism will generally require some morphological specialisation for their storage (McKey 1979). The specialised gland cells of Asparagopsis armata are completely enveloped by their parent cells. While the gland cell is not part of the epidermal layer, cells presumably release their contents through mechanical contact with the outer region of the parent cell (Fig. 2.5). The structure linking the gland cell and its parent cell has not previously been characterised. Chapter 2 Chemical defence against bacteria 39

Transmission electron microscopy of the gland cells and associated structures in A. armata has provided further insight into the ultrastructure of the major storage vesicle and the nature of the gland cell (Chapter 3). In other algae, specialised cells are predominantly found in the epidermal cell layer (Young and West 1979, Dworjanyn et al. 1999). If an alga is to release bioactive metabolites to the exterior, these specialised cells should be proximal to the algal surface. This is the case for A. armata, as a single cell wall separates the gland cell from the exterior. A gland cell, as the name implies, is generally secretory, although this is not always the case (Young 1979). Other types of storage structures are exclusively subcellular, e.g. the refractile inclusions (corp en cerise) of the genus Laurencia (Young et al. 1980). These internally packaged compounds cannot function in surface-mediated roles (de Nys et al. 1998), but can play a role in feeding deterrence (Hay et al. 1987b). It appears that some types of inclusions can function in multiple roles. For example, brown algal physodes, structures containing phenolics typically regarded as key defensive metabolites, also have important roles in cell wall formation (Schoenwaelder 2002). These are all examples of how structure can determine function. Clearly, sufficient ultrastructural evidence (Young 1978) or the harvest, and if possible, quantification, of metabolites from the plant surface (Schmitt et al. 1995, de Nys et al. 1998), is crucial to attributing surface-mediated roles to extracted natural products. The release of bromoform and DBA from Asparagopsis armata represents a mechanism for the interaction of these compounds with surface microbes. Many algae are known to release small halogenated compounds (Nightingale et al. 1995, Marshall et al. 1999), although it is not usually known how the compounds are released or if they are active against microbes. Similar to Marshall et al. (1999), I demonstrated the release of bromoform from the tetrasporophyte of A. armata. The measured rate of release for bromoform (1110 ng g[DW]–1 h–1 ) is comparable to previous reports using purge-and- trap techniques (32 to 1650 ng g[DW]–1 h–1; Marshall et al. 1999 ). This is a higher rate of release than most other algae (Nightingale et al. 1995). Analysis of the culture medium through liquid-liquid partition, rather than through purging, esterified the Chapter 2 Chemical defence against bacteria 40

haloacids, and demonstrated that the other major product (DBA) is also released (539 ng g[DW]–1 h–1). Release of haloacids from an alga has not previously been reported. The release rates of both bromoform and DBA were not correlated with the metabolite levels inside the algae. There was also no correlation between the rates of release of the 2 compounds, despite a strong positive correlation in the internal levels of these compounds. While there was no significant relationship between the external and internal amounts of the metabolites, the data were highly variable. Brominated metabolites are common in red algae (Fenical 1975). However, bromide is not recognised as an essential ion for either algal growth (Fries 1966) or reproduction (McLachlan 1977). Wolk (1968) demonstrated that Asparagopsis armata grown in the absence of bromide ions no longer formed the major vesicle in gland cells, and that morphological effects aside from this were negligible. In this study, GC-MS analysis also indicated qualitative similarities of non-halogenated metabolites in bromide (+)- and bromide (-)-cultured algae. The ability to manipulate the presence of halogenated metabolites in Bonnemaisoniaceae provides a unique experimental system for testing the ecological roles of algal metabolites. Correlations between bacterial densities on living surfaces and levels of metabolites (Sieburth and Conover 1965, Wahl et al. 1994, Maximilien et al. 1998) are potentially constrained by differences in structure, age and chemistry of the study organisms. By testing Asparagopsis armata grown with and without halogenated metabolites, I demonstrated that these compounds were active in maintaining lower densities of bacteria on the surface of the alga. There was an antibacterial effect on both young (apical) and older (mature) regions, although it is possible that apical effects contributed to the lower bacterial densities on the older region. The filamentous algae, that is, the tetrasporophyte of A. armata and Bostrychia moritziana, had lower densities of bacteria on apical regions than on older regions. However, there was no difference between the bacterial densities on younger and older tissue in the gametophytes of A. armata. This could be related to the slower growth rate of the gametophyte compared to a filamentous alga. Chapter 2 Chemical defence against bacteria 41

Bacteria isolated from the surface of bromide (-) algae were less tolerant to halogenated metabolites than those isolated from bromide (+) algae. This suggests that the microbial community on metabolite-producing algae includes bacteria that have colonised the surface but are inhibited by the compounds. In the absence of these compounds, the previously inhibited bacteria may out-compete other resident bacteria. If important symbioses between bacteria and macro-algae exist (Fries 1970), then certain bacteria will be more tolerant to algal natural products than others. As Asparagopsis armata releases relatively large amounts of halogenated metabolites, the dominant epiphytic bacteria should have the capacity to tolerate, or even breakdown, such compounds. The implied change in bacterial community structure suggests that the interaction between A. armata and its epiphytic bacteria is regulated by halogenated metabolites that limit particular bacteria in addition to reducing total surface density. There are other mechanisms that may constrain the interpretation of the antibacterial effect seen in Asparagopsis armata. A high growth rate may minimise fouling on young tissue, although the halogenated metabolites were still effective at these actively-growing regions. Alternatively, chemical defence against microbes is not limited to the production of secondary metabolites. Oxidative bursts can provide an effective defence against pathogens that breach the cell walls of some algae (Weinberger and Friedlander 2000, Bouarab et al. 2001). While the 2 major halogenated metabolites of A. armata (bromoform and DBA) were antibacterial, it is possible that other compounds contributed to the observed antibacterial effect of bromide (+) algae. However, the relatively high levels of production and release of bromoform and DBA in particular, strongly imply that these are the principal antibacterial agents of A. armata.

Conclusions

Considering the constant interaction of algae with potential bacterial epibionts, regulating the attachment or growth of bacteria (bacterial fouling) through the production of secondary metabolites may give an alga an ecological advantage (Wahl Chapter 2 Chemical defence against bacteria 42

1989, Steinberg and de Nys 2002). For Asparagopsis armata, the determined release of halogenated metabolites, their antibacterial activity against relevant bacteria, and the increased densities of bacteria on the surface of metabolite-free algae, indicated a probable role for these metabolites in defence against epiphytic bacteria. The different effect of the halogenated metabolites on bacteria isolated from metabolite-producing algae compared to bacteria from metabolite-free algae implied an impact beyond a general antibiotic effect. While a large proportion of epiphytic bacteria can be isolated from some seaweeds (Jensen et al. 1996), the non-culturable bacteria on A. armata represent a crucial issue for the culture-based component, and the ecological importance of the isolated bacteria remains unknown. Molecular tools that examine community- scale microbial interactions (such as denaturing gradient gel electrophoresis; Dahllöf 2002), will be integral to measuring metabolite-induced changes to the bacterial community, and promise to identify epiphytic bacteria that are ecologically-important and practical for use in manipulative assays. CHAPTER 3

Ultrastructure of the gland cells of the red alga Asparagopsis armata (Bonnemaisoniaceae)*

Introduction

Cellular inclusions of algae have diverse functions. They are important in metabolic reserves (Pueschel 1992), cell wall construction (Schoenwaelder and Clayton 1998, 1999) and chemical defence (Young et al. 1980, Dworjanyn et al. 1999). Many types of inclusions are located free in the cytoplasm (Pueschel 1992, Pedersén et al. 1980, Pueschel and Korb 2001). However, inclusions can also be distinct bodies associated with particular cellular features, such as the cell wall, cuticle or chloroplasts (Young 1979a, Pedersén et al. 1980, Pallaghy et al. 1983, Schoenwaelder and Clayton 1998). In many red algae, inclusions containing refractile material are found in specialised cells (Wolk 1968, Young 1979b, Young and West 1979, Dworjanyn et al. 1999). Such specialised cells – typically known as vesicle or gland cells – are common in the Ceramiales, Bonnemaisoniales and Gigartinales (Womersley 1996, 1998). Functions attributed to these cells include light-collecting bodies, and nutritional, excretory and defence roles (summarised in detail by Young and West 1979). Some red algae produce chemicals that have potent biological effects (Fenical 1975, Fenical

* This chapter is published in the Journal of Phycology (2006) 42:637-645 See Appendix One Chapter 3 Gland cells 44

1982), and thus these specialised storage cells are of some interest, particularly with respect to providing safe storage for harmful substances. However, apart from work of Young and co-workers (1978, 1979b, and, Young and West 1979), fine ultrastructural details of these cells are scarce, as is evidence for their biological significance (Murray and Dixon 1992). The sensitivity of active compounds to traditional, chemical fixation can impede the ultrastructural examination of algae (Clayton and Beakes 1983). As an alternative to chemical fixation, a combination of cyrofixation and freeze substitution has proven useful in preserving sensitive features, such as cytoskeleton or fungal vacuoles, for TEM (Orlovich and Ashford 1993, Chretiennot-Dinet et al. 1997, Babuka and Pueschel 1998). To date, cryotechniques have not been used to examine the ultrastructure of specialised cells or localisation of secondary metabolites in red algae. Although refractile inclusions are a common feature in cells of red algae, their simple presence gives little indication of the types or roles of stored compounds. Differences in histochemistry of stored material and also the TEM identification of organelles for secretion, for example abundant smooth endoplasmic reticulum, provide some evidence for the function of vesicle cells (Young 1977, Young 1978, Young and West 1979). Further means to localise natural products in the red algae rely on the inherent properties of the stored compounds and include the X-ray analysis of organically-bound halogens (Pedersén et al. 1980, Young et al. 1980) or the autofluorescence of known UV-active metabolites using epifluorescence microscopy (Dworjanyn et al. 1999). Relatively few studies, however, have adequately related the localisation of known natural products to specific ecological functions in red algae (Young et al. 1980, de Nys et al. 1998, Paul et al. 2006). For surface-mediated interactions, such as the regulation of micro- and macro- fouling, the presence of algal metabolites at the surface is essential (Steinberg and de Nys 2002). But release of metabolites is not a standard function of specialised cells in red algae (Young 1977, Young 1979b, Young and West 1979) and has rarely been determined (Young 1978, de Nys et al. 1998). Accordingly, Chapter 3 Gland cells 45

the structure and function of the specialised cells in most red algae remain as yet unknown (Murray and Dixon 1992). Members of the Bonnemaisoniaceae form specialised cells (Wolk 1968, Young 1977, Womersley 1996) and also produce a diverse array of halogenated metabolites (Fenical 1975, McConnell and Fenical 1977, de Nys et al. 1998). The internal localisation of these halogenated metabolites is currently limited to the identification of halogens in the specialised cells of Bonnemaisonia nootkana and A. armata by electron microprobe analysis (Wolk 1968), and the localisation of furanones in the gland cells of Delisea pulchra by epifluorescence microscopy (Dworjanyn et al. 1999). Although A. armata was presumed to release its cellular contents (McConnell and Fenical 1977), perhaps through cytolysis (as suggested for B. nootkana by Young 1977), there was no empirical evidence for extra-cellular release of metabolites until recently (Marshall et al. 2003, Paul et al. 2006). In the present study, I investigate for the first time the localisation and potential mechanism of extra-cellular release of the natural products of the tetrasporophyte of Asparagopsis armata using TEM of cryo-fixed, freeze-substituted material, together with light and epifluorescence microscopy.

Materials and Methods

Study organism

Tetrasporophytes of Asparagopsis armata were collected from the shallow subtidal at Bare Island (33° 59’ 38” S, 151° 14’ 00” E), Sydney, Australia. The filamentous tetrasporophyte is found throughout the year as an epiphyte (predominantly on the coralline turf). Algae were maintained in culture at 19ºC in sterile seawater with 16 h: 8 h light: dark cycle under 20 μmol photons m-2 s-1 with 36W daylight globes prior to ultrastructural examination. Freshly collected pieces were used for light and epifluorescence microscopy. Chapter 3 Gland cells 46

To examine the presence of gland cells in young tetrasporophytes, female gametophytes with cystocarps were collected from the field. Carpospores were released after excising the cystocarp and placing it in sterile seawater overnight. Germinating carpospores were monitored over subsequent days.

Light Microscopy

Asparagopsis armata was viewed and photographed using a a Leica DM LB microscope fitted with a Leica DC100 digital camera. Samples were mounted (unstained) in seawater and viewed with bright field optics to examine the location of gland cells, as well as the size (volume) of gland cells. Gland cell size can readily be determined by the size of its refractile inclusion. The formation of gland cells was examined in apical regions of collected individuals, and in carpospores both pre- and post-germination. To determine when gland cells are produced relative to apical cell division, 15 individuals were collected and the position of the first observed gland cells from the apical cell was measured in one filament of each. The frequency of pericentral cells that contained a gland cell was also measured for 20 pericentral cells in each of 15 individuals. Furthermore, the volume that the gland cells occupied in the alga was measured by calculating the volume of the refractile vesicle in the gland cell relative to the algal volume of each tier of cells (comprised of 3 pericentral cells and one axial cell). The average volume for 6 randomly selected cell tiers were used to calculate the overall average of percent volume of gland cells in 7 individuals.

Manipulation of gland cells

The development of the vesicle in the gland cells of algae cultured in two types of artificial culture medium (Provasoli 1968), with bromine (bromide ions at 65 mg L-1: Burton 1996) or without bromine, was observed using light and fluorescence Chapter 3 Gland cells 47

microscopy. Algae were cultured at 19ºC with 16: 8 h light: dark cycle under 20 μmol photons m-2 s-1 with 36W daylight globes in the two different artificial media. For light microscopy, the presence or absence of the refractile vesicle in the gland cell was observed in algae cultured with or without bromine. In order to determine whether any absence of the vesicle was correlated with an absence of the gland cell, fluorescence microscopy of DAPI - stained nuclei was used on A. armata cultured with and without bromine (Goff and Coleman 1990). Here, algae were microwaved for 15 s (x2) in seawater containing 0.5 μg/ ml DAPI, with 1 min between treatments. Samples were left for ½ h in darkness before viewing with a fluorescence microscope using a UV filter set.

Epifluorescence microscopy

Small pieces of the tetrasporophyte of A. armata were mounted on a glass slide and imaged using a Leica DM LB microscope fitted with a 100W Hg fluorescence lamp. A UV filter set was used to examine the gland cells, as certain metabolites will autofluoresce under UV excitation (O’Brien and McCully 1981).

TEM of cryofixed and freeze-substituted material

Small pieces of tissue from the tetrasporophyte of A. armata were plunge-frozen in liquid propane cooled to -190ºC with liquid nitrogen. Samples were freeze substituted in 2 % osmium tetroxide in dry acetone at -85ºC for 72 h, -30ºC for 24 h and room temperature for 1 h (warmed up at 5° / h between temperatures). Samples were washed in acetone (4 times, 15 mins) to remove osmium tetroxide, and infiltrated with Spurr’s resin (Spurr, 1969) in 1 h steps of 1:3, 1:1, 3:1 (Spurr’s:acetone) and one exchange overnight in Spurr’s resin only. Infiltrated tissue was flat-dish embedded in resin between two PolyTetraFluoroEthylene (PTFE ™) -coated slides separated by Chapter 3 Gland cells 48

coverslips, and polymerized at 60ºC for 14 h. Samples were excised from resin and ultrathin sections were cut (90 nm) in the transverse and longitudinal planes with a diamond knife on a Reichert Ultracut ultramicrotome (Moc Inc, Valley Cottage, NJ). Sections stained with toluidine blue were previewed by light microscopy. Selected sections were post-stained with uranyl acetate (10 min) and lead citrate (Reynolds 1963) and examined on a Hitachi H7000 transmission electron microscope (100kV).

Results

Light microscopy

The tetrasporophyte of Asparagopsis armata is a multiseriate filamentous alga with indeterminate branching, comprised of a central axial cell and three pericentral cells (Figs. 3.1A&B). Pericentral cells produce a gland cell from their inner cell wall (Fig. 3.1B). These gland cells are formed on average 3.6 (±1.5 SD) pericentral cells from the apical cell, but are frequently observed in the cell layer adjacent to the apical cell (Fig. 3.1C). The frequency of gland cells per pericentral cell beyond the apical region in field collected material is 99 % (±2 SD). Vesicles were not observed in non-germinated carpospores. However, at germination, the carpospore material divides, forming apical (arrow) and rhizoidal (not shown) filaments (Fig. 3.1D). The cells within the original spore material have prominent vesicles post-germination (Fig. 3.1D: arrow head). These vesicles can be larger than those in cells in the adjacent pericentral cells of the new branches (Fig. 3.1D: arrows). In detail, gland cells characteristically contain a large refractile vesicle, the surface of which appears dimpled, similar to that of a golf ball. The internal space that the gland cells occupy was on average 3.77 % ( ± 1.02 SD) of the algal volume at each Chapter 3 Gland cells 49

cell tier (comprised of one axial cell plus the three surrounding pericentral cells). The gland cell vesicle enlarges as the alga grows; the largest gland cell measured was 1075 μm3. The contents of the vesicles appeared to be highly toxic to the pericentral cell; if the gland cell wall is breached the surrounding tissue quickly becomes severely disrupted. As gland cells are formed at the internal wall of the pericentral cell (Fig. 3.1A), in order for their metabolites to be released to the surface some additional transport mechanism is required. A stalk-like structure connects the gland cell to the opposite (outer) wall of the pericentral cell (Fig. 3.2A&B). This stalk can either be a fine structure (Fig. 3.2A) or a more rigid feature (Fig. 3.2B). The latter often occurs when the relative size of the gland cell to the pericentral cell is large. It appears that the stalk is connected near the outer wall of the pericentral cell (Fig. 3.2A). In some large gland cells, the vesicle seems to be tightly appressed to the outer cell wall of the pericentral cell, excluding the contents of the cytoplasm from this region of the pericentral cell (Fig. 3.2C).

Bromine manipulation

When A. armata was cultured in artificial media with bromine, the major vesicle of the gland cell was the dominant structural feature (Fig. 3.3A). Algae grown in bromide-free media no longer produced the refractile vesicle (Fig. 3.3B). But the absence of bromine did not limit the formation of the gland cells. DAPI-stained nuclei of the gland cells were observed in apical regions of algae grown with and without bromine (data not shown). The nuclei of these gland cells became harder to visualize in older parts of the alga. However, the stalk-like structures associated with the gland cells were also formed in the absence of bromide ions (Fig. 3.3B). Chapter 3 Gland cells 50

Epifluorescence microscopy

The vesicle inside the gland cell fluoresced blue under UV excitation (Fig. 3.4), consistent with the presence of metabolites with conjugated double bonds. Some of the main compounds in A. armata, including 3,3-dibromoacrylic acid, have conjugated double bonds (McConnell and Fenical 1977).

Transmission electron microscopy

The ultrastructure of gland cells in A. armata was well preserved by cryofixation and freeze substitution, as shown by TEM (Figs. 3.5-3.7). As predicted by light microscopy, TEM showed that the gland cells are attached to the inner wall of the pericentral cell (Figs. 3.5 & 3.6A). Within the gland cell a large vesicle is anchored to the cell wall (Fig. 3.5, arrow). This vesicle is highly vacuolated, containing many electron-translucent vacuoles (v) embedded in an electron-opaque matrix (Fig. 3.6A). These electron- translucent vacuoles are consistent with saturated compounds such as bromoform and dibromochloromethane. Carboxylic acids on the other hand may show a small degree of osmiophilic stain due to the presence of double bonds. Some of the vacuoles appear to partially osmiophilic (Fig. 3.6A, arrow heads), although these more likely represent artifacts due to a glancing section of the vacuole membrane. The periphery of the vesicle in the gland cells is dimpled at points where vacuoles are pressing into the cytoplasm (Fig. 3.6A: white arrows). The vacuoles of large gland cells are generally larger than those of smaller cells (Fig. 3.7A: small (s) and large (L) gland cells). Metabolite synthesis likely continues throughout the life of the gland cell, as vesicles contain vacuoles of a variety of sizes, up to 700 nm in diameter (Fig. 3.7B). Where cross walls have formed for branching (Fig. 3.7A), some pericentral cells do not appear to contain a gland cell. The refractile vesicles of gland cells were well preserved throughout, even where some pericentral cells showed signs of freezing damage (i.e. granulation in pericentral cell: Fig. 3.7A). Chapter 3 Gland cells 51

There was no evidence of the connective structures (Fig. 3.2) connecting the gland cell to the outer wall by TEM. This may represent the sensitivity of the stalk to the preparatory process. The relatively thin walls of the gland cell (Figs. 3.5 – 3.7) would facilitate the transfer of materials to the connective stalk. In some cases, the gland cell wall is tightly appressed to the outer wall of the pericentral cell (Fig. 3.7B), confirming similar predictions by light microscopy (Fig. 3.2C). The gland cell was largely filled by a single vesicle, however organelles including chloroplasts, mitochondria and a nucleus were also present (Fig. 3.6A&B). I also observed some ultrastructural characteristics of the axial and pericentral cells. The major features of the pericentral cells were chloroplasts and large electron-translucent vacuoles (Fig. 3.7A&B). Floridean starch was often observed in pericentral cells (Fig. 3.6A), but was not observed in either gland cells or axial cells. The axial cell appeared to contain numerous mitochondria and relatively few chloroplasts (Fig. 3.6A). Chapter 3 Chemical defence against bacteria 52

AB

p g

a g p p

D

C

ap

Figure 3.1. Light micrographs of the tetrasporophyte of Asparagopsis armata. (A) Branching form of A. armata showing position of gland cells (arrows). (B) Transverse section showing three pericentral cells (p), surrounding the axial cell (a). In this section, gland cells (g) are present in two of the pericentral cells. (C) Apical meristem showing the location of the gland cells (arrow heads) in the cell layer adjacent to the apical cell (ap). (D) Recently germinated carpospores produce gland cells in the initial spore material (arrow head), as well as in subsequent new growth of filament (arrow). Scale bars: A,B &D = 10μm, C = 5μm. Chapter 3 Chemical defence against bacteria 53

A

B

g

C

g

Figure 3.2. Light micrographs of gland cells in A. armata. Material is living, unstained thalli. (A) A stalk-like structure connecting the gland cell (arrow) to the surface region of the pericentral cell (arrow head). (B) In other cells, the structure appears more rigid (arrow heads). (C) For large gland cells (g), the cell wall can be tight against the outer wall of the pericentral cell (p) (arrow head). Scale bars; A,B = 10 μm, C = 5 μm. Chapter 3 Chemical defence against bacteria 54

A

B

Figure 3.3. The effects of bromine manipulation on the cell structure of the tetrasporophyte of Asparagopsis armata. (A) Br [+] algae. Pericentral cells contain a gland cell with a large refractile vesicle (arrows). (B) Br [-] algae. This vesicle is no longer produced in cells grown without bromine, although the structures associated with the gland cell are still formed (arrows). Scale bars = 10μm. Chapter 3 Chemical defence against bacteria 55

Figure 3.4. Epifluorescence micrograph of the gland cell in Asparagopsis armata using a UV filter set. The contents of the refractile vesicle of the gland cell (arrow) emit blue autofluorescence under UV excitation. Scale bar = 10 μm. Chapter 3 Chemical defence against bacteria 56

g g

a p p

g

Figure 3.5. Transmission electron micrograph of the tetrasporophyte of A. armata in longitudinal section. Gland cells (g) are formed inside the pericentral cells (p) from the inner wall, surrounding the central axial cell (a). The large vesicle in the gland cell is attached to the inner gland cell wall (arrow). Scale bar = 5 μm. Chapter 3 Chemical defence against bacteria 57

A

v

cw ew

c v m

c fs c

B

n

m

c

Figure 3.6. Transmission electron micrographs (extended legend overleaf). Figure 3.6. Images from TEM of A. armata. (A) Gland cell is attached at the inner wall (cw) of the pericentral cell. Many electron-translucent vacuoles (v) are present in the central vesicle. Some vacuoles appear to be lightly stained (arrow heads). The vesicle membrane is dimpled (arrows) where the vacuoles are pressing against it. The pericentral cell contains numerous chloroplasts (c) and grains of floridean starch (fs) in the cytoplasm. The axial cells have numerous organelles, in particular mitochondria (m). (B) Details of a gland cell showing chloroplasts (c) and a nucleus (n). Some chloroplasts have electron-translucent inclusions (arrow). Note the relatively thin wall of gland cell (arrow head) compared to other cells. (A&B) Scale bars = 1μm. ew, external cell wall. Chapter 3 Chemical defence against bacteria 58

A

L

pv pv

S

B

Figure 3.7. (A) Vacuoles in the vesicle vary in size; small gland cells (s) generally have small vacuoles. Larger gland cells (L) have larger vesicles and vacuoles. Pericentral cells contain typical vacuoles (pv). (B) The cell wall of a large gland cell appears to be tightly appressed to the surface wall of the pericentral cell wall (arrow head). Scale bars = 5 μm Chapter 3 Gland cells 59

Discussion

Structure / function relationships are crucial to marine chemical ecology, as the ecological role of a natural product is essentially dictated by the nature of the structure in which it is produced. This structure / function relationship is particularly important for surface-mediated roles such as antifouling (Steinberg and de Nys 2002). I recently provided empirical evidence for the storage and release of halogenated metabolites from the gland cells of Asparagopsis armata, and showed that these same metabolites inhibit epiphytic bacteria (Paul et al. 2006). I now examine in detail the cellular features involved in both the storage and release of natural products in A. armata, correlating light and epifluorescence microscopy, TEM and media manipulations with the localisation of halogenated metabolites in the specialised gland cells. Gland cells are formed soon after apical cell division in this filamentous alga. The ultrastructure of gland cells was well preserved by cryofixation and freeze substitution, suggesting that this preparation method is useful for the TEM preparation of algae with active natural products. Gland cells in A. armata are produced from the inner wall of the pericentral cells, and occupy space inside these cells. Although it is completely enveloped by the pericentral cell, the gland cell is discrete and contains chloroplasts, mitochondria and a nucleus. These organelles are confined to the peripheral cytoplasm, as the internal space of each gland cell is dominated by a large vesicle. When A. armata was cultured without bromine the large, refractile vesicle typical of the gland cells was no longer formed, indicating that the vesicle is the site of halogenated metabolite storage. Culture manipulations on other Bonnemaisoniaceae algae have demonstrated an integral role for bromine in the vesicle metabolism of the gland cells (Wolk 1968, Dworjanyn et al. 1999). Furthermore, algae that lack specialised cells can be cultured in the absence of bromine with minimal effects on their growth or reproduction (Fries 1969, McLachlan 1977). Such data infer that the specialised bromine metabolism in the Bonnemaisoniaceae has a secondary function. Chapter 3 Gland cells 60

The major vesicle of the gland cells in A. armata was the distinguishing cellular feature and was comprised of many small vacuoles. This vesicle was anchored to the cell wall, suggesting that its contents must be tightly secured, as would be the case for bioactive compounds. Most of the vacuoles within the vesicle were electron-translucent. This absence of vacuolar staining is consistent with the presence of some of the halogenated compounds of A. armata, including bromoform which is present at levels of up to 4 % of the dry mass of the alga (Chapter 2). Some gland cell vesicles contained vacuoles of various sizes, suggesting that new vacuoles (i.e. accumulation sites) form as the cell ages. The protrusion of the vacuoles at the vesicle periphery explains the dimpled (golf ball-like) appearance of the vesicle surface, as seen using light microscopy. An electron-opaque matrix contains and separates the vacuoles in the vesicle of the gland cells of A. armata. This intensely stained matrix could represent protein-rich regions (Pedersén et al. 1980, O’Brien and McCully 1981), such as the enzymes and ribosomes required for the production of the vacuolar contents, but may also consist of the metabolic precursors for the halogenated compounds – metabolic pathways are described in McConnell and Fenical (1977). The ultrastructure of the specialised cells in A. armata differs to Bonnemaisonia nootkana (Bonnemaisoniaceae), in which the cells have a single electron-translucent vacuole with no such matrix (Young 1977). The vesicle cells of B. nootkana may die as a consequence of accumulating brominated compounds (Young 1977). However, it appears that the gland cells of A. armata effectively store similar compounds without senescence. The protrusion of vacuoles at the vesicle periphery indicates that the osmiophilic matrix is not a rigid frame, and a potential for metabolites to cross the structure exists. It is likely that the vesicle is both the site of synthesis and storage of metabolites. However, ultrastructural comparison with similar algae is limited. Other algal metabolites are synthesized at the chloroplasts (Pohnert and Jung 2003) or on endoplasmic reticula (Schoenwaelder 2002). This may also be true for the halogenated compounds of A. armata, but variable sizes of the individual vacuoles inside the vesicle suggest it is not the case. Chapter 3 Gland cells 61

Histochemical stains (Young 1977, Pueschel 1992), electron-microprobe analysis (Wolk 1968, Pedersén et al. 1980, Young et al. 1980) and epifluorescence microscopy (Clayton and Ashburner 1994, Schoenwaelder and Clayton 1999, Dworjanyn et al. 1999) utilize the properties of natural products to aid in their localisation in the alga. I have demonstrated that the vesicle in the gland cells of A. armata autofluoresce blue under ultra-violet excitation. As some of the compounds in A. armata will autofluoresce, it provides further evidence for their localisation in the vesicle. Alternatively, microprobe analyses have shown that certain algal cells - including the gland cells of A. armata - accumulate halogens at levels much higher than surrounding tissue (Wolk 1968, Young et al. 1980). Such halogens are typically presumed to be in an organic form (Pedersén et al. 1980, Young et al. 1980), as appears to be the case for the gland cells of A. armata. The gland cells of A. armata are still formed in bromide-free culture media, but merely without a vesicle. This was also described by Wolk (1969). Marshall et al. (2003) inferred that the frequency of gland cells in Asparagopsis sp. was correlated to the level of bromide ions in the culture media, however, it was not clear whether the densities of non-vesiculate gland cells were affected (i.e. the number of gland cells in algae cultured with bromide-free media). It would be surprising if gland cell production is controlled by bromine, considering that bromide ions do not vary greatly in seawater (Burton 1996). Furthermore, I saw no evidence that the maintenance of the vesicle was affected by culturing without bromine, as gland cells from algae originally cultured in seawater did not lose their vesicles when transferred to bromide-free media. In any respect, the absence of the gland cell vesicle – not an absence of the gland cell – is consistent with results of Delisea pulchra (Bonnemaisoniaceae) grown without bromine (Dworjanyn et al. 1999). Specialised cells and structures are typically found in the surface cell layers of red algae, even those structures that do not function in surface release (Young 1979, Young et al. 1980). Most of the gland cells within the Bonnemaisoniaceae are formed in the surface cell layer (Womersley 1996). Interestingly, the gland cells in A. armata do Chapter 3 Gland cells 62

not abut the surface. However, an intricate relationship exists between the gland cell and its parent pericentral cell. Stalk-like structures connect the gland cell to the outer wall region of the pericentral cell and likely facilitate the movement of metabolites for their release to the surface (Paul et al. 2006). Such a structure would avoid large-scale exposure (i.e. autotoxicity) of the pericentral cell to the halogenated metabolites produced and released by the gland cells. Considering the presupposition by McConnell and Fenical (1977) that metabolites diffuse from the gland cell to the exterior and the detailed descriptive work on Asparagopsis spp. by Bonin and Hawkes (1987), it is surprising that these structures have not been described previously. Unfortunately, the stalk-like connection was not observed in the TEM sections, and whether it is an extension of the cytoplasm or cell wall was indistinguishable from light micrographs. However, it is difficult to determine transient features using TEM, such as the release of metabolites from the gland cells, and it is further possible that these structures were sensitive to the preparatory method. TEM did highlight the relatively thin walls of the gland cells, which could aid in the transfer of material from the gland cell to the stalk. Furthermore, the cell wall of larger gland cells can be tightly appressed to the outer wall of the pericentral cell. The presence of connective structures linking the gland cell to the surface, as well as the appressed cell wall of some large gland cells, indicate a potential mechanism for the release of gland cell contents. Much of the previous interest in the specialised cells and structures of red algae has related to their taxonomic value. For instance, the position and development of gland cells (Moe and Silva 1980, Womersley 1998), crystal morphology (Pueschel 1992) and the numbers of cellular inclusions can be distinguishing taxonomic features, particularly in the Ceramiales. The term “gland cell” is often used in taxonomic descriptions (Womersley 1998) but may lead to the incorrect assumption that all gland cells are secretory (Young 1978). Consequently, the specialised cells of A. armata have been referred to as “vesicle” cells, as a secretory function was not previously known (Wolk 1968, Young 1977, Marshall et al. 2003). However, recent works (Marshall et al. 2003, Paul et al. 2006) indicate that the specialised cells are true gland cells, a Chapter 3 Gland cells 63

terminology consistent with other members of the Bonnemaisoniceae (Womersley 1996). Detailed work on ultrastructure has proven integral to structure / function relationships in algae. This is highlighted by the primary roles for phenolic compounds in cell wall construction in the brown algae (Clayton and Ashburner 1994, Schoenwaelder and Clayton 1999, Schoenwaelder 2002). Prior to these studies, phenolics were principally considered as secondary products due to their anti-herbivore properties (Ragan 1976, Schoenwaelder 2002). Therefore, further developing structure / function relationships for the many specialised cells of red algae should also prove important in elucidating the ecological functions of their natural products. CHAPTER 4

Seaweed-herbivore interactions at small scales: a direct test of feeding deterrence using filamentous algae

Introduction

Functional form or life form models have been used to provide an explanatory framework for the distribution and success of plants with differing growth strategies (Feeny 1976, Littler and Littler 1980, McNaughton 1983, Steneck and Dethier 1994, Haukioja and Koricheva 2000). In terrestrial systems, comparisons have often been drawn between woody and herbaceous plants with large, long-lived and apparent trees contrasting to smaller, fast-growing herbs with short, opportunistic live histories (Feeny 1976, Haukioja and Koricheva 2000). Life form models have also been important in the development of theory for plant-herbivore interactions (Rhoades 1979, Bryant et al. 1983). For example, tolerance of herbivory is typically associated with herbaceous plants due to their greater capacity for compensatory growth (McNaughton 1983), although some woody plants also have potential for compensatory growth (Haukioja and Koricheva 2000). Resistance traits, as an alternative to tolerance, feature in both trees and herbs for their protection against herbivores (Feeny 1976), and the production of secondary metabolites as chemical defence is common in many types of terrestrial plants (Feeny 1976, Fritz and Simms 1992). Chapter 4 Chemical defence against herbivores 65

In the marine environment, macroalgae differ substantially in size and architecture, with life forms analogous to trees and herbs. Some algae are long-lived and reach lengths up to 50 m in height (North 1971) but co-exist with small, often ephemeral, filamentous algae that grow on the scale of centimetres and less. These filamentous algae are often a highly preferred food in the marine environment (Littler and Littler 1980, Steneck and Dethier 1994) and are important primary producers in benthic communities (Klumpp and McKinnon 1992). Generally, filamentous algae are thought to reduce the impacts of herbivory by escaping consumers in space, time or through high growth rates (Lubchenco and Gaines 1981). These traits are similar to those of many herbaceous plants in terrestrial systems (McNaughton 1983, Haukioja and Koricheva 2000). However, the role for chemical defences in filamentous algae has seldom been tested, despite that herbivore deterrents are common in their terrestrial analogues in herbaceous plants (Fritz and Simms 1992). Most studies on chemical defence in macroalgae have been on relatively large, robust algae, in particular kelps or fucoids (Steinberg 1984, Van Alstyne et al. 2001, Kubanek et al. 2004, Toth et al. 2005), but also other foliose or fleshy algae (Paul and Hay 1986, Hay et al. 1987a, 1987b & 1988a, Wright et al. 2004). These macroalgae produce a broad range of chemical defences (Hay and Fenical 1988, Hay 1996, Paul and Puglisi 2004). Perhaps because of the assumed high susceptibility of filamentous algae to herbivores, examples of chemical defence in filamentous forms are limited (but see Paul et al. 1990). There is evidence that some cyanobacteria are chemically defended (Pennings et al. 1997, Nagle and Paul 1999). However, these organisms are not true algae, despite some shared characters. Filamentous algae are subject to herbivory by many types of consumers, similar to other marine algae (Lubchenco and Gaines 1981, Steneck and Dethier 1994). Chemical defence against herbivores could be important for filamentous algae if the scale of herbivory is comparably small. But tests of chemical defence by seaweeds against herbivores have typically focussed on large consumers such as fish and urchins (Paul and Hay 1986, Hay et al. 1987a & 1987b), which can remove a filamentous Chapter 4 Chemical defence against herbivores 66

individual in a single bite. However, many small herbivores – mesograzers (Brawley 1992) – also consume filamentous algae, and often preferentially graze them (Brawley and Adey 1981). As mesograzers make feeding choices at a fine scale and can be present at high densities (Brawley and Adey 1981, Duffy and Hay 2000) it is possible that these smaller herbivores exert a strong selective pressure on filamentous algal communities. To date, the principal means of testing algal chemical defence are by feeding assays where either (1) extracts and pure compounds are incorporated into or coated onto artificial diets (Paul and Hay 1986, Hay et al. 1987a & 1987b, Cruz-Rivera and Hay 2003), or (2) by correlations between metabolite levels and herbivory within individuals (Steinberg 1984, Hay et al. 1988, Van Alstyne et al. 2001). Neither approach is entirely suitable for filamentous algae given that the structural complexity of artificial diets will not reflect the subtleties of the filamentous morphology and that within plant comparisons are not practical as a result of their simple architecture. In this chapter, I test whether filamentous algae can chemically defend themselves from consumption by small herbivores. Firstly, I compare consumption rates of five different filamentous algae by an amphipod. Three of these algae form specialised structures that could indicate the presence of bioactive natural products. I then focus specifically on chemical defence against herbivores by the filamentous stage of Asparagopsis armata, as its halogenated metabolites can be manipulated to produce algae with halogenated metabolites [bromide (+)] and algae without halogenated metabolites [bromide (-)]. I test the feeding response of three mesograzers (an amphipod, an abalone and a sea hare) using these manipulated plants. This allows me to examine for the first time the role of chemical defence in a seaweed whilst maintaining its structural form. I also explore whether differences in consumption of algae which lack halogenated metabolites are attributable to specific halogenated metabolites in A. armata by incorporating its natural products into artificial diets. Bioassays using artificial diets are performed on the same herbivores that are tested with the bromide (+) and bromide (-) A. armata. Chapter 4 Chemical defence against herbivores 67

Materials and methods

Feeding assay with an amphipod

Much of the previous work on chemical defences in macroalgae has focused on large macroalgae such as kelps and foliose varieties (Hay and Fenical 1988, McClintock and Baker 2001), while filamentous species have generally been overlooked. I collected three filamentous red algae that form specialised cells or structures, which can be indicative of chemical defence. These were Anotrichium tenue Nägeli (Ceramiaceae), Asparagopsis armata Harvey (tetrasporophyte) (Bonnemaisoniaceae) and Balliella amphiglanda Huisman & Kraft (Ceramiaceae) [Note: see Figs. 6.1 & 6.2 for micrographs of these algae]. I also collected a common filamentous red alga, Callithamnion korfense Millar (Ceramiaceae), which did not have any obvious storage structures, as well as a green alga from the genus Enteromorpha (Ulvaceae). All algae were collected from Bare Island (33° 59’ 38” S, 151° 14’ 00” E), Sydney, Australia. The consumption rates of the five algae were measured in a no-choice feeding assay using the generalist amphipod Hyale nigra. Consumption rates of each alga were determined by measuring changes in surface area because variations in mass were too small to be quantified. Changes in surface area as a result of consumption were quantified by digital imaging. Algae were spread across a slide and flattened under a glass coverslip, taking care not to damage the plants. Digital images of the algal surface area were made on an Olympus BX100 stereomicroscope under 4X magnification. Image analysis was performed with Image J (Scion Corporation) public domain software. Treatment algae (N = 10, with herbivores) and control algae (N = 10, without herbivore) were run for each of the five species for 36 hours (over two nights) in 30 mL static seawater at 19°C. One amphipod was added to each treatment dish. Data were analysed using 2-factor ANOVA (Peterson and Renaud 1989), where a significant interaction term (between Algal species and Herbivore presence/absence) Chapter 4 Chemical defence against herbivores 68

indicates a differential feeding effect between algal species. Post-hoc comparisons were made with Tukey’s HSD multiple comparisons at P = 0.05.

Collection and culturing of Asparagopsis armata

Asparagopsis armata was selected to further investigate potential chemical defence against herbivores as it is possible to manipulate the presence of the halogenated metabolites in cultured algae (Chapter 2). Culturing A. armata with bromine produced algae with halogenated metabolites [bromide (+)] and culturing without bromine produced algae without halogenated metabolites [bromide (-)]. This provided a unique test of the role of the halogenated metabolites in the interaction with herbivores. The filamentous tetrasporophyte of A. armata was collected from the shallow subtidal (0.25 – 2 m depth) at Bare Island. Clean, apical sections were cultured in sterile seawater for approximately 4 weeks, after which those individuals that grew well were selected for subsequent culturing. Algae were chopped finely and then cultured in one of two artificial media, identical except for the presence or absence of bromide ions at natural concentrations (details in Chapter 2). Algae were cultured in 500 mL of media in glass dishes under 20 μmol photons m-2 s-1 using 36W daylight globes with 16: 8 h light: dark cycle at 19º C. Culture medium was changed every 2 to 3 weeks. Algae were grown until they were of sufficient size for use in feeding assays (~ 5 mg fresh weight [FW]). Only individuals that grew well in the artificial culture media and were of normal morphology were harvested and sub-cultured. This ensured that algae used for the feeding assays were structurally similar.

Effects of bromide (+) and bromide (-) artificial media on algae

Previously (Chapter 2), I have demonstrated that removing bromine from the artificial growth medium prevents the production of the halogenated secondary metabolites of Chapter 4 Chemical defence against herbivores 69

A. armata. The levels of the major halogenated metabolites in bromide (+) cultured algae are 0.3 – 0.5 % dry weight for bromoform, and, 0.07 - 0.1 % dry weight for dibromoacetic acid (Chapter 2). These values are at the lower end of the natural range of concentrations present in the field. No halogenated metabolites are present when A. armata are cultured without bromine. To quantify the effect of bromine removal on components of plant biology other than secondary metabolites, the carbon and nitrogen composition were measured in bromide (+) and bromide (-) algae. Samples were taken from each of seven plants grown in either bromide (+) or bromide (-) media. As individual filaments were small, a number of pieces were pooled for each replicate to give ~ 5 mg dry weight [DW] after freeze drying. Carbon and nitrogen analyses were performed by Ms S Wood at the Research School of Biological Sciences, Australian National University, Canberra, Australia. Percent carbon and nitrogen dry masses were measured for both bromide (+) and bromide (-) algae. Previous chemical analyses of bromide (+) and bromide (-) algae have shown that the production of non-halogenated metabolites is unaffected by culturing without bromine (Chapter 2).

Herbivores

I tested whether the halogenated metabolites in Asparagopsis armata deterred consumption by three herbivores that are associated with the alga, the amphipod Hyale nigra Smith, the sea hare Aplysia parvula Mörch and the black-lip abalone Haliotis rubra Leach.

Hyale nigra is considered a generalist amphipod that mostly consumes filamentous epiphytes on its host algae (Poore 1999). H. nigra is not common on A. armata (N Paul pers. observ.), and in order to obtain sufficient numbers for feeding assays, animals were collected from Sargassum linearifolium, a large brown macroalga at Bare Island. Chapter 4 Chemical defence against herbivores 70

Adult abalone Haliotis rubra readily consume the gametophyte of A. armata in the field (Shepherd & Steinberg 1992) but the preference of juvenile abalones is not known. Juvenile H. rubra were provided by Dr S Dworjanyn at NSW Fisheries, Port Stephens, New South Wales, where they had been settled and grown until they were capable of feeding on macroalgae (~ 1 – 3 months old, 3 – 10 mm in length). A group of larger abalone (100 – 200 mm) was used in the feeding assays with artificial diets as smaller animals did not consume agar-based diets.

The sea hare Aplysia parvula is regularly found on chemically defended algae, including the gametophyte of A. armata, at Bare Island, (Rogers 2000). A. parvula (0.1 – 0.9 g fresh weight [FW]) were collected from algae, predominantly from the genera Asparagopsis, Sargassum and Delisea on which the sea hares aggregate.

All herbivores were maintained at constant temperature (19°C) and were fed on a diet of Ulva sp. for at least 24 hours prior to assays, in order to reduce potential prejudice from recent feeding.

Feeding assays - Whole plants

To test if the halogenated metabolites in bromide (+) Asparagopsis armata function as herbivore deterrents, I ran pairwise-choice and no-choice feeding assays with each herbivore (amphipods < 6 mm length (head to tail), abalone < 400 mg FW, sea hares < 900 mg FW). A choice assay between bromide (+) and bromide (-) A. armata was run for the amphipod, abalone and sea hare. A no-choice assay was also run, but only for the amphipod and abalone (sea hares did not make a choice so were omitted). Prior to running the feeding assays in seawater, I assessed whether new (apical) growth of Asparagopsis armata resumed production of halogenated metabolites over the duration of the assays (from 1 – 2 nights). A. armata stores the halogenated metabolites as a refractile inclusion in the gland cells (Chapter 3). This inclusion can be easily Chapter 4 Chemical defence against herbivores 71

viewed with a compound microscope. After 12 h, no gland cells were obvious in the apical sections of the bromide (-) filaments that had been left in seawater. After 36 h, some small gland cells had formed between 0 and 4 cell tiers beneath the apical cell, indicating the synthesis of halogenated metabolites. As this represented only a small fraction of the overall tissue, it was unlikely to impact on the assay. Feeding assays with H. nigra, A. parvula and the H. rubra choice assay were run over 12 h. The no-choice assay with H. rubra was run over 36 h, as a result of low consumption rates for smaller animals. For all assays, a herbivore was removed from a replicate dish if it had consumed a substantial (i.e. visible) portion of the diet. Feedings assays were run in 30 mL of static seawater at 19°C. Once initial area measurements were made (see Feeding assay with an amphipod (above) for details), a single herbivore was placed in each replicate dish. At the end of experiment, algae were removed and final areas measured. The identity of tissue in the choice assays [bromide (+) or bromide (-)] was determined using a compound microscope. The area of pieces scattered during feeding were pooled with the correct tissue type for image analysis. Autogenic controls (N = 5 – 15, as indicated in results) were run to adjust for any changes in algal size that did not result from herbivory. As some autogenic controls increased in size over the experiment, the figures for the feeding assays display consumption as the change from an adjusted initial size. This initial size was determined from changes in autogenic controls using the equation: Consumption = Areainitial. x

(Control Areafinal / Control Areaintial) – Areafinal . For choice assays, a significant preference for either bromide (+) or bromide (-) algae was determined by a two sample t-test. The t-test was calculated using the difference between the mean difference in area consumed for the paired treatment replicates (bromide[+]/[-] with herbivores) and the mean difference in area for the paired autogenic controls (bromide[+]/[-] without herbivores) (Peterson and Renaud 1989). Only animals that fed sufficiently were used in analyses (i.e. consumed more than 1 SD of the change in surface area for the corresponding autogenic controls). The number of replicates in each analysis is reported in Results. Chapter 4 Chemical defence against herbivores 72

For no-choice assays, consumption was expressed as a mean rate (consumption [mm2 surface area] per hour) for each treatment. A significant interaction in the 2-factor ANOVA between Herbivore (presence/absence) and Media (bromide[+]/[-]) indicates different feeding rates. In these assays, a replicate was only removed from the analysis if the animal had died during the assay. As the feeding assay with Hyale nigra was the first to be performed, a third growth medium level (seawater control) was included in order to determine if differences existed between feeding on algae cultured in bromide (+) and those cultured in natural seawater. This seawater treatment was not incorporated into subsequent experiments with the abalone and sea hare.

Abalone size effects. As an additional analysis of the data obtained from the no-choice assay with abalone Haliotis rubra, the effect of abalone size on consumption of Asparagopsis armata was also examined. In addition to the Peterson and Renaud (1989) 2-factor ANOVA on consumption rates (Media (bromide[+]/[-]) x Herbivore [+]/[-]), I also ran an Analysis of Covariance on the same data set. The ANCOVA incorporated abalone size (which ranged from 30 – 400 mg FW) as a covariate. As this analysis required that there were no missing levels for the covariate (Abalone Size) by Media (bromide[+]/[-]) interaction (i.e. the analysis could not be run with data that included the autogenic controls without herbivores), the data used were consumption rates of bromide (+) and bromide (-) algae (with herbivores) after adjustments had been made for their initial size using the adjustment coefficient (see p 71 for equation).

Feeding assays - Artificial diets

I conducted feeding assays with artificial diets, in order to determine whether any differences in consumption of the whole plants (bromide[+]/[-] algae) could be attributed to specific metabolites present in Asparagopsis armata. Feeding assays using artificial diets (often containing agar) have been the traditional determinants of feeding deterrence by algal natural products. I tested whether the natural products of Chapter 4 Chemical defence against herbivores 73

Asparagopsis armata (both crude extracts and pure compounds) also exhibited feeding deterrence in the artificial diet format. Firstly, I ran feeding assays of these natural products in the field, as the previous assays with whole plants were performed only in the laboratory. Secondly, I ran feeding assays in the laboratory using the same herbivores (amphipod, abalone and sea hare) from the whole plant assays. Only pure compounds were tested in the laboratory, as previous assays using bromide (+) and bromide (-) algae had determined that natural products of A. armata were feeding deterrents. Assays in the field, with the amphipod and with the abalone used agar-based artificial diets. However, as the sea hare would not feed on agar, its diet comprised of metabolites coated onto blades of Ulva sp.. All assays were run by incorporating the natural products into the diets at natural concentrations. Crude extract of A. armata was obtained by extracting 9.1 g of freeze- dried material exhaustively in dichloromethane. This yielded 44.2 mg of crude extract. The two major halogenated metabolites were assayed at levels found in field samples of A. armata. As a percentage of plant dry weight, these are 1.4 % DW for bromoform

(CHBr3) and 0.4 % DW for dibromoacetic acid [DBA] (Chapter 2). Both compounds were purchased from Sigma-Aldrich. For agar-based diets (used in field assays and assays with the amphipod and abalone), metabolites were incorporated into a combination of freeze-dried Ulva sp. 0.3 g, agar 0.9 g and ultrapure water 18 mL (larger quantities kept this same ratio). Extract and pure compounds were dissolved in di-ethyl-ether (DEE) and added to finely ground Ulva to give a dry weight equivalent to natural concentrations in A. armata. Agar was added to 15 mL water and heated until boiling in a microwave. 3 ml of water was added to the dried Ulva and made into a slurry, to which the agar was added after cooling to ~ 50°C. Control discs were produced in the same way merely without the addition of metabolites to the DEE. In field assays, diets (discs ~ 5 g in size) were paired (control and treatment) and secured to coralline turf in areas where A. armata was present. The crude extract, bromoform and DBA were tested in field experiments (N = 15 – 20, but see Results for Chapter 4 Chemical defence against herbivores 74

replicates in actual analysis). Prior to deployment, discs were weighed in the laboratory, along with discs to be used as autogenic controls (N = 5 – 9). Each of the pre-weighed paired discs was attached using a small piece of twine tied to a paper clip. As these were suspended slightly above the turf, it was likely that fish were the major consumers (although this was not quantified). Autogenic controls were kept in seawater collected from the site for the same duration as experimental discs. This assay was run overnight for approximately 16 h, after which both experimental and autogenic control discs were blot-dried and final weights obtained. In laboratory assays, the two major metabolites (bromoform and DBA) were run in a pairwise scenario (metabolite versus control) using a single herbivore in each dish (N = 10 – 20). For the abalone and amphipod (agar-based diets), artificial diets were spread over a plastic sheet so that 60 – 80 mg pieces could be excised. A small piece of fishing line was inserted into each food to differentiate between control and treatment discs. The pre-weighed artificial discs were placed in a dish containing ~50 ml static seawater. Autogenic controls (N = 10) were also run for each assay. Feeding assays were run overnight for ~ 12 h, after which the discs were blot-dried and weighed. Only replicates where the herbivore had consumed sufficient amounts were used in analysis (> 5 % of the initial mass of the disc). For laboratory assays with the sea hare A. parvula, metabolites were applied directly to the surface of pieces of excised Ulva sp. (3 cm2). Ulva-discs were cored from fresh samples, spread out and air-dried prior to applying compounds. Both DBA and bromoform were dissolved in DEE and added to the surface of the alga to give the same dry weight concentration as the agar diets. Each side of the Ulva-disc was coated with an equal amount of DEE (treatment and control alike). Treatment and control discs were marked by threading different coloured cotton through the centre of each disc. As Ulva sp. is uniformly 2 cells thick, consumption was measured by the size (area) of the feeding scar, which was always at the edge of the disc. As preliminary tests determined that the excised discs did not change overnight, consumption was measured as a loss of area from the initial 3 cm2 digital images pre- and post-consumption. The difference in Chapter 4 Chemical defence against herbivores 75

consumption between controls and treatments discs for each replicate sea hare (Area

Control – Area Treatment) was analysed using a 1-sample t-test tested against a mean of zero.

Statistical analysis

All analyses were made using SYSTAT 10 (SPSS Inc.). Assumptions of homogeneity of variance and normality were assessed by examining the scatterplots and distribution of residuals where required (Quinn and Keough 2002). Experimental designs are outlined in each respective section above. Exceptions (and additional information) are listed below. Comparisons between the carbon and nitrogen content of bromine (+) and bromine (-) plants were made using paired sample t-tests, with plants paired for each media type. For choice assays of whole plants and artificial diets, the differences in consumption were analysed by 2-sample t-tests (Peterson and Renaud 1989). If the variance of the differences in control and herbivore groups were not equal, even after transformation, then a Mann-Whitney U-test was used. As the mean consumption of artificial diets was small for many assays, the figures report the mean total amounts of control and treatment diets consumed after adjustments for changes to initial size of the experimental (plus herbivore) pieces. Chapter 4 Chemical defence against herbivores 76

Results

Feeding assays with the amphipod Hyale nigra

There were significant differences in consumption rates of the five filamentous algae (Table 4.1, ANOVA, Alga x Herbivore, P < 0.001). Figure 4.1 shows consumption rates (mm2 surface area per hour) for each alga after the adjustment was made to compensate for changes in autogenic controls through the course of the experiment. Consumption of the three algae with specialised storage structures [Asparagopsis armata (with specialised gland cells), Anotrichium tenue (an alga with many small refractile bodies), and, Balliella amphiglanda (with vesicle cells)] was significantly less than consumption of the algae without specialised structures, Callithamnion korfense and Enteromorpha sp. (Tukey’s HSD at P = 0.05, Fig. 4.1).

Effects of bromide (+) and bromide (-) artificial media on algal thalli

Growing Asparagopsis armata in media with and without bromine produced algae with and without halogenated compounds, respectively. A lack of the characteristic refractile inclusion in gland cells of A. armata (Chapter 2 & 3) in bromide (-) algae indicated the cessation of secondary metabolite production. There was no significant difference between bromide (+) and bromide (-) algae in either percent DW carbon (paired t-test, P = 0.65, Fig. 4.2A), percent DW nitrogen (paired t-test, P = 0.32, Fig. 4.2B) or C:N ratio (paired t-test, P = 0.26, Fig. 4.2C). Chapter 4 Chemical defence against herbivores 77

Whole-plant feeding assays

The manipulation of the halogenated metabolites in Asparagopsis armata had a strong effect on feeding by two of the three herbivores.

The amphipod Hyale nigra strongly preferred bromide (-) algae over bromide (+) algae in the pairwise-choice experiment (t-test, P = 0.003, Nexperimental = 9, Nauto-controls = 9, Fig 4.3A) and consumed bromide (-) algae at a significantly higher rate than both bromide (+) and seawater cultured algae in the no-choice experiment (Table 4.1, ANOVA, Media x Herbivore, P < 0.001, Fig. 4.3B). No difference in consumption was observed between algae grown in bromide (+) medium and those grown in seawater which contained natural levels of bromide ions (Tukey’s HSD, P = 0.352, Fig. 4.3B).

The abalone Haliotis rubra also preferred bromide (-) algae over bromide (+) algae in a pairwise choice experiment (t-test, P = 0.036, Nexp. = 16, Nauto. = 10, Fig. 4.4A) and consumed bromide (-) algae at a significantly higher rate than bromide (+) algae in a no- choice experiment (Media x Herbivore; F1,24 = 22.44, P < 0.001, Fig. 4.4B). Interestingly, further analysis by ANCOVA showed a significant interaction between the size of the abalone and the type of alga consumed (Table 4.2, Media x Abalone Size, P < 0.001). Abalone smaller than 100 mg FW did not consume bromide (+) algae, but animals larger than this readily consumed both types of algae (Fig. 4.4C, arrow).

The sea hare Aplysia parvula showed no preference for either bromide (-) or bromide (+) algae (t-test, P = 0.842, Fig. 4.5). Chapter 4 Chemical defence against herbivores 78

Artificial-diet feeding assays

The variance in mass changes for autogenic control diets (those minus herbivores) of the agar-based artificial foods was high relative to the consumption of treatment discs with herbivores. This variance, combined with low numbers of replicates with measurable consumption, limited the ability to detect significant effects in some of the analyses.

i) Crude extract

In field feeding assays, the crude extract of Asparagopsis armata deterred feeding

(Mann-Whitney U-test; P = 0.013, Nexp. = 7, Nauto. = 9, Fig. 4.6).

ii) Dibromoacetic acid

The only assay in which DBA was an effective feeding deterrent was with the sea hare A. parvula, which was significantly deterred from discs that were coated with DBA (1- sample t-test, P = 0.047, N = 12, Fig. 4.7). Dibromoacetic acid (DBA) did not deter grazing in the field (Mann-Whitney U-test, P = 0.361, Nexp. = 6, Nauto. = 5). Variance for the field assay was high and the analysis suffered from low replication for plus- herbivore treatment as result of limited feeding. In laboratory assays, DBA did not deter feeding by H. nigra (t-test, P = 0.297, Nexp. = 10, Nauto. = 13) nor by the abalone

H. rubra (t-test, P = 0.804, Nexp. = 29, Nauto. = 16).

iii) Bromoform

Bromoform did not deter feeding in any assay. It was not deterrent in the field (Mann-

Whitney U-test, P = 0.291, Nexp. = 7, Nauto. = 5). There were also no significant effects of feeding in laboratory assays with the amphipod H. nigra (t-test, P = 0.525, Nexp. = 12,

Nauto. = 7), the abalone H. rubra (t-test, P = 0.289, Nexp. = 21, Nauto. = 18) or the sea hare A. parvula (1-sample t-test, P = 0.98, N = 7). Chapter 4 Chemical defence against herbivores 79

0.10 d

0.08 / h) 2

0.06

0.04 a Consumption rate (mm

0.02 bc c b 0 Callithamnion Anotrichium Asparagopsis Balliella Enteromorpha

Algal Species

Figure 4.1. No-choice feeding assay with the amphipod Hyale nigra. Mean consumption rates (meansSE, N = 10) are displayed for five filamentous algae after adjustments were made for changes to autogenic controls (N = 10). Algae that have conspicuous specialised structures (as determined using light microscopy) were consumed at a lower rate than algae that do not have obvious storage structures. H. nigra is a generalist amphipod (inset: feeding on Enteromorpha sp., Scale Bar = 1 mm). Chapter 4 Chemical defence against herbivores 80

35 A. NS 30

25

20

% C (DW) % 15

10

5

0

3 B. NS

2.5

2

1.5 % N (DW) % 1

0.5

0 C. 14

12 NS

10

8

%C:%N 6

4

2

0 Bromide (-) Bromide (+)

Figure 4.2. Carbon and nitrogen values for Asparagopsis armata cultured with and without bromine [bromide(+) and bromide(-)]. A. Percent dry weight (DW) carbon, B. Percent dry weight nitrogen, C. C:N ratio. Results display mean values ± SE (N = 7). NS denotes no significant difference at P = 0.05. Chapter 4 Chemical defence against herbivores 81

A. 0.04 * 0.03 / h) 2

0.02

0.01 Consumption rate (mm

0 Bromide (-) Bromide (+) B. 0.07 a

0.06

/ h ) 0.05 2

0.04

0.03 bb 0.02 Consumption rate ( mm

0.01

0.00 Bromide ( - ) Bromide ( + ) Sea Control Growth Medium

Figure 4.3. Whole plant feeding assays with the amphipod Hyale nigra. A. Pairwise-choice feeding assay showing mean consumption (sSE) of bromide (+) and bromide (-) Asparagopsis armata (N = 9). Significance at P < 0.05 indicated by *. B. No-choice feeding assay showing consumption rates (mean sSE) of A. armata (without halogenated metabolites [bromide (-)] and with halogenated metabolites [bromide (+) and seawater control] (N = 24). Bars sharing the same letter do not differ at P = 0.05 (Tukey’s HSD). Chapter 4 Chemical defence against herbivores 82

A. B. 0.25 0.30

/ h) / h) 0.25

2 0.20 2

0.20 0.15 * * 0.15 0.10 0.10 0.05 0.05 Consumption rateConsumption (mm rateConsumption (mm 0 0 Bromide(-) Bromide(+) Bromide(-) Bromide(+)

C. 18 Bromide (+) algae Bromide (-) algae 15 ) 2

12

9

6 Amount consumed (mm

3

0 0.1 0.2 0.3 0.4 0.5

Abalone Size (g fresh weight)

Figure 4.4. Whole plant feeding assays with the abalone Haliotis rubra. A. Pairwise- choice feeding assay between bromide (+) and bromide (-) Asparagopsis armata after 12 h (N = 16). B. No-choice feeding assay showing consumption rates (mean sSE) of A. armata without halogenated metabolites [bromide (-)] and with halogenated metabolites [bromide (+)] (N = 14). A, B Significance at P < 0.05 indicated by *. C. Single datum plot of the total amounts of algae consumed by individual abalone for no-choice assay across abalone size (N = 14). Arrow indicates the size above which abalone consume bromide (+) algae. Consumption of bromide (+) algae above 100 mg FW produced a significant interaction

(F1,24 = 17.96, P<0.001) between Abalone Size and Media. Chapter 4 Chemical defence against herbivores 83

0.5

/ h) 0.4 2 NS 0.3

0.2

0.1 Consumption rate (mm

0 Bromide (-) Bromide (+)

Figure 4.5. Whole plant feeding assay with the sea hare Aplysia parvula. Pairwise-choice feeding assay showing mean consumption rates (sSE) of algae without halogenated metabolites (bromide [-]) and with halogenated metabolites (bromide [+]) [N = 14]. NS denotes no significant difference at P = 0.05. Chapter 4 Chemical defence against herbivores 84

200

180

160 140 * 120 100

80

60

Mass of diet consumed (mg) 40

20

0 Control Extract

Figure 4.6. Effects of crude extract from Asparagopsis armata on feeding in the field. Mean total consumption (sSE) of treatment and control artificial diets are shown for a pairwise-choice feeding assay (N = 8). Significance at P < 0.05 indicated by *. Chapter 4 Chemical defence against herbivores 85

2.0 ) 2 1.5 * 1.0

0.5 Area of diet consumed (cm

0 Control DBA

Figure 4.7. Effect of dibromoacetic acid on feeding by the sea hare Aplysia parvula. Bars show mean total consumption (area) sSE of each treatment (with herbivore) disc consumed, N = 12. Significance at P = 0.05 is represented by *. Chapter 4 Chemical defence against herbivores 86

Table 4.1. Results of 2-factor ANOVA testing the effect of Alga (Callithamnion, Enteromorpha, Anotrichium, Asparagopsis and Balliella) and Herbivore (with [N = 10] and without [N = 10] amphipod) on changes to algal surface area when fed to the generalist amphipod Hyale nigra. Important significant effects at P < 0.05 are in bold. Tukey’s HSD was used for post-hoc comparisons between consumption rates of the algae.

Source df MS F P

Alga 4 1.1 x 10-3 25.70 «0.001 Herbivore 1 9.5 x 10-4 22.51 <0.001 Alga x Herbivore 4 2.6 x 10-4 4.20 <0.001 Error 90 4.0x 10-5

Table 4.2. Results of 2-factor ANOVA testing the effect of Media (bromide [+], bromide[-] and a seawater control) and Herbivore (with [N = 24] and without [N = 10] amphipod) on changes to Asparagopsis armata when fed to the amphipod Hyale nigra. All data was log-transformed to resolve heterogeneous variances. Important significant effects at P < 0.05 are in bold. Tukey’s HSD multiple comparisons were used for post- hoc analyses (bromide [-] > bromide [+] = sea control).

Source df MS F P

Media 2 9.11 x 105 17.01 «0.001 Herbivore 1 9.61 x 105 17.94 <0.001 Media x Herbivore 2 4.96 x 105 9.25 <0.001 Error 102 5.36 x 104 Chapter 4 Chemical defence against herbivores 87

Table 4.3. Results of 2-factor ANCOVA testing the effect of Media (bromide [+] v bromide [-]) and a co-variate Abalone Size (30 to 400 mg FW) on the consumption of Asparagopsis armata by the abalone Haliotis rubra. Consumption was measured as a no-choice feeding rate after 36 hours (N = 14). ANOVA of the non-adjusted consumption data confirmed a significant interaction between Media (bromide [+]/[-]) and Herbivore (presence/absence) (F1,24 = 22.44, P < 0.001). Important significant effects at P < 0.05 are in bold.

Source df MS F P

Media 1 0.285 54.22 <0.001 Abalone Size 1 0.105 20.01 <0.001 Media x Abalone Size 1 0.093 17.76 <0.001 Error 24 0.005 Chapter 4 Chemical defence against herbivores 88

Discussion

Filamentous algae in marine systems are small and typically have fast growth rates (Klumpp and McKinnon 1992). They are assumed to be highly susceptible to consumers (Littler and Littler 1980, Steneck and Dethier 1994) and to occupy spatial and temporal niches in environments with low to moderate herbivory (Lubchenco and Gaines 1981, Paine 1990). However, here I demonstrate that the different rates of consumption among five filamentous algae could relate to chemical defences, as indicated by the presence of specialised storage structures in those algae that were least consumed. Furthermore, I demonstrate that the filamentous life history stage of Asparagopsis armata produces halogenated natural products that deter feeding by relevant consumers. This unusual combination of chemical defence and a filamentous growth form could be a valuable strategy for A. armata but also for any filamentous alga that produces chemical defence. Filamentous algae represent a highly successful growth form, present in all three macroalgal divisions. However, large macroalgae have been the traditional focus of marine chemical ecology (Hay and Fenical 1988, Paul 1992, Paul and Puglisi 2004). Ecological tests of chemical defence are for the most part absent for filamentous algae (but see Paul et al. (1990) for chemical defence by a filamentous green alga). More is known of the chemical defence of cyanobacteria than of filamentous algae (Pennings et al. 1997, Nagle and Paul 1999). The low incidence of defensive chemicals in filamentous algae is perhaps a combined product of the assumed dependence by these algae on high growth rates (Littler and Littler 1980, Lubchenco and Gaines 1981, Paine 1990) and the difficulty to collect and directly test their natural products whilst maintaining any interaction between defence and the filamentous form. Filamentous turfing communities on benthic reefs are comprised of many species. I have shown here that some filamentous algae were consumed at lower rates by the amphipod Hyale nigra than others. A unifying feature of these algae was the Chapter 4 Chemical defence against herbivores 89

presence of specialised storage structures (for Asparagopsis armata, Balliella amphiglanda and Anotrichium tenue). The fourth red alga Callithamnion korfense contained no structures and was consumed at a significantly higher rate. Although structural and nutritional differences may have influenced the outcome of this experiment, the presence of these specialised structures provides an alternative explanation in chemical defence. Numerous halogenated metabolites have been identified in A. armata (McConnell and Fenical 1977). However, nothing is reported of the natural product chemistry for either B. amphiglanda or A. tenue (MarinLit 2004). The novel technique of manipulating halogenated metabolites in Asparagopsis armata enabled the roles of these metabolites to be separated from the potential effects of other aspects of algal biology on herbivore response. The bromine manipulation did not alter the carbon to nitrogen ratio of the alga and the secondary metabolites of A. armata remained localised in their natural state. This meant that potential differences in concentration or plant nutrition were avoided (Cruz-Rivera and Hay 2003), any synergistic or additive effects of the metabolites were preserved (Hay et al. 1994), and the structural integrity of the filamentous form was maintained (Littler and Littler 1980, Steneck and Dethier 1994). Furthermore, the consumption rates of bromide (+) algae and algae cultured in natural seawater did not differ in the no-choice feeding assay with the amphipod Hyale nigra, indicating that the deterrent effect of the natural products in bromide (+) algae is similar to that of the natural products of algae cultured in seawater. Feeding deterrence by bromide (+) algae (with metabolites) against two of the three mesograzers examined (amphipod and abalone) can be directly attributed to the halogenated metabolites of A. armata. This test is the first demonstration of the direct benefits of secondary metabolites to a filamentous alga. The rationale for focussing on chemical defence against herbivory in filamentous algae at a small scale using small herbivores is that these herbivores may exert a greater selective force on the filamentous algal community than larger herbivores. This is because large herbivores such as fish and urchins can consume an entire filamentous alga with a single bite. However, small herbivores should make choices at a much finer scale. As the group “mesograzers” Chapter 4 Chemical defence against herbivores 90

include animals that spend their entire life as mesograzers (e.g. amphipods and sea hares) and many juvenile versions of large benthic herbivores (e.g. abalone and urchins), together these consumers can be at high densities and have a substantial impact on algal biomass (Brawley 1992, Duffy and Hay 2000). The amphipod Hyale nigra was highly deterred from feeding upon bromide (+) algae. Small, mobile mesograzers – such as amphipods – are important consumers of algae but are also relatively transient (Brawley and Adey 1981, Poore 1999). Amphipods will make choices at the scale of single branches or even cells within algal filaments as a consequence of their small mouthparts. As amphipods are selective of their food source (Poore and Steinberg 1999) they could be important selective agents on the structure of filamentous algal communities. Abalone selected non-defended bromide (-) algae over bromide (+) algae in the choice feeding assay. Considering the potential negative effect of defensive compounds on abalone growth (Winter and Estes 1992), the choice of algae without halogenated metabolites by small abalone was perhaps an expected outcome. However, adult Haliotis rubra are not deterred from eating the gametophyte of A. armata in the field (Shepherd and Steinberg 1992). As the halogen chemistry is the same for both the gametophyte and tetrasporophyte of A. armata (Chapter 2) this difference in choice between adult and juvenile cannot be explained solely by chemical defences. I found a clear size-dependent effect of the halogenated metabolites from A. armata on the consumption by abalone. Abalone smaller than 100 mg FW readily consumed bromide (-) algae, but did not consume bromide (+) algae, despite the absence of an alternative food source. Although abalone display a number of diet shifts post-settlement (Hooker and Morse 1985, Tutschulte and Connell 1988), little is known of the role that algal chemical defence has on different stages of an abalone’s life cycle. This is the first demonstration of a size-dependant shift in feeding preference by a marine herbivore that is mediated by secondary metabolites. There is some variability between adult and juvenile preferences in fish feeding on extracts of algae (Paul et al. Chapter 4 Chemical defence against herbivores 91

1990), although in this case it was the adults that were more susceptible to the chemical defence. The change in consumption by abalone around 100 mg FW may represent an ontogenetic shift (Werner and Gilliam 1984, Gosselin 1997). Further evidence for this is that large Haliotis rubra are known to consume Asparagopsis armata (Shepherd and Steinberg 1992). An increase in abalone body size may correlate with a delayed development of certain digestive enzymes or gut flora. For larger abalone, such features may be required for a drift-feeding strategy (Hooker and Morse 1985), as a mixed diet would dilute the impacts of occasionally feeding on a chemically rich alga. Alternatively, small abalone are cryptic and predominantly consume attached algae (Tutschulte and Connell 1988). These small individuals should be more selective than adults, as the repercussion of eating a chemically defended food would be a higher dose (per unit weight) relative to consumption by an adult. The soft-bodied opisthobranch molluscs (e.g. sea hares) often consume chemically rich algae and sequester host plant chemistry for their own defence (Pennings 1990, Paul and Pennings 1991, Ginsburg and Paul 2001, Rogers et al. 2002). Considering the occurrence of the sea hare Aplysia parvula on Asparagopsis armata and other red algae that produce brominated metabolites (Rogers et al. 2002), it is interesting that bromide (+) algae were not the preferred food, i.e. that feeding was not stimulated by halogenated metabolites. But other sea hare species also show no preferences for seaweeds that have high concentrations of active metabolites (Ginsburg and Paul 2001). A lack of preferences suggests that sea hares may cue to other aspects of the algae or even arrive at random. However, it is possible that sea hares which actively choose defended algae could be beyond the size range examined, as changes in feeding preference can occur with age (Pennings 1990). Feeding tests of the halogenated metabolites from A. armata using the bromide (+) and bromide (-) manipulation demonstrated that they are effective deterrents of feeding but do not inform the specific metabolites involved. Assays with artificial diets also revealed that crude extracts of A. armata deterred consumers in the Chapter 4 Chemical defence against herbivores 92

field. However, assays with the two major compounds – bromoform and dibromoacetic acid [DBA] – gave mixed results in both the field and the laboratory. DBA had no significant effects on feeding in three of the four tests. Perhaps surprisingly, DBA only significantly deterred feeding by the sea hare, the herbivore that was not deterred from feeding upon bromide (+) algae. Other sea hares that appear to gain advantage by sequestration of algal natural products have also been shown to be deterred from feeding by extracts of the same seaweeds (Ginsburg and Paul 2001). It remains possible that DBA is responsible for some of the feeding deterrence in A. armata. However, bioassay-guided fractionation may reveal minor compounds that have been overlooked as a result of focussing on the major components (Kubanek et al. 2004). Bromoform, the major metabolite, did not deter feeding in any of the tests. Many algae produce and release bromoform (Nightingale et al. 1995), even highly preferred algae from the genus Ulva (Manley and Barbero 2001). For this reason bromoform may be ineffective as a herbivore deterrent. Furthermore, brominated compounds are not uniformly active as feeding deterrents (Kicklighter et al. 2004). But to my knowledge, neither DBA nor bromoform are present in any other alga at concentrations as high as that of A. armata and it is possible that these high levels relate to other types of defence (e.g. bacterial antifouling, Chapter 2). However, there were some methodological issues with the artificial diets that may have affected the efficacy of bromoform and DBA in the feeding assays. As metabolites will leach from artificial diets (Paul and Hay 1986), particularly when water soluble, it is difficult to quantify the relative importance of the major metabolites without determining leach rates. This would be compounded if any synergistic effects exist for the compounds (Hay et al. 1994). The small herbivores used in the feeding assays in the laboratory typically consumed little of either diet, such that, the autogenic changes in mass of control (minus herbivore) diets were relatively large. The analysis for pairwise-choice experiments is not tolerant of this variance (Peterson and Renaud 1989). Feeding assays that test the major metabolites across a range of concentrations and that quantify metabolite release from the diets are required. However, it is also Chapter 4 Chemical defence against herbivores 93

possible that the combination of the nutrition of the artificial diet (comprising a green alga) and that of these halogenated metabolites from a red alga, influenced the results with the pure compounds. This is why the feeding assays with bromide (+) and bromide (-) A. armata provide such compelling results for feeding deterrence by this filamentous alga, as these test are removed from any nutritional or structural constraints that arise with the use of artificial diets. More than 80 halogenated compounds are produced by A. armata (McConnell and Fenical 1977). This seemingly large number is a result of the varying degrees of halogenation and halogen combinations on simple carbon structures such as carboxylic acids, ketones and methanes. The efficacy of secondary metabolites, in general, cannot be categorised on a structure/function basis, as similar metabolites have varied effects on consumers (Hay et al. 1987a, 1987b & 1988a, Paul et al. 1990, Duffy and Paul 1992). Furthermore, stimulatory effects are not uncommon, even for compounds that are in other cases deterrent (Duffy & Paul 1992, Hay et al. 1987a & 1987b, Cruz-Rivera and Hay 2003). It is possible that the large variety of halogenated compounds in A. armata has evolved in response to a diverse range of natural enemies. Evidence for this is that although the major metabolite bromoform was not active against herbivores it is active against epiphytic bacteria (Chapter 2). Although many plants exist in environments with intense grazing, the traits that facilitate persistence may not be always clear, especially if a plant exhibits a combination of resistance and tolerance strategies (Mauricio 2000, Koricheva et al. 2004). Chemical defence and strategic growth is a feature of some marine algae (Hay et al. 1988b), and many marine organisms have multiple resistance traits, for instance a combination of chemical and structural defences (Paul and Hay 1986, Hill et al. 2005) or one of chemical and nutritional defences (Duffy and Paul 1992, Cruz-Rivera and Hay 2003). However, chemical defence in filamentous algae represents a largely unexplored facet of marine chemical ecology. If the requirement for defence is low and resources are adequate then algal communities may be dominated by high producers (Lubchenco and Gaines 1981). But as no environment is completely free of herbivores (Paine 1990), Chapter 4 Chemical defence against herbivores 94

a niche may exist for algae with relatively high growth rates (i.e. tolerance) and defence. For these organisms, even if grazing intensities were beyond the efficacy of their defences they may ultimately rely on the tolerance traits characteristic of the filamentous form. CHAPTER 5

Trade-offs between growth and chemical defence in the red alga Asparagopsis armata

Introduction

Resource allocation to secondary metabolites is critical to many plants for the mediation of plant-herbivore interactions (Feeny 1976) and there are a number of models that attempt to explain the evolution of plant chemical defences in the context of resource allocation (Herms and Mattson 1992, Stamp 2003). These models differ in their emphases on the degree to which variation in secondary metabolites is driven by environmental resources (Bryant et al. 1983, Coley et al. 1985, Herms and Mattson 1992) or by herbivore pressure (McKey 1979, Herms and Mattson 1992). However, despite their diversity, all of these models hinge on the assumption that the production of secondary metabolites is costly (Rhoades 1979, Herms and Mattson 1992, Simms 1992). That is, investment in defence leads to a reduction in plant fitness in the absence of natural enemies, as the resources allocated to chemical defences cannot be used for primary processes. Costs of chemical defence can manifest in a number of ways. The energy and material required to synthesise secondary metabolites (Gershenzon 1994), and the investment in specialised storage and transfer structures (McKey 1979, Bjorkman et al. 1992) represent allocation costs. However, costs are not solely a strict function of Chapter 5 Resource allocation 96

resource allocation. They may further be manifested as an opportunity cost if resources are diverted away from primary processes such as growth or reproduction. Ecological costs can occur if secondary metabolites also deter mutualists or attract other natural enemies (Strauss et al. 2002). Each of these costs can result in a reduction in plant fitness (Simms 1992, Bergelson and Purrington 1996). A negative correlation between the level of chemical defence and plant fitness in the absence of natural enemies is considered evidence for cost of resistance (Berenbaum et al. 1986, Mauricio 1998, Koricheva 2002). However, correlations are influenced by resource levels and other factors, and may not always be representative of allocation costs (Simms and Rausher 1987). Additional evidence for cost is that the concentrations of secondary metabolites are often higher in tissues most important to plant fitness (e.g. reproductive tissue) (McKey 1979, Rhoades 1979, Zangerl and Bazzaz 1992). For many species, however, costs of resistance have not been directly measured or perhaps do not exist (but see Koricheva (2002) for a meta-analysis that indicates cost when averaged across many studies). Some similar constraints for the assumptions of the terrestrial plant defence models exist in their application to marine systems. The secondary metabolites of marine algae provide a broad range of defences including herbivore deterrence, antifouling and allelopathy (Hay and Fenical 1988, Paul 1992, McClintock and Baker 2001). The concentration of these metabolites can vary across resource gradients (Yates and Peckol 1993, Steinberg 1995, Cronin and Hay 1996a, Puglisi and Paul 1997) but also within and among individuals (Steinberg 1984, Pavia et al. 2002, Jormalainen and Honkanen 2004, Toth et al. 2005, Dworjanyn et al. in press). The variation in levels of secondary metabolites is implicit of a cost for these chemical defences (Rhoades 1979). However, few empirical studies have confirmed that the secondary metabolites of marine algae are costly (Steinberg 1995, Pavia et al. 1999, Dworjanyn et al. in press). Brown macroalgae have received the most attention with respect to resource allocation and the cost of chemical defence in macroalgae (Steinberg 1984, Tuomi et al. 1989, Pfister 1992, Cronin and Hay 1996a & 1996b, Pavia et al. 1999, Van Alstyne et Chapter 5 Resource allocation 97

al. 1999, Pavia et al. 2002). Despite the diversity of natural products in red algae (Fenical 1975, Faulkner 1996, Paul and Puglisi 2004), comparatively little information exists on the chemical ecology of defence for these organisms, particularly with respect to cost (Puglisi and Paul 1997, Dworjanyn et al. in press). However, the red algae are attractive candidates for testing resource allocation and costs of chemical defence. Firstly, many species have simple filamentous body plans (Cole and Sheath 1990), for which growth is by apical meristems and growth rates are typically high. Secondly, the storage sites of ecologically active metabolites in many red algae are single specialised cells that are easily observed (Murray and Dixon 1992). Furthermore, the movement of material within individuals is limited by an absence of specialised transport structures (Cole and Sheath 1990). Because of these characteristics, any cost of producing secondary metabolites in the specialised cells of red algae is likely to be isolated to the cellular region in which they are formed. Interestingly, few studies on resource allocation in marine algae have considered allocation patterns involving chemical defences at the cellular scale (but see Dworjanyn et al. in press). Due to detection limits for quantitative measurements of secondary metabolites, the use of storage sites may prove to be a useful alternative measure for testing fine-scale aspects of resource allocation in marine algae. In this chapter, I examine resource allocation to growth or defence and the costs of secondary metabolite production in the tetrasporophyte of Asparagopsis armata, a filamentous red alga that produces halogenated metabolites. These metabolites are ecologically active (Chapters 2 & 4) and are stored in specialised structures known as gland cells (Chapter 3). Firstly, I examine variation in resource allocation to defence and growth across a gradient of light in the laboratory. I then test whether there is a trade-off between growth and secondary metabolite production, in particular determining if it is costly for individuals to produce halogenated metabolites under fixed levels of resources. I then consider the trade-off between growth and defence at the cellular scale by using the size of the specialised gland cells in A. armata as a surrogate measure of investment in defence. I subsequently integrate whole algal chemistry and cellular Chapter 5 Resource allocation 98

measurements and examine whether trends in allocation to chemical defence observed in laboratory experiments are consistent with field samples.

Methods

Study organism

Asparagopsis armata Harvey (Bonnemaisoniaceae) concentrates halogenated metabolites in specialised gland cells (Wolk 1969, Chapters 2 & 3, Fig. 5.1A). It has a heteromorphic life history that alternates between a filamentous tetrasporophyte and a pseudoparenchymatous, plumose gametophyte (Womersley 1996). The tetrasporophyte was used in all experiments in this Chapter. Asparagopsis armata has apical growth from a single apical cell. Upon apical cell division, an axial cell and three surrounding pericentral cells are produced (Fig. 5.1A). Each of the pericentral cells may contain a gland cell (Fig. 5.1A). These specialised gland cells store the major halogenated metabolites (Fig. 5.1B) which can comprise up to 5 % of algal dry weight [DW] (Paul et al. 2006, Chapter 2). This chapter uses the term “cell tier” to refer to the group of cells comprised of a single axial cell and three surrounding pericentral cells with their associated gland cells (Fig. 5.1A). Algae were collected from the shallow subtidal at Bare Island (33° 59’ 38” S, 151° 14’ 00” E) and Long Bay (33° 59’ 10” S, 151° 14’ 15” E), Sydney, Australia. The filamentous tetrasporophyte is found throughout the year at these sites. When tetrasporophytes were collected for culture, individuals were returned to the laboratory and small, apical sections were excised. Areas with obvious fouling were avoided. In all culture experiments, algae were grown in sterile seawater with ½ strength Provasoli Enrichment Solution (PES) (Provasoli 1968), and 10μg l-1 of germanium dioxide was added to reduce diatom growth. Antibiotics were not added as many red algae do not grow well in the absence of bacteria (Fries 1975). Previous experiments suggested that this is true for A. armata (N. Paul, unpublished data). Chapter 5 Resource allocation 99

Chemical analyses

Methods for quantifying the major halogenated natural products using gas chromatography – mass spectrometry (GC-MS) are outlined in detail in Chapter 2. In brief, there are five major non-polar metabolites (bromoform, dibromoacetic acid (DBA), bromochloroacetic acid (BCA), dibromochloromethane, and, 3, 3- dibromoacrylic acid) that are produced and stored at high levels in A. armata (~ 25 μg mg-1 DW). While more than 80 different metabolites are produced by A. armata (McConnell and Fenical 1977), I focussed on the major metabolites as extracted from freeze-dried material. Freeze-drying of samples provides a more efficient extraction of metabolites but may release volatile components. However, previous extraction of freeze-dried material versus wet material (Chapter 2) and the reports by McConnell and Fenical (1977) and Woolard et al. (1979) suggest that the non-volatile metabolites are the major natural products in this alga. Ion fragments characteristic of these metabolites were monitored in selective ion mode on the GC-MS. Concentrations were determined by reference to standard curves produced from pure compounds purchased from Sigma- Aldrich (bromoform and dibromochloromethane) and Chem Service Inc. (DBA and BCA); no standards were obtainable for 3,3 dibromoacrylic acid. All algae were freeze-dried, weighed and extracted in MeOH with 10 μg ml-1 naphthalene as an internal standard. Mass spectrometry was performed on a HP 5972 Mass Selective Detector (MSD). Standards were run at regular intervals throughout the analysis and samples were randomly ordered in the injection queue to ensure that there was no sampling bias. Masses of target compounds are expressed as mass of compound per unit surface area (μg mm-2), as measures were made with small pieces of algae. Chapter 5 Resource allocation 100

Growth and chemistry across a light gradient

It has been predicted that plants alter their investment in chemical defence in response to changes in available resources (Herms and Mattson 1992). As light is an important parameter for algal growth, I examined how growth and production of chemical defence in A. armata varied with changes in light. This was done by measuring changes in growth and the concentration of the major halogenated metabolite (bromoform) in four individuals at different light levels in a laboratory culture experiment. Replicates of each individual (N = 6 – 8) were grown in 5 mL Petri dishes under 4, 10, 20 and 60 μmol-1 photons m-2 s-1 (Note: μmol-1 photons m-2 s-1 has been abbreviated to μE for the following sections). Growth of sublittoral algae in the field is typically light limited below 100 μE (Lüning 1981). Algae were cultured in sterile seawater with the addition of ½ PES and the culture medium was changed weekly. Due to their small size (< 0.1 mg fresh weight [FW]), growth was quantified as changes to algal surface area over time. The projected area (surface area) of algae was measured on digital images captured with low magnification using a stereomicroscope. Area calculations were made with the image analysis software Image J (Scion Corporation). Growth was monitored weekly for 4 weeks, after which final measurements were made and the levels of the major halogenated metabolites in each replicate were analysed by GC-MS. Analysis of growth and bromoform levels were made by Model II 2-factor ANOVA with Light (N = 4) as the fixed factor and Plant (N = 4) as the blocking factor.

Correlations between growth and metabolite production in the laboratory

I subsequently examined whether growth and metabolite concentration co-varied among individuals. A negative correlation between these variables in the absence of natural enemies is supportive of an allocation cost for the secondary metabolites (Koricheva Chapter 5 Resource allocation 101

2002). The two light levels (low, 10μE and high, 35 μE) used to test for correlations across individuals were selected from the previous light experiment, in order that neither defence nor growth was limited. Light was assumed to be the major limiting factor as excess nutrients were added to the seawater. This is important as a cost of chemical defence may not be detected where the level of resources limits growth more than it limits photosynthesis (Herms and Mattson 1992). Plants (N = 15) were collected from two different sites (Bare Island and Long Bay) to minimise the chance of sampling clones. Algae were cultured from apical sections (initially 3 individual apices per plant) in the laboratory at two light levels, 10μE (low) and 35 μE (high). Algae were grown for 4 weeks in sterile seawater (+ ½ PES), with weekly changes to culture medium. Genetic correlations (Lynch and Walsh 1998) were calculated between growth and both the total and single metabolite concentrations using the mean values (N = 2-3) for each plant (N = 15). Bonferroni significance values are reported for the Pearson’s correlations. These data were further analysed by 2-factor ANOVA with light and plant as factors. This allowed me to examine whether there was an interaction between plant (genotype) and light level (environment). Growth rate, combined (total) and single metabolite concentrations were analysed separately by Model II ANOVA, with plant ( N = 15) as a blocking factor.

Use of gland cell size to inform allocation to defence

As A. armata is a small filamentous alga some of the finer aspects of resource allocation to defence, for example allocation between cells, cannot be examined by quantifying defence using the chemical methods described above. Information regarding the commencement of production of metabolites in the apex or trends in investment along filaments is also difficult to gather. Furthermore, whole plant quantification of secondary metabolites may not truly represent chemical defence trade-offs, especially if within-plant variation in investment exists as a result of age or growth (Herms and Chapter 5 Resource allocation 102

Mattson 1992) or if movement of resources within plants is limited – as is the case for red algae (Cole and Sheath 1990). I used the relative size of the gland cells compared to the somatic cells as a surrogate for halogenated metabolite levels. My rationale for the use of the gland cells in this way is as follows: previous ultrastructural work demonstrated that the gland cell contains a refractile inclusion that is filled with natural products (Chapters 2 & 3); the production of halogenated secondary metabolites is explicitly related to the presence of this inclusion in the gland cell. Thus, when A. armata is grown without bromine the halogenated metabolites are no longer produced and neither is the refractile inclusion characteristic of the gland cells (Chapters 2 & 3). These features suggest that measuring the size of the refractile inclusion will allow investment in chemical defence and growth along the axis of A. armata to be tested at the cellular scale. The size (volume) of the refractile inclusion in the gland cell was taken as a measure of allocation to defence and the size of the whole cell tier to represent allocation to growth (each “cell tier” is the axial cell and the surrounding three pericentral cells with their gland cells: Fig. 5.4B). Three measurements were required to calculate the volume of the each tissue type: pericentral cell length (Fig. 5.4B) and filament width at the centre of the cell tier were used to calculate cell tier volume, and the diameter of a single gland cell was used to calculate total gland cell volume (i.e. the volume of 3 gland cells) for each cell tier (Fig. 5.4B).

Resource allocation at the cellular scale in the laboratory

I then tested if the correlations between growth and chemistry in the laboratory light experiment could be explained by resource allocation at the cellular scale. Gland cell and cell tier sizes (volumes) were measured in algae cultured under 10 and 35 μE from the previous laboratory experiment. I divided the alga into three growth regions (youngest to oldest) from the apical tip (Fig. 5.4A): Apex (0-300 μm), Mid (700- 1000 μm) and Mature (1500-2000 μm). From these data I quantified the mean position Chapter 5 Resource allocation 103

of the first pericentral cell containing a gland cell from the apex and the mean percent volume occupied by gland cells in cell tiers from each region. Measurements were taken for 7 algae from each light treatment. Between 6 and 10 cell tiers were measured within each growth region (apex, mid and mature cells). ANOVA was used to analyse differences in the mean percent volume of gland cells (defence) within cell tiers (somatic growth). Light (10 and 35 μE) and Region (Apex, Mid and Mature) were fixed factors and Plant was the blocking factor in an unreplicated block design. Correlations between the percent of cell tier volume occupied by gland cells and cell tier volume were run separately, incorporating all cell tiers from each light level. Correlations across a spectrum of cell tier volumes would offer a different perspective on investment in defence and growth than that from the mean values of the three growth regions. Separate correlations with cell tiers only from the apical regions of algae from both light regimes were also run.

Comparison of laboratory-cultured and field-sampled algae

As the experiments described above were with laboratory cultures, I examined whether the cellular trends were relevant to algae from the field. Although the tetrasporophyte is filamentous, individual balls or tufts are formed that are comprised of intermingled filaments up to 5 g FW. I collected separate individuals of A. armata (N = 13; between 1 – 5 g FW) from the heterogenous algal turfing community in the field and separated each individual into two types of tissue, an inner and an outer portion. These two types of tissues should be exposed to different environmental resources, as resources available to the inner portion of an individual will differ to that received by the outer portions (for example, self-shading will reduce light quantity that reaches the inner portion). Inner and outer tissues were sampled by taking a peripheral sample and a central sample of an individual tuft, respectively. Chapter 5 Resource allocation 104

I could not directly compare the two tissue treatments from the field with the two light treatments in laboratory (i.e. high light z outer tissue and low light z inner tissue). However, I have previously shown that the concentration of halogenated metabolites in laboratory cultures of the algae were within the range of that found in the field (Chapter 2). I now compare the branching rate in apical sections and the range in size of the cell tiers in both sets of algae from the laboratory and field to further examine whether the laboratory growth results are relevant to the biology of the field samples.

Cellular correlates between defence and growth in the field

1) Defence-Growth correlates for outer and inner tissues

The cell sizes of filaments sampled from two tissue types (outer and inner as per above) from 7 individuals were measured at three different regions; Apex (0-500 μm), Mid (1400-1900 μm) and Mature (2800-3300 μm). Measurements for the field samples were taken further along the axis as the individual filaments were much larger in length than those cultured in the laboratory. The relative size of gland cell to cell tiers was correlated to the size of the cell tiers along the filament. Separate correlations were made for inner and outer tissue. Further correlations were made using only those cells from the actively growing apical regions of both inner and outer tissues.

2) ANOVA of mean values for defence-growth trade-off

Mean percent volume of cell tiers occupied by gland cells for each growth region in field samples (inner and outer) were calculated. Measurements were made on two filaments for each region (apex, mid and mature) of each tissue type in 7 individuals and the means of these two values was used in analysis. The percent volume of gland cells Chapter 5 Resource allocation 105

inside cell tiers was analysed by Model II 3-factor ANOVA with Tissue and Region as fixed factors, and Plant as a blocking factor. This was an unreplicated block design with no estimate of the Tissue x Region x Plant interaction (Quinn and Keough 2002). In order to assess whether a single replicate for each region within the blocking factor (Plant) could be used, ANOVA was used to test the potential interaction between Plant (blocking factor) and Tissue (fixed factor). A 2-factor ANOVA was run with data obtained by subtracting the percent volume of the gland cell in cell tiers in the apical region from the percent volume of the mature region in the same filament. Two filaments were used per individual (N = 7). As there was no significant interaction between tissue and plant ( F1,19 = 0.003, P = 0.956), an unreplicated block design was acceptable. I also measured the position from the apical cell of the first cell tier (i.e. pericentral cell) that contained a gland cell. Comparisons between inner and outer tissue were made using a paired t-test.

Statistical analysis

Experimental designs are outlined in each respective section. All data were analysed using Systat 10 (SPSS). The assumptions for homogeneity of variance and normality required by ANOVA were verified by examining the scatterplots of residuals versus mean values and histogram of residual values, respectively (Quinn and Keough 2002). Transformations were made where appropriate. Multiple comparisons for ANOVA for were made using Tukey’s HSD at a significance level of P = 0.05. Chapter 5 Resource allocation 106

Results

Growth and chemistry across a light gradient

Light strongly influenced both the growth rate (P < 0.001) and concentration of bromoform ( P < 0.001) in A. armata cultured in the laboratory (Table 5.1), as increasing light levels lead to higher growth rates and higher concentrations of bromoform (Fig. 5.2). There was no significant Light x Plant interaction for bromoform levels, however, a marginal value for the interaction in the analysis of growth rates (P = 0.051) indicates that individuals may respond differently to changes in light levels.

Covariance between growth and metabolite production in the laboratory

Significant genetic correlations between growth rate and total metabolite concentration existed for both low and high light levels but differed in sign (Fig. 5.3A). Under low light, total metabolite concentration negatively covaried with growth (Pearson’s coefficient = -0.57, P = 0.014). The decreasing growth rate with increasing concentrations of metabolites across plants is indicative of an allocation cost for halogenated secondary metabolite production. Under high light, there was a positive correlation between growth and metabolite production (Pearson’s coefficient = 0.44, P = 0.013). There was also variation in the sign of genetic correlations among the correlations of single metabolite concentrations against growth rate (Fig. 5.3B-F). The two major metabolites (as a proportion of unit area) in the alga were bromoform and DBA. Levels of the major metabolites in the laboratory cultures were again dominated by bromoform, accounting for approximately 60 % of the total load in both low and high light treatments. Under low light, the concentrations of three metabolites negatively covaried with growth, including bromoform (P = 0.001, Fig. 5.3B), CHBr2Cl (P = Chapter 5 Resource allocation 107

0.044, Fig. 5.3C) and dibromoacrylic acid (P = 0.004, Fig. 5.3D). DBA and BCA showed no significant correlation. When algae were cultured under high light, three metabolites mirrored the positive correlation between growth and metabolite concentration observed for total metabolite levels, including bromoform (P = 0.009, Fig. 5.3B), DBA (P = 0.011, Fig. 5.3D) and BCA (P = 0.001, Fig. 5.3E). The correlation between dibromoacrylic acid and growth remained negative (P = 0.026, Fig. 5.3F). No correlation existed for CHBr2Cl (Fig. 5.3C). ANOVAs testing variation in halogenated metabolite concentrations between the two light levels showed that light had a significant effect in most univariate tests. There were no significant interactions between the two factors of Plant (genotype) and Light level (environment) in any of the analyses. Algae cultured under the high light treatment had higher growth rates ( F1,14 = 97.96, P < 0.001) and higher total levels of metabolites

( F1,14 = 21.92, P < 0.001), as well as higher levels of bromoform ( F1,14 = 21.00, P <

0.001), CHBr2Cl ( F1,14 = 6.77, P = 0.012) and DBA ( F1,14 = 5.15, P = 0.04).

Dibromoacrylic acid had lower levels in the high light treatment than the low ( F1,14 = 16.96, P < 0.001). BCA showed no difference between treatments.

Variation in cellular allocation to defence and growth at two light levels

Apical regions had significantly lower mean volumes of the cell tier occupied by gland cells than both mid and mature regions in these laboratory cultures (Table 5.2, F2,12 = 12.36, P = 0.001, Fig. 5.5). There was no difference between light levels and no interaction between the light level and region (Table 5.2). These cellular data provided a means to discuss the increased levels of metabolites with increasing light from the previous experiment in the context of changes to algal biology along the growth axis. Despite the trend for an increase in mean percent of gland cells from the apex to the mid and mature growth regions in the ANOVA (Table 5.2), there were no significant positive correlations between the size of the cell tier and the percent of the cell tier that was occupied by the gland cells (Fig. 5.6A&B). If only cell tiers from the apical section Chapter 5 Resource allocation 108

are considered, then a negative correlation exists with increasing cell tier size for apices from both low and high light Fig. 5.6C&D). These negative correlations at the apex along the axis of growth (increasing cell tier size) show that there are small-scale trends overshadowed using mean values for the different regions in the ANOVA. This result means that the size of the gland cells in recently formed cells is higher than the larger (and older) cells further along the axis. However, as these cell tiers age the size of the gland cell relative to the somatic cell must increase in the mid and mature growth regions (i.e. for cell tiers >50000 μm3 in volume) consistent with the higher relative levels of defence in the above ANOVA. While the density of pericentral cells containing a gland cell is high beyond the apex (>97%, Chapter 3), there was some variation in the distribution of gland cells at the apex and sub-apical cells. For instance, no gland cell was present in the apical cell itself. The first pericentral cell (or cell tier) that contained a gland cell was three to four cells from the apical cell and this did not differ between low and high light treatments (2- sample t-test, P = 0.198).

Comparison of field-sampled and laboratory-cultured algae

The mean level of metabolites in algae from the laboratory experiments was higher in the high light treatment than in the low light treatment. The field samples also differed in total metabolite levels between the two tissue types (outer and inner). Outer (peripheral) samples had 1.5 times the amount of halogenated metabolites than the inner (central) samples (paired sample t-test, P = 0.042). The major metabolite on a dry weight basis was bromoform which comprised 62 % ± 17 (mean ± SD) of the total load. This was also the case in the laboratory experiment. No direct measurements of the quanta or quality of light were made to differentiate between the two tissues (inner and outer) sampled from individuals in the field. However, contrasts of the biology of the inner and outer treatments for field individuals were similar to those between the low and high light treatments in the Chapter 5 Resource allocation 109

laboratory. For instance, there were significant differences in the branching rates between each treatment in the laboratory (Paired sample t-test, P = 0.024) and also between each treatment in the field (F1,14 = 16.19, P = 0.026). The mean branching rates were similar for high light (35 μE; 2.22 branches per 20 cells ± 0.34 SEM) and outer tissues (1.9 branches per 20 cells ± 0.25 SEM) and also similar for low light (10μE; 0.91 branches per 20 cells ± 0.26 SEM) and inner tissues (1.1 branches per 20 cells ± 0.20 SEM). Furthermore, most of the cell tier sizes in each of the four treatments did not exceed 10 4 μm3 in volume (Figs. 5.6 & 5.8), although the upper maxima of the cell tier sizes was higher in the high light laboratory cultures than for the other three treatments (Fig. 5.6).

Allocation of resources along the growth axis in field samples

1) Correlations along the axis of growth

Correlations between the percentage volume of each cell tier occupied by gland (defence) cells and the actual cell tier volume differed between the inner and outer tissue (Fig. 5.7). Filaments sampled from the inner portions of the algae showed a significant decrease in the relative volume of gland cell to somatic cells, although it was not a strong correlation (Pearson’s coefficient = -0.129, P = 0.014, Fig. 5.7A). There was no significant correlation between the relative size of gland cells to somatic cell volume in filaments sampled in outer tissue, although this was tending positive (Pearson’s coefficient = 0.074, P = 0.185, Fig. 5.7B). In the inner tissues of an individual the cell tiers in the apical section had higher values of defence relative to somatic tissue (Pearson’s coefficient = -0.26, P < 0.001; Fig. 5.7C), as was also the case for apical cell tiers from the outer tissue (Pearson’s coefficient = -0.21, P = 0.009; Fig. 5.7D). Chapter 5 Resource allocation 110

2) ANOVA of the defence-growth trade-off

There was a significant interaction between Tissue and Region (P < 0.001 , Fig. 5.8) in the ANOVA on size of the gland cells as a percent of the cell tier volume (Table 5.4). This interaction was a result of two main features: (1) in the cell tiers of the apex, gland cells occupied higher amounts of the volume of cell tiers from the inner filaments (low resource) than in those from outer filaments (high resource); (2) in cell tiers of mature sections, the gland cells occupied higher amounts of volume in the outer filaments compared with the inner filaments (Fig. 5.8). There was no difference in the position of the first (sub-apical) pericentral cell that contained a gland cell (paired sample t-test, P = 0.067). The pericentral cell was on average part of the cell tier positioned 3.3 cells (± 0.33 SE) beneath the apical cell. Gland cells were observed in the first sub-apical pericentral cell and sometimes were not produced until the 5th sub-apical pericentral cell. However, gland cells were never observed in the apical cell itself. Chapter 5 Resource allocation 111

A

p g

a g p p

B Halogenated compounds

i) CH Br3 (bromoform)

ii) CH Br2 – COOH (dibromoacetic acid)

iii) CH Br2 Cl (dibromochloromethane) iv) CH Br Cl – COOH (bromochloroacetic acid)

v) C Br2= CH- COOH (3,3 dibromoacrylic acid)

Figure 5.1. Location of the gland cells in Asparagopsis armata (A) and the compounds (B) that are stored within the refractile inclusion. (A) Transverse section of the tetrasporophyte of A. armata. The filament is comprised of a single axial cell (a), three pericentral cells (p) and gland cells (g) within each of the pericentral cells (only 2 shown). Scale bar, 10 μm. (B) Major halogenated metabolites in A. armata: i) bromoform, ii) dibromoacetic acid, iii) dibromochloromethane, iv) bromochloroacetic acid, v) 3,3 dibromoacrylic acid. Chapter 5 Resource allocation 112

A. 6 d 5 )

2 c

4

3

2 b

Total growth (mm Total growth a 1

0 4 μE 9 μE 18 μE 60 μE Light level

B. 3.0 d 2.5 ) -2 2.0

(μg mm 1.5 3 c 1.0 CHBr b 0.5 a

0 4 μE 9 μE 18 μE 60 μE Light level

Figure 5.2. Differences in mean total growth over 20 days (A) and mean bromoform concentration (B) (means ± SE, N = 4) across four light levels in laboratory cultures of the tetrasporophyte of A. armata. Bars sharing the same letter do not differ at P = 0.05 (Tukey’s HSD). Chapter 5 Resource allocation 113

A. B. ) -2 7 5 ) -2 g mm

P 6 g mm

P 4

5

3 4

3 2

2

1 Low Light (10 PE), P = 0.014 1 Low Light (10 PE), P = 0.010 High Light (35 PE), P = 0.013 High Light (35 PE), P = 0.009 Mean Bromoform concentration ( Mean TotalMean ( concentration metabolite Significant correlation Significant correlation 0 0 0123 01232 Growth ( mm2 ) Growth ( mm )

C. D. 0.12 2.0 ) -2 ) -2 0.10 g mm P g mm

P 1.5 0.08

0.06 1.0 Cl concentrationCl ( 2 0.04

0.5

0.02 Low Light (10 PE), P = 0.044 ( concentration DBA Mean Low Light (10 PE), P = 0.790 Mean CHBr High Light (35 PE), P = 0.866 High Light (35 PE), P = 0.011 Signficant correlation Significant correlation 0.00 0.0 01230123 Growth ( mm2 ) Growth ( mm2 )

E. F. 0.05 ) G -2 ) -2 Low Light (10 PE), P = 0.004 0.04 F High Light (35 E), P = 0.026

g mm P

P Signficant correlation

0.03 E

0.02 D

0.01 C Mean BCA concentration ( concentration BCA Mean 0.00 Low Light (10 PE), P = 0.362 B High Light (35 PE), P = 0.001 Significant correlation

Mean Dibromoacrylic concentration (units mm Mean Dibromoacrylic A 01230123 Growth ( mm2 ) Growth ( mm2 )

Figure 5.3. Genetic correlations between metabolite concentration [total (A), bromoform (B),

CHBr2Cl (C), DBA (D), BCA (E) and dibromoacrylic acid (F)] and total new growth (over 4 weeks). Significant correlations (P<0.05) are indicated by the regression lines in the figures, N = 15. Note: dibromoacrylic acid could not be standardised as a concentration per unit area. Chapter 5 Resource allocation 114

A.

Apex

Mature

Mid

B.

gland cell diameter cell tier length cell tier breadth

Figure 5.4. (A) Habit of the tetrasporophyte of A. armata. Different regions of growth (Apex, Mid and Mature) are highlighted along one axis. (B) A single “cell tier” in a filament. Each cell tier is comprised of a longitudinal plane (the pericentral cell each represent cell tier length) and a transverse plane (width across the pericentral cells and axial cell gives cell tier breadth). The gland cells are effectively spherical, and their volume can be determined by measuring the size of the refractile inclusion (gland cell diameter). Scale bars, A = 200 μm B = 10 μm. Chapter 5 Resource allocation 115

4 b b 3

a 2

1 Percent cell tier volume cell Percent (%) by gland cells occupied

0 Apex Mid Mature

Growth region

Figure 5.5. Mean percent volume (± SE) of the algal filament that is occupied by gland cells in the three different growth regions. Results are the combined values for algae cultured under low and high light (10 μE and 35 μE). Bars sharing the same letter do not differ at P = 0.05 (Tukey’s HSD). Chapter 5 Resource allocation 116

A. B. 10 10 10 μE 35 μE 8 8

6 6

4 4

2 2 % volume occupied by gland cell % volume occupied by gland cell

0 0 50000 100000 150000 200000 250000 50000 100000 150000 200000 250000 Cell tier volume (μm3) Cell tier volume (μm3)

C. D. 10 10

8 8 10 μE 35 μE

6 6

4 4

2 2 % volume occupied by gland cell % volume occupied by gland cell

0 0 10000 20000 30000 40000 50000 10000 20000 30000 40000 50000 Cell tier volume (μm3) Cell tier volume (μm3)

Figure 5.6. Relative change in size of gland cells compared with the pericentral cell as pericentral cell size increases for the two different light levels; (A) low light (10 μE) and (B) higher light (35 μE). No correlations existed across regions for algae cultured under the low light level (Pearson’s coefficient = -0.03, P = 0.651, N = 192), nor for algae grown under higher light (Pearson’s coefficient= 0.085, P = 0.211, N = 221). (C) & (D) For cell tiers only from the apical sections (under 50000 μm3) the trends showed a decrease in % occupied by gland cell with size (i.e. with cell age); 10 μE apex, Pearson’s co-efficient = -0.502, N=95, P < 0.001; 35μE apex, Pearson’s co-efficient = -0.512, N = 118, P < 0.001. Chapter 5 Resource allocation 117

A. B. 12 10

10 Inner 8 Outer

8 6

6

4 4

% occupied by gland cell 2 2 % occupied by gland cell

0 0 20000 40000 60000 80000 100000 20000 40000 60000 80000 100000

Cell tier volume (μm3) Cell tier volume (μm3)

C. D. 10 10 Inner Outer 8 8

6 6

4 4 % occupied by gland cell

2 % occupied by gland cell 2

0 0 20000 40000 60000 20000 40000 60000 Cell tier volume (μm3) Cell tier volume (μm3)

Figure 5.7. Relative change in size of gland cells as a percent of cell tier volume across cell tier length for the two different tissues, (A) Inner tissue and (B) outer filaments, from field collected tetrasporophytes. Correlations indicated that there is a slight decrease in ratio with size (i.e. with cell age) for internal tissue (P = 0.014, N = 366), however for external tissue there were no differences (P = 0.185, N = 322). Inner tissue had higher values in younger parts (i.e. smaller pericentral cells) relative to older parts. (C) & (D) For cell tiers only from the apical sections (under 60000 μm3) the trends showed a decrease in % occupied by gland cell with size (i.e. with cell age); Inner apex cell tiers, Pearson’s co-efficient = -0.26, P < 0.001, N = 184; Outer apex cell tiers, Pearson’s co-efficient = -0.21, P = 0.009, N = 157. Chapter 5 Resource allocation 118

5

4

Growth region

3 Apex Mid Mature

2 % occupied by gland cell

1 Inner Outer

Tissue

Figure 5.8. Interaction plot of mean percent volume (±SE) of the alga (cell tier) that is occupied by gland cells in two different tissues (inner and outer) at three different regions along the filament (apex, mid and mature). There

was a significant Tissue x Region interaction (F2,12= 16.86, P < 0.001). There was a trend within both tissues (between apex, mid and mature) for older (mid and mature) cell tiers to have higher proportions of volume occupied by gland cells than young (apical) cell tiers. Chapter 5 Resource allocation 119

Table 5.1 Results for ANOVAs on growth rate and metabolite concentration in A. armata (N = 4) cultured under four light levels. All data was log-transformed. Significant results at P < 0.05 are in bold.

Growth rate Bromoform concentration

Source df MS F P MS F P

Light 3 20.709 280.90 <0.001 61.478 105.89 <0.001 Plant 3 0.645 16.95 <0.001 1.961 5.04 0.002 Light x Plant 9 0.074 1.94 0.051 0.581 1.49 0.156 Error 148 0.038 0.389

Table 5.2. Resource allocation along an axis in laboratory cultures. Results for ANOVA on percentage of cell tier volume that is occupied by gland cells in algae cultured under high and low light. Main effects only are listed. F-ratio for light was calculated with 1 & 6 df with Error MS of 0.586. F-ratio for region was calculated with 2 & 12 df with Error MS of 0.613. Factors for which there are no F-tests (Quinn and Keough 2002) are not listed. Significant results at P < 0.05 are in bold.

Source df MS F P

Light 1 0.960 1.64 0.248

Region 2 7.587 12.36 0.001

Light x Region 2 1.098 2.75 0.104

Error 12 0.399 Chapter 5 Resource allocation 120

Table 5.3. Resource allocation along an axis in field samples. Results for ANOVA on percentage of cell tier volume that is occupied by gland cells in inner and outer tissues of field sampled plants. Error is reported as Tissue x Region x Plant. Factors for which there are no F-tests (Quinn and Keough 2002) are not listed. Significant results at P < 0.05 are in bold.

Source df MS F P

Tissue 1 0.235 0.20 0.668

Region 2 9.609 18.12 <0.001

Tissue x Region 2 1.157 16.86 <0.001

Error 12 0.343 Chapter 5 Resource allocation 121

Discussion

Both growth rate and production of secondary metabolites in the red alga Asparagopsis armata increased in response to increasing light availability. However, the relationship among individuals changed between different resource levels. For individuals at a low light level, there was a negative correlation between growth rate and total levels of the halogenated metabolites. This trade-off between growth and defence is evidence for an allocation cost to secondary metabolite production at low resource levels. However, at the higher light level, individuals that grew faster had higher concentrations of secondary metabolites. The difference in sign for the correlations between growth and whole algal metabolite concentration under low and high light was explained by the accumulation of metabolites in the gland cells of non-apical (older) cell tiers of algae from the high light treatment. A combination of chemical and cellular analyses of secondary metabolite production proved essential to explaining the patterns of resource allocation in A. armata.

Trade-offs between growth and defence in Asparagopsis armata

Asparagopsis armata is a useful organism for testing hypotheses regarding resource allocation between growth and secondary metabolite production in the context of the plant defence models (Rhoades 1979, Herms and Mattson 1992, Stamp 2003). It is a fast-growing, filamentous alga that produces large amounts of secondary metabolites which are stored in specialised cells throughout its tissue. The existence of an allocation cost for secondary metabolite production is a major assumption for all plant defence models (Rhoades 1979, Herms and Mattson 1992, Stamp 2003). However, these costs have not always been identified for chemical defences in terrestrial plants (Simms and Rausher 1987, Simms 1992). Chapter 5 Resource allocation 122

The allocation of resources to secondary metabolite production reduced the fitness (growth) of Asparagopsis armata under low light in the absence of natural enemies. Cost was manifested as a negative genetic correlation between growth rates and whole algal concentrations of secondary metabolites across individuals. Relatively few empirical studies on marine algae have demonstrated costs of production of chemical defence (Steinberg 1995, Pavia et al. 1999, Dworjanyn et al. in press). The present lack of evidence for such costs in marine algae represents one limitation for the use of the terrestrial plant defence models in marine systems. Numerous studies attempting to correlate changes in growth with levels of chemicals in terrestrial plants have proven unsuccessful (see Simms 1992). A lack of identifiable costs could be a result of i) the types of compounds quantified (Koricheva 2002, Riipi et al. 2002); ii) that not all chemical defences have been quantified (Koricheva 1999, Riipi et al. 2002); iii) that concentrations are not the appropriate unit for investigation (Koricheva 1999); or iv) that the measure of fitness is not tightly coupled to the production of these secondary metabolites (Simms 1992, Koricheva 2002). For example, single metabolites in A. armata had varying degrees of costs associated with them. Some were negatively correlated with growth under low light

(bromoform, CHBr2Cl and dibromoacrylic acid) but others showed no such correlation (DBA and BCA). Variation in the trends of different classes of chemicals has been explored in terrestrial systems (Bryant et al. 1984, Koricheva 1999, Riipi et al. 2004). However, most of the studies on marine algal chemical defences have tested brown algal phenolics (Steinberg 1984, Van Alstyne et al. 1999, Pavia et al. 1999, Arnold and Targett 2003), while the dynamics of many other classes of compound are largely unknown. Evidence for costs of halogenated metabolite production across individuals of A. armata was not detected at the high light level. In this case, growth and total metabolite level positively covaried across individuals. Positive correlations between growth and secondary metabolites have been found in other algae (Dworjanyn et al. in press) and also for terrestrial plants (Riipi et al. 2002). The presence of positive Chapter 5 Resource allocation 123

correlations could be interpreted as a synergistic effect of growth on defence production or vice versa (Arnold and Targett 2003). However, cellular analyses of the structures involved in storage of secondary metabolites in A. armata revealed that the positive correlation may be explained by an increase in the relative amount of resources that is invested in growth and defence along the filamentous axis of growth.

Cellular correlates of allocation to chemical defence

I measured cellular aspects of chemical defence within A. armata using the percentage gland cell volume in cell tiers as a proxy for secondary metabolite levels. This demonstrated that the mean percent volume of the cell tier occupied by gland cells was significantly lower in the apical region than the two older regions (mid and mature). If a cell tier has reached its maximal size (or even reduced its expansion rate) then the resources available to defence may increase, such that the relative amounts of defence to somatic tissue increases with age along a filament. A similar trend was evident within individuals cultured under low light. Given there was a negative correlation between growth and metabolite concentration at low light, the ratio of young (growing) cells to old cells in individuals must be larger in algae cultured under low light than those cultured under high light. The relationship between growth region and tissue type (resource levels) was more complex for field samples, which is likely an effect of variables other than light. In the field, an interaction existed between growth region and tissue, although again older cells had higher relative values of defence. The consistency of this trend in field samples indicated that it was not an artefact of laboratory cultures. Correlations between the cell tier size and its percentage volume occupied by gland cell were negative across cells sampled only from apical regions. A negative slope implies that the concentration of metabolites is higher in smaller (younger) cells, despite greater opportunity costs that would be expected for these cells (Herms and Mattson 1992). This relationship changes as cells become older, as was demonstrated by the higher mean values of mid and mature sections than the values for the apex. Again, a Chapter 5 Resource allocation 124

similar trend existed in field samples of A. armata. The difference in slopes between tissues from the field implies that the relative investment in defence compared to growth varies with respect to resource levels. Changes in allocation to secondary metabolite production with respect to growth stage are important to resource-based defence models, in particular the growth differentiation balance hypothesis (Herms and Mattson 1992). For A. armata, however, the differences may be a result of ontogenetic changes rather than actual changes in allocation to defence (i.e. changes to investment do not relate to optimal defence) (McConnaughay and Coleman 1999). For instance, the gland cells in A. armata may continue to accumulate material rather than receive the benefits of a “shift” in resource allocation to the production of defence. Determining the effects of ontogenetic variation in levels of defence will be difficult with respect to optimal defence and resource-based models, especially if interactions between the age of cell tiers, their size and possible physiological constraints on secondary metabolite production exist. The size of the gland cells (defence) relative to the size of the surrounding somatic cells (cell tier) allowed me to test aspects of chemical ecology at a cell-to-cell scale. Most previous studies on resource allocation in marine algae have not considered allocation patterns at this scale, although in some cases different tissues have been examined with respect to growing and non-growing regions (Cronin and Hay 1996a, Dworjanyn et al. 1999) or growth and differentiation (Steinberg 1984, Pfister 1992, Van Alstyne et al. 1999). Allocation between cells (such as diffusion of resources) may be important for red algae due to their limited capacity to translocate material throughout individuals (Schmitz 1981, Cole and Sheath 1990). Costs of defence in A. armata may be ameliorated by reducing the demand for resources by gland cells but also by varying defence investment within plants (Rhoades 1979). For instance, the gland cells of A. armata have chloroplasts (Chapter 3) and may not rely on the associated somatic (pericentral) cells for their photosynthetic products. This would mean that cells in apical regions (i.e. fast growing cells) can invest in defence with reduced allocation costs, perhaps as a basal amount of defence in new Chapter 5 Resource allocation 125

growth is necessary (Rhoades 1979). Furthermore, the production of secondary metabolites is ultimately delayed, as neither the apical cell nor the immediate subapical pericentral cells contain gland cells. If there are no physiological constraints for metabolite production, then this delayed investment in defence at the apex also implies a cost of producing secondary metabolites in A. armata (McKey 1979, Rhoades 1979). There were three different measures of allocation trade-offs in A. armata: the negative correlation across plants (indicating costs in low light), the within-plant variation in mean values across growth regions, and the within-plant variation across cells from the apical region. Taken together, they show that while it may cost to produce compounds at the apex for all individuals, fitness costs will only manifest across plants (i.e. summed across all growth regions). However, any increase in growth rate for older individuals as a result of lower defence investment in the early, important axial growth (i.e. cost amelioration) could be concealed if measuring whole plant chemistry. This would mean that the costs of defence may be difficult to detect in adult populations if entire individuals are sampled (but see Pavia et al. 1999).

Application of plant defence models to A. armata

Since their inception, plant defence models have rarely avoided controversy (Hamilton et al. 2001). Most studies invoking these models have fitted their data post-hoc, with varying degrees of success. This is perhaps due to difficultly in directly testing any of the plant defence models (Stamp 2004), but may also result from misinterpretations of their predictions (Stamp 2003), or that the study variables are not appropriate (Koricheva 1999). Arguably, if the predictions of a model are not met on numerous occasions then this should preclude its use, as has been suggested for the carbon-nutrient balance hypothesis (Hamilton et al. 2001). The resource-based plant defence models have been applied to marine algae with mixed results; some compatible (Cronin and Hay 1996b) and others not (Yates and Peckol 1993, Cronin and Hay 1996a & 1996b, Puglisi and Paul 1997). Resource-based Chapter 5 Resource allocation 126

models are important to the production of defence in A. armata, considering that growth and defence levels vary in response to resource levels in both laboratory and field. The accumulation of chemical defence in more mature growth regions in A. armata is consistent with one of the major predictions of the growth-differentiation balance hypothesis (GDBH) (Herms and Mattson 1992, Stamp 2003). Furthermore, outer tissues of field samples were relatively less defended in the apex than in older cell tiers compared to the inner (low resource) tissue. The GDBH predicts that a plant in a high resource environment invests relatively less in the defence of fast growing parts and more in the slow growing parts compared to a lower resourced plant. This also appears to be the case for field samples of A. armata. However, the negative correlations between defence and growth across cells in the apical region contradict the GDBH, and perhaps are more suited to an optimal defence argument for the protection of new growth (Rhoades 1979). Most tests of optimal defence in marine algae have been on large algae with distinct morphologies (Steinberg 1984, Tuomi et al. 1989, Van Alstyne et al. 1999, Pavia et al. 2002). These and similar studies that have examined within-plant variation of algal defences have focussed on the trend for higher concentrations of the defences in plant parts more important to fitness (Rhoades 1979). Higher levels of defence in reproductive structures relative to somatic tissues in marine algae conform to predictions of optimal defence (Steinberg 1984, Pfister 1992, Van Alstyne et al. 1999; but see Tuomi et al. 1989, Pavia et al. 2002 for examples that do not). The filamentous form of A. armata does not have distinct tissues but does vary within plants with respect to cell size and age. However, it is unlikely that the higher concentrations present in older cell tiers of A. armata represent an evolutionary response to herbivores that may target these regions.

Marine versus terrestrial systems

In order to demonstrate that the plant defence models are applicable to marine algae, more evidence is required for their basic assumption of cost for chemical defence. Most Chapter 5 Resource allocation 127

work on cost in marine algae has involved constitutive defences, however, the constitutive defences of terrestrial plants often do not negatively correlate with a measure of plant fitness (Simms 1992, Bergelson and Purrington 1996). It is possible that, as for terrestrial plants, strong evidence for costs of secondary metabolite production will come from inducible defence systems (Baldwin et al. 1990). Recent advances with the induction of chemical defences in marine algae (Pavia and Toth 2000, Rohde et al. 2004, Toth et al. 2005) should provide more thorough tests for the assumption of cost. Biological differences between algae and terrestrial plants may influence predictions of plant defence models, perhaps for the resource-based models (which predict changes in secondary metabolite production over ecological time) more than optimal defence (in which variation in secondary metabolites is driven over evolutionary time). A model that incorporates the transport constraints throughout algae could also incorporate the need to produce some defence in early growth (sensu optimal defence) but with the majority of production occurring further along the axis (sensu GDBH). An age-based model of defence in A. armata could be more suitable, especially as changes in age and size of cell tiers appear to affect the relative investment in chemical defence more than actual shifts in allocation, as suggested by the GDBH. As foliose red algae are essentially comprised of filamentous building blocks, this pattern of growth and allocation may be applied to other red algae despite superficial differences in thallus form. There may be difficulties, however, in comparing the allocation of resources between algal divisions, as some brown algae do have rudimentary transport systems (Lee 1999). Terrestrial chemical ecologists have long been aware that defence metabolites may not solely be secondary in function (Seigler and Price 1976, Berenbaum 1995). This notion is particularly important for correlative and other indirect measures of the cost for chemical defence, and has received recent attention in macroalgae (Arnold and Targett 2003). For example, brown algal phenolics have anti-herbivore properties (Steinberg 1984, Toth et al. 2005) but also primary roles in cell wall construction and Chapter 5 Resource allocation 128

metabolic turnover (Schoenwaelder 2002, Arnold and Targett 2003). If plant secondary metabolites evolved from metabolic waste products (Whittaker and Feeny 1971), then the possibility remains that some metabolites still function in waste metabolism (Seigler and Price 1976). Waste metabolism has also been suggested as a role for the halogen metabolism of many red algae (Pedersén et al. 1996). However, most algae do not concentrate halogenated metabolite production in specialised cells. If an alga does not have specialised cells or structures where metabolites can be localised, then it may be difficult to separate allocation to chemical defence from allocation to primary metabolism, as is required by the plant defence models.

Conclusions

The cost of allocating resources to secondary metabolite production in Asparagopsis armata was manifested at different scales. Integrating results for resource manipulations across individuals with the cellular analyses of growth and defence trade-offs enabled a thorough insight into resource allocation to chemical defence in this alga. The cost of chemical defence was only manifested as a reduction in growth of algae that produced higher levels of secondary metabolites when cultured under low light. However, it is likely that costs are still present under higher resource levels, but are not observed at the scale of whole plants due to the accumulation of metabolites in mature cell tiers of larger individuals. Within-plant distribution of gland cells provided inferential evidence for costs, in particular, the delayed formation of secondary metabolites at the apices of filaments and the accumulation of metabolites in older cells beyond the apical region. If similar trade-offs exist between defence and growth at the cellular scale in other algae, then whole plant concentrations of metabolites should not be the only scale at which the plant defence models are applied. Size and age could both be incorporated into the predictions for chemical defence of algae with apical growth. However, the relationship between size, age and growth rate may not be straightforward, as branching rates also increase with increasing light. Many different compounds in marine algae Chapter 5 Resource allocation 129

remain to be examined in the context of the plant defence models. Studies that incorporate this wealth of chemical diversity with the biological features of these algae will contribute substantially to the plant defence literature. CHAPTER 6

General Discussion

In this thesis I used the red alga Asparagopsis armata (Bonnemaisoniaceae) to address key questions in marine chemical ecology. I established that the halogenated secondary metabolites produced by A. armata have multiple ecological roles against natural enemies, preventing bacterial fouling and deterring consumption by herbivores. Notable aspects of this research included i) a definitive test of bacterial antifouling at the surface of A. armata (Chapter 2), ii) the integration of the localisation of the halogenated metabolites with the ecology of algal chemical defences (Chapters 2, 3 & 5), iii) the determination of an ecological role of chemical defence in a filamentous alga (Chapter 4) and iv) the detection of a cost of secondary metabolite production for A. armata (Chapter 5). A quantitative method for the natural products chemistry, an understanding of the biology integral to chemical defence and the high growth rate of A. armata facilitated tests of resource allocation in the context of the terrestrial plant defence models. While there are fitness benefits associated with the reduced impact of natural enemies (for example, herbivores and fouling organisms) on A. armata, there are also costs imposed by the production of the halogenated chemical defences. However, evidence for cost was only detected under limiting resource conditions. The biology of A. armata also influenced interpretations of plant defence models due to measurements made at a whole-alga scale. These results indicate that the specifics of algal biology – their morphology, ultrastructure, and patterns of growth and age – are crucial to tests of the models that describe the ecology and evolution of chemical defences. Chapter 6 General Discussion 131

Chemical defences of marine algae

Marine macroalgae are exposed to many natural enemies, of which herbivores (Lubchenco and Gaines 1981, Choat 1982) and fouling organisms (Wahl 1989, Steinberg and de Nys 2002) are two of the most important. Chemical mediation of the interactions with herbivores and fouling organisms can reduce the detrimental impacts of these organisms on marine algae (Hay and Fenical 1988, Wahl 1989, Paul 1992, McClintock and Baker 2001). This was demonstrated for Asparagopsis armata: not only did herbivores avoid chemically defended algae, but bacterial densities were lower on the surface of algae that released secondary metabolites. Chemical defences of marine algae against herbivores are rarely absolute (Hay et al. 1987a & 1987b, Hay et al. 1988a, Steinberg et al. 1991, Van Alstyne et al. 2001). This perhaps relates to an inability to defend their tissues from the diverse variety of herbivores that consume macroalgae (Underwood 1980, Choat 1982, Hay and Fenical 1988, Brawley 1992). Some of the consumers that are not deterred by bioactive metabolites may be specialists (Hay 1992). For instance, many opisthobranchs (such as sea hares) feed upon algae that contain otherwise deterrent chemicals (Pennings 1990, Paul and Pennings 1991) and the sea hare Aplysia parvula was similarly not deterred from feeding on Asparagopsis armata (Chapter 4). However, the ability of a herbivore to consume a chemically defended alga may depend not only on the type of herbivore, but also its size. This was demonstrated by a consumer-size effect for the response of the abalone Haliotis rubra to the halogenated metabolites in A. armata. The presence of a threshold size above which a herbivore is no longer deterred from feeding upon a chemically defended alga has not previously been reported, and will further complicate generalisations regarding herbivore responses to macroalgae. Although specialisation by marine herbivores has long been a focus of marine chemical ecology (Hay 1992), the specialisation of microbial symbionts on algal surfaces is almost completely unexplored. Bacterial communities on A. armata grown with and without halogenated metabolites varied in their susceptibility to brominated Chapter 6 General Discussion 132

metabolites. This was indicated by some bacteria that had greater degrees of tolerance to the halogenated compounds than others (Chapter 2). Because growth of red algae can be positively influenced by bacteria (Fries 1966), understanding the relationship between algae and their epiphytic flora could prove to be a fruitful area of investigation, especially if chemical mediation selects for positive associations in addition to selecting against negative ones. Just as the testing of algal-microbial interactions now extends beyond the presence of in vitro activity of algal natural products against laboratory bacteria (Maximilien et al. 1998, Hellio et al. 2001), the relationships that exist between chemical defences against bacteria and their benefits to algal fitness could also be incorporated into future studies of these interesting interactions. The manipulation of the halogenated metabolites in Asparagopsis armata provided a unique empirical test of the ecological roles of these compounds. To date, similar types of tests have not been utilised in marine algae. However, it is possible that the ecology of other algae with halogenated chemistries could be tested in this way. For example, culture manipulation of metabolites has recently been used to measure the costs of chemical defence in Delisea pulchra (Dworjanyn et al. in press). As the genetic control of secondary metabolism is not yet available to marine algae (the genetics of secondary metabolisms remain poorly understood, even with model organisms such as Arabidopsis thaliana, Kliebenstein 2004), the manipulation of compounds using culture techniques could provide new angles for research in marine systems. In the present work, manipulating metabolites in A. armata was not only important for testing the antifouling role of these compounds against bacteria at the algal surface, but also for demonstrating that small filamentous algae can deter feeding by consumers.

Large versus small macroalgae

The functional form models of Littler & Littler (1980) and Steneck & Dethier (1994) are the most prominent attempts to relate the structural form of algae to their susceptibility to herbivores. These models predict that filamentous algae should be more susceptible to Chapter 6 General Discussion 133

feeding by herbivores than larger macroalgae and/or macroalgae with lower productivity. However, ecological predictions based on functional form have not always held true (Padilla and Allen 2000). This is also the case for chemical defence in terrestrial plants, where plant form does not necessarily correlate with the presence of either tolerance or resistance traits (Mauricio 2000, Koricheva et al. 2004). Considering the large diversity of macroalgae in marine environments, it is perhaps not surprising that some organisms do not fit this paradigm. Even Ulva species, widely considered to be free of resistance traits due to their edibility, can produce some chemicals that deter herbivores (Van Alstyne et al. 2003). Categorisation of palatability and the assumed lack of defence in similarly “susceptible” algae (including filamentous varieties) needs to be re-examined. In this respect, previously overlooked algae may represent new sources of bioactive metabolites. Given that small algae may defend themselves chemically and these small individuals are often subject to the same herbivores as large macroalgae, then the relative differences in size of consumer and prey for small and large algae should further influence the efficacy of a chemical defence. For instance, large herbivores (such as fish) cannot differentiate between food at the scale of cells, as many cells are consumed simultaneously by the bite of a fish. Consequently, any effects of metabolites localised in surface cells (as appears to be the case for many red algae; Fig. 6.1) may be diluted if a bite includes internal cells that are not defended. Alternatively, feeding by smaller herbivores, such as amphipods, will be more selective (possibly feeding from cell to cell). But in this case, a single bite (e.g. of the surface layer) may be entirely of a gland cell and its contents, and thus highly deterrent. Questions regarding size of different herbivores have been an important feature in the chemical defence literature (Hay et al. 1987a, Duffy and Hay 1994), and incorporating these “scale” aspects of defence from the algal perspective may also prove interesting. Chapter 6 General Discussion 134

Structure/function relationships

One of the main themes in this thesis has been the use of biological structure to inform the ecology of chemical defence, including the ecological functions of secondary metabolites and the allocation of resources to production of these natural products. While localisation of metabolites in an alga is particularly important for ecological roles such as antifouling (which require that metabolites are delivered to the surface), it may also provide insight into chemical defence against herbivores. This will be important if herbivores graze at fine scales and differentiate between algal cells that contain deterrent compounds and those that do not (e.g. Poore 1994). In Chapter 5, I further demonstrated that the specialised cells of A. armata can be used to examine aspects of allocation to chemical defence that cannot be explored using quantitative chemistry. If specialised cells can be used to predict levels of secondary metabolites then focussing on the actual storage structure will enable fine-scale tests of resource allocation and plant-herbivore interactions for many other algae. The halogenated metabolites of Asparagopsis armata were localised to the gland cells, which themselves occupied space inside dermal cells. A stalk-like feature was identified that likely provides a means to release these compounds to the algal surface. There are many other varieties of specialised storage cells and sub-cellular structures in marine algae that vary in type and position (Figs. 6.1 & 6.2). For example, the vesicle cells of members of the Ceramiaceae vary in position on the filaments (Fig 6.1D-F), but it is not known whether this variation correlates to ecological function or merely relates to growth form or physiology. This diversity in the types of specialised structures in marine algae may relate to the similarly diverse functions of marine natural products (see Hay 1996, Paul 1992, McClintock and Baker 2001). The ultrastructure and position of these specialised structures will ultimately determine the potential kinds of ecological roles that the metabolites serve (Young and West 1979, Young et al. 1980). The position of the specialised cells in the red algal family Bonnemaisoniaceae also varies amongst its members (Womersley 1996), as does the natural products Chapter 6 General Discussion 135

chemistry (Fenical 1975). In a phylogenetic study that included the Bonnemaisoniaceae (Ni Chualain et al. 2004), the genera Asparagopsis and Bonnemaisonia grouped closely, as did Ptilonia and Delisea. This may also be interpreted from the perspective of specialised cells, because the two genera with gland cells that maintain a structural association with the parent cells grouped in the phylogeny (the gametophytes of Asparagopsis [Fig. 6.1C] and Bonnemaisonia) as did the two genera in which gland cells are single entities surrounded by a rosette of dermal cells (e.g. Delisea [Fig. 6.2C] and Ptilonia [Fig. 6.1B]). Nonetheless, algae from both groups can release metabolites that are antibacterial at the algae surface (Maximilien et al. 1998, Nylund et al. 2005, Paul et al. 2006). It is interesting to consider the evolution of these specialised cells, in particular the gland cells of A. armata, for which the pericentral cell almost entirely surrounds the storage cell (Chapter 2 & 3). This tight association between the cells could almost be described as endosymbiosis. From a natural products perspective, the presence of specialised structures could also be a useful directive tool for those pursuing organisms with novel, active metabolites. As the natural products and chemical defence of many macroalgae with larger morphologies have been thoroughly examined (McClintock and Baker 2001), a fruitful source of new compounds may prove to be filamentous algae, in particular those of the Ceramiaceae (Fig. 6.2B&F).

Resource allocation and costs of algal chemical defence

A number of terrestrial plant defence models have been developed to explain the evolution of chemical defences in the context of resource allocation (Herms and Mattson 1992, Stamp 2003). These terrestrial defence models have been adopted by studies on resource allocation in marine systems (Cronin 2001). But, considering the large catalogue of natural products in marine macroalgae (Fenical 1982, Faulkner 1996, MarinLit 2004, Paul and Puglisi 2004), there is still comparatively little information Chapter 6 General Discussion 136

regarding the sources of variation in levels of defence for algae. In view of the diversity of secondary chemicals and plant growth forms in marine systems, a wider understanding of algal chemical ecology could contribute substantially to the plant defence models. However, the results for resource allocation in Asparagopsis armata suggest that there may be limitations to the application of the plant defence models to algae. Firstly, measures of defence in algae may not be appropriate at a whole plant scale. Secondly, it may be difficult to specifically relate some of the plant defence models to algae based on differences in biology between marine and terrestrial organisms. As discussed in detail in Chapter 5, the presence of a “cost” of secondary metabolite production is the central assumption for all plant defence models (Simms 1992). This concept of cost is a widely accepted assumption for both terrestrial and marine chemical ecology, yet evidence for costs of secondary metabolites in marine algae remains limited (Steinberg 1995, Pavia et al. 1999, Dworjanyn et al. in press). Empirical tests of allocation (or opportunity) costs for terrestrial plants should correlate plant fitness with whole plant chemistry (Stamp 2003). However, the whole plant level may not be the appropriate scale for some algae due to their lack of internal transport. Research that focuses solely on the whole alga scale may limit the prospect of detecting negative correlations between algal fitness and secondary metabolite concentration. This issue aside, the red algae provide an excellent system for measuring costs of chemical defence, because their simple anatomy makes it relatively easy to differentiate between growth and defence processes. Even the seemingly complex, large-bodied algae of the Florideophycidae (Rhodophyta) have a basic body plan with filamentous construction (Coomans and Hommersand 1990), a feature which makes them suitable candidates for modelling the trade-offs between primary processes and chemical defence. Differences in biology between terrestrial plants and macroalgae may also limit the application of specific defence models. Recent work has demonstrated that the most valuable parts of an alga (with respect to optimal defence) are not necessarily the reproductive structures (Pavia et al. 2002). As filamentous algae are totipotent (i.e. Chapter 6 General Discussion 137

individual cells can create whole new individuals, Waaland 1990), development does not only occur by reproductive propagules. This makes defining apriori the algal parts that are “optimal” to fitness more difficult. The resource-based models developed for terrestrial plants may not be directly applicable to algae, not only because of reduced translocation within plants (but see Cronin and Hay 1996b), but also with respect to differentiating between defence and growth processes. For instance, some algae form specialised cells that sequester the chemical defence (e.g. the gland cells of Asparagopsis armata), whereas others do not (e.g. phenolic compounds are present in many types of brown algal cells; Schoenwaelder and Clayton 1998). Incorporating the specific aspects of an alga’s biology and its secondary metabolism will help interpret patterns of algal chemical defences in the context of the plant defence models (Arnold and Targett 2003).

Asparagopsis armata – a model organism for marine chemical ecology?

The genus Asparagopsis has been much studied for its natural products chemistry (McConnell and Fenical 1977, Woolard et al. 1979) and its interesting cellular features (Wolk 1969, Codomier et al. 1977). Prior to this thesis, however, there was little empirical evidence for the ecological roles of these compounds, although it was known that the halogenated metabolites were bioactive in a general sense (McConnell and Fenical 1979). Furthermore, because Asparagopsis armata is an invasive species (Feldmann 1942, Bouderesque and Verlaque 2002) and has a heteromorphic life history (Feldmann and Feldmann 1939), these two features make it a particularly attractive candidate for future research in marine chemical ecology. Asparagopsis armata was one of the first organisms in European waters to be identified as introduced (Feldman 1942). Since then it has become widely invasive throughout European coastlines, most likely because it has no natural predators (Boudouresque and Verlaque 2002). Research on the chemical ecology of A. armata, Chapter 6 General Discussion 138

which considers interactions with herbivores and the evolution of feeding preference (e.g. specialisation and niches for chemically tolerant herbivores, Trowbridge and Todd 2001), should prove very interesting in these invaded environments. Heteromorphic life histories in macroalgae are important in mediating the interaction with herbivores (Lubchenco and Cubit 1980, Searles 1980). For Asparagopsis armata, the heteromorphic life history permits comparison of the chemical defence provided by the same halogenated natural products between the morphologically distinct life stages. This comparison is relevant as adults of both stages can coexist. Future studies that address the presence of a triphasic life history in the context of the plant defence models should also be rewarding. It would be interesting to compare trade-offs between growth and defence for the two different life stages, in particular examining selection for chemical defence between generations. Taken together, the complex life history and multiple roles of secondary metabolites in Asparagopsis armata should provide further unique tests of the ecology and evolution of chemical defence in marine algae. Chapter 6 General discussion 139

A B C

D E F

G H I

Figure 6.1. Specialised storage structures of some marine red algae. Images are of the surface layer of cells in all algae. A-F: Vesicle cells (arrows) are found in foliose algae, e.g. Phacelocarpus peperocarpus (A), Ptilonia australasica (B) and Asparagopsis armata (C). Members of the Ceramiaceae have vesicle cells in many positions; mid-filament - Elisiella australis (D), towards the tip - Antithamnion hanovoides (E) and even terminal - Antithamnion pinnafolium (F). Refractile inclusions are also common but are components of somatic cells. These may still occupy large amount of internal volume; as is the case for the single inclusions in Laurencia rigida (corp en cerise) (G) and Plocamium leptophyllum (H) or the multiple smaller inclusions of Anotrichium tenue (I). Chapter 6 General discussion 140

A B

C D

E F

Figure 6.2 Light and corresponding epifluorescence micrographs of different storage structures in three different algae; Balliella amphiglanda (A & B), Delisea pulchra (C & D), Elisiella australis (E & F). Each of these specialised cells contained material that was excited by UV light using epifluorescence microscopy (arrows), indicating the presence of compounds that are chromaphores (i.e. metabolites with conjugated double bonds). Some non-polar compounds extracted from algae also autofluoresce, a property that has been attributed to bioactive metabolites (Paul and Hay 1986). 141

References

Al-Ogily SM, Knight-Jones EW (1977) Antifouling role of antibiotics produced by marine algae and bryozoans. Nature 265:728–729 Arnold TM, Targett NM (2003) To grow and defend: lack of tradeoffs for brown algal phlorotannins. Oikos 100:406–408 Babuka SJ, Pueschel CM (1998) A freeze-substitution ultrastructural study of the cytoskeleton of the red alga Antithamnion kylinii (Ceramiales). Phycologia 37:251– 258 Baldwin I, Sims CL, Kean SE (1990) The reproductive consequences associated with inducible alkaloidal responses in wild tobacco. Ecology 71:252–262 Berenbaum MR (1995) Turnabout is fair play: secondary roles for primary compounds. J Chem Ecol 21:925–940 Berenbaum MR, Zangerl AR, Nitao JK (1986) Constraints on chemical coevolution: Wild parsnips and the parsnip webworm. Evolution 40: 1215–1228 Bergelson J, Purrington CB (1996) Surveying costs of resistance in plants. Am Nat 148: 536–558 Björkman C, Larsson S, Gref R (1991) Effects of nitrogen fertilization on pine needle chemistry and sawfly performance. Oecologia 86:202–209 Bonin DR, Hawkes MW (1987) Systematics and life histories of New-Zealand Bonnemaisoniaceae (Bonnemaisoniales, Rhodophyta): I. The genus Asparagopsis. N-Z J Bot 25:577–590 Bouarab K, Potin P, Weinberger F, Correa J and Kloareg B (2001) The Chondrus crispus-Acrochaete operculata host-pathogen association, a novel model in glycobiology and applied phycopathology. J Appl Phycol 13:185–193 Boudouresque CF, Verlaque M (2002) Biological pollution in the Mediterranean Sea: invasive versus introduced macrophytes. Mar Poll Bull 44:32–38 Brawley SH (1992) Mesoherbivores. In: John DM, Hawkins SJ, Price JH (eds) Plant- animal interactions in the marine benthos. Clarendon, Oxford, p 235–263 Brawley SH, Adey WH (1981) The effect of micrograzers on algal community structure in a coral reef microcosm. Mar Biol 61:167–178 Bryant JP, Chapin III FS, Klein DR (1983) Carbon/nutrient balance of boreal plants in relation to vertebrate herbivory. Oikos 40:357–36 References 142

Burton JD (1996) The ocean: a global chemical system. In: Summerhayes CP, Thorpe SA (eds) Oceanography: An Illustrated Guide. Manson Publishing Ltd., London, p 166 Cancho B, Ventura F, Galceran MT (1999) Behavior of halogenated disinfection by– products in the water treatment plant of Barcelona, Spain. Bull Environ Contamin Toxicol 63:610–617 Carpenter RC (1990) Competition among marine macroalgae: a physiological perspective. J Phycol 26:6–12 Choat JH (1982) Fish feeding and the structure of benthic communities in temperate waters. Ann Rev Ecol Syst 13:423–449 Chretiennot–Dinet M, Giraud–Guille M, Vaulot D, Putaux J, Saito Y, Chanzy H (1997) The chitinous nature of filaments ejected by Phaeocystis (Prymnesiophyceae). J Phycol 33:666–672 Clayton MN, Ashburner CM (1994) Secretion of the phenolic bodies following fertilisation in Durvillaea potatorum (Durvillaeales, Phaeophyta). Eur J Phycol 29:1–9 Clayton MN, Beakes GW (1983) Effects of fixatives on the ultrastructure of physodes in vegetative cells of Scytosiphon lomentaria (Scytosiphonaceae, Phaeophyta). J Phycol 19:4–16 Clayton MN, King RJ (1990) Biology of marine plants. Longman Cheshire, Melbourne Codomier L, Bruneau Y, Combaut G, Teste J (1977) Etude biologique et chimique d’Asparagopsis armata et de Falkenbergia rufolanosa (Rhodophycees, Bonnemaisoniales). C R Acad Sci Paris 284:1163–1165 Cole KM, Sheath RG (1990) Biology of the red algae. Cambridge University Press, Cambridge Coley PD, Byrant JP, Chapin III FS (1985) Resource availability and plant antiherbivore defense. Science 230:895–899 Coomans RJ, Hommersand MH (1990) Vegetative growth and organization. In: Cole KM, Sheath RG (eds) Biology of the Red Algae. Cambridge University Press, Cambridge, p 275–304 Correa JA, Sanchez PA (1996) Ecological aspects of algal infectious diseases. Hydrobiologia 326/327:89–95 Cronin G (2001) Resource allocation in seaweeds and marine invertebrates: chemical defense patterns in relation to defense theories. In: McClintock JB, Baker BJ (eds) Marine Chemical Ecology. Marine Science Series – CRC Press, p 325-353 Cronin G, Hay ME (1996a) Effects of light and nutrient availability on the growth, secondary chemistry, and resistance to herbivory of two brown seaweeds. Oikos 77:93–106 References 143

Cronin G, Hay ME (1996b) Within–plant variation in seaweed palatability and chemical defenses: optimal defense theory versus the growth-differentiation balance hypothesis. Oecologia 105:361–368 Cruz–Rivera E, Hay ME (2003) Prey nutritional quality interacts with chemical defenses to affect consumer feeding and fitness. Ecol Monogr 73:483–506 Dahllöf I (2002) Molecular community analysis of microbial diversity. Curr Opin Biotech 13:213–217 de Nys R, Coll JC, Price IR (1991) Chemically mediated interactions between the red alga Plocamium hamatum (Rhodophyta) and the octocoral Sinularia cruciata (Alcyonaceae). Mar Biol 108:315–320 de Nys R, Dworjanyn SA, Steinberg PD (1998) A new method for determining surface concentrations of marine natural products on seaweeds. Mar Ecol Prog Ser 162:79– 87 de Nys R, Steinberg PD, Rogers CN, Charlton TS, Duncan MW (1996) Quantitative variation of secondary metabolites in the sea hare Aplysia parvula and its host plant, Delisea pulchra. Mar Ecol Prog Ser 130:135–146 de Nys R, Steinberg PD, Willemson P, Dworjanyn SA, Gabelish CL, King RJ (1995) Broad spectrum effects of secondary metabolites from the red alga Delisea pulchra in antifouling assays. Biofouling 8: 259–271 Duffy JE, Hay ME (1994) Herbivore resistance to seaweed chemical defense: the roles of mobility and predation risk. Ecology 75:1304–1319 Duffy JE, Hay ME (2000) Strong impacts of grazing amphipods on the organization of a benthic community. Ecol Monogr 70:237–263 Duffy JE, Paul VJ (1992) Prey nutritional quality and the effectiveness of chemical defenses against tropical reef fishes. Oecologia 90:333–339 Dworjanyn SA, de Nys R, Steinberg PD (1999) Localisation and surface chemistry of secondary metabolites in the red alga Delisea pulchra. Mar Biol 133:727–736 Dworjanyn SA, Wright JT, Paul NA, de Nys R, Steinberg PD (In Press) Cost of chemical defence in the red alga Delisea pulchra. Oikos Feeny P (1976) Plant apparency and chemical defense. Rec Adv in Phytochem 10:1–40 Feldmann J (1942) Les algues marines de la cote des Alberes IV - Rhodophycees (suite). Rev Algol 12:77–100 Feldmann J, Feldmann G (1939) Sur le developpement des carpospores et l’alternance de generations de l’Asparagopsis armata Harvey. C R Hebd Seanc Acad Sci Paris 208:1240–1242 Fenical W (1975) Halogenation in the Rhodophyta: a review. J Phycol 11:245–259 Fenical W (1982) Natural products chemistry in the marine environment. Science 215:923–928 References 144

Fineblum WL, Rausher MD (1995) Tradeoff between resistance and tolerance to herbivore damage in the morning glory. Nature 377:517–519 Fries L (1966) Influence of iodine and bromine on growth of some red algae in axenic culture. Physiol Planta 19:800–808 Fries L (1970) The influence of microamounts of organic substances other than vitamins on the growth of some red algae in axenic culture. Br Phycol J 5:39–46 Fritz RS, Simms EL (1992) Plant Resistance to Herbivores and Pathogens, University of Chicago Press, Chicago Gershenzon J (1994) Metabolic costs of terpenoid bioaccumulation in higher plants. J Chem Ecol 20:1281–1328 Ginsburg DW, Paul VJ (2001) Chemical defenses in the sea hare Aplysia parvula: importance of diet and sequestration of algal secondary metabolites. Mar Ecol Prog Ser 215: 261–274 Goff LJ, Coleman AW (1990) Microspectrofluorometric analysis of algal DNA. In: Cole KM, Sheath RG (eds) Biology of the Red Algae. Cambridge University Press, Cambridge, p 43–72 Gosselin LA (1997) An ecological transition during juvenile life in a marine snail. Mar Ecol Prog Ser 157:185–194 Gribble GW (1994) Natural Organohalogens – Many more than you think. J Chem Ed 71: 907–911 Hamilton JG, Zangerl AR, DeLucia EH, Berenbaum MR (2001) The carbon-nutrient balance hypothesis: its rise and fall. Ecol Lett 4:86-95 Haukioja E, Koricheva J (2000) Tolerance to herbivory in woody vs. herbaceous plants. Evol Ecol 14:551–562 Hay ME (1992) The role of seaweed chemical defences in the evolution of feeding specialisation and in the mediation of complex interactions. In: Paul VJ (ed) Ecological roles of marine natural products, Comstock, Ithaca, New York, p 93– 118 Hay ME (1996) Marine chemical ecology: what is known and what is next? J Exp Mar Biol Ecol 200:103–134 Hay ME, Duffy JE, Fenical W, Gustafson K (1988a) Chemical defense in the seaweed Dictyopteris delicatula: differential effects against reef fishes and amphipods. Mar Ecol Prog Ser 48:185–192 Hay ME, Duffy JE, Pfister CA (1987a) Chemical defense against different marine herbivores: are amphipods insects equivalents? Ecology 68:1567–1580 Hay ME, Fenical W (1988) Marine-plant herbivore interactions: the ecology of chemical defense. Annu Rev Ecol Syst 19:111–145 References 145

Hay ME, Fenical W, Gustafson K (1987b) Chemical defense against diverse coral-reef herbivores. Ecology 68:1581–1591 Hay ME, Kappel QE, Fenical W (1994) Synergisms in plant defenses against herbivores: interactions of chemistry, calcification, and plant quality. Ecology 75:1714–1726 Hay ME, Paul VJ, Lewis SM, Gustafson K, Tucker J, Trindell R (1988b) Can tropical seaweeds reduce herbivory by growing at night?: diel patterns of growth, nitrogen content, herbivory, and chemical versus morphological defenses. Oecologia 75:233–245 Hellio C, De La Broise D, Dufosse L, Le Gal Y, Bourgougnon N (2001) Inhibition of marine bacteria by extracts of macroalgae: Potential use for environmentally friendly antifouling paints. Mar Environ Res 52:231–247 Herms DA, Mattson WJ (1992) The dilemma of plants: to grow or defend. Q Rev Biol 67:283–335 Hill MS, Lopez NA, Young KA (2005) Anti-predator defenses in western North Atlantic sponges with evidence of enhanced defense through interactions between spicules and chemicals. Mar Ecol Prog Ser 291:93–102 Hooker N, Morse DE (1985) Abalone: the emerging development of commercial cultivation in the United States. In: Huner JV, Brown EE (eds) Crustacean and mollusk cultivation in the United States. AVI Publishing Co., Westport, USA, p 365–413 Hornsey IS, Hide D (1974) The production of antimicrobial compounds by British marine algae I. Antibiotic producing marine algae. Br Phycol J 9:353–361 Huisman JM (2000) Marine Plants of Australia. University of Western Australia Press, Nedlands, Western Australia Jensen PR, Kauffman CA, Fenical W (1996) High recovery of culturable bacteria from the surfaces of marine algae. Mar Biol 126:1–7 Jormalainen V, Honkanen T (2004) Variation in natural selection for growth and phlorotannins in the brown alga Fucus vesiculosus. J Evol Biol 17:807–820 Karban R, Baldwin I (1997) Induced Responses to Herbivory. Univ. Chicago Press, Chicago Kicklighter CE, Kubanek J, Hay ME (2004) Do brominated natural products defend marine worms from consumers? Some do, most don’t. Limnol Oceanogr 49:430– 441 Kliebenstein DJ (2004) Secondary metabolites and plant/environment interactions: a view through Arabidopsis thaliana tinged glasses. Plant Cell Environ 27:675–684 References 146

Klumpp DW, McKinnon AD (1992) Community structure, biomass and productivity of epilithic algal communities on the Great Barrier Reef: Dynamics at different spatial scales. Mar Ecol Prog Ser 86:77–89 Koricheva J (1999) Interpreting phenotypic variation in plant allelochemistry: problems with the use of concentrations. Oecologia 119: 467–473 Koricheva J (2002) Meta-analysis of sources of variation in fitness costs of plant antiherbivore defenses. Ecology 83:176–190 Koricheva J, Nykänen H, Gianoli E (2004) Meta-analysis of trade-offs among plant antiherbivore defenses: are plants jacks-of-all-trades, masters of all? Am Nat 163:64–75 Kubanek J, Jensen PR, Keifer PA, Sullards MC, Collins DW, Fenical W (2003) Seaweed resistance to microbial attack: A targeted chemical defense against marine fungi. Proc Natl Acad Sci U.S.A. 100:6916–6921 Kubanek J, Lester SE, Fenical W, Hay ME (2004) Ambiguous role of phlorotannins as chemical defenses in the brown alga Fucus vesiculosus. Mar Ecol Prog Ser 277:79– 93 Kushmaro A, Rosenberg E, Fine M, Loya Y (1997) Bleaching of the coral Oculina patagonica by Vibrio AK–1. Mar Ecol Prog Ser 147:159–165 Kylin H (1927) Uber die Blazenzellen der Flordieen. Bot Not 1927:274–285 Lee RE (1999) Phycology, 3rd Edition, Cambridge University Press, Cambridge Littler MM, Littler DS (1980) The evolution of thallus form and survival strategies in benthic marine macroalgae: field and laboratory tests of a functional form model. Am Nat 116:25–44 Littler MM, Littler DS (1995) Impact of the CLOD pathogen on Pacific coral reefs. Science 267:1356–1360 Lubchenco J, Cubit JD (1980) Heteromorphic life histories of certain marine algae as adaptations to variations in herbivory. Ecology 61:676–687 Lubchenco J, Gaines SD (1981) A unified approach to marine plant-herbivore interactions. I. Populations and communities. Annu Rev Ecol Syst 12:405–437 Lüning K (1981) Light. In: Lobban CS, Wynne MJ (eds) The Biology of Seaweeds. Blackwell Scientific, p 326–355 Lynch M, Walsh JB (1998) Genetics and Analysis of Quantitative Traits. SinauerAssoc. Inc., Sunderland, MA Maier I, Muller DG (1986) Sexual pheremones in algae. Biol Bull 170:145–175 Manefield M, de Nys R, Kumar N, Read R, Givskov M, Steinberg P, Kjelleberg S (1999) Evidence that halogenated furanones from Delisea pulchra inhibit acylated homoserine lactone (AHL)–mediated gene expression by displacing the AHL signal from its receptor protein. Microbiology 145:283–291 References 147

Manley SL, Barbero PE (2001) Physiological constraints on bromoform (CHBr3) production by Ulva lactuca (Chlorophyta). Limnol Oceanogr 46:1392–1399 Mann KH (1973) Seaweeds: their productivity and strategy for growth. Science 182:975–981 MarinLit (2004) Marine Natural Products Database. University of Canterbury, Christchurch, New Zealand Marshall RA, Hamilton JTG, Dring MJ, Harper DB (2003) Do vesicle cells of the red alga Asparagopsis (Falkenbergia stage) play a role in bromocarbon production? Chemosphere 52:471–5 Marshall RA, Harper DB, McRoberts CW, Dring MJ (1999) Volatile bromocarbons produced by Falkenbergia stages of Asparagopsis spp. (Rhodophyta). Limnol Oceanogr 44:1348–1352 Mauricio R (1998) Costs of resistance to natural enemies in field populations of the annual plant Arabidopsis thaliana. Am Nat 151:20–28 Mauricio R (2000) Natural selection and the joint evolution of tolerance and resistance as plant defenses. Evol Ecol 14:491–507 Maximilien R, de Nys R, Holmström C, Gram L, Givskov M, Crass K, Kjelleberg S, Steinberg PD (1998) Chemical mediation of bacterial surface colonisation by secondary metabolites from the red alga Delisea pulchra. Aquat Microb Ecol 15:233–246 McClintock JB, Baker BJ (2001) Marine Chemical Ecology. Marine Science Series – CRC Press McConnaughay KDM, Coleman JS (1999) Biomass allocation in plants: ontogeny or optimality? A test along three resource gradients. Ecology 80:2581–2593 McConnell OJ, Fenical W (1977) Halogen chemistry of the red alga Asparagopsis. Phytochemistry 16:367–374 McConnell OJ, Fenical W (1979) Antimicrobial agents from marine red algae of the family Bonnemaisoniaceae. In: Hoppe HA, Levring T, Tanaka Y (eds) Marine Algae in Pharmaceutical Science. Walter de Gruyter, New York, p 403–427 McKey D (1979) The distribution of secondary compounds within plants. In: Rosenthal GA, Janzen DH (eds) Herbivores: their interaction with secondary plant metabolites. Academic Press, New York, p 55–133 McLachlan, J (1977) Effects of nutrients on growth and development of embryos of Fucus edentatus (Phaeophyceae, Fucales). Phycologia 16:329–338 McNaughton SJ (1983) Compensatory growth as a response to herbivory. Oikos 40:329–336 References 148

Moe RL, Silva PC (1980) Morphological and taxonomic studies on Antarctic Ceramiaceae (Rhodoohyceae). II. Pterothamnion antarcticum (Kylin) Comb. Nov. (Antithamnion antarcticum Kylin). Br Phycol J 15:1–17 Murray SM, Dixon PS (1992) The Rhodophyta: some aspects of their biology. III. Oceanogr Mar Biol Annu Rev 30:1–148 Nagle DG, Paul VJ (1999) Production of secondary metabolites by filamentous tropical marine cyanobacteria: ecological functions of the compounds. J Phycol 35:1412– 1421 Ni Chualain F, Maggs CA, Saunders GW, Guiry MD (2004) The invasive genus Asparagopsis (Bonnemaisoniaceae, Rhodophyta): molecular systematics, morphology, and ecophysiology of Falkenbergia isolates. J Phycol 40: 1112–1126 Nightingale PD, Malin G, Liss PS (1995) Production of chloroform and other low molecular weight halocarbons by some species of macroalgae. Limnol Oceanogr 40:680–689 North WJ (1971) The biology of giant kelp beds (Macrocystis) in California. Beihefte Nova Hedwegia 32:1–600 Nylund GM, Cervin G, Hermansson M, Pavia H (2005) Chemical inhibition of bacterial colonization by the red alga Bonnemaisonia hamifera. Mar Ecol Prog Ser 302:27– 36 O’Brien TP, McCully ME (1981) The Study of Plant Structure: Principles and Selected Methods. Termarcarphi Pty Ltd, Melbourne p 2.27–2.28 Orlovich DA, Ashford AE (1993) Polyphosphate granules are an artefact of specimen preparation in the ectomycorrhizal fungus Pisolithus tinctorius. Protoplasma 173:91–102 Padilla DK, Allen BJ (2000) Paradigm lost: reconsidering functional form and group hypotheses in marine ecology. J Exp Mar Bio Ecol 250:207–221 Paine RT (1990) Benthic macroalgal competition: complications and consequences. J Phycol 26:12–17 Pallaghy CK, Minchinton J, Kraft GT, Wetherbee R (1983) Presence and distribution of bromine in Thysanocladia densa (Solieriaceae, Gigartinales), a marine red alga from the Great Barrier Reef. J Phycol 19:204–208 Paul NA, Cole L, de Nys R, Steinberg PD (2006) Ultrastructure of gland cells of the red alga Asparagopsis armata (Bonnemaisoniaceae). J Phycol 42:637-645 Paul NA, de Nys R, Steinberg PD (2006) Chemical defence against bacteria in the red alga Asparagopsis armata: linking structure with function. Mar Ecol Prog Ser 306:87-101 Paul VJ (1992) Ecological Roles of Marine Natural Products. Comstock, Ithaca, New York References 149

Paul VJ, Hay ME (1986) Seaweed susceptibility to herbivory: chemical and morphological correlates. Mar Ecol Prog Ser 33:255–264 Paul VJ, Nelson SG, Sanger HR (1990) Feeding preferences of adult and juvenile rabbitfish Siganus argenteus in relation to chemical defenses of tropical seaweeds. Mar Ecol Prog Ser 60:23–34 Paul VJ, Pennings SC (1991) Diet-derived chemical defenses in the sea hare Stylocheilus longicauda. J Exp Mar Biol Ecol 151:227–243 Paul VJ, Puglisi M (2004) Chemical mediation of interactions among marine organisms. Nat Prod Rep 21:189–209 Pavia H, Toth GB (2000) Inducible chemical resistance to herbivory in the brown seaweed Ascophyllum nodosum. Ecology 81:3212–3225 Pavia H, Toth GB, Åberg P (1999) Trade-offs between phlorotannin production and annual growth in natural populations of the brown seaweed Ascophyllum nodosum. J Ecol 87:761–771 Pavia H, Toth GB, Åberg P (2002) Optimal defense theory: elasticity analysis as a tool to predict intraplant variation in defenses. Ecology 83:891–897 Pedersén M, Collen J, Abrahamsson K, Ekdahl A (1996) Production of halocarbons from seaweeds: an oxidative stress reaction? Scientia Marina 60:257–263 Pedersén M, Roomans GM, Hofsten AV (1980) Cell inclusions containing bromine in Rhodomela confervoides (Huds.) Lamour. and Polysiphonia elongata Harv. (Rhodophyta; Ceramiales). Phycologia 19:153–158 Pennings SC (1990) Size–related shifts in herbivory: specialization in the sea hare Aplysia californica. J Exp Mar Biol Ecol 142:43–61 Pennings SC, Pablo SR, Paul VJ (1997) Chemical defenses of the tropical benthic marine cyanobacterium Hormothamnion enteromorphoides: diverse consumers and synergisms. Limnol Oceanogr 42:911–917 Peterson CH., Renaud PE (1989) Analysis of feeding preference experiments. Oecologia 80:82–86 Pfister CA (1992) Costs of reproduction in an intertidal kelp: patterns of allocation and life history consequences. Ecology 73:1586–1596 Pohnert G, Jung V (2003) Intracellular compartmentation in the biosynthesis of caulerpenyne: study on intact macroalgae using stable-isotope-labeled precursors. Org Lett 5:5091–5093 Poore AGB (1994) Selective herbivory by amphipods inhabiting the brown alga Zonaria angustata. Mar Ecol Prog Ser 107:113–123 Poore AGB (1999) Ecology and evolution of host plant use in herbivorous marine amphipods, Ph. D. Thesis, University of New South Wales, Sydney, Australia References 150

Poore AGB, Steinberg PD (1999) Preference–performance relationships and effects of host plant choice in an herbivorous marine amphipod. Ecol Monogr 69:443–464 Potin P, Bouarab K, Salaün J, Pohnert G, Kloareg B (2002) Biotic interactions of marine algae. Curr Opin Plant Biol 5:1–10 Provasoli L (1968) Media and prospects for the cultivation of marine algae. In: Watanabe A, Hattori A (eds) Culture and Collections of Algae. Proceedings of the U.S.-Japanese Conference, Hakone, Japanese Society of Plant Physiology, Tokyo, p 63–75 Pueschel CM (1992) An ultrastructural survey of the diversity of crystalline, proteinaceous inclusions in red algal cells. Phycologia 31:489–499 Pueschel CM, Korb RE (2001) Storage of nitrogen in the form of protein bodies in the kelp Laminaria solidungula. Mar Ecol Prog Ser 218:107–114 Puglisi MP, Paul VJ (1997) Intraspecifc variation in the red alga Portieria hornemannii: monoterpene concentrations are not influenced by nitrogen or phosphorous enrichment. Mar Biol 128:161–170 Quinn GP, Keough MJ (2002) Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge Ragan MA (1976) Physodes and the phenolic compounds of brown algae: Composition and significance of physodes in vivo. Bot Mar 19:145–154 Reynolds ES (1963) The use of lead citrate at high pH as an electron-opaque stain in electron microscopy. J Cell Biol 17:208–212 Rhoades DF (1979) Evolution of plant chemical defense against herbivores. In: GA Rosenthal and DH Janzen (eds) Herbivores: their interaction with secondary plant metabolites. Academic Press, New York, New York, p 3–54 Riipi M, Ossipov V, Lempa K, Haukioja E, Koricheva J, Ossipova S, Pihlaja K (2002) Seasonal changes in birch leaf chemistry: are there trade-offs between leaf growth and accumulation of phenolics? Oecologia 130:380–390 Rogers CN (2000) Cost and benefits of acquired chemical defence. Ph.D. Thesis, University of New South Wales, Sydney, Australia Rogers CN, de Nys R, Steinberg PD (2002) Effects of algal diet on the performance and susceptibility to predation of the sea hare Aplysia parvula. Mar Ecol Prog Ser 236:241–254 Rohde S, Molis M, Wahl M (2004) Regulation of anti–herbivore defence by Fucus vesiculosus in response to various cues. J Ecol 92:1011–1018 Sand–Jensen K (1977) Effects of epiphytes on eelgrass photosynthesis. Aquat Bot 3: 55–63 References 151

Schmitt TM, Hay ME, Lindquist N (1995) Constraints on chemically mediated coevolution: multiple functions for seaweed secondary metabolites. Ecology 76:107–123 Schmitz K (1981) Translocation. In: Lobban CS, Wynne MJ (eds) The Biology of Seaweeds. Blackwell, Oxford, p 534–558 Schoenwaelder MEA (2002) The occurrence and cellular significance of physodes in brown algae. Phycologia 41:125–139 Schoenwaelder MEA, Clayton MN (1998) Secretion of phenolic substances into the zygote wall and cell plate in embryos of Hormosira and Acrocarpia (Fucales, Phaeophyceae). J Phycol 34:969–980 Schoenwaelder MEA, Clayton MN (1999) The role of the cytoskeleton in brown algal physode movement. Eur J Phycol 34:223–229 Searles RB (1980) The strategy of the red algal life history. Am Nat 115:113–120 Seigler D, Price PW (1976) Secondary compounds in plants: primary functions. Am Nat 110:101–105 Shepherd SA, Steinberg PD (1992) Food preferences of three Australian abalone species with a review of the algal food of abalone. In: Shepherd SA, Tegner MJ, Guzman del Proo SA (eds) Abalone of the world: Biology, Fisheries and Culture. Proceedings of the 1st International Symposium on Abalone, Marston Book Services Ltd, Oxford, p 169–181 Sieburth JM, Conover JT (1965) Sargassum tannin, an antibiotic that retards fouling. Nature 208:52–53 Simms EL (1992) Cost of herbivore resistance to herbivory. In: Fritz RS, Simms EL (eds) Plant Resistance to Herbivores and Pathogens. University of Chicago Press, Chicago p 392–425 Simms EL, Rausher MD (1987) Costs and benefits of plant resistance to herbivory. Am Nat 130:570–581 Spurr AR (1969) A low-viscosity epoxy resin embedding medium for electron microscopy. J Ultrastruct Res 26:31–43 Stamp N (2003) Out of the quagmire of plant defense hypotheses. Q Rev Biol 78:23–55 Stamp N (2004) Can the growth–differentiation balance hypothesis be tested rigorously? Oikos 107:439–448 Steinberg PD (1984) Algal chemical defense against herbivores: allocation of phenolic compounds in the kelp Alaria marginata. Science 223:405–406 Steinberg PD (1995) Seasonal variation in the relationship between growth rate and phlorotannin production in the kelp Ecklonia radiata. Oecologia 102:169–173 Steinberg PD, de Nys R (2002) Chemical mediation of colonization of seaweed surfaces. J Phycol 38:621–629 References 152

Steinberg PD, Edyvane K, de Nys R, Birdsey R, van Altena IA (1991) Lack of avoidance of phenolic-rich brown algae by tropical herbivorous fishes. Mar Biol 109:335–343 Steneck RS, Dethier MN (1994) A functional group approach to the structure of algal- dominated communities. Oikos 69:476–498 Strauss SY, Rudgers JA, Lau JA, Irwin RE (2002) Direct and ecological costs of resistance to herbivory. Trends Ecol Evol 17:278–285 Toth GB, Langhamer O, Pavia H (2005) Inducible and constitutive defenses of valuable seaweed tissues: consequences for herbivore fitness. Ecology 86:612–618 Toth GB, Pavia H (2000) Waterborne cues induce chemical defense in a marine alga (Ascophyllum nodosum). Proc Natl Acad Sci USA 97:14418–14420 Trowbridge CD, Todd CD (2001) Host-plant change in marine specialist herbivores: ascoglossan sea slugs on introduced macroalgae. Ecol Monogr 71:219–243 Tuomi J, Ilvessalo H, Niemela P, Siren S, Jormalainen V (1989) Within-plant variation in phenolic content and toughness of the brown alga Fucus vesiculosus L. Bot Mar 32:505–509 Tutschulte TC, Connell JH (1988) Feeding behavior and algal food of three species of abalones (Haliotis) in southern California. Mar Ecol Prog Ser 49:57–64 Underwood AJ (1980) The effects of grazing by gastropods and physical factors on the upper limits of distribution of inter-tidal macroalgae. Oecologia 46:201– 213 Van Alstyne KL, Houser LT (2003) Dimethylsulfide release during macroinvertebrate grazing and its role as an activated defense. Mar Ecol Prog Ser 250:175–181 Van Alstyne KL, McCarthy JJI, Hustead CL, Kearns LJ (1999) Phlorotannin allocation among tissues of northeastern Pacific kelps and rockweeds. J Phycol 35:483–492 Van Alstyne KL, Whitman SL, Ehlig JM (2001) Differences in herbivore preferences, phlorotannin production, and nutritional quality between juvenile and adult tissues from marine brown algae. Mar Biol 139:201–210 Waaland SD (1990) Development. In: Cole KM, Sheath RG (eds) Biology of the Red Algae. Cambridge University Press, Cambridge, p 259–273 Wahl M (1989) Marine epibiosis. I. Fouling and antifouling: some basic aspects. Mar Ecol Prog Ser 58:175–189 Wahl M, Hay ME, Enderlein P (1997) Effects of epibiosis on consumer – prey interactions. Hydrobiologia 355:49–59 Wahl M, Jensen PR, Fenical W (1994) Chemical control of bacterial epibiosis on ascidians. Mar Ecol Prog Ser 110:45–57 Weinberger F, Friedlander M (2000) Response of Gracilaria conferta (Rhodophyta) to oligoagars results in defense against agar-degrading epiphytes. J Phycol 36:1079– 1086 References 153

Werner EE, Gilliam JF (1984) The ontogenetic niche and species interactions in size- structured populations. Annu Rev Ecol Syst 15:393–425 Whittaker RH, Feeny PP (1971) Allelochemics: chemical interactions between species. Science 171:757–770 Winter FC, Estes JA (1992) Experimental evidence for the effects of polyphenolic compounds from Dictyonerum californicium Ruprecht (Phaeophyta, Laminariales) on feeding rate and growth in the red abalone Haliotis rufescens Swainson. J Exp Mar Biol Ecol 155:263–277 Wolk CP (1968) Role of bromine in the formation of the refractile inclusions of the vesicle cells of the Bonnemaisoniaceae (Rhodophyta). Planta (Berl) 78:371–378 Womersley HBS (1996) The Marine Benthic Flora of Southern Australia: Rhodophyta Part IIIB. Australian Biological Resources Study, Canberra Womersley HBS (1998) The Marine Benthic Flora of Southern Australia: Rhodophyta Part IIIC. State Herbarium of Southern Australia, Adelaide Woolard FX, Moore RE, Roller PP (1979) Halogenated acetic and acrylic acids from the red alga Asparagopsis taxiformis. Phytochemistry 18:617–620 Wright JT, de Nys R, Poore AGB, Steinberg PD (2004) Chemical defenses in a marine alga: heritability and the potential for selection by herbivores. Ecology 85:2946– 2959 Yates JC, Peckol P (1993) Effects of nutrient availability and herbivory on polyphenolics in the seaweed Fucus vesiculosus. Ecology 74:1757–1766 Young DN (1977) Comparative fine structure and histochemistry of vesiculate cells in selected red algae. Ph.D Dissertation, University of California, Berkeley, U.S.A Young DN (1978) Ultrastructural evidence for a secretory function in the gland cells of the marine red alga Botryocladia pseudodichotoma (Rhodymeniaceae). Protoplasma 94:109–126 Young DN (1979a) Ontogeny, histochemistry and fine structure of cellular inclusions in vegetative cells of Antithamnion defectum (Ceramiaceae, Rhodophyta). J Phycol 15:42–48 Young DN (1979b) Fine structure of the ‘gland cells’ of the red alga Opuntiella californica (Soleriaceae, Gigartinales). Phycologia 18:288–95 Young DN, Howard BM, Fenical W (1980) Subcellular localization of brominated secondary metabolites in the red alga Laurencia snyderae. J Phycol 16:182–185 Young DN, West JA (1979) Fine structure and histochemistry of the vesicle cells of the red alga Antithamnion defectum (Ceramiaceae). J Phycol 15:49–57 Zangerl AR, Bazzaz FA (1992) Theory and pattern in plant defense allocation. In: Fritz RS, Simms EL (eds) Plant Resistance to Herbivores and Pathogens. University of Chicago Press, Chicago, p 361–393 Appendix One J. Phycol. 42, 637–645 (2006) r 2006 by the Phycological Society of America DOI: 10.1111/j.1529-8817.2006.00226.x

ULTRASTRUCTURE OF THE GLAND CELLS OF THE RED ALGA ASPARAGOPSIS ARMATA (BONNEMAISONIACEAE)1

Nicholas A. Paul2 School of Biological, Earth and Environmental Sciences, and Centre for Marine Biofouling and Bio-Innovation, University of New South Wales, Sydney 2052, Australia Louise Cole Australian Key Centre for Microscopy and Analysis, Electron Microscope Unit, University of Sydney, Sydney 2006, Australia Rocky de Nys School of Marine Biology and Aquaculture, James Cook University, Townsville 4811, Australia and Peter D. Steinberg School of Biological, Earth and Environmental Sciences, and Centre for Marine Biofouling and Bio-Innovation, University of New South Wales, Sydney 2052, Australia

Localization of natural products in the gland Key index words: chemical defense; cryofixation; cells of the tetrasporophyte of Asparagopsis armata freeze substitution; refractile inclusion; release; Harvey was examined using light microscopy, epi- secondary metabolite; TEM; vesicle cell fluorescence microscopy, and TEM. A. armata pro- duces a range of halogenated metabolites that deter herbivores and inhibit bacterial fouling. The halo- genated metabolites accumulate as a refractile in- clusion inside specialized gland cells and this Cellular inclusions of algae have diverse functions. inclusion was no longer produced when the alga They are important in metabolic reserves (Pueschel was cultured without bromine. Gland cells are 1992), cell wall construction (Schoenwaelder and Clay- formed soon after the apical division and can occu- ton 1998, 1999), and chemical defense (Young et al. py a large portion of the algal volume, up to 10% of 1980, Dworjanyn et al. 1999). Many types of inclusions some parts of the filament. TEM was carried out on are located free in the cytoplasm (Pederse´n et al. 1980, cryofixed and freeze-substituted samples. Ultra- Pueschel 1992, Pueschel and Korb 2001). However, structure studies revealed that gland cells are posi- inclusions can also be distinct bodies associated with tioned inside the pericentral cell, originating from particular cellular features, such as the cell wall, cuticle, the axial cell wall. The refractile inclusion of these or chloroplasts (Young 1979a, Pederse´n et al. 1980, gland cells is comprised of numerous electron- Pallaghy et al. 1983, Schoenwaelder and Clayton translucent vacuoles enclosed by an electron-opa- 1998). In many red algae, inclusions containing ref- que matrix. Some contents of the inclusion autoflu- ractile material are found in specialized cells (Wolk oresced under UV excitation by epifluorescence 1968, Young and West 1979, Young 1979b, Dworjanyn microscopy. Light microscopy further revealed et al. 1999). that stalk-like structures connected the gland cell Such specialized cells, typically known as vesicle or to the outer wall of the pericentral cell. These stalk- gland cells, are common in the Ceramiales, Bonnemai- like structures may provide the mechanism for soniales, and Gigartinales (Womersley 1996, 1998). metabolite transfer to the algal surface. Gland cell Functions attributed to these cells include light-collect- walls are relatively thin, which in turn would aid the ing bodies, and nutritional, excretory, and defense transfer of metabolites to the stalk-like structure. roles (summarized in detail by Young and West These features of the gland cells provide essential 1979). Some red algae produce chemicals that have clues to the production and storage of the halogen- potent biological effects (Fenical 1975, 1982), and thus ated metabolites in A. armata and offer new insights these specialized storage cells are of some interest, par- into a possible mechanism for their release. ticularly with respect to providing safe storage for harmful substances. However, apart from the work of Young (Young 1978, 1979b, Young and West 1979), 1 Received 6 October 2005. Accepted 22 February 2006. fine ultrastructural details of these cells are scarce, as is 2Author for correspondence and current address: School of Marine Biology and Aquaculture, James Cook University, Townsville 4811, evidence for their biological significance (Murray and Australia. e-mail [email protected]. Dixon 1992).

637 638 NICHOLAS A. PAUL ET AL.

The sensitivity of some active natural products (such together with light and epifluorescence microscopy, as phenolics) to traditional chemical fixation can im- and examine the cellular features that may be involved pede the ultrastructural examination of algae (Clayton in the extracellular release of these compounds. and Beakes 1983). As an alternative to chemical fixa- tion, a combination of cyrofixation and freeze substi- tution has proven useful in preserving sensitive MATERIALS AND METHODS features, such as fungal vacuoles or cytoskeleton, for Study organism. Tetrasporophytes of A. armata were col- TEM (Orlovich and Ashford 1993, Chretiennot-Dinet lected from the shallow subtidal at Bare Island (3315903800S, et al. 1997, Babuka and Pueschel 1998). To date, cryo- 15111400000E), Sydney, Australia. The filamentous tetra- techniques have not been used to examine the sporophyte is found throughout the year as an epiphyte (pre- ultrastructure of specialized cells or localization of sec- dominantly on the coralline turf). Algae were maintained in culture at 191 C in sterile seawater with a 16:8 light:dark cycle ondary metabolites in red algae. under 20 mmol photons m 2 s 1 with 36 W daylight bulbs Although refractile inclusions are a common feature before ultrastructural examination. Freshly collected pieces in cells of red algae, their simple presence gives little were used for light and epifluorescence microscopy. indication of the types or roles of stored compounds. To examine the presence of gland cells in young tetras- Differences in histochemistry of stored material and porophytes, female gametophytes with cystocarps were collect- TEM identification of organelles for secretion (includ- ed from the field. Carpospores were released after excising the ing abundant smooth endoplasmic reticulum) provide cystocarp and placing it in sterile seawater overnight. Germi- nating carpospores were monitored over subsequent days. some evidence for the function of vesicle cells (Young Light microscopy. A. armata was viewed and photographed 1977, 1978, Young and West 1979). Further means to using a Leica DM LB microscope fitted with a Leica DC100 localize natural products in the red algae rely on the digital camera (Leica Microsystems, Inc., Bannockburn, IL, inherent properties of the stored compounds and in- USA). Samples were mounted (unstained) in seawater and clude the X-ray analysis of organically bound halogens viewed with bright-field optics to examine the location of (Pederse´n et al. 1980, Young et al. 1980) or the auto- gland cells, as well as the size (volume) of gland cells. Gland fluorescence of known UV-active metabolites using cell sizes were readily determined by the size of the refractile inclusion. The formation of gland cells was examined in api- epifluorescence microscopy (Dworjanyn et al. 1999). cal regions of collected individuals, in carpospores, and in Relatively few studies, however, have adequately re- recently metamorphosed tetrasporophytes. lated the localization of known natural products to To determine when gland cells were produced relative to specific ecological functions in red algae (Young et al. apical cell division, 15 individuals were collected and the po- 1980, de Nys et al. 1998, Paul et al. 2006). For surface- sition of the first observed gland cells from the apical cell was mediated interactions, such as the regulation of micro- measured in one filament of each. The frequency of pericen- tral cells that contained a gland cell was also measured for 20 and macro-fouling, the presence of algal metabolites at pericentral cells in each of 15 individuals. Furthermore, the the surface is essential (Steinberg and de Nys 2002). volume that the gland cells occupied in the alga was measured However, the release of metabolites is not a standard by calculating the volume of the refractile inclusion in the function of specialized cells in red algae (Young 1977, gland cells relative to the volume of the corresponding tier of 1979b, Young and West 1979) and has rarely been de- cells (a cell tier is comprised of three pericentral cells and one termined (Young 1978, de Nys et al. 1998, Paul et al. axial cell). The average volume for six randomly selected cell tiers was used to calculate the overall average of percent 2006). Accordingly, the structure and function of the volume of gland cells in seven individuals. specialized cells in most red algae remain as yet un- Manipulation of gland cells. The development of the ref- known (Murray and Dixon 1992). ractile inclusion in the gland cells of algae cultured in two Members of the Bonnemaisoniaceae form special- types of artificial culture medium (Provasoli 1968), with bro- ized cells (Wolk 1968, Young 1977, Womersley 1996) mine (bromide ions at 65 mg/L: Burton 1996) or without and also produce a diverse array of halogenated met- bromine, was observed using light and fluorescence micro- abolites (Fenical 1975, McConnell and Fenical 1977, de scopy (culture conditions as above). For light microscopy, the presence or absence of the ref- Nys et al. 1998). The internal localization of these ractile inclusion in the gland cell was observed in algae cul- halogenated metabolites is currently limited to the tured with or without bromine. In order to determine whether identification of halogens in the specialized cells of any absence of the refractile inclusion was correlated with an Bonnemaisonia nootkana and Asparagopsis armata by elec- absence of the gland cell, fluorescence microscopy of 40,6-di- tron microprobe analysis (Wolk 1968), and the locali- amidino-2-phenylindole (DAPI)-stained nuclei was used on A. zation of furanones in the gland cells of Delisea pulchra armata cultured with and without bromine (Goff and Coleman 1990). Here, algae were microwaved for 15 s (2) in seawater by epifluorescence microscopy (Dworjanyn et al. containing 0.5 mg mL 1 DAPI, with 1 min between treat- 1999). Although A. armata was presumed to release ments. Samples were left for 0.5 h in darkness before viewing its cellular contents (McConnell and Fenical 1977), with a Leica DM LB fluorescence microscope. perhaps through cytolysis (as suggested for B. nootkana Epifluorescence microscopy. Small pieces of the tetras- by Young 1977), there was no empirical evidence for porophyte of A. armata were mounted on a glass slide and extracellular release of metabolites until recently (Mar- imaged using a Leica DM LB microscope fitted with a 100 W shall et al. 2003, Paul et al. 2006). In the present study, Hg fluorescence lamp. A UV filter set (excitation wavelength from 360 to 395 nm) was used to examine the gland cells, as we investigate for the first time the localization of the certain metabolites autofluoresce under UV excitation natural products in the tetrasporophyte of A. armata (O’Brien and McCully 1981). This may include some of the using TEM of cryo-fixed, freeze-substituted material, major compounds present in A. armata that are known to GLAND CELL ULTRASTRUCTURE 639 have conjugated double bonds, such as 3,3-dibromoacrylic adjacent to the apical cell (Fig. 1C). The frequency of acid (McConnell and Fenical 1977). gland cells per pericentral cell beyond the apical re- TEM of cryofixed and freeze-substituted material. Small pieces gion in field collected material was 99% ( 2 SD). of tissue from the tetrasporophyte of A. armata were plunge- Refractile inclusions were not observed in carpo- frozen in liquid propane cooled to 1901 C with liquid ni- trogen. Samples were freeze substituted in 2% osmium tetr- spores. However, after germination, the carpospore oxide in dry acetone at 851 C for 72 h, 301 C for 24 h, material divided, forming apical (arrow) and rhizoi- androomtemperaturefor1h(warmedupat51/h between dal (not shown) filaments (Fig. 1D). The gland cells temperatures). Samples were washed in acetone (four times, within the original spore material had prominent in- 15 min) to remove osmium tetroxide, and infiltrated with clusions post-germination (Fig. 1D, arrowhead). Spurr’s resin (Spurr 1969) in 1 h steps of 1:3, 1:1, 3:1 (resin: These inclusions were sometimes larger than those acetone), and one exchange overnight in Spurr’s resin only. Infiltrated tissue was flat-dish embedded in resin between in gland cells in the adjacent pericentral cells of the two PolyTetraFluoroEthylene (PTFEt)-coated slides separat- new branches (Fig. 1D, arrows). ed by coverslips, and polymerized at 601 C for 14 h. Samples In detail, gland cells characteristically contained a were excised from resin, and ultrathin sections were cut large refractile inclusion, whose surface appears dim- (90 nm) in the transverse and longitudinal planes with a dia- pled, similar to that of a golf ball. The internal space mond knife on a Reichert Ultracut ultramicrotome (Moc that the gland cells occupy was on average 3.77% Inc., Valley Cottage, NJ, USA). Sections stained with tolui- dine blue were previewed by light microscopy. Selected sec- ( 1.02 SD) of the algal volume at each cell tier (com- tions were post-stained with uranyl acetate (10 min) and lead prised of one axial cell plus the three surrounding citrate (Reynolds 1963) and examined on a Hitachi H7000 pericentral cells). The gland cell inclusion enlarged as transmission electron microscope (100 kV) (Hitachi High the alga grew; the largest gland cell measured was Technologies America, Inc., Pleasanton, CA, USA). 1075 mm3. The contents of this inclusion appeared to be highly toxic to the pericentral cell, because if the RESULTS gland cell wall was breached, the surrounding peri- Light microscopy. The tetrasporophyte of A. armata central cell quickly becomes severely disrupted. is a multiseriate filamentous alga with indeterminate As gland cells were formed at the axial wall of the branching, comprised of a central axial cell and three pericentral cell (Fig. 1A), in order for their metabolites pericentral cells (Fig. 1, A and B). Pericentral cells to be released to the surface, some additional transport produced gland cells from the axial face of their cell mechanism was required. A stalk-like structure con- walls (Fig. 1B). These gland cells were formed on av- nected the gland cell to the opposite (outer) wall of the erage 3.6 ( 1.5 SD) pericentral cells from the apical pericentral cell (Fig. 2, A and B). This stalk-like struc- cell, but are frequently observed in the tier of cells ture was either a fine structure (Fig. 2A) or a thicker

FIG. 1. Light micrographs of the tetrasporophyte of Asparago- psis armata. (A) Branching form of A. armata showing the position of the gland cells (arrows). (B) Transverse section showing three pericentral cells (p), sur- rounding the axial cell (a). In this section, two gland cells (g) are ensheathed by pericentral cells. (C) Apical meristem show- ing the location of the gland cells (arrowheads) in the cell layer adjacent to the apical cell (ap). (D) Recently settled tetrasporo- phytes produce gland cells in the original spore material (arrow- heads), as well as in subsequent new growth of filament (arrow). Scale bars: A, B, and D 5 10 mm; C 5 5 mm. 640 NICHOLAS A. PAUL ET AL.

FIG. 3. Effects of bromine (Br) manipulation on cell struc- ture. (A) Br [ þ ] algae. Pericentral cells contain a gland cell with a large refractile inclusion (arrows). (B) Br [ ] algae. The inclu- sion is no longer produced in cells grown without bromine, al- though the structures associated with the gland cell are still formed (arrows). Scale bars, 10 mm.

Bromine manipulation. When A. armata was cul- tured in artificial media with bromine, the refractile inclusion of the gland cell was the dominant struc- tural feature (Fig. 3A). Algae grown in bromide- free media no longer produced refractile inclusions (Fig. 3B). However, the absence of bromine did not limit the formation of the gland cells. The DAPI- stained nuclei of the gland cells were observed in

FIG. 2. Light micrographs of gland cells. Material is living, apical regions of algae grown with and without bro- unstained thalli. (A) A stalk-like structure (arrowhead) connect- mine (not shown). The nuclei of these gland cells be- ing the gland cell (arrow) to the surface region of the pericentral came harder to visualize in older parts of the alga. cell. Scale bar, 10 mm. (B) In other cells, the structure appears However, the stalk-like structures associated with thicker (arrowheads). Scale bar, 10 mm. (C) For large gland cells (g), the cell wall can be tight against the outer wall of the peri- the gland cells were also formed in the absence of central cell (p) (arrowhead). Scale bar, 5 mm. bromide ions (Fig. 3B). Epifluorescence microscopy. The contents of the in- feature (Fig. 2B). The latter often occurred when the clusion inside the gland cell fluoresced blue under relative size of the gland cell to the pericentral cell was UVexcitation (Fig. 4), consistent with the presence of large. It appeared that the stalk-like structure was con- some metabolites with conjugated double bonds. nected near the outer wall of the pericentral cell TEM. The ultrastructure of gland cells in A. ar- (Fig. 2A). In some large gland cells, the refractile in- mata was well preserved by cryofixation and freeze clusion seemed to be tightly appressed to the outer cell substitution, as shown by TEM (Figs. 5–7). The TEM wall of the pericentral cell, excluding the contents of micrograph showed that the gland cells were formed the cytoplasm from this region of the pericentral cell inwardly from the axial wall of the pericentral (Fig. 2C). cell (Figs. 5 and 6A). Within the gland cell, a large GLAND CELL ULTRASTRUCTURE 641

FIG. 4. Epifluorescence micrograph of a gland cell using a UV filter set. The contents of the refractile inclusion of the gland cell (arrow) emit blue autofluorescence under UV excitation. m Scale bar, 10 m. FIG. 5. Transmission electron micrograph of Asparagopsis ar- mata in longitudinal section. Gland cells (g) are formed inside the pericentral cells (p) from the inner wall, surrounding the central inclusion was anchored to the cell wall (Fig. 5, arrow). axial cell (a). The large inclusion in the gland cell is attached to m This inclusion was highly vacuolated, containing the inner gland cell wall (arrow). Scale bar, 5 m. many electron-translucent vacuoles (v) embedded in an electron-opaque matrix (Fig. 6A). Some of the vacuoles appeared to be partially osmiophilic (Fig. cases, the gland cell wall was tightly appressed to the 6A, arrowheads), although these more likely repre- outer wall of the pericentral cell (Fig. 7B), confirming sent artifacts due to a glancing section of the vacuole similar predictions by light microscopy (Fig. 2C). membrane. The gland cell was largely filled by a single refractile The periphery of the refractile inclusion in the inclusion; however, organelles including chloroplasts, gland cells was dimpled at points where vacuoles mitochondria, and a nucleus were also present (Fig. 6, were pressing into the cytoplasm (Fig. 6A, white ar- A and B). We also observed some ultrastructural rows). The protrusion of these vacuoles at the periph- characteristics of the axial and pericentral cells. The ery of the inclusion indicated that the osmiophilic major features of the pericentral cells were chloroplasts matrix was not a rigid frame. The refractile inclusions and large electron-translucent vacuoles (Fig. 7, A and contained vacuoles of a variety of sizes, up to 700 nm in B). Floridean starch was often observed in pericentral diameter (Fig. 7). The vacuoles of large gland cells cells (Fig. 6A), but was not observed in either gland were generally larger than those of smaller cells (Fig. cells or axial cells. The axial cell appeared to contain 7A, small (s) and large (L) gland cells). Where cross numerous mitochondria and relatively few chloro- walls had formed for branching (Fig. 7A), some peri- plasts (Fig. 6A). central cells did not appear to contain a gland cell. The refractile inclusions of gland cells were well preserved throughout, even where some pericentral cells showed DISCUSSION signs of freezing damage (i.e. granulation in pericen- Structure–function relationships are crucial to ma- tral cell, Fig. 7A). rine chemical ecology, as the ecological role of a natural There was no evidence of the stalk-like structures product is essentially dictated by the nature of the (Fig. 2) connecting the gland cell to the outer wall by structure in which it is produced. This structure–func- TEM. This may have represented the sensitivity of the tion relationship is particularly important for surface- stalk-like structure to the preparatory process. The cell mediated roles such as antifouling (Steinberg and de walls of the gland cells were relatively thin compared Nys 2002). We recently provided empirical evidence with those of surrounding cells (Figs. 5–7). In some for the storage and release of halogenated metabolites 642 NICHOLAS A. PAUL ET AL.

FIG. 6. Images from TEM of the gland cells of Asparagopsis armata. (A) The gland cell is attached at the inner wall (cw) of the pericentral cell. Many electron-translucent vacuoles (v) are present in the central inclusion. Some vacuoles appear to be lightly stained (arrowheads). The membrane sur- rounding the inclusion is dimpled (ar- rows) where the vacuoles are pressing against it. The pericentral cell contains numerous chloroplasts (c) and grains of floridean starch (fs) in the cytoplasm. The axial cells have numerous organ- elles, in particular mitochondria (m). (B) Details of a gland cell showing chloroplasts (c) and a nucleus (n). Some chloroplasts have electron-trans- lucent inclusions (arrow). Note the rel- atively thin wall of gland cell (arrowheads) compared with other cells. (A and B) Scale bars, 1 mm. ew, external cell wall.

from the gland cells of A. armata, and showed that these When A. armata was cultured without bromine, the same metabolites inhibit epiphytic bacteria (Paul et al. large, refractile inclusion typical of the gland cells was 2006). Here, we examined in detail the cellular fea- no longer formed, indicating that the inclusion is the tures involved in both the storage and release of nat- site of halogenated metabolite storage. Culture mani- ural products in A. armata, correlating light and pulations on other Bonnemaisoniaceae algae have epifluorescence microscopy, TEM, and media manip- demonstrated an integral role for bromine in the me- ulations with the localization of halogenated meta- tabolism of the gland cells (Wolk 1968, Dworjanyn bolites in the specialized gland cells. et al. 1999). Furthermore, algae that lack specialized Gland cells were formed soon after apical cell divi- cells can be cultured in the absence of bromine with sion in the filamentous tetrasporophyte of A. armata. minimal effects on their growth or reproduction (Fries GLAND CELL ULTRASTRUCTURE 643

FIG. 7. (A) Vacuoles in the inclusion vary in size; small gland cells (s) generally have small vacuoles. Larger gland cells (L) have larger inclusions and larger vacuoles. Pericentral cells contain typical vacuoles (pv). (B) The cell wall of a large gland cell appears to be tightly appressed to the surface wall of the pericentral cell wall (arrowheads). Scale bars, 5 mm.

1966, McLachlan 1977). Such data infer that the spe- the small number of organelles in the cytoplasm of the cialized bromine metabolism in the Bonnemaisonia- gland cell implies that this is not the case. ceae has a secondary function. An electron-opaque matrix contains and separates The ultrastructure of the gland cells in A. armata was the vacuoles in the inclusion of the gland cells of A. well preserved by cryofixation and freeze substitution, armata. This intensely stained matrix could represent suggesting that this preparation method is useful for protein-rich regions (Pederse´n et al. 1980, O’Brien the TEM preparation of algae with active natural prod- and McCully 1981), such as the enzymes and ribo- ucts. Gland cells in A. armata are produced from the somes required for the production of the vacuolar axial wall of the pericentral cells, and occupy space in- contents, but may also consist of the metabolic precur- side these cells. Although it is ensheathed by the peri- sors for the halogenated compounds—metabolic path- central cell, the gland cell is discrete and contains ways are described in McConnell and Fenical (1977). chloroplasts, mitochondria, and a nucleus. These or- The ultrastructure of the specialized cells in A. armata ganelles are confined to the peripheral cytoplasm, as differs from those of Bonnemaisonia nootkana (Bonne- the internal space of each gland cell is dominated by a maisoniaceae), in which the cells have a single electron- large inclusion. Previously, the halogenated natural translucent vacuole with no such matrix (Young 1977). products in the ‘‘vesicle’’ cells of A. armata had been The vesicle cells of B. nootkana may die as a conse- assumed to accumulate in a refractile vesicle or vacuole quence of accumulating brominated compounds (Wolk 1968, Young 1977, Marshall et al. 2003). How- (Young 1977). However, it appears that the gland cells ever, our TEM results indicate that the ‘‘vesicle’’ of the of A. armata effectively store similar compounds with- gland cells in A. armata is an inclusion, which itself is out senescence. comprised of many small vacuoles. Histochemical stains (Young 1977, Pueschel 1992), Most of the vacuoles within the inclusion of the electron-microprobe analysis (Wolk 1968, Pederse´n gland cells were electron-translucent. This absence of et al. 1980, Young et al. 1980), and epifluorescence vacuolar staining is consistent with the presence of microscopy (Clayton and Ashburner 1994, Dworjanyn some saturated halogenated compounds in A. armata, et al. 1999, Schoenwaelder and Clayton 1999) utilize including bromoform, which can be present at levels the properties of natural products to aid in their local- up to 4% of the dry mass of the alga (Paul et al. 2006). ization in the alga. We have demonstrated that the in- It is unclear from the ultrastructural examination clusion in the gland cells of A. armata autofluoresces whether the inclusion represents the sites of synthesis blue under ultra-violet excitation. As some of the com- for these metabolites or whether the vacuoles are the pounds in A. armata will autofluoresce, it provides fur- accumulation sites for metabolites formed elsewhere in ther evidence for their localization in the refractile the cell. Other algal metabolites are synthesized in the inclusion of the gland cells. Alternatively, microprobe chloroplasts (Pohnert and Jung 2003) or on endoplas- analyses have shown that certain algal cells—including mic reticula (Schoenwaelder 2002). This could also be the gland cells of A. armata—accumulate halogens at true for the halogenated compounds of A. armata,but levels much higher than surrounding tissue (Wolk 644 NICHOLAS A. PAUL ET AL.

1968, Young et al. 1980). Such halogens are typically never devoid of their refractile inclusion when cul- presumed to be in an organic form (Pederse´netal. tured in seawater, and considering that epiphytic bac- 1980, Young et al. 1980), as appears to be the case for teria are inhibited by newly formed and older gland the gland cells of A. armata. cells of the filamentous tetrasporophyte of A. armata We have demonstrated that the gland cells of A. ar- (Paul et al. 2006), it is likely that the metabolite release mata are still formed in bromide-free media but with- is a continuous process rather than a large, single out a refractile inclusion. This was also described by event. However, the precise mechanism of release re- Wolk (1968). Marshall et al. (2003) inferred that the mains to be determined. frequency of gland cells in Asparagopsis sp. was corre- Much of the previous interest in the specialized cells lated to the level of bromide in the culture media; and structures of red algae has been related to their however, it was not clear whether the densities of non- taxonomic value such as the position and development vesiculate gland cells were affected (i.e. the number of of gland cells (Moe and Silva 1980, Womersley 1998), gland cells in algae cultured in bromide-free media). It crystal morphology (Pueschel 1992), and the numbers would be surprising if gland cell production is control- of cellular inclusions can be distinguishing taxonomic led by bromine, considering that bromide concentra- features, particularly in the Ceramiales. The term tions do not vary greatly in seawater (Burton 1996). ‘‘gland cell’’ is often used in taxonomic descriptions Furthermore, we saw no evidence that the mainte- (Womersley 1998) but may lead to the incorrect as- nance of the inclusion was affected by culturing with- sumption that all gland cells are secretory (Young out bromine, as gland cells from algae originally 1978). Consequently, the specialized cells of A. armata cultured in seawater did not lose the inclusion when have been referred to as ‘‘vesicle’’ cells, as a secretory transferred to bromide-free media. In any respect, the function was not previously known (Wolk 1968, Young absence of the gland cell inclusion, not an absence of 1977, Marshall et al. 2003). However, recent works the gland cell, is consistent with the results of Delisea (Marshall et al. 2003, Paul et al. 2006) indicate that the pulchra (Bonnemaisoniaceae) grown without bromine specialized cells are true gland cells, a terminology (Dworjanyn et al. 1999). consistent with other members of the Bonnemaisoni- Specialized cells and structures are typically found ceae (Womersley 1996). in the surface cell layers of red algae, even when not Detailed work on ultrastructure has proven integral involved with surface release (Young 1979b, Young to structure–function relationships in algae. This is et al. 1980). Most of the gland cells within the Bonne- highlighted by the primary roles for phenolic com- maisoniaceae are formed in the surface cell layer (Wo- pounds in cell wall construction in the brown algae mersley 1996). Interestingly, the gland cells in A. (Clayton and Ashburner 1994, Schoenwaelder and armata do not abut the surface. However, the pericen- Clayton 1999, Schoenwaelder 2002). Before these tral cell does not represent a barrier to the release of studies, phenolics were principally considered as sec- metabolites from the gland cell that is located on its ondary products due to their anti-herbivore properties axial wall. Stalk-like structures connect the gland cell to (Ragan 1976, Schoenwaelder 2002). Therefore, fur- the outer wall region of the pericentral cell and likely ther developing structure–function relationships for facilitate the movement of metabolites for their release the many specialized cells of red algae should also to the surface (Paul et al. 2006). Such a structure would prove important in elucidating the ecological functions avoid large-scale exposure (i.e. autotoxicity) of the of their natural products. pericentral cell to the halogenated metabolites pro- duced and released by the gland cells. Considering the We would like to thank D. Davies for assistance with the TEM presupposition by McConnell and Fenical (1977) that preparatory work. We also thank two anonymous reviewers for comments that contributed substantially to the manuscript. metabolites diffuse from the gland cell to the exterior and the detailed descriptive work on Asparagopsis spp. by Bonin and Hawkes (1987), it is surprising that these Babuka, S. J. & Pueschel, C. M. 1998. A freeze-substitution ultra- structures have not been described previously. structural study of the cytoskeleton of the red alga Antitham- Unfortunately, the stalk-like connection was not ob- nion kylinii (Ceramiales). Phycologia 37:251–8. served in the TEM sections. However, it would be dif- Bonin, D. R. & Hawkes, M. W. 1987. Systematics and life histories of New-Zealand Bonnemaisoniaceae (Bonnemaisoniales, Rho- ficult to determine transient release of metabolites dophyta). I. The genus Asparagopsis. N.Z. J. Bot. 25:577–590. from the gland cells using TEM and it is possible that Burton, J. D. 1996. The ocean: a global chemical system. In Sum- these structures were sensitive to the preparatory merhayes, C. P. & Thorpe, S. A. [Eds.] Oceanography: An Illus- method. The relatively thin walls of the gland cells trated Guide. Manson Publishing Ltd., London, 166 pp. (observed by TEM) could aid in the transfer of material Chretiennot-Dinet, M., Giraud-Guille, M., Vaulot, D., Putaux, J., Saito, Y. & Chanzy, H. 1997. The chitinous nature of fila- from the gland cell to the stalk-like structure. Further- ments ejected by Phaeocystis (Prymnesiophyceae). J. Phycol. 33: more, the cell walls of larger gland cells were tightly 666–72. appressed to the outer wall of the pericentral cell. The Clayton, M. N. & Ashburner, C. M. 1994. Secretion of the phenolic presence of connective structures linking the gland cell bodies following fertilisation in Durvillaea potatorum (Durvillae- ales, Phaeophyta). Eur. J. Phycol. 29:1–9. to the surface, as well as the appressed cell wall of some Clayton, M. N. & Beakes, G. W. 1983. Effects of fixatives on the large gland cells, indicate a potential means for the re- ultrastructure of physodes in vegetative cells of Scytosiphon lease of gland cell contents. Because gland cells were lomentaria (Scytosiphonaceae, Phaeophyta). J. Phycol. 19:4–16. GLAND CELL ULTRASTRUCTURE 645 de Nys, R., Dworjanyn, S. A. & Steinberg, P. D. 1998. A new meth- Pohnert, G. & Jung, V. 2003. Intracellular compartmentation in od for determining surface concentrations of marine natural the biosynthesis of caulerpenyne: study on intact macroalgae products on seaweeds. Mar. Ecol. Prog. Ser. 162:79–87. using stable-isotope-labeled precursors. Org. Lett. 5:5091–3. Dworjanyn, S. A., de Nys, R. & Steinberg, P. D. 1999. Localization Provasoli, L. 1968. Media and prospects for the cultivation of and surface chemistry of secondary metabolites in the red alga marine algae. In Watanabe, A. & Hattori, A. [Eds.] Culture Delisea pulchra. Mar. Biol. 133:727–36. and Collections of Algae. Proceedings of the U.S.-Japanese Confer- Fenical, W. 1975. Halogenation in the Rhodophyta: a review. ence, Hakone. Japanese Society of Plant Physiology, Tokyo, J. Phycol. 11:245–59. pp. 63–75. Fenical, W. 1982. Natural products chemistry in the marine envi- Ragan, M. A. 1976. Physodes and the phenolic compounds of ronment. Science 215:923–8. brown algae: composition and significance of physodes in vivo. Fries, L. 1966. Influence of iodine and bromine on growth of some Bot. Mar. 19:145–54. red algae in axenic culture. Physiol. Plant. 19:800–8. Reynolds, E. S. 1963. The use of lead citrate at high pH as an Goff, L. J. & Coleman, A. W. 1990. Microspectrofluorometric anal- electron-opaque stain in electron microscopy. J. Cell. Biol. ysis of algal DNA. In Cole, K. M. & Sheath, R. G. [Eds.] Biology 17:208–12. of the Red Algae. Cambridge University Press, Cambridge, pp. Schoenwaelder, M. E. A. 2002. The occurrence and cellular signif- 43–72. icance of physodes in brown algae. Phycologia 41:125–39. Marshall, R. A., Hamilton, J. T. G., Dring, M. J. & Harper, D. B. Schoenwaelder, M. E. A. & Clayton, M. N. 1998. Secretion of phe- 2003. Do vesicle cells of the red alga Asparagopsis (Falkenbergia nolic substances into the zygote wall and cell plate in embryos stage) play a role in bromocarbon production? Chemosphere of Hormosira and Acrocarpia (Fucales, Phaeophyceae). J. Phycol. 52:471–5. 34:969–80. McConnell, O. J. & Fenical, W. 1977. Halogen chemistry of the red Schoenwaelder, M. E. A. & Clayton, M. N. 1999. The role of the alga Asparagopsis. Phytochemistry 16:367–74. cytoskeleton in brown algal physode movement. Eur. J. Phycol. McLachlan, J. 1977. Effects of nutrients on growth and develop- 34:223–9. ment of embryos of Fucus edentatus Phaeophyceae Fucales. Spurr, A. R. 1969. A low-viscosity epoxy resin embedding medium Phycologia 16:329–38. for electron microscopy. J. Ultrastruct. Res. 26:31–43. Moe, R. L. & Silva, P. C. 1980. Morphological and taxonomic stud- Steinberg, P. D. & de Nys, R. 2002. Chemical mediation of coloni- ies on Antarctic Ceramiaceae (Rhodoohyceae). II. Pterotham- zation of seaweed surfaces. J. Phycol. 38:621–9. nion antarcticum (Kylin) Comb. Nov. (Antithamnion antarcticum Wolk, C. P. 1968. Role of bromine in the formation of the refractile Kylin). Br. Phycol. J. 15:1–17. inclusions of the vesicle cells of the Bonnemaisoniaceae (Rho- Murray, S. M. & Dixon, P. S. 1992. The Rhodophyta: some aspects dophyta). Planta (Berlin) 78:371–8. of their biology. III. Oceanogr. Mar. Biol. Annu. Rev. 30:1–148. Womersley, H. B. S. 1996. The Marine Benthic Flora of Southern Aus- O’Brien, T. P. & McCully, M. E. 1981. The Study of Plant Structure: tralia: Rhodophyta Part IIIB. Australian Biological Resources Principles and Selected Methods. Termarcarphi Pty Ltd, Mel- Study, Canberra. bourne, pp. 2.27–2.28. Womersley, H. B. S. 1998. The Marine Benthic Flora of Southern Aus- Orlovich, D. A. & Ashford, A. E. 1993. Polyphosphate granules are tralia: Rhodophyta Part IIIC. State Herbarium of Southern Aus- an artefact of specimen preparation in the ectomycorrhizal tralia, Adelaide. fungus Pisolithus tinctorius. Protoplasma 173:91–102. Young, D. N. 1977. Comparative Fine Structure and Histochemistry of Pallaghy, C. K., Minchinton, J., Kraft, G. T. & Wetherbee, R. 1983. Vesiculate Cells in Selected Red Algae. Ph.D. dissertation, Univer- Presence and distribution of bromine in Thysanocladia densa sity of California, Berkeley. (Solieriaceae, Gigartinales), a marine red alga from the Great Young, D. N. 1978. Ultrastructural evidence for a secretory func- Barrier Reef. J. Phycol. 19:204–8. tion in the gland cells of the marine red alga Botryocladia pseu- Paul, N. A., de Nys, R. & Steinberg, P. D. 2006. Chemical defence dodichotoma (Rhodymeniaceae). Protoplasma 94:109–26. against bacteria in the red alga Asparagopsis armata: linking Young, D. N. 1979a. Ontogeny, histochemistry and fine structure structure with function. Mar. Ecol. Prog. Ser. 306:87–101. of cellular inclusions in vegetative cells of Antithamnion defectum Pederse´n, M., Roomans, G. M. & Hofsten, A. V. 1980. Cell inclu- (Ceramiaceae, Rhodophyta). J. Phycol. 15:42–8. sions containing bromine in Rhodomela confervoides (Huds.) Young, D. N. 1979b. Fine structure of the ‘‘gland cells’’ of the red Lamour. and Polysiphonia elongata Harv. (Rhodophyta; Cerami- alga Opuntiella californica (Soleriaceae, Gigartinales). Phycologia ales). Phycologia 19:153–8. 18:288–95. Pueschel, C. M. 1992. An ultrastructural survey of the diversity of Young, D. N. & West, J. A. 1979. Fine structure and histochemistry crystalline, proteinaceous inclusions in red algal cells. Phycolo- of the vesicle cells of the red alga Antithamnion defectum gia 31:489–99. (Ceramiaceae). J. Phycol. 15:49–57. Pueschel, C. M. & Korb, R. E. 2001. Storage of nitrogen in the Young, D. N., Howard, B. M. & Fenical, W. 1980. Subcellular form of protein bodies in the kelp Laminaria solidungula. Mar. localization of brominated secondary metabolites in the red Ecol. Prog. Ser. 218:107–14. alga Laurencia snyderae. J. Phycol. 16:182–5.