Effects of Macroalgal Hosts on the Growth and Epiphytic Behavior

EFFECTS OF MACROALGAL HOSTS ON THE GROWTH AND EPIPHYTIC BEHAVIOR

OF FIVE GAMBIERDISCUS SPECIES FROM THE GREATER CARIBBEAN REGION

A Thesis

Presented to

The Faculty of the College of Arts and Sciences

Florida Gulf Coast University

In Partial Fulfillment

Of the Requirement for the Degree of

Master of Science

Lacey Kay Rains

2015 APPROVAL SHEET

This thesis is submitted in partial fulfillment of

the requirements for the degree of

Master of Science

______

Lacey Kay Rains

Approved:

______Michael Parsons, Ph.D. Committee Chair / Advisor

______Ai Ning Loh, Ph.D.

______Mindy Richlen, Ph.D.

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline. i

ACKNOWLEDGMENTS

I would like to thank my major advisor Dr. Michael Parsons for his continuous support and inspiration throughout my entire time as a graduate student at FGCU. His guidance, expertise, and continued confidence in my abilities throughout my research and writing process were invaluable. Thanks to his support and friendship, along with the various travel and field opportunities he provided, I could not imagine a better experience as a graduate student than mine was. I would also like to thank my committee members, Dr.

Ai Ning Loh and Dr. Mindy Richlen, for their insight and support during my writing process, as well as the commitment to be a part of my work, even from long distances.

Both Alex Leynse and Ashley Brandt played a large role in my research experiences, as we all began on this ciguatera journey, alongside Mike, together (as team Firecrest). Both above and below the water, they both were a tremendous help with every aspect of my lab work and field collections, and they always made the trips to the Florida Keys enjoyable, even under some of the coldest and windiest conditions.

The support, encouragement, and friendship of so many other graduate students have gotten me to this point. Rheannon Ketover was helpful in too many ways to mention, and

Laura Markley was invaluable for her help with culturing over the years. The friendship and guidance from both Lesli Haynes and Erin Rasnake over the past 15 years has played a large role in where I am today.

I would like to recognize the ECOHAB funding grant for making my research possible, as it funded the field work that allowed for field collections that were vital for my research.

Finally, I owe my largest thanks to my Mom and Dad, who have always supported me enormously in my educational journey. My Dad provided with a great sense of adventure and love for the outdoors and the water that has led me to where I am today, and I could not be more thankful for that. Thank you to my entire family and my husband for their continued support and encouragement in everything that I do.

iii

ABSTRACT

Ciguatera fish poisoning is caused by toxins produced by dinoflagellates of the genus

Gambierdiscus. This genus has recently been revised and new research on the physiology and ecology of the revised species is needed. While it has been demonstrated that

Gambierdiscus spp. are predominately epiphytic, there is also evidence that they are conditional, not obligate, epiphytes and that not all algae hosts are preferred equally by

Gambierdiscus populations. This experiment exposed five Caribbean species of

Gambierdiscus to living conditions among 8 different species of macroalgal hosts, and their epiphytic behavior (attachment vs. non-attachment) and growth were monitored over 29 days. Additionally, the experiment was carried out under two separate nutrient conditions, ambient versus enriched Florida Keys seawater.

Results demonstrate variable responses in epiphytic behavior within the Gambierdiscus genus to different macroalgal hosts, and stimulation and/or inhibition of Gambierdiscus growth by different macroalgal hosts. Attachment data indicate that Gambierdiscus populations prefer attachment to hosts that have a filamentous structure, but those species did not always support high cell abundances. Certain algal host species appear to stimulate growth of some Gambierdiscus species while others seem to inhibit the growth of other species. Control treatments (no host) reveal that some Gambierdiscus reached higher cell abundances without the presence of any host algae, suggesting inhibition by the hosts. No overall attachment preference was observed among Rhodophytes,

Chlorophytes, and Phaeophytes, and no phyla stimulated growth or cell abundance more than the others. There was also no difference in attachment to hosts that were nutrient- enriched versus those that were not. The variability of growth responses and attachment

iv behavior to different hosts by the various Gambierdiscus species in this experiment add complexity to our understanding of the epiphytic nature of Gambierdiscus and the dynamics of blooms.

TABLE OF CONTENTS

Acknowledgments ...... i Abstract ...... iii Table of Contents ...... v List of Tables ...... vi List of Figures ...... vii

Chapter 1: General Introduction ...... 1

Chapter 2: Macroalgal host effects on the growth and epiphytic behavior of Gambierdiscus ...... 17 Introduction ...... 18 Research Purpose ...... 22 Methods ...... 23 Results ...... 27 Discussion ...... 42

Chapter 3: Ambient vs. enriched host nutrient condition effects on the growth and epiphytic behavior of Gambierdiscus ...... 57 Introduction ...... 58 Research Purpose ...... 62 Methods ...... 63 Results ...... 67 Discussion ...... 86

Chapter 4: Overall discussion ...... 90

References ...... 105

Appendices ...... 119

LIST OF TABLES

Chapter 2 Table 2-1 Description of host algae used in experiments ...... 24

Table 2-2 Growth rates of the various Gambierdiscus species for each of the host treatments...... 30

Table 2-3 Abundance (# of cells at end of experiment, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments ...... 34

Table 2-4 Average attachment, growth, and abundance data for each Gambierdiscus, with the host that supported the highest attachment ...... 34

Table 2-5 Average attachment (% of total cells that were attached to host, averaged across triplicates and throughout entirety of experiment) of the various Gambierdiscus species for each of the host treatments ...... 37

Table 2-6 Averaged (by triplicate) attachment, cell abundance, and growth rate data is categorized for each of the host treatments ...... 39

Table 2-7 Percent biomass changes for each host algae, averaged across all treatments ...... 41

Chapter 3 Table 3-1 Average growth rate values for all ambient and enriched treatments ...... 74

Table 3-2 Average percent attached cells for all ambient and enriched treatments ...80

Table 3-3 Average growth rate and cell attachment data for cluster groups ...... 81-81

Table 3-4 Cluster groupings ranked from highest values to lowest values, for both growth rate and average attachment percentages ...... 83

vii

LIST OF FIGURES

Chapter 2 Figure 2-1 Well plates used for experiments ...... 26 Figure 2-2 Gambierdiscus cells attached to host ...... 27

Figure 2-3 Abundance (# of cells, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments over time ...... 29

Figure 2-4 Abundance (# of cells, averaged across triplicates) of each Gambierdiscus species across each of the host treatments over time ...... 30

Figure 2-5 Abundance (# of cells at end of experiment, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments ...... 32

Figure 2-6 Abundance (# of cells at end of experiment, averaged across triplicates) for each Gambierdiscus species across the host treatments ...... 33

Figure 2-7 Average attachment (% of total cells that were attached to host, averaged across triplicates and throughout entirety of experiment) of the various Gambierdiscus species for each of the host treatments ...... 37

Figure 2-8 Average start and end weights for each host ...... 41

Figure 2-9 Percent change in biomass data plotted against average attached cells, average end cells, and average growth rates...... 42

Chapter 3

Figure 3-1 Illustration of the preparation and usage of „ambient‟ and „enriched‟ seawater prior to and throughout the duration of the experiment ...... 65

Figure 3-2 G. belizeanus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment ...... 69

Figure 3-3 G. caribaeus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment ...... 70

Figure 3-4 G. carolinianus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment ...... 71

Figure 3-5 G. carpenteri cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment ...... 72

viii

Figure 3-6 G. yasumotoi cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment ...... 73

Figure 3-7 Percent of G. belizeanus cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment ...... 76

Figure 3-8 Percent of G. caribaeus cells attached to host over 29 day period for75ambient and enriched conditions, for each host treatment ...... 76

Figure 3-9 Percent of G. carolinianus cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment ...... 77

Figure 3-10 Percent of G. carpenteri cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment ...... 77

Figure 3-11 Percent of G. yasumotoi cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment ...... 78

Figure 3-12 Percentage of cell attachment across all days and all triplicates for each host and for ambient and enriched nutrient conditions ...... 79

Figure 3-13 Dendrogram illustrating the hierarchical clustering of samples, grouped by similarity of growth rates and attachment values ...... 84

5Figure 3-14 Differences in growth rate (div/day) values for cluster groups...... 84

Figure 3-15 Differences in average attachment (%) values for cluster groups...... 85

Appendix

Figure A1 Abundance and attachment data for G. belizeanus in ambient conditions…...... 120

Figure A2 Abundance and attachment data for G. caribaeus in ambient conditions ...... 121

Figure A3 Abundance and attachment data for G. carolinianus in ambient conditions ...... 122

Figure A4 Abundance and attachment data for G. carpenteri in ambient conditions ...... 123

Figure A5 Abundance and attachment data for G. yasumotoi in ambient conditions…...... 124

CHAPTER 1:

INTRODUCTION

Ciguatera fish poisoning (CFP) is the most common phycotoxin-related seafood poisoning worldwide (Ragelis 1984, Fleming et al., 1998). This syndrome affects people who have consumed fishes containing ciguatoxins (CTX), the precursors of which are produced by dinoflagellates from the genus Gambierdiscus. Gambierdiscus cells are epiphytic in nature and are consumed by herbivorous fish and invertebrates grazing upon the macroalgae that host them; once introduced into the food web, ciguatoxins may biomagnify and bioaccumulate in the food web.

For years, researchers have reported that CFP flare-ups oftentimes follow disturbances to coral reefs, such as hurricanes, dredging, and shipwrecks (Cooper 1964; Bagnis 1994; deSylva 1994). The general consensus on this observation is that coral degradation results in dead coral surfaces that can be colonized by macroalgae, therefore providing more substrate for Gambierdiscus populations. Although coral reef degradation has been increasingly documented on reefs worldwide, the implications for CFP frequency are unknown and largely unstudied. Reefs experiencing coral losses and increases in algal cover have been reported in the Hawaiian Islands, the Great Barrier Reef, the Seychelle

Islands, the Maldives, Belize, Jamaica, Panama, the Florida Keys, the Indo-Pacific, and many other locations, particularly throughout the Greater Caribbean Region (Bruno et al.,

2009). Although much work has focused on the ecological and environmental factors that affect Gambierdiscus populations, very little work has considered how macroalgal hosts influence their distribution. As decreases in coral cover and increases in algal biomass are becoming a common occurrence to coral reefs worldwide, the effects of those changes in benthic community on incidences of ciguatera cases represent a major gap in ciguatera research.

Background and history of ciguatera

CFP has been recognized for centuries, first reported by Martyr (1555) in the West

Indies. Captain James Cook (1777) reported outbreaks of fish poisoning in the South

Pacific in 1774 with symptoms closely coinciding with those known for CFP today

(Doherty 2005). Early on, Halstead (1978) suggested that more than 400 species of fish may be ciguatoxic; however, highest toxins levels are generally found in carnivorous fishes. Although predominantly a (sub) tropical problem, CFP can affect people globally due to the export of fish from tropical regions to markets worldwide (Sanner et al., 1997).

However, CFP is of highest concern to small island communities, particularly those in the

South Pacific, who depend heavily on locally caught fish as the main source of protein in their diet. CFP symptoms vary by individual and by region, but include gastrointestinal and neurological effects. Symptoms will onset within hours of fish consumption and can last for weeks to months. Death is rare, but a recent mortality and subsequent study in the

Pacific demonstrated the potential of tropical reef fish to accumulate enough CTXs to be lethal to humans, especially when the liver and viscera are consumed as part of the meal

(Hamilton et al., 2010).

Scheuer et al. (1967) first isolated the toxin responsible for CFP from moray eel livers and named it ciguatoxin. Gambierdiscus toxicus was eventually identified as the source of the toxin by Yasumoto et al. (1977), and named after its place of discovery in the

Gambier Islands, French Polynesia. The structure of CTX was later identified by Murata et al. (1989). In the years following the discovery of G. toxicus and the isolation of CTXs, much work focused on the ecology of G. toxicus and environmental conditions affecting growth and toxicity. Early research demonstrated that G. toxicus was predominately

4 epiphytic (Yasumoto et al., 1979) and that cells attached themselves to host macroalgae by a mucus secretion (Yasumoto et al., 1980; Ballantine et al., 1988; Bomber et al.,

1988).

Accumulation of CTX in fish through diet, as opposed to production of the toxins by the fish themselves, was first proposed by Randall in 1958, nearly two decades before

Gambierdiscus was implicated as the causative dinoflagellate. Randall proposed that the toxins originated in algal materials and were passed into the food web once ingested by herbivorous fishes. Helfrich and Banner (1963) demonstrated that the toxins could be transferred from toxic fish to non-toxic fish and later (1968) hypothesized that ciguatera outbreaks were linked to bursts of ciguatoxin production.

Studies involving the growth and toxin production of G. toxicus under different environmental conditions have produced contradictory results, indicating that different strains of G. toxicus grow and produce toxins differently under varying conditions.

Despite these variations, G. toxicus was assumed to be the only species in the genus until

Faust (1995) described the second species, G. belizeanus, collected from Belize. In the time since, additional species have been described as well: G. yasumotoi(Singapore;

Holmes 1998), G. polynesiensis, G. pacificus, G. australes (French Polynesia, Chinain et al., 1999), G. caribaeus, G. carolinianus, G. carpenteri, and G. ruetzleri (Belize, North

Carolina, Belize, Belize, respectively; Litaker et al., 2009), G. excentricus (Canary

Islands, Fraga 2011), G. silvae (Canary Islands, Fraga 2014), and G. scabrosus (Japan,

Nishimura et al., 2014). However, the genus is in a state of change, as new species have been discovered almost annually in recent years, while others are in the process of being moved out of the genus. For example, recent work has suggested that both G. yasumoto

5 and G. ruetzleri belong in a separate genus (Fukuyoa; Gomez et al., 2015), based on both morphological and molecular data.

Parsons et al. (2012) give a comprehensive review of past work including biogeography, substrate preferences, and the effects of light, temperature, salinity, nutrients, wave energy, and seasonal patterns on Gambierdiscus growth, citing numerous variations in reported results. They suggested that many of the contradictory results cited in their review could be attributed to the false assumption that G. toxicus was used in the earlier studies, when in fact multiple species were likely used (prior to the discovery and description of additional species). With a total of 13 species and several ribotypes within the genus now identified, it remains unknown how genetic differences (and resulting physiological variation) among these species contribute to the temporal and spatial patchiness of CFP.

Toxin production by Gambierdiscus

At least some Gambierdiscus species produce the toxins that cause CFP. These toxins were first extracted from wild and cultured cells in French Polynesia by Bagnis et al.

(1980). Early on, there were reports of variation in toxin production among different strains of Gambierdiscus (Bomber et al., 1989; Holmes et al., 1991; Legrand et al., 1990;

Babinchak et al., 1996). In addition, Holmes et al. (1991) demonstrated that not all

Gambierdiscus strains produced toxins and that toxin production was higher in wild cells than cultured cells. Morton et al. (1993) found that light, salinity, and temperature conditions can have dramatic effects on the toxin production of Gambierdiscus cells in

6 culture (up to 200-fold difference) and that manipulation of these environmental conditions can result in inhibition of toxin production altogether.

Reports on the variability of toxin production under different nutrient conditions have been inconsistent, with both early and recent studies indicating that nutrients do have an effect (Lechat et al., 1985; Sperr and Douchette 1996; Lartigue et al., 2009), while others found no effect (Durant-Clement 1986). Lartigue et al. (2010) found that the source of nitrogen had no apparent impact on the toxin production of two different Gambierdiscus strains, but that nitrogen limitation may promote an increase in toxicity in some strains.

However, it is not known whether the increase in toxicity observed in this study was solely due to an increase in ciguatoxin, or if maitotoxin (a water-soluble toxin also produced by Gambierdiscus) may have been present in the extracts as well.

With the description of new species, it is apparent that variation in toxin production exists within the genus. For example, Chinain et al. (1999, 2010) found that G. polynesiensis was more ciguatoxic than other Gambierdiscus species tested. Rhodes et al. (2014) similarly reported that G. polynesiensis was more ciguatoxic than other Gambierdiscus isolates from the Cook Islands (G. australes and G. pacificus). Cultures now known to be Gambierdiscus ribotype 2 were more ciguatoxic than G. caribaeus in a study conducted by Lartigue et al. (2009). A 1000-fold range in ciguatoxicity was exhibited in the isolates tested (i.e., G. polynesiensis versus G. caribaeus). This range assumes that the toxin quantification methods in the studies are quasi-comparable; i.e., the N2A assay used in Lartigue et al. (2009; standardized to C-CTX-1 equivalents) gives similar results to the receptor-binding assay used by Chinain et al. (2010; standardized to P-CTX-3C equivalents), acknowledging that P-CTX-3C is approximately twice as potent as C-CTX-

1 (Lewis 2001). Based on these assumptions, of the Caribbean isolates tested, G. belizeanus (STB-1 tested by Chinain et al., 2010) is approximately three-times more ciguatoxic than G. ribotype 2 (CCMP 1655), and 10-100 times more toxic than G. caribaeus (CCMP 1651). Therefore, while G. polynesiensis appears to be the primary toxin producer putatively responsible for Pacific cases of ciguatera (based on published data), the current front-runner for the “super-producing” Caribbean species is G. belizeanus, albeit at levels of toxin production >10 times lower than G. polynesiensis.

Ciguatoxins are lipophilic; as they are ingested by herbivorous fishes and invertebrates, they are stored in muscle tissue. As the toxins are passed through the food chain, they are metabolized and oxidized into other ciguatoxin congeners (e.g., CTX-1; Lewis 2001) by fishes, becoming more polar and more toxic as they reach the highest trophic levels

(Legrand et al., 1990; Murata et al., 1990; Legrand et al., 1992; Lewis and Sellin 1992;

Pottier et al., 2002). Toxin profiles vary greatly among fish species and also among individual fish. Because ciguatoxins are digested and metabolized differently by various herbivorous, detritivorous, and piscivorous fish, a single fish can have over 10 different ciguatoxin congeners present in their tissue (Legrand et al., 1990; Pottier et al., 2002;

Dechraoui et al., 2005).

Toxin flux into the food web

Herbivorous fishes are largely responsible for the transfer of ciguatoxins to carnivorous fish, with surgeonfish (Acanthuridae) and parrotfish (Scaridae) being the key vectors

(Lewis 2001). Surgeonfish (Acanthuridae), parrotfish (Scaridae), and damselfish

(Pomacentridae), all mostly herbivorous, collectively form the largest portion of the fish

8 biomass of most coral reefs (Thresher 1980). Herbivorous fishes can be classified as either grazers, who pick up inorganic substrate while non-selectively feeding by scraping or sucking food affixed to substrate, or browsers, who selectively bite or tear macroalgae and rarely ingest any inorganic material (Jones 1968). However, as explained by Cruz-

Rivera and Villareal (2006), macroalgal hosts of Gambierdiscus are not homogenous or equally grazed by fishes, and the ecology of host algae may be a determining factor in the route and flux of ciguatoxins into the food web. Few experimental studies have examined the relationship between Gambierdiscus and their algal hosts (Grzebyk et al., 1994;

Nakahara et al., 1996; Parsons et al., 2011).

Palatability of host macroalgae plays an important role in ciguatoxin transfer within a food web (i.e., all macroalgae species are not equally consumed by herbivores; Cruz-

Rivera and Villareal 2006). Some macroalgae have mechanisms to defend themselves against grazing, including chemical defenses, low nutritional values, and structural defenses. For example, some calcified algae, such as Halimeda, have tough leathery exteriors and experience reduced grazing because they are less susceptible to grazers with weak mouthparts (Cruz-Rivera and Villareal 2006). Many species produce chemicals to deter grazers while others are non-nutritional and therefore too costly energy-wise for herbivores to consume. Algae with low nutritional value are oftentimes subjected to lower grazing pressures when more nutritional algae are available (Atsatt and O‟Dowd

1976). Therefore, a preference for a more nutrient-enriched host may result in increased flux of toxins in the food web, if that host harbors significant numbers of toxic

Gambierdiscus cells.

With herbivorous fishes such as Acanthurids being a major vector of toxins in Caribbean food webs (Lewis 2001; A. Robertson, University of South Alabama, pers. comm.) a better understanding of the diets and feeding preferences of these fish, and subsequent effects of host preferences and host effects on Gambierdiscus populations is needed. In general, the mechanisms and pathways of toxin flux at the base of the food web need further investigation. The different preferences and/or growth responses exhibited by

Gambierdiscus add complexity to toxin flux scenarios; e.g., host palatability, different host preferences and epiphytic behavior among the Gambierdiscus species, and different levels of toxin precursor production among the Gambierdiscus species. A better understanding of these parameters will clarify the role of the macroalgae community in the production and transfer of ciguatoxins to higher trophic levels, the pulse/burst of which can lead to flare-ups of CFP.

Patchiness of ciguatera

Incidences of CFP have been largely described as being „sporadic‟ and „patchy‟ occurrences, with a large amount of spatial and temporal variability. The factors influencing the occurrence of ciguatera outbreaks and the dynamics controlling both toxin production and flux of toxin into the food web are still poorly understood. In the

Pacific, there are areas known to be relatively free of ciguatera, sometimes located immediately adjacent to areas of high risk (Lewis 2006). Cooper (1964) describes reef areas of the Pacific that had been non-toxic for many years, but then suddenly have toxic fish. Others also report of areas in which toxicity has changed over time as well (Banner et al., 1966).

Chinain et al. (1999) conducted a five year survey in which Gambierdiscus spp. populations were monitored weekly in French Polynesia. During this time, 58 blooms were recorded, with 65% occurring within two of the five years. During those two years, seasonality in the population dynamics of Gambierdiscus spp. was observed, with maximum abundance occurring at the beginning and end of the hot season. Aside from this study, little long-term data exist for any intense monitoring of Gambierdiscus populations, making it difficult to better understand the specific factors involved in blooms. However, this study did reveal a seasonal cycle related to cell abundances, which has also been reported by Gillespie et al. (1985, Australia) and Bomber et al. (1988,

Florida Keys). It has also been observed that abundances are positively correlated with water temperatures (Bomber et al., 1988; Morton et al., 1992; Hokama et al., 1996;

Richlen et al., unpub. data). Modeling scenarios based on field data in Hawaii by Parsons et al. (2010) indicated that maximum abundances resulted from nutrient enrichment and warming sea surface temperatures. It has been reported that positive correlations exist between toxicity of fish and both El Niño events (Hales et al., 1999), and warming sea surface temperatures (Tosteson 2004; Tester et al., 2010).

Although research has identified these factors to be related to Gambierdiscus dynamics, many of these factors or human disturbances may also occur without triggering any

Gambierdiscus blooms or cases of CFP. This suggests that a bloom large enough to initiate a CFP flare-up is triggered by dynamics much more complex than just changing environmental conditions. If incidences of CFP are to be compared to Gambierdiscus cell abundances, some factors to consider include differences in toxicity within the

Gambierdiscus genus, rates of introduction of the toxins into the food web, and pathways of toxin flux once within the food web.

Climate change and the future of ciguatera

Recently, ciguatera research has focused heavily on climate change and the implications for increased CFP. Various studies have reported that outbreaks of CFP are associated with tropical storm frequency (Gingold et al., 2014) and warm water temperatures (Hales et al., 1999; Llewellyn 2010; Tester et al., 2010), both linked to global climate change.

On coral reefs worldwide, community shifts are occurring in response to overfishing, increased nutrient loading, coral bleaching, warming seas, and other factors (Bruno et al.,

2009; Norstrom et al., 2009). The resulting shift from a coral-dominated system to a macroalgae-dominated system (termed a „phase shift‟) could be a contributing factor in ciguatera flare-ups. This could occur in response to the increased availability of host algae or increased dominance or enhanced distribution of particular growth (and/or toxin production) stimulating species of host macroalgae.

In the Caribbean, loss of coral cover began in the early 1980s, and was mostly attributed to anthropogenic factors as well as several major hurricanes, coral disease outbreaks, overfishing, habitat degradation, and the massive die-off of the herbivorous sea urchin,

Diadema antillarum, followed by a proliferation of macroalgae (Hughes et al., 1987;

Gardner et al., 2003; Schutte et al., 2010; Jackson et al., 2014).

The widespread shift to algal-dominated reefs may have more implications for ciguatera than simply an increased algae cover, depending on the cause(s) of the shift. Many phase shifts are occurring due to anthropogenic effects such as coastal eutrophication or overfishing, but are then further exacerbated by rising ocean temperatures or increased storm activity.

Increased temperatures have been shown to be favorable for Gamberidiscus growth.

Tester et al. (2010) reports that the highest incidences of CFP in the Caribbean coincide with the region of the warmest and least variable temperatures. They also report growth experiments in which five species of Gambierdiscus showed positive or maximum growth rates at temperatures greater than 29°C. More recent temperature studies have showed maximum growth to occur between 26.5 and 31.1°C for most species of

Gambierdiscus (Kibler et al., 2012; Xu et al., in prep.). Coral bleaching most commonly occurs in the Caribbean when water temperatures exceed their normal range by 1 or 2°C for at least four weeks (Wilkinson et al., 2008). Bleaching events may create ideal scenarios for Gambierdiscus growth in two ways: increased temperatures, and new bare substrate for turf or macroalgae growth.

Descombes et al. (2015) predict that within this century, habitat suitability for coral reefs could increase by up to 16% globally, as warming temperatures create suitable water temperatures at higher latitudes than are currently present today. Currently, ciguateric fish are restricted to the (sub) tropical areas region of 35° northern and 35° southern latitude (Johnson and Jong 1983; Lange, 1987). Rising ocean temperatures will likely allow the range of Gambierdiscus to spread further into temperate zones, increasing the

13 risk of ciguatera globally, as more communities would potentially be exposed to ciguateric fish.

Ciguatera in Florida

Although CFP is a more common occurrence in those most tropical locations, such as the

Caribbean or the South Pacific islands, the syndrome is still of concern in the United

States, particularly in South Florida. The most recent data on CFP occurrences in Florida are provided by Radke et al. (2015), in which they collect data from both the Florida

Department of Health as well as from a large survey given to fisherman throughout the state, and account for underreporting. They calculate that CFP in Florida occurs annually at an incidence of 65.6 per 100,000, which means that just over 1,000 people are estimated to be affected by ciguatera each year in the state. They also state that the majority of these cases are caused by fish caught in either southern Florida, or brought back from the Bahamas.

Gambierdiscus populations in the Florida Keys

Many early studies (Babinchak et al., 1986; Bergmann et al., 1981; Bomber et al., 1988;

Morton et al., 1992) report the presence of G. toxicus during field collections in the

Florida Keys, although new research identifying the variety of species within the genus suggests that several different species may have been collected (but probably not G. toxicus). Analyses by Litaker et al. (2010) and Parsons (unpublished data) indicate that G. belizeanus, G. caribaeus, G. carolinianus, and G. carpenteri are all present in the Florida

Keys. Cell populations have been collected from many various substrate types including drifting macroalgae, seagrass blades, carbonate sediment, and benthic macroalgae

(Bomber 1985; Bomber 1987; Bomber et al., 1989; Parsons, unpublished data), with benthic macroalgae hosting the highest cell abundances (Bomber 1985).

In the Florida Keys, species belonging to the genera Halimeda (Chlorophyta) and

Dictyota (Phaeophyta) are the most dominant macroalgae (Chiappone and Sullivan 1997;

Hanisak and Overdorf 1998; Lirman and Biber 2000), representing as much as 77-99% of the macroalgal biomass in some areas (Lirman and Biber 2000). Both Halimeda and

Dictyota are well-known as hosts to Gambierdiscus in this region (Bomber 1985;

Ballantine et al., 1988; Carlson and Tindall 1985; personal observations). Collections in the Florida Keys by Bomber et al. (1989) also found Gambierdiscus populations associated with other genera including Heterosiphonia, Penicillus, Udotea, Laurencia,

Caulerpa, Padina, Acetabularia, Thalassia, and Sargassum. Bomber et al. (1989) suggest that Chlorophytes may support the largest biomass of dinoflagellates in the Florida Keys, as the phylum often dominates, accounting for as much as 95% of the macroalgal community in intertidal and subtidal zones (Bomber 1985), encompassing the optimum depth ranges for Gambierdiscus populations (Carlson et al., 1984; Bomber et al., 1988).

The benthic macroalgae community in the Florida Keys varies considerably from one location to the next (personal observations; Crowley and Dawes 1970), likely due to a multitude of geographical and environmental factors including proximity to land, depth, light levels, salinity, and season. The seasonal, temporal, and spatial variability in macroalgae cover is a likely factor in the abundance of Gambierdiscus that has received little attention thus far in the field of ciguatera research.

Research Purpose

The pioneering work of Lobel et al. (1988) and St. Martin et al. (1988) demonstrated that studies of the dynamics of Gambierdiscus substrate attachment and preference are feasible in the laboratory and that valuable information can be obtained by observing the behavior of Gambierdiscus in response to different macroalgal hosts. The investigation into the role that host macroalgae play in the flux of toxins into the food web and the resulting flare-ups of CFP has clearly been understudied thus far in the progression of ciguatera research.

The studies described here are largely based on the methods and experimental design used by Parsons et al. (2011), encouraged by their findings that some macroalgae stimulated/inhibited cell growth and/or influenced epiphytic behavior. As numerous new species have been described within the Gambierdiscus genus, it is important to understand the differences that exist within the genus; the current study is the first known work examining epiphytic behavior and host preferences among different Gambierdiscus species. The purpose of this experiment was to ascertain whether differences exist in host preferences and epiphytic behavior among Gambierdiscus species; important factors when considering pathways of CTX flux in the food web that play a role in ciguatera flare-ups.

Objective 1: Determine if growth rates vary among five Gambierdiscus species in

the presence of different macroalgal hosts.

Hypothesis: Growth rate will vary among different species of

Gambierdiscus in the presence of different macroalgal hosts.

Objective 2: Determine if these Gambierdiscus species have different epiphytic behaviors (attachment vs. non-attachment) in the presence of different macroalgal hosts.

Hypothesis: Different Gambierdiscus species will exhibit different

epiphytic behaviors in the presence of different macroalgal hosts.

Objective 3: Determine if a nutritionally-enriched macroalgal host affects

Gambierdiscus growth rates in comparison to a non- or less-enriched host.

Hypothesis: A nutritionally-enriched macroalgal host will support higher

Gambierdiscus growth rates in comparison to a non- or less-enriched host.

Objective4: Determine if a nutritionally-enriched macroalgal host affects

Gambierdiscus epiphytic behavior in comparison to a non- or less-enriched host.

Hypothesis: A nutritionally-enriched macroalgal host will have increased

Gambierdiscus attachment in comparison to a non- or less-enriched host.

CHAPTER 2:

MACROALGAL HOST EFFECTS ON THE GROWTH AND EPIPHYTIC

BEHAVIOR OF GAMBIERDISCUS

INTRODUCTION

Macroalgae of the phyla Rhodophyta, Chlorophyta, and Phaeophyta are known to host epiphytic dinoflagellates of the genus Gambierdiscus. However, the physical (e.g., surface area, texture, morphology), chemical (e.g., production of metabolites, release/uptake of available nutrients), and ecological (e.g., epiphytic community, seasonal fluctuations) characteristics that influence the abundance of Gambierdiscus populations on any given macroalgae individual, are widely unknown.

Epiphytic behavior of Gambierdiscus

The epiphytic relationship between Gambierdiscus cells and their host macroalgae may be advantageous for the dinoflagellates in various ways. Villareal and Morton (2002) demonstrated that Gambierdiscus may utilize the three-dimensional structure of an algal host to minimize light exposure, allowing them to thrive in shallow, well-lit tropical locations, despite their intolerance to high light levels. It has been demonstrated that cells actively swim during daylight hours and rapidly attach to hosts upon darkness (Nakahara et al., 1996). Nakahara et al. (1996) also suggest that cells normally swim around near the macroalgae thalli, but quickly attach to the algal surface when a sudden disturbance or strong water motion occurs, presumably to avoid being dispersed. Some researchers have observed Gambierdiscus to form a mucilaginous matrix over the thallus of the host macroalgae and aggregate within it (Yasumoto et al., 1980; Ballantine et al., 1988). Other studies have observed the cells to attach to their host by a mucus thread, tethering themselves to the algae, sometimes with a rotating motion (Besada et al., 1982; Nakahara

19 et al., 1996; personal obs.). Additionally, Bomber et al. (1988) demonstrated that

Gambierdiscus may utilize drift algae as a means for attachment and transport, a likely dispersal mechanism that has resulted in their circumtropical distribution. Another possible advantage of epiphytism is the nutritional value that algal hosts may offer to

Gambierdiscus cells (Grzebyk et al., 1994), especially those living in an otherwise oligotrophic environment, such as a coral reef.

Gambierdiscus cells themselves may stimulate their own growth (i.e., quorum sensing).

For example, Sakamoto et al. (1996) found that G. toxicus cultures grown in media conditioned from dense-growing cultures resulted in enhanced growth. They suggested that gambieric acid-A, a polyether compound produced by Gambierdiscus that was abundant in the conditioned media, is an endogenous growth enhancer of G. toxicus.

Epiphytic preferences of Gambierdiscus

Early studies investigating Gambierdiscus host preferences were conducted by Lobel et al. (1988) and St. Martin et al. (1988). St. Martin et al. determined that Gambierdiscus prefer to affix themselves on algae rather than inorganic substrates, although cells did colonize dead parts of the experimental macroalgae. Their results also suggested that the preference of Gambierdiscus for macroalgae is independent of macroalgal phylum. The authors concluded that the mechanism behind the attraction to the macroalgae is unknown, but suggested that it may be associated with the production or diffusion of one or more substances produced by the macroalgae which may be necessary for the

Gambierdiscus growth.

Lobel et al. (1988) assessed the colonization behavior of Gambierdiscus on Dictyota versus Galaxaura. Their results were inconclusive, but revealed complications related to measuring dinoflagellate abundances; Gambierdiscus abundance was always higher on

Dictyota when examined per gram of host algae, but Galaxaura was the preferred host when the abundances were normalized to host surface area instead of weight.

Gambierdiscus abundance has been shown to be positively correlated with surface area of macroalgal hosts (Bomber et al., 1989); however, Gambierdiscus abundances are most often reported as “cells per gram macroalgae.” The discrepancies demonstrated by Lobel et al. (1988) suggest that “cells per surface area” is the appropriate enumeration to be used when comparing cell abundances among different macroalgal host species.

However, the difficulties associated with determining surface area values from a macroalgae sample have constrained field surveys and Gambierdiscus abundances have continued to be reported as “cells per gram macroalgae” in most cases.

Gambierdiscus cells have been found on various types of substrates, although the highest abundances have been reported for highly foliose rhodophytes and phaeophytes

(Gillespie et al., 1985; Bomber and Aikman 1989; Cruz-Rivera and Villareal 2006). The previously mentioned review by Parsons et al. (2012) demonstrates that numerous conflicting results on substrate preferences of Gambierdiscus exist. While the genus

Halimeda has been reported to commonly host Gambierdiscus (Florida Keys, Bomber et al., 1988; French Polynesia, Chinain et al., 2010; Cook Islands, Rhodes et al., 2010), others report finding no cells (Great Barrier Reef, Heilet al., 1998). Acanthophora in

Cuba was found to not host any epiphytic dinoflagellates (Delgado et al., 2006), whereas it is otherwise reported to host high abundances elsewhere (BVI, Carlson 1984; Belize,

Morton and Faust 1997). Differences were also reported for preferences for Dictyota as a host, supporting both dense populations (Cuba, Delgado et al., 2006; Caribbean,

Ballantine et al., 1988 and Carlson and Tindall 1985) and no populations at all (Tahiti,

Nakahara et al., 1996). Nakahara et al. (1996) noted that although G. toxicus cells were detected on a variety of coral reef macroalgae species, their preference for associating with a particular species varied across different geographical areas.

Parsons et al. (2011) examined how the epiphytic relationship between Gambierdiscus toxicus (BIG 12) and host algae varied among 24 different macroalgal species from

Hawaii. Their results indicated that G. toxicus will attach to some prospective host species, while completely avoiding others, suggesting that Gambierdiscus spp. may not be obligate epiphytes. In addition, some host species allowed for proliferation of G. toxicus cells, while others appeared to inhibit growth. However, it was suggested by the authors that the conditions used in the experiment may have exposed the cells to an environment influenced by stressed algal hosts, rather than an environment with a healthy host.

Using semi-quantitative polymerase chain reaction (qPCR) assays, Vandersea et al.

(2012) found as many as four species on a single macrophyte sample. Interestingly, some of their (high-energy) sampling locations revealed over half of the macrophytes sampled to contain at least three species of Gambierdiscus, while the majority of their sites had no more than one species per macrophyte sample. In that particular study, no significant preference for a specific macrophyte was apparent.

Other epiphytic micro-organisms (e.g., diatoms, other dinoflagellates, bacteria) may also influence the epiphytic behavior of Gambierdiscus (Yasumoto et al., 1980).For example,

Nakahara et al. (1996) suggested that differences in the epiphytic behavior of

Gambierdiscus when exposed to field samples of Jania and Amphiroa (attachment) versus unialgal (axenic) cultures (unattached cells) was due to stimulatory exudates produced by the microbial community on the field samples versus the cultures. It has also been suggested that Gambierdiscus may be capable of utilizing organic materials originating from bacteria as a nutrient source and/or that organic materials produced by bacteria might function as metal chelators and stimulate Gambierdiscus growth (Sakami et al., 1999). In addition to bacteria, the epiphytic community on any given substrate may consist of many other organisms, including other dinoflagellate species, fungi, cyanobacteria, diatoms, and other protozoans; the effects of this community, and of the macroalgal host itself, on Gambierdiscus are largely unstudied (Yasumoto et al., 1980).

Research Purpose

The purpose of the experiments described in this chapter was to ascertain whether differences exist in host preferences and epiphytic behavior among Gambierdiscus species.

Objective 1: Determine if growth rates vary among five Gambierdiscus species in

the presence of different macroalgal hosts.

Objective 2: Determine if these Gambierdiscus species have different epiphytic

behaviors (attachment vs. non-attachment) in the presence of different macroalgal

hosts.

METHODS

Gambierdiscus culturing

Five species of Gambierdiscus known to inhabit the Florida Keys were used. Three of the species, G. carolinianus, G. carpenteri, and G. yasumotoi, were isolated from the Florida

Keys and genotyped by the Anderson Lab at Woods Hole Oceanographic Institute

(WHOI). The other two species used, G. caribaeus and G. belizeanus, were provided by

WHOI, previously isolated from St. Thomas, USVI, as at the time of the experiment, no known isolates from the Florida Keys had been confirmed for those species. All cultures were maintained in 35 salinity filtered Florida Keys seawater with modified Keller‟s medium (without TRIS, Cu, or Si; with GeO2, to inhibit diatom growth), at 23-24°C on a

12/12 light/dark cycle at ~100 lum/ft2.

Algae collection

In the months prior to the start of experimentation, large portions of macroalgae were collected by scuba diving and snorkeling near Long Key, in the Florida Keys.

Macroalgae species collected were chosen based on either their common presence in the

Florida Keys, particularly those growing in habitats known to harbor Gambierdiscus populations, and/or by their previously known associations with Gambierdiscus populations in past studies. Once the algae were transported back to the lab, they were shaken vigorously to remove epiphytes, and placed under the same growth conditions as the Gambierdiscus cultures (as described above) with air flow and weekly water changes.

Algae were identified to species level, to the best of our ability using taxonomic keys and

24 descriptions by Dawes and Mathieson (2008) and Littler and Littler (2000). Species were tentatively identified as Laurencia intricata, Dasya crouaniana, Polysiphonia ramantacea, Ulva fasciata, Dictyota cervicornis, Acanthophora spicifera, Derbesia marina, and Caulerpa mexicana. The eight species chosen represent all three major phyla of macroalgae and include various morphological characteristics; one Phaeophyte, four

Rhodophytes, and three Chlorophytes; two sheet algae, three filamentous algae, and three coarsely branched algae (Table 2-1).

Table 2-1.Description of host algae used in experiments, with data on palatability and physical structure.

Functional- Algal host genus Phyla Palatability form Group Dictyota Phaeophyta Chemically defended, consumed by Sheeta some herbivores b, h Acanthophora Rhodophyta Palatable 2b Coarsely branched a Laurencia Rhodophyta Chemically defended, but can be Coarsely highly palatable 2b branched a Dasya Rhodophyta Chemically defended i, low Filamentousk palatability j Polysiphonia Rhodophyta Highly palatable, especially to some Filamentousk damselfish g Ulva Chlorophyta Chemically defended, still palatable i Sheet a

Derbesia Chlorophyta Filamentous

Caulerpa Chlorophyta Chemically defended f Coarsely branched a a Littler et al., 1983; b Cruz-Rivera and Villareal, 2006; c Hay and Steinberg, 1992; d Paul and Fenical, 1986; f Norris and Fenical, 1982; g Hata, 2010; h Bolser and Hay, 1996; i De Lara-Isassi et al.,2000; j Arthur et al.,2009; k Mackay, 1836;

Two weeks prior to the start of the experiment, a portion of each of the algae growing in the conditions described above were divided and grown separately under the same physical conditions as the Gambierdiscus cultures, but in ambient filtered seawater

(salinity 35) from the Florida Keys, without Keller‟s medium.

Experimental design and setup

Gambierdiscus populations were placed in close living association with the various host macroalgae species in order to monitor growth and behavior in response to the hosts. This experiment included five treatments of Gambierdiscus species, eight treatments of macroalgae host species, and a control (Gambierdiscus cells living with no host algae).

Five days prior to Day 0 (start of experiment), macroalgae fragments of ~25 mg were placed individually into separate wells of a six well culture plate (Figure 2-1) containing

8 ml of ambient Florida Keys seawater for each treatment to be used in the experiment in triplicate (8 host algae + 1 control * 5 Gambierdiscus species * triplicates = 135 wells).

These fragments acclimated in the growth media under the experimental conditions for five days, with frequent (almost daily) water changes, to allow for any harmful exudates resulting from the cutting of algal thalli (to make the 25 mg fragments) to dissipate and be removed before the Gambierdiscus cells were added.

On Day 0, new wells of 8 ml ambient Florida Keys seawater were prepared for all treatments. Twenty-five cells of the appropriate species of Gambierdiscus were added to each well, and then the appropriate acclimated algae fragments were added. Triplicate

26 controls for each Gambierdiscus species (containing no algae) were also prepared under the same conditions.

Figure 2-1 (left) Well plates used for experiments, with 25 mg algae fragments growing in 8 ml seawater. Figure2-2 (right) Gambierdiscus cells attached to host Acanthophora spicifera, in preliminary experiments, (A) showing an unattached cell and (B) showing an attached cell.

Cells were counted on Days 1, 8, 15, 22, and 29 (preliminary experiments determined that these weekly counts would provide sufficient data for analysis), the frequency of which was intended to capture both the growth and interaction dynamics over a full growth cycle of Gambierdiscus. In order to account for any possible behavioral changes over the course of a day, the order in which counts were done was changed each time.

Each cell was categorized as dead, alive and unattached to host, or alive and attached to host (Figure 2-2). Water changes were performed 2-3 times weekly by slowly removing 4 ml of the water by transfer pipet (with a loss of <1% of cells when done carefully), and replaced with 4 ml filtered ambient seawater. At the end of the experiment, after all counts were completed, macroalgae fragments were wet-weighed and recorded.

Data Analysis

Prior to statistical analysis, the cell counts from the five Gambierdiscus species were converted to relative abundance values (% of total cells dead, % attached and alive, % unattached and alive). Cell count data gathered over the course of the experiment were used to calculate growth rates using the following equation:

# cells on last day of exponential growth Ln # cells on first day of exponential growth Growth rate = # days of exponential growth

In order to determine how the different Gambierdiscus species interacted with the different host algal species, growth rates were determined for each treatment and statistically analyzed using multivariate analysis. End cell counts and maximum attachment percentages were also analyzed by two-way ANOVA with Tukey‟s post-hoc analysis. End wet weights were compared to beginning wet weights in order to determine

% biomass either gained or lost throughout the duration of the experiment and correlation analyses were run to determine if any correlation exists among changes in biomass weights and growth rates, end cells, or attachment percentages.

RESULTS

Do macroalgal hosts affect the growth of the different Gambierdiscus species?

In each of the five Gambierdiscus species tested, all showed significantly higher growth

(max cells) in some algae treatments versus others. For most species, at least one host algae particularly stood out as having stimulating effects on cell growth. Acanthophora treatments showed the highest growth for G. carolinianus, G. caribaeus, and G.

28 yasumotoi, while Caulerpa treatments showed the highest growth for G. belizeanus, and

Laurencia for G. carpenteri (Table 2-4, Figure 2-6).

Gambierdiscus carolinianus controls showed significantly higher growth than all of the treatments with algae (Figure 2-6), suggesting that the algae hosts may actually be suppressing growth for that species. Gambierdiscus belizeanus also showed high growth in the controls, but slightly higher in Caulerpa treatments (Figure 2-6). For the other three Gambierdiscus species, controls showed more growth than some host treatments, but less than others (Figure 2-4).

Gambierdiscus carpenteri showed the highest growth rates (div/day) of the experiment, when cultured with hosts Dasya (0.222), Polysiphonia (0.220), and Laurencia (0.211). G. carolinianus showed the highest growth rate of all the control treatments (0.211); these treatments also yielded the highest cell counts of all the treatments by the end of the experiment. Gambierdiscus caribaeus had the highest growth rates for Acanthophora

(0.199), Dictyota (0.170), Ulva (0.143), and Dasya (0.126) treatments, but showed no significant differences among any of its host treatments. Growth rates for G. belizeanus were highest in Caulerpa treatments (0.184, Figure 2-4) and G. yasumotoi never had the highest growth rates among any of the host treatments (Table 2-2).

When averaged across all treatments, however, G. belizeanus had the highest overall growth rate (0.153), followed by G. caribaeus (0.145), G. carpenteri (0.136), G. carolinianus (0.117), and G. yasumotoi(0.081). For only one Gambierdiscus species, G. carolinianus, did the control treatment yield a higher number of cells than all 8 of the algal host treatments.

1200 1200 1200 Acanthophora Caulerpa Dasya 900 900 900 600 600 600 300 300 300 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

1200 1200 1200 Derbesia Dictyota Laurencia 900 900 900 600 600 600 300 300 300

Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 1200 1200 1200 Polysiphonia Ulva Control 900 900 900 600 600 600 300 300 300 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

G. belizeanus G. caribaeus G. carolinianus G. carpenteri G. yasumotoi

Figure 2-3. Abundance (# of cells, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments over time. Tukey groupings and standard deviations for the cell growth represented here can be found in Table 2-2.

1200 1200 1000 G. belizeanus 1000 G. caribaeus 800 800 600 600 400 400 200 200 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

1200 1200 1000 G. carolinianus 1000 G. carpenteri 800 800 600 600 400 400 200 200 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

1200 G. yasumotoi 1000 800 600 400 200 0 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 2-4.Abundance (# of cells, averaged across triplicates) of each Gambierdiscus species across each of the host treatments over time. Tukey groupings and standard deviations for the cell growth represented here can be found in Table 2-2.

Table 2-2. Growth rates (div/day) of the various Gambierdiscus species for each of the host treatments. Values shown are averages ± standard deviation, averaged to the nearest whole number. The letters under each value indicate Tukey groupings within each row. Under each Gambierdiscus species, the data in bold represents the host treatment where the highest growth rate occurred.

G. belizeanus G. caribaeus G. carolinianus G. carpenteri G. yasumotoi

0.193 ± 0.007 0.143 ± 0.005 0.211 ± 0.003 0.089 ± 0.031 0.156 ± 0.011 Control B A D B ABC 0.160 ± 0.000 0.199 ± 0.011 0.158 ± 0.011 0.130 ± 0.003 0.177 ± 0.009 Acanthophora AB A CD B BC 0.184 ± 0.011 0.104 ± 0.129 0.148 ± 0.023 0.083 ± 0.019 0.008 ± 0.005 Caulerpa B A ABC B ABC 0.150 ± 0.005 0.189 ± 0.038 0.095 ± 0.041 0.222 ± 0.006 0.184 ± 0.015 Dasya AB A AB C C 0.110 ± 0.019 0.126 ± 0.014 0.074 ±0.005 0.095 ± 0.004 -0.015 ± 0.082 Derbesia A A B B ABC 0.164 ± 0.057 0.170 ± 0.018 0.094 ± 0.008 0.010 ± 0.059 -0.030 ± 0.132 Dictyota AB A AB B AB 0.148 ± 0.026 0.132 ± 0.015 -0.048 ± 0.059 0.211 ± 0.021 -0.053 ± 0.142 Laurencia AB A A C A 0.134 ± 0.007 0.136 ± 0.020 0.138 ± 0.020 0.220 ± 0.017 0.187 ± 0.006 Polysiphonia AB A ABC C C 0.108 ± 0.013 0.143 ± 0.008 0.100 ± 0.027 0.000 ± 0.000 0.054 ± 0.065 Ulva A A AB A ABC

Average 0.150 0.149 0.107 0.118 0.074

G. belizeanus G. caribaeus G. carolinianus G. carpenteri G. yasumotoi

Figure 2-5.Abundance (# of cells at end of experiment, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments, with standard error bars. The letters above each data point indicate Tukey groupings among the same-host treatments.

G. belizeanus G. caribaeus

G. carolinianus G. carpenteri

G. yasumotoi

Figure 2-6.Abundance (# of cells at end of experiment, averaged across triplicates) for each Gambierdiscus species across the host treatments, with standard error bars. The letters above each data point indicate Tukey groupings among the same-host treatments.

Table 2-3 Abundance (# of cells at end of experiment, averaged across triplicates) of the various Gambierdiscus species for each of the host treatments. Values shown are averages ± standard deviation, averaged to the nearest whole number. The letters under each value indicate Tukey groupings within each row. Under each Gambierdiscus species, the data in bold represents the host treatment where the abundance occurred.

G. belizeanus G. caribaeus G. carolinianus G. carpenteri G. yasumotoi

668 ± 75 86 ± 33 1022 ± 76 197 ± 64 237 ± 30 Control B A D AB B 162 ± 106 479 ± 31 335 ± 66 274 ± 40 638 ± 47 Acanthophora A C C BC C 744 ± 80 197 ± 161 82 ± 42 170 ± 8 29 ± 8 Caulerpa B AB AB AB A 115 ± 16 335 ± 97 144 ± 32 543 ± 60 267 ± 120 Dasya A BC B D B 151 ± 42 199 ± 13 76 ± 6 155 ± 21 27 ± 26 Derbesia A AB AB AB A 274 ± 141 286 ± 24 131 ± 13 186 ± 203 7 ± 13 Dictyota A ABC B AB A 194 ± 92 164 ± 49 13 ± 13 807 ± 114 41 ± 72 Laurencia A AB A E A 139 ± 26 146 ± 88 121 ± 26 510 ± 66 266 ± 38 Polysiphonia A AB AB CD B 98 ± 29.95 159 ± 57 76 ± 10 0 ± 0 36 ± 27 Ulva A AB AB A A

Average 283 228 222 316 172

Do Gambierdiscus species have different epiphytic behaviors (attachment vs. non- attachment) in the presence of different macroalgal hosts?

For all Gambierdiscus species, the algal host that supported the highest cell growth was never the same host in which the highest attachment was observed. While Polysiphonia and Derbesia had significantly higher attachment percentages, Caulerpa, Acanthophora and control treatments promoted greater cell growth (Table 2-4). Except for in the case of

G. yasumotoi, Polysiphonia and Derbesia consistently showed the two highest percent cell attachment values. Caulerpa and Acanthophora consistently showed cell attachment in the lowest third of all treatments for all five Gambierdiscus species (Table 2-4; Figure

2-7).

Table 2-4. Average percent attachment and average growth rates for each Gambierdiscus, is shown, along with the host that supported the highest attachment, highest cell abundance, and highest growth rates.

Overall Host with Host with Highest growth Species average highest average highest growth rate (div/day) attachment attachment (# cells at end) Polysiphonia, Caulerpa Control G. belizeanus 15.74 % 33.43 % 744 0.193 Polysiphonia, Acanthophora Acanthophora G. caribaeus 19.21 % 35.93 % 479 0.199

Polysiphonia, Control Control G. carolinianus 20.40 % 44.50 % 1022 0.211

Derbesia, Laurencia Dasya G. carpenteri 16.51 % 50.03% 807 0.222

Derbesia Acanthophora Polysiphonia G. yasumotoi 4.56 % 12.29% 638 0.184

Gambierdiscus carolinianus showed the highest overall attachment (all treatments averaged across all days, 20.40%). Attachment to host was highest in Polysiphonia (max

76.05%, avg44.50%), Dasya (max 58.74%, avg 27.64%), and Derbesia (max 58.80%, avg 41.72%) treatments. The average overall attachment of G. caribaeus was 19.12%, with the highest attachment in Polysiphonia (max 53.19%, avg 35.93%), Derbesia (max

42.86%, avg 29.53%), and Dasya (max 39.71%, avg 25.56%) treatments. Gambierdiscus carpenteri (overall attachment 16.51%) attachment to host algae was highest in Derbesia

(max 60.45%, avg 50.03%) and Polysiphonia (max 58.74%, avg 35.62%) treatments. G. belizeanus (overall attachment 15.74%) attachment was highest in Polysiphonia (max

67.99%, avg 33.43%) and Dasya (max 51.75%, avg 21.96%) treatments. G. yasumotoi showed significantly lower attachment for all eight macroalgal hosts when compared with the other four Gambierdiscus species. Very little attachment to any host treatments occurred overall for this species (4.56%). The maximum attachment occurred in Derbesia treatments (max 20.96%, avg 12.29%). G. carolinianus and G. caribaeus show the highest degree of attachment overall, and Polysiphonia was commonly attached to the most (Table 2-4 and Figure 2-7).

80 80 80 70 G. belizeanus 70 G. caribaeus 70 G. carolinianus 60 60 60 50 50 50 40 40 40 30 30 30 20 20 20 % cells attached cells % 10 10 10 0 0

Ulva Ulva

Dasya Dasya

Ulva Ulva

Derbesia Derbesia

Dictyota Dictyota Dictyota

Caulerpa Caulerpa

Dasya

Laurencia Laurencia

Derbesia

Dictyota Dictyota

Caulerpa

Polysiphonia Polysiphonia

Laurencia Laurencia

Acanthophora Acanthophora

Polysiphonia Acanthophor

80 80 70 G. carpenteri 70 G. yasumotoi 60 60 50 50 Average 40 40 attached 30 30 20 Max

% cells attached cells % 20 attached 10 10

0 0

Ulva Ulva

Dasya

Derbesia

Dictyota Dictyota

Caulerpa

Derbesia

Dictyota Dictyota

Caulerpa

Laurencia Laurencia

Polysiphonia

Acanthophora Acanthophora

Figure 2-7. Average attachment (% of total cells that were attached to host, averaged across triplicates and throughout entirety of experiment) of the various Gambierdiscus species for each of the host treatments. Maximum attachment (after averaging treatments, the single day that showed the highest attachment) is also shown for each treatment. Standard error values are shown by error bars. Tukey groupings for these data are provided in Table 2-5.

Table 2-5. Average attachment (% of total cells that were attached to host, averaged across triplicates and throughout entirety of experiment) of the various Gambierdiscus species for each of the host treatments. Data is presented from lowest attachment to highest. Letters indicate groupings from Tukey‟s post-hoc tests.

G. belizeanus G. caribaeus

Average % Tukey Average % Tukey Host Algae Host Algae attachment grouping attachment grouping Laurencia 7.05 AB Acanthophora 2.41 A Caulerpa 7.50 AB Caulerpa 4.16 AB Acanthophora 9.45 ABC Dictyota 13.84 ABC Ulva 12.88 BCD Laurencia 20.96 ABCD Dictyota 13.58 BCD Ulva 21.30 ABCD Derbesia * 20.18 CD Dasya * 25.56 BCD Dasya * 21.96 DE Derbesia * 29.53 CD Polysiphonia * 33.43 E Polysiphonia * 35.93 D 10.

G. carolinianus G. carpenteri

Average % Tukey Average % Tukey Host Algae Host Algae attachment grouping attachment grouping Caulerpa 5.47 A Ulva 0.0 A Acanthophora 7.50 A Caulerpa 4.02 AB Laurencia 10.07 AB Acanthophora 4.66 AB Dictyota 11.37 AB Laurencia 9.21 BC Ulva 14.91 AB Dictyota 12.94 BC Dasya * 27.64 BC Dasya * 15.66 C Derbesia * 41.72 C Polysiphonia * 35.62 D Polysiphonia * 44.50 C Derbesia * 50.03 E

G. yasumotoi

Average % Tukey Host Algae attachment grouping * Data marked with a star indicate host species Caulerpa 0.63 A that have been described as filamentous in Acanthophora 0.72 A structure. Notice that these three algae Dasya * 2.62 AB consistently host the most Gambierdiscus Laurencia 2.94 AB attachment for all but one species. Dictyota 3.81 AB Polysiphonia * 3.92 AB Ulva 9.49 BC Derbesia * 12.29 C

Table 2-6Averaged (by triplicate) attachment, cell abundance, and growth rate data is categorized for each of the host treatments. The lowest three values in each treatment were categorized as „low‟, values that fell in the middle three (or two when no control is included) were categorized as „mid,‟ and the highest three values were categorized as „high.‟

Acanthophora Caulerpa Dasya Derbesia Dictyota Laurencia Polysiphonia Ulva Control G. belizeanus End # cells Mid High Low Mid High Mid Low Low High Growth rate Mid High Mid Low High Mid Low Low High % attachment Low Low High High Mid Low High Mid G. caribaeus End # cells High Mid High Mid High Mid Low Low Low Growth rate High Low High Low High Low Mid Mid Mid % attachment Low Low High High Low Mid High Mid G. carolinianus End # cells High Mid High Low Mid Low Mid Low High Growth rate High High Mid Low Low Low Mid Mid High % attachment Low Low High High Mid Low High Mid G. carpenteri End # cells Mid Low High Low Low High High Low Mid Growth rate Mid Low High Mid Low High High Low Mid % attachment Low Low High High Mid Mid High Low G. yasumotoi End # cells High Mid High Mid Low Low High Low Mid Growth rate High Mid High Low Low Low High Mid Mid % attachment Low Low Low High Mid Mid High High

In order to simplify the data as a way to review which algae were overall beneficial or non-beneficial within the Gambierdiscus genus, categories were created. Data was categorized in a value of „high, „medium‟, or „low‟ for average attachment, cell abundance, and growth rate data, and for each of the host treatments. For each Gambierdiscus species, each of the variables was reviewed, from lowest to highest values (Table 2-5). The lowest three values for each were categorized as „low‟, values that fall in the middle three (or two when no control was included) were categorized as „mid‟, and the highest three values were categorized as „high‟. Polysiphonia and Derbesia ranked high in cell attachment for all five species, while Acanthophora and Caulerpa ranked low for all five. Acanthophora did not rank low for growth rate or cell abundance for any of the five. Dasya ranked high in all three variables for G. carpenteri and G. caribaeus and Polysiphonia ranked high for all three variables for G. carpenteri and G. yasumotoi. Gambierdiscus belizeanus was the most variable, with its highest and lowest values for growth and attachment not occurring simultaneously in any of the treatments (Table 2-6).

Changes in wet weight of host algae

Wet weights taken at the start of the experiment (all ~25 g) were compared to the end of experiment wet weights (see Table 2-7 and Figure 2-8). Four host algae showed a positive change in biomass (Dictyota81.53%, Laurencia 8.42 %, Polysiphonia14.32%, and Ulva 77.39%), and four algae showed a negative change (Acanthophora -12.33%,

Caulerpa -36.69%, Dasya -18.94%, and Derbesia -43.98%). When compared to average growth rates, average end cells, and average attachment percentages, no correlations were present (Figure 2-9).

Table 2-7. Percent biomass changes for each host algae, averaged across all treatments.

Host Algae % Biomass ∆

Acanthophora - 12.33 Caulerpa - 36.69

Dasya - 18.94 Derbesia - 43.98 Dictyota + 81.53 Laurencia + 8.42 Polysiphonia + 14.32

Ulva + 77.39

50 Average start wet weight Average end wet weight 40

30 wet weight weight (g) wet 20

Ulva Ulva

Dasya

Derbesia

Dictyota Dictyota

Caulerpa

Laurencia

Polysiphonia Acanthophora

Figure 2-8. Average start and end weights for each host

1000 0.3 60 R² = 0.0234 800 R² = 0.0585 0.2 R² = 0.0117 40 600 0.1 400 0 20 Div/day

200 -0.1 Average # end # Average cells Avg % attached cells Avg 0 0 -0.2 -100 0 100 200 -100 0 100 200 -100 0 100 200 % biomass change % biomass change % biomass change Figure 2-9. Percent change in algal biomass versus average attached cells (left panel), average end cells (middle panel), and average growth rates (right panel).

DISCUSSION

Although a great deal of work has focused on the ecological and environmental factors that affect Gambierdiscus populations, very little work has considered the role that the macroalgal hosts play. The results of this experiment demonstrate variable responses in epiphytic behavior of Gambierdiscus to different macroalgal hosts (Figure 2-7 and Table

2-5) and stimulation and/or inhibition of Gambierdiscus growth by different macroalgae hosts (Figures 2-3, 2-4, 2-5; Tables 2-2, 2-3, 2-4). While previous research has suggested similar results within the Gambierdiscus genus (Parsons et al., 2011; Grzebyk et al.,

1994; Bomber et al., 1989), this is the first study to demonstrate that these phenomena vary between different Gambierdiscus species.

Variations in growth within the Gambierdiscus genus

The significant variations in cell growth within the Gambierdiscus genus (Figures 2-3, 2-

6; Table 2-2, 2-4) and within the host treatments (Figures 2-4 and 2-5) show that there is

43 great variability within the genus. Although increased growth of certain Gambierdiscus species may be supported by certain host species (i.e., showing more growth than the control treatment; Figure 2-6), not all Gambierdiscus will be the same and some may even be inhibited by that same host species. G. carolinianus showed the highest growth rate in the control treatment (with no host algae), while other species showed moderate or low growth in the controls, indicating that while some species were stimulated in the host treatment, G. carolinianus was inhibited by the presence of host algae in general. This might be explained by different assimilation rates within the genus. The other species exhibited faster growth in some treatments with hosts, possibly because the algae were releasing nutrients or other chemicals that supported cell growth. However, if G. carolinianus has a slower assimilation rate, it may have been absorbing the same chemicals and/or nutrients, but was unable to use them as quickly, and demonstrate the higher growth that the other species showed. Another possible explanation is that G. carolinianus is an inferior competitor for nutrients (compared to the other four

Gambierdiscus), and the algae and/or the microbial community were able to outcompete the cells, resulting in low growth.

Variations in epiphytic behavior (cell attachment) within the Gambierdiscus genus

Macroalgal hosts Polysiphonia (Rhodophyte), Derbesia (Chlorophyte), and Dasya

(Rhodophyte) consistently had the most attachment overall by Gambierdiscus cells in this experiment (Table 2-5, see *). Hosts Acanthophora (Rhodophyte) and Caulerpa

(Chlorophyte) had the least attachment. These results suggest that the algal phylum does not play a role in the host preference by Gambierdiscus. Instead, the host structure has

44 more of an effect as Polysiphonia, Derbesia, and Dasya were the only filamentous species used and all three species showed very high levels of attachment by

Gambierdiscus in four out of five species, the fifth being G. yasumotoi, which showed low epiphytic behavior in all host treatments (Figure 2-7, Table 2-5). As visually observed throughout this experiment, these filamentous algae species provide a thinner/smaller substrate, and we hypothesize that this algae structure may provide more locations conducive for Gambierdiscus cells to attach to, compared with larger/flatter surfaces that are provided by the more coarsely branched algae genera, such as Dictyota,

Laurencia, Caulerpa, and Acanthophora; or by the flatter, sheet-like algae such as Ulva.

These results were highly reproducible, as demonstrated by the low variation in cell counts between replicates for most treatments, resulting in low standard deviation values, and small error bars as seen on Figures 2-5, 2-6, and 2-7 (also see Appendices 1-5). This low within-treatment variability contrasts sharply with the large amount of variation seen among hosts, exemplifying the importance of the between-treatment effects.

It is important to mention that these experiments were carried out using a single strain for each of the Gambierdiscus species, and the differences seen within the results may represent strain level differences. However, it is unknown how much variation exists between strains within these five each single species of Gambierdiscus. Previously, it has been reported that G. caribaeus grew the fastest, in experiments of eight separate species, including all of the ones used here, except for G. yasumotoi (Kibler et al., 2012). Here, the growth rate of G. caribaeus was lower than three (under control conditions) of the five species. This discrepancy could be explained by differences in growth conditions

45 between the 2 experiments, or possibly differences in growth existing between strains of the same species.

Attachment to host macroalgae alone may not accurately represent Gambierdiscus’ preference for those algae. For example, although Acanthophora treatments showed the highest growth for three of the five Gambierdiscus species (compared to the other algae), there was very little attachment by all Gambierdiscus species to Acanthophora (Figure 2-

7; Table 2-4). In many cases, the hosts with the lower attachment percentages were the treatments that showed the most Gambierdiscus cell growth (Figures 2-8 – 2-12). This may suggest that conditions of the microenvironment around host algae (e.g., nutrient levels, stimulatory or inhibiting metabolites) may be influencing without physical attachment to that host.

In hindsight, a calculation of surface area may have been useful for comparisons of how much area was available for attachment between hosts, however, it was not possible to determine surface area after experimentation was over. It is important to state that there may have been significant differences in the surface area available in each treatment, which could have had an effect on the attachment. Gambierdiscus abundance has been shown to be positively correlated with surface area of macroalgal hosts (Bomber et al.,

1989), and Lobel et al. (1988) suggest that surface area is the most appropriate for abundance comparisons. Although, as mentioned previously, Gambierdiscus abundances are typically reported as “cells per gram macroalgae” due to difficulties associated with determining surface areas.

Some hosts, such as Dictyota and Ulva, took up much more space in the treatment wells, meaning that in those treatments, any single cell at any given time may have been physically closer to the host than those cells in treatment wells in which the algae physically took up less space in the well (e.g., Figure 2-1). However, with a well diameter of 3.64 cm, no cell was any further than 1.5 cm from a portion of a host at any time, and in most cases, much closer. The swimming abilities of Gambierdiscus would suggest that this distance would not be an obstacle for attachment, but there is the possibility that some cells were content with attaching to the plastic sides of the well plate if they were physically closer to it. There did not appear to be any lack of free space available for attachment in any of the treatment wells. Observations suggest that the Gamberidiscus will attach in close association with one another. This observation along with the large areas of host algae that did not have attached cells, suggests that the cells were not unattached due to a lack of free space.

Differences in Gambierdiscus yasumotoi

Gambierdiscus yasumotoi showed the lowest attachment of all five species, yet growth rates of the cells did not appear to be reduced by the presence of the macroalgae; for most host treatments the G. yasumotoi growth rate was not significantly lower than the other species, indicating that conditions were optimal for cell growth, but not for attachment.

As previously mentioned, it has been suggested that G. yasumotoi be moved to a different genus (Gomez et al., 2015). These experiments took place prior to this information.

However, it was beneficial still that the species was included in the experiments, since the differences in epiphytic behavior are now able to support the notion that the species

47 behaves differently compared to others within the Gambierdiscus genus. While the other

4 species had average attachments of between 15.74 and 20.40 percent overall for this experiment, G. yasumotoi had a significantly lower overall attachment of 4.56 percent. In addition, for each of the eight macroalgae hosts, the lowest attachment percentage occurred within in the G. yasumotoi treatment (expect for Ulva, which had 0% attachment in Ulva, due to complete mortality in the treatment).

Gambierdiscus yasumotoi also exhibited different physical behaviors throughout the experiment as well, which were observed during the microscopy. Aside from being notably smaller and of a more globular shape (Holmes, 1998; Gomez et al.,2015), the cells were observed much more often to be actively swimming in the water column compared to the other four species in these studies. Their behavior stood out noticeably as being different, and these observations along with the data presented here fully support the move of the species to a separate genera.

Macroalgae selection

The macroalgae species used were chosen based on their common presence in the Florida

Keys and their ability to be easily collected and kept alive in culture prior to and during the duration of the experiment. All of the host species were collected from locations that have been previously known to harbor Gambierdiscus populations. Some species that are also common in the collection areas, and are considered to be good host genera for

Gambierdiscus, were unable to be used due to physical characteristics that would have not have worked under the experimental conditions used. For example, Halimeda,

Galaxaura, Padina, Penicillus, and Udotea would have been good choices, but the

48 calcareous structure and/or thickness of the algae would not allow for observations under the microscope. Recent work by Parsons (unpublished data) has also found the seagrass

Thalassia testudinum to be a good host as well. Unfortunately, seagrass blade fragments would not have survived the 29 days of experimentation. Future studies investigating the host preferences of some of these species would be beneficial, as some of them, particularly the calcareous Chlorophytes, make up a large portion of the benthic community throughout the Caribbean, including the Florida Keys.

Phaeophytes

Macoralgal species of the genus Dictyota have been generally considered to be good hosts for Gambierdiscus due to studies presenting high cell abundances (Carlson et al.,

1984; Ballantine et al., 1985; Bomber et al., 1989; Delgado et al., 2006; Carlson and

Tindall, 1985) and also the wide distribution throughout the Caribbean. No other

Phaeophytes were used in this experiment, as Dictyota was the only one common enough for field collection that would also survive in culture.

Parsons et al. (2011) reported that >99% of G. toxicus cells exposed to Dictyota died within a short time, and presumed it to be a result of either harmful exudates or other chemical conditions within the experiment. In the work presented here, Dictyota treatments showed average, or lower, growth in comparison to other host treatments, but not necessarily high levels of mortality. The differences in these results may be attributed to water changes in this experiment, or differences in G. toxicus versus the Caribbean species used here. Attachment was also low or average for Dictyota treatments for all

Gambierdiscus species.

Dictyota is a commonly sampled macroalgae for field studies aiming to determine

Gamberidiscus abundances, mostly because it is common on reefs over a wide geographical area and it is present year round. These attributes allow for consistent comparisons in abundance to be made across different locations, as well as seasonally.

However, this study, along with the results of Parsons et al. (2011) demonstrate that further consideration should be given as to which species are most commonly collected, to obtain the most accurate outlook of Gambierdiscus abundance on a reef.

At least 230 chemical compounds have been isolated from various species within the

Dictyota genus (Paul et al., 2001). However, despite their production of secondary metabolites that act as a chemical defense against grazing (Hata, 2010; Norris and

Fenical, 1982), it has been demonstrated that Dictyota may be heavily grazed upon on

Caribbean reefs (Littler et al., 1983). Interestingly, this study found the grazing intensity on Dictyota to be very high when presented to grazers suspended midway in the water column or as a drifting algae, but low when exposed on a benthic grid, which is the closest representation of that study to their natural presence on a reef.

Chlorophytes

Many Chlorophytes are known to be good hosts for Gambierdiscus. This experiment used species of the genera Ulva, Derbesia, and Caulerpa. Although few Gambierdiscus cell abundances have been published for field collections of any these genera (Vandersea et al., 2012), Caulerpa has been collected for Gambierdiscus cells the past (McMillan et al., 1986; Holmes 1998; Kuno et al., 2010; Vandersea et al., 2010; Richlen and Lobel,

2011). While cell attachment was high in Derbesia treatments for all Gambierdiscus

50 species, attachment was very low in all Caulerpa treatments and average in Ulva treatments. Cell abundances were also low for all three of these Chlorophytes in comparison to other hosts.

Gambierdiscus belizeanus, however, had its highest cell abundance in the Caulerpa treatment. Attachment percentages were not high for G. belizeanus on Caulerpa, but the significantly higher cell abundances in those treatments versus all other host species, suggest that Caulerpa was somehow stimulating growth for G. belizeanus, but for no other species. This could be credited to nutrient uptake rates or assimilation rates, or it could be a result of G. belizeanus responding differently than the other species to secondary metabolites being produced by the host. It has been demonstrated that although

Ulva lactuca is chemically defended (De Lara-Issassi et al., 2000), it is still heavily grazed upon (Littler et al., 1983). With G. belizeanus being of particular interest at this time in the Caribbean, known to be the most toxic species in the region currently, the relationship between G. belizeanus and Ulva is interesting to consider.

Aside from the growth stimulation of Caulerpa for G. belizeanus, and the high attachment percentages in all Derbesia treatments, results from this study did not suggest

Chloropytes as a group to be an especially good host for Gambierdiscus, as has been previously suggested (Carlson et al., 1984; Bomber et al., 1989). However, the inclusion of calcareous species or any other Chlorophytes may have yielded different results.

Rhodophytes

It has been suggested that Rhodophytes are the best hosts for Gambierdiscus (Taylor

1979; Yasumoto et al., 1979; Yasumoto et al., 1980). Most of the Rhodophyte genera

51 used in this research have previously been shown to be hosts for Caribbean species/strains of Gambierdiscus (Laurencia - Carlson et al., 1984; Bomber et al., 1989;

Vandersea et al., 2012; Acanthophora - Carlson 1984; Vandersea et al., 2012;

Polysiphonia – Vandersea et al., 2012).

Dasya and Polysiphonia both supported the highest percentages of attachment by

Gambierdiscus cells throughout the experiment (with the exception of G. yasumotoi on

Dasya), and Dasya also supported high cell abundances for both G. carpenteri and G. caribaeus. Although having both high attachment and high cell abundance may indicate a host supportive of Gambierdiscus populations, Dasya is chemically defended (De Lara-

Isassi et al., 2000), and has been shown to be of low palatability to herbivorous fishes

(Arthur et al., 2009).Conversely, while Polysiphonia hosted high percentages of attached cells, the cell abundances were not high in comparison to other host treatments, except somewhat in G. carpenteri and G. yasumotoi when compared to control treatments. Both

Polysiphonia and Dasya were likely supportive of high attachment by cells because of their filamentous structure, not their classification of Rhodophyta.

Laurencia only appeared to be growth stimulating to G. carpenteri and attachment was low for all species except G. caribaeus. More than 570 chemical compounds have been isolated from the genus (Paul et al., 2001).Although Litter et al. (1983) did not use

Laurencia intricata in field experiments, they did use two other species in the genus which yielded significantly different high/low levels of grazing, indicating that differences exist in palatability within the genus. However, it has been reported that the genus in general is not palatable.

Acanthophora, another common collection species for field studies of Gambierdiscus, typically had low numbers of attached cells throughout this experiment, but it did have high overall cell abundances for some species, notably G. caribaeus and G. yasumotoi, in which the Acanthophora treatments had the highest cell abundance of all treatments. G. carolinianus also had the highest cell abundances in the host treatment Acanthophora, but less than in the control treatment. Acanthophora are not known to be chemically defended, and grazing intensity has shown to be high (Littler et al., 1983).

No overall preference by Gambierdiscus for Rhodophytes, Chlorophytes, or Phaeophytes occurred for attachment and no phyla stimulated growth or cell abundance more than the others. However, statistical analysis was not ideal, due to only one Phaeophyte being used. More species from all phyla would need to be tested in order to determine if any real preference or growth stimulation is occurring.

Host structure

Results suggest that host preference (high attachment) by Gambierdiscus is independent of host phylum, and more dependent on host structure, as previously suggested by Taylor and Gustavson (1983) and Parsons and Preskitt (2007). The highly foliose, or filamentous, algae (Polysiphonia and Derbesia) consistently hosted the most attached cells in this experiment, and for all five Gambierdiscus species. Growth was too variable among Gambierdiscus to determine whether any specific phyla or structural group of algae was most supportive of growth.

Significance of host palatability

Regardless of how many Gambierdiscus cells a host may harbor, or how toxic those cells may be, those toxins only enter the food web if an herbivore consumes that host. As previously discussed, some of the hosts in this study do produce types of chemical compounds that deter grazing, therefore making them unpalatable to some grazers. Even with these chemical defenses, most are still palatable to some extent (Cruz-Rivera and

Villareal, 2006; Littler et al., 1983), but there are certainly some host algae that are more preferably eaten by herbivorous fishes.

Some host algae may end up acting as a sort of refuge, or safe harbor, for Gambierdiscus populations, meaning they may be present in high abundances on a particular reef, but there is no danger of an outbreak of ciguatera, because the toxins are not entering the food web. However, in such a situation, there could be potential for shifts in the ecosystem to result in changes to the grazing behaviors of fish. For example, if the algal cover changes, then grazers may be forced to consume algae that they typically don‟t prefer. Or changes in the fish community could occur as a result of overfishing or disease, leading to new grazers moving onto a reef, with a different grazing preference.

Both represent scenarios in which toxic Gambierdiscus are finding refuge on a particular host algae population that experienced little/no grazing pressure (possibly leading to high abundances over that time), followed by a shift to grazing of that host, resulting in a new vector for high amounts of toxin flux into the food web.

It has been demonstrated that herbivores prefer enriched over ambient macroalgae, and that nutrient enrichment on a coral reef will increase overall macroalgal palatability, even

54 for species that are normally unpalatable (Chan et al., 2012). Another scenario of high toxin flux might be under a situation such as this one, in which the enriched nutrient conditions are also influencing growth and toxin production of the Gambierdiscus, in addition to influencing the palatability of hosts.

Changes in health of host macroalgae

Of the host algae used in this experiment, Dictyota and Ulva gained the most biomass throughout the experiment (Figure 2-8). However, both of these hosts supported populations of Gambierdiscus throughout the experiment that showed low growth rates and low-mid attachment, when compared with the other host species (Table 2-6), suggesting that a healthy, growing host does not necessarily create optimal habitat or growing conditions for Gambierdiscus.

As host algae become unhealthy and/or begin deteriorating, there are potential effects on epiphytes, such as: a) unhealthy algae may become defensive to any type of predation or epiphytism and begin producing chemical compounds to deter grazers and/or epiphytes; b) deteriorating algae fragments may release chemical compounds from within as they are broken down; c) unhealthy algae may begin taking up nutrients at a faster rate, trying to gain nutrition, therefore leaving less available for surrounding organisms; d) a healthy microbial community that is busy with decomposition processes may begin outcompeting other epiphytes for available resources in the water; or even e) there is so much decomposition occurring that concentrations of nutrients being cycled back in the water are supportive of other epiphytes.

Healthy hosts, on the other hand, also have a range of potential effects on the epiphytic community, such as: a) being healthy, they have plenty of nutrition, and begin taking up nutrients slower, leaving more available for other organisms; b) the healthier they are, the faster they are growing, and the nutrients they need to absorb, outcompeting other organisms; c) the healthier they become, the more aggressive they can be, producing more chemical compounds to deter grazing/epiphytism, or d) they healthier they become, the less they need to be defensive, therefore they produce less chemical compounds. As varied as these potential effects may be, there may be just as much variation among different genera of host algae. Variations such as these may account for the differences in the results of this experiment, for both cell growth and attachment.

For some of the species that had a loss of biomass over the 28 days, the decline in health was visually apparent in the treatment wells. Caulerpa and Dasya, for example, both appeared to be dying during the experiment, and this was supported in the wet weight data (Table 2-8). Both hosts supported Gambierdiscus populations that had moderate levels of cell abundance, but Caulerpa showed low attachment for all Gambierdiscus species. Dasya, however, showed high attachment for all species except G. yasumotoi.

This variation makes it unclear how host health (growing or dying) may affect epiphyte populations, but possibly suggests that the physical structure (coarsely branched versus filamentous) was more important criteria for attachment than host health.

Unknown factors

It is unknown what nutrient flux is occurring in the treatments over the time of the experiment. The physical appearance of the algae over the course of the experiment indicates that some algae were continuing to grow, while others slowly began dying.

These two separate scenarios likely lead to two very different nutrient fluxes occurring; while growing algae might be taking up nutrients from the water (which was constantly being replenished), the dying algae may be releasing nutrients, as well as other compounds. Although an entirely different experiment is needed to fully understand the nutrient dynamics that may have occurred in the treatment, the next chapter of this thesis investigates how differing nutrient enrichment of the host algae affects the growth and behavior of the Gambierdiscus.

CHAPTER 3:

AMBIENT VERSUS ENRICHED HOST NUTRIENT CONDITION EFFECTS ON

THE GROWTH AND EPIPHYTIC BEHAVIOR OF GAMBIERDISCUS

Nutrient effects on Gambierdiscus growth

Gambierdiscus populations are commonly found in the oligotrophic waters of coral reefs, an environment in which nutrient concentrations are low in the water column and can be limiting. Gambierdiscus populations are not limited to coral reef environments, however, and can also be found in areas capable of having much higher nutrient fluxes (i.e., mangrove lagoons, enclosed bays, seagrass beds). Lartigue et al. (2010) established that growth responses among different Gambierdiscus strains varied with different nitrogen sources, indicating that the various Gambierdiscus species likely have different nutrient uptake kinetics. It has yet to be determined whether different species within the genus inhabit different habitat types based on their varying nutrient physiologies and growth responses to temperature, light, and salinity.

Various field studies have found no correlation between nutrients and Gambierdiscus cell densities (Yasumoto et al., 1980; Ichinotsubo et al., 1994; Parsons and Preskitt 2007; as reviewed by Parsons et al., 2012). A possible explanation may be differential nutrient uptake kinetics between species; it may be that some species of Gambierdiscus prefer low nutrient conditions, while other species prefer medium or higher level conditions, and therefore no correlations will exist without species-specific studies. Carlson et al.

(1984), however, did find a correlation between several nutrients and Gambierdiscus populations, although both increased along an inshore-offshore gradient, so the relationship may be naturally coincidental.

Changes in nutrient inputs can lead not only to increased macroalgal cover, but may alter the algal composition of an ecosystem, becoming dominated by species which thrive in

59 nutrient-rich conditions. Surveys conducted in the Florida Keys report data that suggests that coral reefs in the region are undergoing a phenomenon referred to as a phase-shift, in which the reefs show a reduction in coral cover and biodiversity, and a consequential increase of benthic macroalgal cover (Maliao et al., 2008). This phenomenon is widespread throughout the Caribbean region (Gardner et al., 2003; Hughes 1994).

However, the effects of these changing macroalgal communities on the abundance and toxicity of Gambierdiscus have yet to be investigated.

The physiology of macroalgal species often changes under nutrient-rich versus oligotrophic conditions (e.g., higher tissue nitrogen content in the former). If macroalgal tissues are enriched in nutrients (N and P), it is likely that they will be more favorable hosts for Gambierdiscus, or the hosts may release these nutrients, which makes them available for uptake by the epiphytic community (Russell et al., 2005), including

Gambierdiscus populations. Although it has been suggested that epiphytic dinoflagellates may live in close association with macroalgal species that provide higher concentrations of nutrients (Steidinger 1983), few studies have investigated whether Gambierdiscus cells are receiving nutrients from their hosts.

Marine secondary metabolites

A wide diversity of secondary metabolites are produced by many benthic marine macroalgae, many of which are effective at deterring grazing pressures from fishes and invertebrates and may affect micro-organisms as well (Hay and Fenical 1988). Over 500 secondary metabolites have been described from marine algae, although few studies have

60 investigated the effects that these compounds have on epiphytic communities. Many of the studies on algal secondary metabolites have focused on the interactions with herbivores (Hay et al., 1987; Hay and Fenical 1988; Hay and Steinberg 1992), as it is easy to measure the concentration of these compounds within the algae tissue.

Determining the levels of compounds present on the surfaces of the algae and understanding the effects on epiphyte populations is much more difficult (Jennings and

Steinberg 1997). Ragan et al. (1980) suggest that some of the various compounds released by macroalgal hosts may benefit the growth of phytoplankton.

Red algae produce a variety of compounds, the most common of which are terpenes and phenols. Algae of the genus Laurencia produce over 250 secondary metabolites (Erikson

1983), some of which are known to have antibiotic effects (Vairappan 2003). Dictyota produces and releases several secondary compounds, dictyols for example, into the surrounding water (see Table 2-1; Cronin et al., 1995; Walters et al., 1996; Targett and

Arnold 1998), which have been shown to deter feeding by many species of fish, sea urchins and some amphipods (Hay et al., 1988).

Various green algae, including Caulerpa, Halimeda, Udotea, Penicillus, and

Rhipocephalus are particularly resistant to coral reef herbivores, and are less palatable than other coral reef algal species due to the compounds they produce (Paul and Hay

1986; Wylie and Paul 1988; Meyer et al., 1994).For example, chlorophytes of the order

Caulerpales have been shown to produce compounds that inhibit the growth of microorganisms, inhibit development of fertilized urchin eggs, and show toxic effects on larval stages of potential herbivores (Paul and Fenical 1986). The effects of these compounds on specific microorganisms such as Gambierdiscus are unknown, but the

61 toxic effects shown on other organisms give reason to believe that these compounds may be capable of hindering growth.

While a certain compound produced by a macroalgal host may have stimulatory effects, it is possible the same host may produce other compounds which have inhibiting effects.

Negative effects of the metabolites may include deterring epiphytes from attaching to the host, inhibiting growth, or hindering physiological responses, such as production of toxins or other compounds. The effects on the physiology of Gambierdiscus specifically, if any, are unknown aside from a few studies.

Grzebyk et al. (1994) conducted experimental assays in which the growth of

Gambierdiscus was compared in different seawater treatments previously incubated with macroalgae, demonstrated that the red algae, Portieria hornemanii, stimulated growth while another red algae, Halymenia floresia, inhibited growth. The brown algae

Turbinaria ornata and Sargassum c.f. turbinetifolium both slightly stimulated growth as well. The two red algae released nitrogenous nutrients and phosphates during the incubation, while the two brown algae released phosphate but consumed nitrogenous nutrients. Concurrent field studies found some macroalgae, particularly the rhodophytes, to harbor higher numbers of dinoflagellates, including Gambierdiscus, which led the authors to suggest was the result of the release of stimulating compounds by the red algae. Similar work by Carlson and Tindall (1985) found Gambierdiscus growth in extracts of Dictyota and Turbinaria to be no different than in natural seawater, but extracts of Chaetomorpha supported significantly greater growth.

In contradiction to the results of Grzebyk et al. (1994) described above, Asuncion et al.

(1994) found Halymenia extracts to stimulate Gambierdiscus growth and Parsons et al.

(2011) found that cells did not attach to Portieria hornemanii, and died within a few days of exposure. Although Parsons et al. (2011) used cultures verified to be G. toxicus in their experiments, it is unknown which species were used in the other studies, and variations in growth responses between different Gambierdiscus species are a likely explanation for the different results.

RESEARCH PURPOSE

This portion of the study was conducted simultaneously with the experiment described in the previous chapter, but is presented separately to avoid confusion. Results were presented on the effects of eight different macroalgal hosts on the growth and epiphytic behavior (attachment) of five different Gambierdiscus species. This portion of the research investigates whether hosts with different nutrient enrichment will have a variable effect on those same Gambierdiscus responses.

Objective 3: Determine if a nutritionally enriched macroalgal host affects

Gambierdiscus growth rates and cell abundance in comparison to a non-, or less-

enriched host.

Objective4: Determine if a nutritionally enriched macroalgal host affects

Gambierdiscus epiphytic behavior in comparison to a non-, or less-enriched host.

METHODS

The methods explained in the previous chapter for Gambierdiscus isolations,

Gambierdiscus collections, and macroalgae collection are the same as those used in the portion of the experiment described in this chapter, as subsamples of the same cultures were used for this portion of the experiment. All growth conditions for the

Gambierdiscus cultures were the same as well. The macroalgae hosts described in the previous chapter, were grown in what will from here on be referred to as „ambient‟ conditions. Simultaneously, there were also portions of macroalgae hosts that were conditioned in „enriched‟ conditions. Two weeks prior to the start of these experiments, portions of macroalgae, then all growing in ambient conditions, were divided into smaller portions and both grown separately in ambient and enriched conditions. The enriched water conditions consisted of the same filtered water from the Florida Keys, with modified Keller‟s medium (without TRIS, Cu, or Si; with GeO2, to inhibit diatom growth) added.

This experiment was conducted in the same way as the one described in the previous chapter, only the algae was enriched prior to the start, and the water changes throughout the experiment were done using the enriched water solution and not the ambient water solution, thus keeping the macroalgae hosts enriched with higher levels of nutrients throughout the entirety of the experiment.

Five days prior to Day 0 (start of experiment), macroalgae fragments of 25 mg were placed individually into separate wells containing 8 ml of enriched seawater for each treatment to be used in the experiment in triplicate (8 host algae + 1 control * 5

Gambierdiscus species * triplicates = 135 wells). These fragments acclimated for five days, with frequent water changes, to allow for any harmful exudates to dissipate. See

Figure 3-1for an illustration of the usage of ambient vs. enriched seawater throughout the experiment.

On Day 0, new wells of 8 ml enriched Florida Keys seawater were prepared for all treatments. Twenty-five cells of the appropriate species of Gambierdiscus were added to each well, and then the appropriate enriched algae fragments were added. Triplicate controls for each Gambierdiscus species (containing no algae) were also prepared.

Cells were counted on Days 1, 8, 15, 22, and 29 and each cell was categorized as dead, alive and unattached to host, or alive and attached to host (Figure 3-2). Water changes were performed 2-3 times weekly by slowly removing 4 ml of the water by transfer pipet

(with a loss of <1% of cells when done carefully), and replaced with 4 ml filtered enriched seawater.

Figure 3-1.Illustration of the preparation and usage of „ambient‟ and „enriched‟ seawater prior to and throughout the duration of the experiment.

Data Analysis

All data from the experimentation described in the previous chapter will hereby be referred to as „ambient‟ data and everything from this portion of the experiment as

„enriched.‟ Prior to statistical analysis, the cell counts from the five Gambierdiscus species were converted to relative abundance values (% of total cells dead, % attached and alive, % unattached and alive). Cell count data gathered over the course of the experiment was used to calculate growth rates using the same equations presented in the previous chapter.

The 90 treatments (9 host treatments x 5 Gambierdiscus species x 2 nutrient conditions) were grouped according to their similarities using the variables of growth rate, maximum cells, average percent attachment, and maximum percent attachment (2 growth variables and 2 attachment variables, for equal weight) using group-averaged cluster analysis

(CLUSTER) based on Euclidean distance, coupled with similarity profile (SIMPROF) testing for group differences at α = 0.05. Groups determined through the

CLUSTER/SIMPROF procedures were then analyzed using the similarity percentages

(SIMPER) procedure to characterize how the groups differed from one another. The group differences were then visualized using Multi-Dimensional Scaling (MDS) analysis.

All values were normalized prior to analysis, and these multivariate procedures were conducted using PRIMER 7.

RESULTS

Enrichment effects on growth

For the large majority of treatments, there was either no significant difference in growth between the ambient and enriched host conditions, or there was only a slight difference

(see Figures 3-2 through 3-6). No host algae that was enriched consistently provided better conditions for growth or attachment than its ambient treatment (Figure 3-12).

For G. belizeanus treatments, growth was similar under enriched conditions for most treatments as those presented previously for ambient conditions, except that by days 22 and 29 cell abundance was significantly higher in the Acanthophora, Dictyota, and control (Day 29 only) treatments, but lower in the Caulerpa treatments. Under enriched conditions, cell abundance was highest in the control treatment, followed by the

Acanthophora treatments, and highest in the Caulerpa treatment under ambient conditions (Figure 3-2).

Gambierdiscus caribaeus showed little variation between ambient and enriched treatment among the different hosts. For Days 1 through 22, there were no significant differences in cell abundances. By the last week (Day 29), the enriched treatments of Acanthophora and the control showed slightly larger abundances, and Caulerpa showed a much higher abundance in the enriched treatment. In all three of these, the ambient treatments showed a decrease in growth after Day 22. The growth rates in all of the nine treatments were the same for Day 1 through 22, but the exponential growth slowed/stopped after Day 22 for the three treatments mentioned. Under both ambient and enriched conditions, cell abundance was the highest in Acanthophora treatments (Figure 3-3).

Gambierdiscus carolinianus showed similar results, with no significant differences between Days 1 through 22 for any of the treatments. By Day 29, enriched Caulerpa treatments showed higher abundance than the ambient treatment, in which the population crashed by that time. Cell abundance was the largest within the Caulerpa and control treatment under enriched conditions, and significantly higher in the control treatments ambient conditions (Figure 3-4).

G. carpenteri treatments showed no differences between cell abundance between ambient and enriched conditions for Days 1 through 22, except for Dasya treatments, which had higher abundances under ambient conditions by Day 15. By Day 29, Dasya and

Laurencia showed significantly higher abundances in ambient treatments. Only Dictyota treatments showed higher abundances in enriched conditions. The highest cell abundance under ambient conditions occurred in Laurencia treatments and under enriched conditions in Dictyota treatments (Figure 3-5).

G. yasumotoi did not show higher abundances under any of the enriched conditions, but did show significantly higher abundances in the ambient conditions for Acanthophora,

Dasya, Polysiphonia, and control treatments. These four treatments are the only ones that showed much growth at all for the species. The highest abundance occurred in ambient conditions of the Acanthophora treatment (Figure 3-6).

800 Acanthophora 800 Caulerpa 800 Dasya 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 Derbesia 800 Dictyota 800 Laurencia 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 800 Polysiphonia Ulva 1600 Controls 1400 600 600 1200 1000 400 400 800 600 200 200 400 Cell abundance Cell 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22Day 29

Figure 3-2.Gambierdiscus belizeanus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

800 Acanthophora 800 Caulerpa 800 Dasya 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 Derbesia 800 Dictyota 800 Laurencia 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 800 Polysiphonia Ulva 800 Controls 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-3.Gambierdiscuscaribaeus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

1000 Acanthophora 1000 Caulerpa 1000 Dasya 800 800 800 600 600 600 400 400 400 200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day Day Day Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15Day 22Day 29 15 22 29

1000 Derbesia 1000 Dictyota 1000 Laurencia 800 800 800 600 600 600 400 400 400 200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day Day Day Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 15 22 29

1000 Polysiphonia 1000 Ulva 1000 Controls 800 800 800 600 600 600 400 400 400 200 200 Cell abundance Cell 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15Day 22Day 29 Figure 3-4. Gambierdiscus carolinianus cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

800 Acanthophora 800 Caulerpa 800 Dasya

600 600 600

400 400 400

Cell abundance Cell 200 200 200

0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 Derbesia 800 Dictyota 800 Laurencia

600 600 600

400 400 400 200 200 Cell abundance Cell 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

800 Polysiphonia 800 Ulva 800 Controls 600 600 600 400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-5. Gambierdiscus carpenteri cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

800 Acanthophora 800 Caulerpa 800 Dasya 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15Day 22Day 29

800 Derbesia 800 Dictyota 800 Laurencia 600 600 600

400 400 400

200 200 200 Cell abundance Cell 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15Day 22Day 29

800 Polysiphonia 800 Ulva 800 Controls 600 600 600

400 400 400 200

Cell abundance Cell 200 200 0 0 0 Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Figure 3-6. Gambierdiscus yasumotoi cell abundances presented over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

Table 3-1.Average growth rate (div/day) for each ambient and enriched treatment.

Ambient Enriched treatment treatment (div/day) (div/day) G. belizeanus Acanthophora 0.111 0.129 Caulerpa 0.127 0.106 Dasya 0.134 0.136 Derbesia 0.104 0.077 Dictyota 0.076 0.089 Laurencia 0.114 0.149 Polysiphonia 0.103 0.091 Ulva 0.093 0.091 G. caribaeus Acanthophora 0.075 0.074 Caulerpa 0.138 0.141 Dasya 0.072 0.110 Derbesia 0.099 0.095 Dictyota 0.131 0.109 Laurencia 0.087 0.087 Polysiphonia 0.118 0.120 Ulva 0.092 0.092 G. carolinianus Acanthophora 0.094 0.077 Caulerpa 0.099 0.086 Dasya 0.109 0.104 Derbesia 0.102 0.117 Dictyota 0.146 0.125 Laurencia 0.066 0.085 Polysiphonia 0.052 0.079 Ulva 0.065 0.086 G. carpenteri Acanthophora -0.034 0.054 Caulerpa 0.095 0.088 Dasya 0.069 0.065 Derbesia 0.090 0.076 Dictyota 0.058 0.084 Laurencia 0.062 0.059 Polysiphonia 0.154 0.104 Ulva 0.066 0.086 G. yasumotoi Acanthophora 0.076 0.106 Caulerpa 0.146 0.132 Dasya 0.153 0.112 Derbesia 0.000 0.062 Dictyota 0.123 0.006 Laurencia 0.006 -0.007 Polysiphonia 0.108 -0.018 Ulva 0.127 0.048

Enrichment effects on epiphytic behavior

Within the G. belizeanus treatments, the enriched treatments showed overall higher attachment percentages than the ambient treatments, except for in Polysiphonia treatments which had higher attachment during Days 8, 15, 22, and 29. No other host treatments in ambient conditions showed higher attachment for more than for a single day

(Figure 3-7).

Gambierdiscus caribaeus had similar results, with only Polysiphonia treatments showing more attachment under the ambient conditions for the duration of the experiment.

Laurencia treatments showed higher attachment on some days under ambient conditions

(Figure 3-8).

Gambierdiscus carolinianus had higher percentages of attachment under ambient conditions for Days 8, 15, 22, and 29 for treatments of Dasya, Derbesia, and

Polysiphonia. (Figure 3-9).

Higher percentage of attachment for G. carpenteri was observed for all days in

Polysiphonia and Ulva treatments, and for most days (8, 15, 22, 29) in Derbesia treatments (Figure 3-10).

Gambierdiscus yasumotoi treatments all showed very little growth and very little attachment in both enriched and ambient conditions. Only the Derbesia treatments showed a higher percentage of attachment in the ambient conditions for more days than in the enriched treatment (Figure 3-11).

100 Acanthophora 100 CaulerpaCaulerpa 100 DasyaDasya 80 80 80 60 60 60 40 40 40 20 20 20

% cells attached %cells 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 DerbesiaDerbesia 100 DictyotaDictyota 100 LaurenciaLaurencia 80 80 80 60 60 60 40 40 40 20 20 20 % cells attached %cells 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 PolysiphoniaPolysiphonia 100 Ulva 80 80 Ulva 60 60 40 40 20 20

% cells attached %cells 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-7. Percent of G. belizeanus cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

100 AcanthophoraAcanthophora 100 CaulerpaCaulerpa 100 DasyaDasya 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 % cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 Derbesia 100 Dictyota 100 Laurencia 80 Derbesia 80 Dictyota 80 Laurencia 60 60 60 40 40 40 20 20 20

% cells attached %cells 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 Polysiphonia 100 Ulva 80 Polysiphonia 80 Ulva 60 60 40 40 20 20

% cells attached %cells 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-8. Percent of G. caribaeus cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

100 Acanthophora 100 CaulerpaCaulerpa 100 DasyaDasya 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 % cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 DerbesiaDerbesia 100 DictyotaDictyota 100 LaurenciaLaurencia 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 % cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Polysiphonia 100 Polysiphonia 100 UlvaUlva 80 80 60 60 40 40 20 20 0 0

% cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-9. Percent of G. carolinianus cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

100 AcanthophoraAcanthophora 100 CaulerpaCaulerpa 100 DasyaDasya 80 80 80 60 60 60 40 40 40 20 20 20

0 0 0 % cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 Polysiphonia 100 Ulva 80 Polysiphonia 80 Ulva 60 60 40 40 20 20 0 0

% cells attached %cells Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-10. Percent of G. carpenteri cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

100 Acanthophora 100 Caulerpa 100 Dasya 80 80 80 60 60 60 40 40 40 20 20 20

% cells attached %cells 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 Derbesia 100 Dictyota 100 Laurencia 80 80 80 60 60 60 40 40 40 20 20 20

% cells attached %cells 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

100 Polysiphonia 100 Ulva 80 80 60 60 40 40 20 20

% cells attached %cells 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Figure 3-11. Percent of G. yasumotoi cells attached to host over 29 day period for ambient and enriched conditions, for each host treatment. Error bars show standard error.

Some host treatments showed a decline in the percentage of attached cells from the beginning to the end of the experiment. Dasya and Caulerpa treatments, both ambient and enriched, showed this trend more consistently than the other host treatments (Figures

3-7 through 3-11). Dasya and Caulerpa were also the hosts that visually appeared to be declining in health (breaking apart, losing color, and getting thinner) the most over the 29 days. For many ambient treatments, the population was crashing sometime between Day

22 and Day 29 while growth continued through Day 29 for their enriched counterparts.

The enriched treatments were receiving a higher introduction of nutrients with each water change than the ambient treatments which is a plausible reason as to why the growth cycle was lasting longer in many of those treatments.

80 G. belizeanus 60

80 G. caribaeus 60

80 G. carolineanus 60 40 20

% cell attachment cell % 0

80 G. carpenteri 60 40 20 0

80 G. yosumotoi 60 40 20

Ulva Ulva

Dasya

Derbesia

Caulerpa

Dictyota

Laurencia

Polysiphonia Acanthophora

enriched ambient

Figure 3-12. Percentage of cell attachment across all days and all triplicates for each host and for ambient and enriched nutrient conditions.

Table 3-2. Average percent attachment values for each ambient and enriched treatment.

Ambient Enriched treatment treatment (%) (%) G. belizeanus Acanthophora 9.4 14.4 Caulerpa 7.5 21.5 Dasya 22.0 15.1 Derbesia 20.1 35.2 Dictyota 13.6 17.3 Laurencia 7.1 11.6 Polysiphonia 33.4 15.6 Ulva 12.9 24.2 G. caribaeus Acanthophora 2.4 11.0 Caulerpa 4.2 0.0 Dasya 25.6 33.5 Derbesia 29.5 43.7 Dictyota 13.8 14.0 Laurencia 21.0 14.5 Polysiphonia 35.9 18.3 Ulva 21.3 15.4 G. carolinianus Acanthophora 7.5 3.7 Caulerpa 5.5 0.0 Dasya 27.6 22.0 Derbesia 41.7 31.8 Dictyota 11.4 15.2 Laurencia 10.1 12.6 Polysiphonia 44.5 19.1 Ulva 14.9 17.7 G. carpenteri Acanthophora 4.6 3.7 Caulerpa 4.0 0.0 Dasya 15.7 18.7 Derbesia 50.0 30.7 Dictyota 12.9 10.9 Laurencia 9.2 8.2 Polysiphonia 35.6 21.0 Ulva 0.0 12.0 G. yasumotoi Acanthophora 0.7 1.6 Caulerpa 0.6 0.0 Dasya 2.6 9.3 Derbesia 12.3 11.1 Dictyota 3.8 3.1 Laurencia 2.9 1.8 Polysiphonia 3.9 9.9 Ulva 9.5 5.3

CLUSTER groupings

Fourteen groups were defined in the CLUSTER/SIMPER analysis based on growth and attachment values of the 90 treatments (Table 3-3; Figure 3-13).

Group A contains only 2 treatments, controls for G. carolineanus (ambient) and G. belizeanus (enriched), and the group is characterized by the highest growth (0.14 div/day,

Table 3-4). Group B consists of those treatments with the highest attachment percentages

(an average of 35%), and includes 12 treatments, all with either Polysiphonia, Derbesia, or Dasya as the host algae.

Table 3-3. Average growth rate and cell attachment data for cluster groups, with a list of which group each of the ambient and enriched treatments belong to.

Average Average Growth Ambient treatments Enriched treatments Group attachment rate included in this group included in this group (%) (div/day) A 0.14 0.0 G. carolineanus control G. belizeanus control

G. belizeanus Polysiphonia G. caribaeus Polysiphonia G. belizeanus Derbesia G. carolineanus Dasya G. caribaeus Dasya B 0.09 37.0 G. carolineanus Derbesia G. caribaeus Derbesia G. carolineanus Polysiphonia G. carolineanus Derbesia G. carpenteri Derbesia G. carpenteri Derbesia G. carpenteri Polysiphonia

G. yasumotoi Dictyota C -0.03 3.4 G. yasumotoi Laurencia

G. yasumotoi Acanthophora G. carpenteri Ulva G. yasumotoi Caulerpa D 0.00 0.5 G. yasumotoi Caulerpa G. yasumotoi control

G. carolineanus Laurencia G. yasumotoi Derbesia E -0.01 9.7 G. yasumotoi Derbesia G. yasumotoi Ulva

G. belizeanus Dictyota F 0.13 8.2 G. carpenteri Dasya G. caribaeus Caulerpa G. carolineanus control

G. belizeanus Caulerpa G. belizeanus control G. caribaeus Acanthophora G 0.13 5.7 G. carpenteri Laurencia G. yasumotoi Acanthophora

G. belizeanus Dasya G. belizeanus Laurencia G. belizeanus Dasya G. belizeanus Polysiphonia G. belizeanus Derbesia G. belizeanus Ulva G. caribaeus Derbesia G. caribaeus Laurencia H 0.08 18.4 G. caribaeus Laurencia G. caribaeus Polysiphonia G. caribaeus Ulva G. caribaeus Ulva G. carolineanus Ulva G. carolineanus Dasya G. carpenteri Dictyota G. carolineanus Dictyota G. carolineanus Polysiphonia G. carolineanus Ulva

G. belizeanus Acanthophora G. belizeanus Caulerpa G. caribaeus control G. belizeanus Dictyota G. caribaeus Dictyota G. caribaeus Dasya I 0.11 14.0 G. carolineanus Caulerpa G. caribaeus Dictyota G. carpenteri Dasya G. carpenteri Dictyota G. carpenteri Polysiphonia

G. belizeanus Acanthophora G. belizeanus Laurencia G. carpenteri Laurencia J 0.11 7.5 G. carolineanus Acanthophora G. carolineanus Caulerpa

G. caribaeus Acanthophora G. caribaeus control G. carpenteri Acanthophora G. carolineanus Acanthophora K 0.11 2.5 G. yasumotoi control G. yasumotoi Dasya G. yasumotoi Polysiphonia

G. yasumotoi Dictyota

L 0.05 2.4 G. yasumotoi Laurencia

M 0.07 1.9 G. carpenteri control G. carpenteri Acanthophora

G. carolineanus Laurencia G. belizeanus Ulva G. carpenteri Caulerpa G. caribaeus Caulerpa G. carpenteri control G. carolineanus Dictyota N 0.06 7.8 G. carpenteri Ulva G. carpenteri Caulerpa G. yasumotoi Dasya G. yasumotoi Ulva G. yasumotoi Polysiphonia

Table 3-4. Cluster groupings ranked from highest values to lowest values, for both growth rate and average attachment.

Growth rate Average (div/day) attachment (%) Group A (0.14) Group B (37) Group F (0.13) Group H (18.4) Group G (0.13) Group I (14) Group I (0.11) Group E (9.7) Group J (0.11) Group F (8.2) Group K (0.11) Group N (7.8) Group B (0.09) Group J (7.5) Group H (0.08) Group G (5.7) Group M (0.07) Group C (3.4) Group N (0.06) Group K (2.5) Group L (0.05) Group L (2.4) Group D (0.0) Group M (1.9) Group E (-0.01) Group D (0.5) Group C (-0.03) Group A (0)

Groups C and D, along with groups M and N, fell within the bottom 6 groups (groups highlighted in red, Table 3-4) for both growth and attachment. These four groups consist of only G. yasumotoi and G. carpenteri treatments.

The only group that ranked close to the top for both growth and attachment was Group I

(growth rate 0.11 div/day, 14% attachment, highlighted in blue, Table 3-4), although the average attachment value was only half of that of the highest ranking group (Group A,

37%). This group consisted of only 3 ambient treatments, and 8 enriched treatments, and all species of Gambierdiscus were represented by this group, except for G. yasumotoi.

A B C D E F G H I J K L M N

Figure 3-13. Dendrogram illustrating the hierarchical clustering of samples, grouped by similarity of growth and attachment values.

Figure 3-14. Differences in growth rate values (div/day) for cluster groups.

Figure 3-15. Differences in average attachment (%) for cluster groups.

The MDS plots provided a good representation of the groupings exhibiting a low stress level of 0.03. When growth (Figure 3-14) and average attachment (Figure 3-15) group averages are overlain, gradients in these variables are evident across the groupings, further demonstrating that there are no dramatic differences among the groups, but rather the treatments exhibit a gradational pattern.

DISCUSSION

It was expected that the cell abundances and the percentage of attached cells would be higher in the enriched treatments, as those hosts would be healthier. However, this was not observed as an overall trend in the experiment (Figure 3-12). Although some of the hosts did show increased growth in the enriched treatments, this was observed for some ambient treatments as well. The majority of the treatments did not show an overall difference between either growth or attachment among the different hosts.

Not surprisingly, the group that exhibited the highest percentages for cell attachment

(Group B, which included 12 total treatments) consisted entirely of the 3 filamentous hosts. Among the different treatments of those 3 host species, there was a high variation of ambient versus enriched conditions having higher attachment, and often there was no significance difference between the two for an individual treatment. No G. yasumotoi treatment was included in this group of attachment behavior, further demonstrating its distinct differences the others in the genus. .

Gambierdiscus belizeanus showed significantly higher growth in the enriched controls compared to the ambient, indicating that for this species optimum growth occurs in

87 nutrient levels higher than those of ambient Florida Keys water (Figure 3-2). Being the most toxic in the genus for the Caribbean region, this could have implications for areas experiencing nutrient enrichment.

Gambierdiscus caribaeus, G. carolinianus, G. carpenteri, and G. yasumotoi all showed much less variation in growth within the control treatments between ambient and enriched conditions, possibly indicating adaptation to oligotrophic environments (Figures

3-3 through 3-6). Gambierdiscus yasumotoi, however, showed significantly higher cell abundances under ambient conditions for four treatments (including control), and no higher abundances in any treatments for enriched conditions (Figure 3-6). Growth was very low in all of the enriched conditions for this species and for all but three ambient treatments (Acanthophora, Polysiphonia, and Dasya). This may indicate that the optimum growth conditions for this species were not present in the experimental conditions, but possibly there was something else stimulating in those three treatments that allowed for growth, such as nutrients being released from the algae or growth- stimulating chemicals being produced by the algae.

In four out of five of the Gambierdiscus species, Polysiphonia showed higher attachment by cells under the ambient conditions than the enriched conditions (Figure 3-12). This was also the case for Derbesia in three of the five species. These two species showed the highest attachment overall for the experiment as well. The other host species showed no differences between ambient and enriched, or more attachment under enriched conditions

(Figure 3-12).

At the start of the experiment (Day 1), there was not any more overall attachment in the enriched treatments compared to the ambient treatments, indicating that when first placed in the treatment wells with their host algae, the Gambierdiscus cells were not more inclined to attach to a host that was of more nutritional value (Figures 3-7 through 3-11).

However, in almost all treatments, the enriched algae visually appeared to be healthier

(darker, thicker branches, etc) than their equivalent ambient treatment. In addition to there being no difference in attachment percentages, there was also no difference in immediate growth of the cells. While there were significant differences in cell abundance by the end of the experiment between enriched and ambient, the growth patterns between the two were not significantly different for most treatments for Day1 through Day 15

(Figures 3-2 through 3-6), again indicating that the nutritional value of the host had no initial effect on either the growth or the attachment behavior.

The highest growth rates occurred for G. belizeanus and G. carolinianus. Although these experiments were not designed to illustrate the differences in growth rates within the genus, the controls do demonstrate which species of Gambierdiscus showed the highest growth without any influence by host algae. Under ambient conditions, G. carolinianus had the highest growth rate, but under the enriched nutrient conditions, G. belizeanus did.

Under both conditions, these two species grew the fastest, compared to the other three species. The only other work that has compared the growth rates within the genus found

G. caribaeus to grow the fastest (Kibler et al., 2012).

Previous work has shown that varying nutrient levels have an effect on the growth of

Gambierdiscus populations (Lartigue et al., 2010). However, the results presented here indicate that there is no strong preference by Gambierdiscus for a more or less nutrient-

89 enriched host. There were no Gambierdiscus species that showed a consistent increase in growth across all hosts for either of the nutrient conditions. This may be because the cells are simply attaching to the host and not dependent on them for any nutritional uptake or that the different hosts were taking up the nutrients at different rates, not allowing for any consistent trends across all host treatments. However, it was expected that different conditions and nutritional value of the hosts would lead to different nutrient levels being released by the host and that variations in growth and attachment would be apparent. In addition, enriched conditions were receiving higher nutrients every 2-3 days during the water changes than what the ambient treatments were receiving. It is possible that the hosts were taking up those nutrients faster than Gambierdiscus.

Although there were no differences within the genus in ambient versus enriched, there was an overall trend observed. For the majority of enriched treatments, the percentage of attached cells declined over the course of the experiment (Figure 3-7 through 3-11). Cell attachment also declined in the ambient conditions in most host treatments, although to a lesser extent. Over the course of the experiment, most algae species lost color. This would indicate that they were becoming unhealthier over the course of the 29 days, but no nutritional data on the host algae was recorded. If the algae were in fact declining in health, there was an impact on the Gambierdiscus, as shown by their declining attachment behavior.

CHAPTER 4:

OVERALL DISCUSSION

Certain macroalgal host species stimulate growth of Gambierdiscus while others seem to inhibit growth, with high variability within the genus of Gambierdiscus. Epiphytic behavior (physical attachment) was less variable between the host species, as certain hosts consistently promoted more attachment by Gambierdiscus than others. If macroalgae are supporting, even stimulating, the growth of Gambierdiscus, this may be an important component in the flux of ciguatoxins into a coral reef food web that eventually leads to an outbreak of ciguatera. Influences by macroalgae host are perhaps a major factor that has thus-far been overlooked in ciguatera research.

SIGNIFICANT CONCLUSIONS

Growth rates of Gambierdiscus in response to different host algae vary within the genus.

Previous research has demonstrated that different species within the Gambierdiscus genus grow differently under varying environmental conditions. It is now clear that there is also a large variability in growth rates within the genus when exposed to different host algae species.

Previous experiments have suggested similar results. Although methods differed, including using experimental water previously incubated with macroalgae (Grzebyk et al., 1994), using macroalgal extracts in water (Carlson and Tindall 1985), and using methods similar to the ones in this study (Parsons et al., 2011), all three report similar results; different algae have significantly different effects on the growth of

Gambierdiscus, some stimulating growth and some inhibiting growth, with no real

92 consistency among the species as to which stimulate and which inhibit, likely because each of the above experiments were using different methodologies, host species, and species/strains of Gambierdiscus.

The results of these studies alongside the current ones presented here all suggest that there is a strong impact of the reef algal community on the Gambierdiscus populations on that reef. Further, if some species of Gambierdiscus are more toxic than others as has been suggested (Bomber et al., 1989; Holmes et al., 1991; Legrand et al., 1990;

Babinchak et al., 1996; Chinain et al., 2010), then there are implications for how host algae can influence the influx of toxins into the food web.

Epiphytic behavior of Gambierdiscus in response to different host algae varies within the genus.

Although the physical (e.g., surface area, texture, morphology), chemical (e.g., production of metabolites, release/uptake of available nutrients), and ecological (e.g., epiphytic community, seasonal fluctuations) characteristics that influence the abundance of Gambierdiscus populations on any given macroalgal individual are still widely unknown, it has been demonstrated here that Gambierdiscus do in fact, prefer to attach to certain macroalgal species, but avoid attachment to others. It is also important to note that there is much inconsistency within the genus regarding these preferences (i.e., preference and avoidance appears to be Gambierdiscus species (or strain)-specific).

While some previous work (Besada et al., 1982; Bomber et al., 1988; Bomber et al.,

1989; Nakahara et al., 1996; Parsons et al., 2011) has focused on the relationship and behavior of Gambierdiscus with host macroalgae, this is the first to experiment with more

93 than one described species within the genus, and to report varying attachment behaviors among those species.

Some Gambierdiscus are more epiphytic than others.

Throughout the experiment, there were large percentages of unattached cells for all species, and among all hosts. This supports the hypothesis previously suggested by Parsons and Preskitt (2007) that Gambierdiscus are not obligate epiphytes. It has been suggested (Nakahara et al., 1996) that cells normally swim around the host, but quickly attach to the surface of the algae when a sudden disturbance or strong water motion occurs. In their wild habitat, Gambierdiscus are certainly exposed to more extreme conditions and disturbances, and it can be presumed that a much higher percentage of cells within a wild population may attach to host algae, particularly during periods of strong wave action or tidal flow. Although the cells did experience minor disturbance during the transfer of the well plates from the incubator to the microscope, this was a minor disturbance in comparison. The cells also had the option to attach to the plastic edges of the well, which may have influenced the attachment behavior to hosts.

The conditions that contribute to physical attachment to host algae need further investigation. This is an important aspect when considering the amount of toxins that are moving into the food web. Gambierdiscus that are physically attached to the host algae will be more likely to be consumed by grazing herbivores than Gambierdiscus that are instead swimming near the algae, although it is likely the swimming ones may be grazed upon as well.

Gambierdiscus yasumotoi exhibited significantly less epiphytic behavior than the other

Gambierdiscus species, and this (along with the different swimming behaviors observed during microscopy), support Gomez et al. (2015) in their suggestion that the species be moved to different genus.

Environmental conditions created by macroalgae are influencing Gambierdiscus, even when they are not attaching to the algae.

In many cases, the hosts that showed the most Gambierdiscus cell growth also showed a low rate of attachment by the cells. Gambierdiscus were benefitting (as shown by higher growth when compared to the control) from the host algae, even without being physically connected to it. Host microenvironment is an important factor in Gambierdiscus growth.

The microenvironment is likely influenced by organisms Chapter other than the macroalgae itself, including other epiphytes, and possibly even the surrounding

Gambierdiscus themselves (Sakamoto et al., 1996).

Nakahara et al. (1996) suggested that differences in the epiphytic behavior of

Gambierdiscus exposed to xenic versus axenic samples of host algae was due to stimulatory exudates produced by the microbial community on the xenic samples

(Gambierdiscus would attach to xenic hosts, but not axenic hosts). Presumably this would have been occurring in all treatments of this experiment, under both ambient and enriched conditions, with possible elevated effects in host treatments which were experimenting tissue degradation and higher levels of bacterial activity. It has also been suggested that

Gambierdiscus may be capable of utilizing organic materials originating from bacteria as

95 a nutrient source and/or that organic materials produced by bacteria might function as metal chelators and stimulate Gambierdiscus growth (Sakami et al., 1999). In addition to bacteria, the epiphytic community on any given substrate may consist of many other organisms, including other dinoflagellate species, fungi, cyanobacteria, diatoms, and other protists; the influences of this community, and of the macroalgal host itself, on

Gambierdiscus are largely unstudied (Yasumoto et al., 1980). Within the limitations of this experimental design, there was no way to account for the influence of all epiphytic micro-organisms (especially microbial), but steps were taken to remove as many other influencing epiphytes from the algae samples as possible. However, the xenic conditions of the treatments were likely influenced by the presence of other epiphytes, including bacteria. The previous research mentioned here suggests that both the attachment behaviors and the growth rates may have been influenced. However, the influencing effect of the epiphytic community versus other conditions within the treatments is unknown. It is likely though, that within same-host treatments, there would have been a similar community structure of epiphytes, as the host algae fragments all came from the same larger piece of algae, growing in the same tank for weeks prior to the start of the experiment.

There was no significant difference in growth or cell attachment under ambient versus enriched conditions.

While some Gambierdiscus species grew better under the enriched control environments, others grew better in the ambient control conditions, but overall, there was no consistent trend where one treatment resulted in better growth than the other. Cell attachment also

96 did not differ. Although some of the host algae did show an increase in biomass in the enriched conditions, there was no resulting effect on the Gambierdiscus. It is possible that the macroalgae were taking up the excess nutrients faster than the Gambierdiscus were able to, or possibly the macroalgae were growing under the enriched conditions and releasing growth-inhibiting chemicals, but more work needs to be done to better test these possibilities. The changing chemistry of the experimental environment was largely unknown and there were likely factors attributing to growth and inhibition occurring that were not discussed here. In addition, there were likely chemical conditions occurring within the small, closed, experimental environment that may not occur in an actual ocean environment, although the frequent water exchanges were conducted to reduce this effect.

Previous research exploring the effects of nutrients has found no correlation between nutrients and Gambierdiscus cell densities (Yasumoto et al., 1980; Ichinotsubo et al.,

1994; Parsons and Preskitt 2007; as reviewed by Parsons et al., 2012), which corroborates the conclusions here; the influence of nutrients on Gambierdiscus growth and behavior is complex and unclear.

Algal phylum does not play a role in the host preference by Gambierdiscus.

As suggested by previous research (Taylor and Gustavson 1983; Lobel et al., 1988; St.

Martin et al., 1988; Parsons and Preskitt 2007), there was no significant preference for cell attachment to phaeophytes, chlorophytes, or rhodophytes by the Gambierdiscus species tested in this experiment. However, with the new knowledge that the preference of host algae is very species-specific depending on the Gambierdiscus species, no generalizations should be made overall for the genus.

The highest levels of attachment occurred in the hosts that were the most filamentous in structure.

This has been suggested before (Taylor and Gustavson 1983; Parsons and Preskitt 2007) and is further supported by these results. The host algae species with the thinner thalli

(Polysiphonia, Dasya, and Derbesia; Table 2-5) supported the most epiphytic behavior

(physical attachment by the cells) throughout the experiments by all but one of the

Gambierdiscus species. Although no turf algae species were used in this experiment, turf algae generally tend to have a small, filamentous structure to them, and have been known to harbor populations of Gambierdiscus. Reefs experiencing algal cover of a more filamentous nature may have more Gambierdiscus being consumed, if there are high numbers of them attached to algae species that being grazed upon.

IMPLICATIONS FOR CIGUATERA ECOLOGY

The different preferences and/or growth responses exhibited within the Gambierdiscus genus adds complexity to toxin flux scenarios; e.g., host palatability, different host preferences and epiphytic behavior among the Gambierdiscus species, different levels of toxin production among the Gambierdiscus species.

Chemical ecology of host algae

Macroalgae will produce harmful metabolites as a method of defense against grazing

(Hay et al., 1987; Paul and Fenical, 1987). Cruz-Rivera and Villareal (2006) give a

98 review of macroalgae and herbivorous fish interactions, pointing out the importance of chemical defenses in deterring grazers. What is less understood, however, is the potential influence of these chemical deterrents on the epiphytic community. As previously discussed, the eight species of host macroalgae used in this study have varying degrees of chemical production. A lack of information prevents any attempts to examine potential relationships between metabolite production and growth/attachment for the species studied here. However, the results presented here do indicate that the chemical ecology of macroalgal hosts is playing a role, by establishing that some macroalgae may produce ideal conditions for Gambierdiscus growth, even if it is not consistent throughout the genus. While some macroalgae stimulated Gambierdiscus growth in these experiments, other macroalgae inhibited growth, and the control treatments provided a standard for comparison. Although microbial community and the nutrient conditions being controlled by the host algae were also potentially influencing, the variation between the hosts suggests that the chemistry of the experimental conditions was being influenced in some way by the individual algae species.

Therefore just as important as the macroalgae community may be on the growth of

Gambierdiscus, it could play an equally important role in controlling and limiting populations.

Toxin production

In the Caribbean, among the species of Gambierdiscus identified thus far, G. belizeanus has been shown to be the most toxic (Chinain et al., 2010). G. belizeanus is influenced by host macroalgae. As the genus does in general, G. belizeanus shows preferential

99 attachment to highly filamentous species of algae (Polysiphonia, Dasya, and Derbesia).

Growth conditions, however, were most optimal for G. belizeanus when exposed to

Caulerpa, but attachment to that host was not high (<10%).The potential toxin production of G. caribaeus, G. carolinianus, G. carpenteri, and G. yasumotoi are less clear.

Host palatability and grazing

The data from these experiments strongly suggest that macroalgae are an important aspect of ciguatera ecology that we must better understand if there is any hope to monitor and predict the risk of ciguatera on reefs. There are three significant findings within this work that suggest a stronger focus on the study of host algae and their palatability (i.e., role in flux of toxins into the food web).

First, there are Gambierdiscus species that grew better under host conditions than they did in their control treatments. This suggests that some unknown condition that the host was creating was stimulating growth. When Gambierdiscus populations inhabit coral reefs or other environments that are dominated by macroalgae that are capable of stimulating their growth, this is a strong factor in the largely unknown process of

Gambierdiscus blooms on reefs. Further, if that same algal species is subject to high levels of grazing, there will be increased levels of toxins entering the food web. It has been reported that just a small group of herbivorous fishes can account for the majority of grazing on a reef (i.e., only 4 fish species accounting for 97% of macroalgae consumption, Rasher et al., 2013). This research implies a very fixed relationship between algae and grazers on coral reefs, suggesting that it may be plausible to

100 distinguish direct routes of trophic transfer between specific macroalgae and specific herbivorous fishes, thereby establishing toxin flux into the food web. The effects of increased macroalgal cover on coral reefs needs a better understanding, given the number of coral reefs globally that are currently undergoing shifts resulting in these conditions.

Secondly, some species of macroalgae, particularly those most filamentous in structure, are likely to be harboring higher populations of Gambierdiscus. Generally, the lack of calcareous structure and their thinner thalli will make these species more palatable to herbivorous fishes. It is known that some algae species will counteract the effects of grazing by rapid growth, particularly turf algae (Carpenter 1986). Rapid consumption followed by rapid growth and more rapid consumption and so on, of a turf algae that is harboring toxin-producing Gambierdiscus, is going to increase toxin transfer into the food web. However, if consumption of the algae is too fast, then there may be no chance for cell abundances to reach high numbers. In the southern Cook Islands, a period of increased disturbances over 14 years (cyclones, crown-of thorns starfish outbreak, and coral bleaching events – which all resulted in increased turf algae cover) corresponded with the highest number of CFP reports for that area. Years later when those disturbances slowed, turf algae decreased, and CFP cases declined (Rongo and van Woesik 2013).

Polysiphonia, the very palatable red turf algae, has been shown to be an early successional species (Agatsuma et al., 1997) and can grow quickly (Campbell, 2001; as reviewed in Cruz-Rivera and Villareal, 2006). If filamentous algae, such as these red turf algae, are be harboring cell abundances under certain conditions, and those algae are highly palatable, this may be an important scenario to consider when identifying high

101 fluxes of toxins into the food web. It is also important to consider as many reefs in the

Caribbean continue to lose coral cover and algae dominance increases.

Lastly, Gambierdiscus cells were more inclined to attach to the hosts in the beginning of the experiment, before the hosts began losing their dark colors, particularly the enriched algae, suggesting thicker, fuller, more robust macroalgae encourage Gambierdiscus attachment. No studies have considered the health or nutritional value of host macroalgae when reporting on cell counts of Gambierdiscus during field collections.

In conclusion of these findings, a reef could potentially be at risk for ciguatera under conditions that include a) host macroalgae with a microenvironment that is enhancing

Gambierdiscus growth; b) hosts that are filamentous, highly palatable, macroalgae or turf algae and therefore promoting high levels of attachment; c) Gambierdiscus populations that are producing toxins; and d) herbivorous grazers consuming those algae that are hosting the toxic cells in high numbers. However, these are just the factors influencing the positive potential for ciguatera. The factors influencing the control and limitations of ciguatera are just as important to consider and are widely unknown.

FUTURE RESEARCH

With each new variation in behavior and/or growth that is found within the

Gambierdiscus genus, ciguatera research becomes more challenging. The missing links in understanding the dynamics influencing toxin production and the path of toxin flux is exemplified each time a new species is discovered or variability in behavior within the genus is revealed. As mentioned previously, it is important to state that these results and

102 the differences reported may represent strain level differences, meaning differences potentially exist within each single species, as well as within the genus; future studies should further investigate this possibility.

Although the results presented here demonstrate new findings and contribute to a better understanding of the role of macroalgae host, they also raise even more new questions that now need answered. The following are a collection of topics/questions that need further research by the ciguatera community in order to understand the role that microalgae hosts contribute in the overall understanding of ciguatera on coral reefs.

Host preference studies

The experimental design of this study only provided one host within each treatment for

Gambierdiscus to either attach to or not attach to. This does not mimic an ocean environment in which dinoflagellates have various options of choosing a „home‟ among differing algae within a small area. When exposed to more than one species, the

Gamberidiscus may be drawn to the chemical or physical aspects of a certain algae over another. This study only observed their behavior under one choice of host. When placed into an environment with more than one choice of hosts, what preferences will

Gambierdiscus have? Further, what is controlling those preferences?

103

Nutrient enrichment

It was expected that the enriched treatments would provide optimal conditions for growth and attachment by Gamberidiscus but that was not the case. This may have been in part due to the design, and enough attention was not placed on monitoring the chemical condition of the experimental water. Further studies with a focus on how the nutritional value of host algae influence Gambierdiscus are necessary for a further understanding of how nutrient enrichment on reefs might play a role in Gambierdiscus population growth.

While previous studies have investigated the direct effects of nutrients on the growth of

Gambierdiscus, none have reported on the effects of an enriched macroalgal hosts versus a nutritionally deprived one.

Toxicity studies

The factors involved in Gamberidiscus toxicity levels are widely understudied. We do not know what environmental factors on a reef lead to increased production of toxins.

Because epiphytes are highly influenced by the microenvironment that they are exposed to, it is probable that there may be conditions within that microenvironment that stimulate toxin production (e.g., metabolites released by host, nutrient uptake/release by host, chemical products of neighboring epiphytes).

IMPLICATIONS FOR CIGUATERA MANAGEMENT

It has been projected that CFP in the United States is likely to increase by 200-400% in this century (Gingold et al., 2014). Currently, there is no method to predict when and

104 where CFP is going to become prevalent. There is no way for local communities to know which reefs are unsafe for fishing, until people get sick. The ecology of ciguatera becomes more and more understood as more data becomes available on the toxin production within the genus, the growth variability within the genus, the discovery of new strains and/or species, the ongoing input into the range of Gambierdiscus, and many other topics. The results presented here strongly suggest that the ecology, the growth dynamics, and the chemistry of host macroalgae all deserve further investigation if we are to fully understand the dynamics of ciguatera on coral reefs.

105

REFERENCES

106

Agatsuma, Y., K. Matsuyama, A. Nakata, T. Kawai, and N. Nishikawa. 1997. Marine algal succession on coralline flats after removal of sea urchins in Suttsu Bay on Japan Sea coast of Hokkaido, Japan. Nippon Suisan Gakkaishi 63: 672-680.

Arthur, K.E., K.M. McMahon, C.J.Limpus, and W.C. Dennison. 2009. Feeding ecology of green turtles (Cheloniamydas) from Shoalwater Bay, Australia. Marine Turtle Newsletter 123:6–12.

Asuncion, D.A, A.Y. Asahina, P. Muthiah, Y. Hokama, T. Higa, and J. Tanaka. 1995. The effects of monosaccharides and natural marine products on the growth of Gambierdiscus toxicus in vitro. In: Hokama, Y., P.J. Scheuer, T. and Yasumoto (Eds.), Proceedings of the International Symposium on Ciguatera and Marine Natural Products, Honolulu, Hawaii, 1994. Asain-Pacific Research Foundataion, Honolulu, Hawaii. pp 99-107.

Atsatt, P.R. and D.J. O‟Dawd. 1976. Plant defense guilds. Science 193: 24-29.

Babinchak, J.A., D.J. Jollow, M.S. Voegtline, and T.B. Higerd. 1986. Toxin production by Gambierdiscus toxicus isolated from the Florida Keys. Marine Fisheries Review 48(4):53- 56.

Babinchak, J.A., G.J. Doucette, R.M. Ball. 1996. Partial characterization of the LSUrRNA gene from the ciguatoxic dinoflagellate, Gambierdiscus toxicus. In: Yasumoto,T., Y. Oshima, and Y. Fukuyo (Eds.), Harmful and Toxic Algal Blooms. Intergovernmental Oceanographic Commission of UNESCO, pp. 459– 462.

Bagnis, R. 1994. Natural versus anthropogenic disturbances to coral reefs – comparison in epidemiological patterns of ciguatera. Memoirs of the Queensland Museum 34:455-460.

Bagnis, R., S. Chanteau, E. Chungue, J.M. Hurtel, T. Yasumoto, T. and A. Inoue. 1980. Origins of ciguatera fish poisoning: a new dinoflagellate, Gambierdiscus toxicus Adachi and Fukuyo, definitively involved as a casual agent. Toxicon 18(2): 199- 208.

Ballantine, D.L., A.T. Bardales, and T.E. Tosteson. 1985. Seasonal abundance of Gambierdiscus toxicus and Ostreopsis sp. in coastal waters of southwest Puerto Rico. In: Delesalle, B., R. Galzon, and B, Salvat (Eds.), Proceedings of the Fifth International Coral Reef Congress, Tahiti, Antenna Museum-EPHE, Moorea (French Polynesia), pp 417-422.

107

Ballantine, D.L., T.E. Tosteson, and A.T. Bardales. 1988. Population dynamics and toxicity of natural populations of benthic dinoflagellates in southwestern Puerto Rico. Journal of Experimental Marine Biology and Ecology 119: 201-212.

Banner, A.H. 1976.Ciguatera: a disease from coral reef fish. In: Jones, A.O., Endean, R. (Eds.), Biology and Geology of Coral Reefs 3: 177-213.

Bergmann, J.S. and M. Alam. 1981. On the toxicity of the ciguatera producing dinoflagellate, Gambierdiscus toxicus, isolated from the Florida Keys, USA. Journal of Environmental Science and Health, Part A: Environmental Science and Engineering 16(5): 493-500.

Besada, E.G., L.A. Loeblich, and A.R. Loeblich III. 1982. Observations on tropical, benthic dinoflagellates from ciguatera-endemic areas: Coolia, Gambierdiscus, and Ostreopsis. Bulletin of Marine Science 32: 723–725.

Bolser, R.C. and M.E. Hay. 1996. Are tropical plants better defended? Palatability and defenses of temperate versus tropical seaweeds. Ecology 77:2269-2286.

Bomber, J.W., 1985. Ecological studies of benthic dinoflagellates associated with ciguatera from the Florida Keys. M.S. Thesis. Department of Oceanography and Ocean Engineering, Florida Institute of Technology.104 pp. Bomber, J.W., 1987.Ecology, genetic variability and physiology of the ciguatera-causing dinoflagellate Gambierdiscus toxicus Adachi & Fukuyo. Ph.D. Dissertation, Department of Oceanography and Ocean Engineering, Florida Institute of Technology, 147 pp.

Bomber, J.W., R.L. Guillard, and W.G. Nelson. 1988. Roles of temperature, salinity, and light in seasonality, growth, and toxicity of ciguatera-causing Gambierdiscus toxicus Adachi Et Fukuyo (Dinophyceae). Journal of Experimental Marine Biology and Ecology 115:53-65.

Bomber, J.W., M.G. Rubio, and D.R. Norris. 1989. Epiphytism of dinoflagellates associated with the disease ciguatera: substrate specificity and nutrition. Phycologia28:360–368.

Bomber, J.W., and K.E. Aikman. 1989. The ciguatera dinoflagellates. Biological Oceanography 6: 291-311.

Bruno, J.F., H. Sweatman, W.F. Precht, E.R. Selig, and V.G.W. Schutte. 2009. Assessing evidence of phase shifts from coral to macroalgal dominance on coral reefs. Ecology 90:1478-1484.

108

Campbell, S. 2001. Ammonium requirements of fast-growing ephemeral macroalgae in a nutrient-enriched marine embayment (Port Phillip Bay, Australia). Marine Ecology Progress Series 209:99-107.

Carlson, R.D., and D.R. Tindall.1985. Distribution and periodicity of toxic dinoflagellates in the Virgin Islands. In: Anderson D.M., A.W. White, and D.G. Baden (Eds.), Toxic Dinoflagellates. New York: Elsevier Science Publishing Co., Inc. pp. 171–177.

Carlson, R.D., G. Morey-Gaines, D.R. Tindall, and R.W. Dickey. 1984. Ecology of toxic dinoflagellates from the Caribbean Sea: effects of macroalgal extracts on growth in culture. In: Ragelis, E.P. (Ed.), Seafood Toxins. American Chemical Society, Washington, D.C. pp. 271-287.

Carpenter, R.C. 1986. Partitioning herbivory and its effects on coral reef algal communities. Ecological Monographs 56(4): 345-363.

Chan, A., K. Lubarsky, K. Judy, and P. Fong. 2012. Nutrient addition increases consumption rates of tropical algae with different initial palatabilities. Marine Ecology Progress Series: 465:25-31.

Chiaponne, M. and K.M. Sullivan. Rapid assessment of reefs in the Florida Keys: Results form a synoptic survey. Proceedings of the 8th International Coral Reef Symposium 2:1509-1514.

Chinain, M, M. Germain, X. Deparis, S. Pauillac, and A.M. Legrand. 1999. Seasonal abundance and toxicity of the dinoflagellate Gambierdiscus spp. (Dinophyceae), the causative agent of ciguatera in Tahiti, French Polynesia. Marine Biology, 135: 259-267.

Chinain, M., H.T. Darius, A. Ung, P. Cruchet, Z. Wang, D. Ponton, D. Laurent, and S. Pauillac. 2010. Growth and toxin production in the ciguatera-causing dinoflagellate Gambierdiscus polynesiensis (Dinophyceae) in culture. Toxicon 56: 739-750.

Cook, J. 1777. A Voyage Towards the South Pole and Around the World, 3rd edition. Cook, J., T. Furneaux, and W.Hodos (Eds.),printed for W. Strahan and T. Cadell, London, pp 112-113.

Cooper, M.J. 1964. Ciguatera and other marine poisoning in the Gilbert Islands. Pacific Science 18: 411-440.

109

Cronin, G., N. Lindquist, M. Hay, and W. Fenical. 1995. Effects of storage and extraction procedures on yields of lipophilic metabolites from the brown seaweeds Dictyota ciliolate and D. menstrualis. Marine Ecology Progress Series 119: 265-273.

Croley, F. and C. Dawes. 1970. Ecology of the algae of a Florida Key. I. A preliminary checklist, zonation, and seasonality. Bulletin of Marine Science 20:165-185.

Cruz-Rivera, E. and T. A. Villareal. 2006. Macroalgal palatability and the flux of ciguatera toxins through marine food webs. Harmful Algae 5:497-525.

Dawes, C.J. and A.C. Mathieson.2008. The seaweeds of Florida. pp. [i]- viii, [1]-591, [592], pls I-LI. Gainesville, Florida: University Press of Florida.

Dechraoui Bottein, M.-Y., Z. Wang, J. Turquet, M. Chinain, T. Darius, P. Cruchet, F.F.Y Radwan, R.W. Dickey, and J.S. Ramsdel. 2005. Biomonitoring of ciguatoxinexposure in mice using blood collection cards. Toxicon 46: 243–251.

De Lara-Isassi, G., S. Álvarez-Hernández,and L. Collado-Vides.2000. Ichthyotoxic activity of extracts from Mexican marine macroalgae. Journal of Applied Phycology 12:45–52.

Delgado, G., C.H. Lechuga-Deveze, G. Popowski, L. Troccoli, and C.A. Salinas. 2006.Epiphytic dinoflagellates associated with ciguatera in the Northwestern coast of Cuba. Revista de Biologia Tropical 54: 299–310.

Descombes , P., M. Wisz, F. Leprieur, V. Parravicini, and C. Heine. 2015. Forecasted coral reef decline in marine biodiversity hotspots under climate change. Global Change Biology 21: 2479-2487.

De Sylva, D.P. 1994. Distribution and ecology of ciguatera fish poisoning in Florida, with emphasis on the Florida Keys. Bulletin of Marine Science 54: 944–954.

Doherty, M.J. 2005.Captain Cook on poison fish. Historical Neurology 65: 1788-1791.

Douglas B. R.,A.S. Hoey, and M.E.Hay.2013. Consumer diversity interacts with prey defenses to drive ecosystem function. Ecology 94:1347–1358

Durand-Clement M. 1986. A study of toxin production by Gambierdiscus toxicus in culture. Toxicon24:1153–1157.

Faust, M.A. 1995. Observation of sand-dwelling toxic dinoflagellates (Dinophyceae) from widely differing sites, including two new species. Journal of Phycology 31:996-1003.

110

Fleming, L.E., D.G. Baden, J.A. Bean, R. Weisman, R., and D.G. Blythe. 1998. Seafood toxin diseases: issues in epidemiology and community outreach. In: Reguera, B., Blanco, J., Fernandez, M.L., Wyatt, T. (Eds.), Harmful Algae. Xunta de Galicia and Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 245– 248.

Fraga, S., R. Francisco, A. Caillaud, J. Dioge`ne, J., N. Raho, and M. Zapata. 2011. Gambierdiscus excentricus sp. nov (Dinophyceae), a benthic toxic dinoflagellate from the Canary Islands (NE Atlantic Ocean) Harmful Algae 11: 10–22.

Fraga, S. and F. Rodriquez. 2014. Genus Gambierdiscus in the Canary Islands (NE Atlantic Ocean) with Description of Gambierdiscus silvae sp. nov., a New Potentially Toxic Epiphytic Benthic Dinoflagellate. Protist, 165: 839-853.

Gardner, T.A., I.M. Côté, J.A. Gill, A. Grant, and A.R. Watkinson.2003 Long-term region-wide declines in Caribbean corals. Science 301:958−960

Grzebyk, D., B. Berland, B.A. Thomassin, C. Bosi, and A. Arnoux. 1994. Ecology of ciguateric dinoflagellates in the coral reef complex of Mayotte Island (S.W. Indian Ocean). Journal of Experimental Marine Biology and Ecology 187:51–66.

Gillespie, N.C., M.J. Holmes, J.B. Burke, and J. Doley. 1985. Distribution and periodicity of Gambierdiscus toxicusin Queensland, Australia. In: Anderson, D.M., A.W. White, and D.G. Baden (Eds.), Toxic Dinoflagellates. Elsevier, New York, pp. 183-188.

Gómez, F., D. Qiu, R.M. Lopes, and S. Lin. 2015.Fukuyoapaulensis gen. et sp. nov., a New Genus for the Globular Species of the Dinoflagellate Gambierdiscus (Dinophyceae). PLoS ONE 10(4): e0119676. doi:10.1371/journal.pone.0119676

Gingold, D.B., M.J. Strickland, and J.J. Hess. 2014. Ciguatera fish poisoning and climate change: analysis of national poison center data in the United States, 2001–2011. Environmental Health Perspectives 122: 580–586

Hales, S., P. Weinstein, and A. Woodward. 1999. Ciguatera (fish poisoning), El Nino, and Pacific sea surface temperatures. Ecosystem Health 5: 20–25

Halstead, B.W. 1978. Poisonous and Venomous Marine Animals of the World, Revised ed. Darwin Press, Princeton.

Hamilton, B., N. Whittle, G. Shaw, G. Eaglesham, G. Moore, and R. Lewis. 2010. Human fatality associated with Pacific ciguatoxin contaminated fish. Toxicon 56: 668-673.

111

Hanisak, M.D., and L.S. Overdorf. 1998. Spatial and temporal variability in the abundance of coral reef algae in the Florida Keys. Journal of Phycological Sciences.34: 52.

Hata, H.H. 2010. Geographic variation in the damselfish-red alga cultivation mutualism in the Indo-West Pacific. BMC Evolutionary Biology 10: 185.

Hay, M. E., and W. Fenical. 1988. Marine plant-herbivore interactions: The ecology of chemical defense. Annual Review of Ecology and Systematics,19:111–145.

Hay, M. E., W. Fenical, and K. Gustafson. 1987. Chemical defense against diverse coral reef herbivores. Ecology, 68:1581–1591. http://dx.doi.org/10.2307/1939850.

Hay, M. E., and P. D. Steinberg. 1992. The chemical ecology of plant-herbivore interactions in marine versus terrestrial communities. In: Herbivores: Their interaction with secondary metabolites, evolutionary and ecological processes, ed. J. Rosenthal and M. Berenbaum, pp. 371–413. San Diego: Academic

Heil, C.A., P. Bird, and W.C. Dennison. 1998. Macroalgal habitat presence of ciguatera dinoflagellates at Heron Island, a coral cay in the southeastern Great Barrier Reef, Australia. In: Reguera, B., J. Blanco, M.L. Fernandez, and T. Wyatt (Eds.), Harmful Algae. Xunta de Galicia and Intergovernmental Oceanographic Commission of UNESCO, Paris, pp. 52–53.

Helfrich, P. 1964. Fish poisoning in Hawaii. Hawaii Medical Journal 22: 361-372.

Helfrich, P. and A.H. Banner. 1963. Experimental induction of ciguatera toxicity in fish through diet. Nature 197: 1025–1026.

Helfrich, P. and A.H. Banner. 1968. Ciguatera fish poisoning. II. General patterns of development in the Pacific. Occasional Papers of the Bernice P. Bishop Museum 23: 371-382.

Hokama, Y., J.S.M. Ebesu, K. Nishimura, S. Oishi, B. Mizuo, M. Stiles, B. Sakamoto, W. Takenaka, and H. Nagai. 1996. Human intoxications from Hawaiian reef fishes associated with diverse marine toxins. Journal of Natural Toxins 5: 235– 247.

Holmes, M.J.1998. Gambierdiscus yasumotoi sp. nov. (Dinophyceae), a toxic benthic dinoflagellate from Southeastern Asia. Journal of Phycology 34:661–668.

Holmes, M.J.,R.J. Lewis, M.A. Poli, and N.C. Gillespie. 1991. Strain dependent production of ciguatoxin precursors (gambiertoxins) by Gambierdiscus toxicus (Dinophyceae) in culture. Toxicon 29: 761–775.

112

Holmes, M.J., R.J. Lewis, M. Sellin, and R. Street. 1994. The origin of ciguatera in Platypus Bay, Australia. Memoirs of the Queensland Museum 34: 505–512.

Hughes, T.P, B.D. Keller, J.B.C. Jackson, M.J. Boyle. 1987. Herbivory on coral reefs: community structure following mass mortality of sea urchins. Journal of Experimental Marine Biology and Ecology 113, 39-59.

Hughes, T.P. 1994. Catastrophes, phase-shifts, and large-scale degradation of a Caribbean coral-reef. Science 265, 1547–1551.

Ichinotsubo, D., A.Y. Asahina, E. Titus, S. Chun, T.W.P. Hong, J.L. Shirai, and Y. Hokama.1994. Survey for ciguatera fish poisoning in west Hawaii. Memoirs of the Queensland Museum34, 513–522.

Jackson, J.B.C., M. Donovan, K. Cramer, and V. Lam. 2014.Status and trends of Caribbean coral reefs 1970–2012. Global Coral Reef Monitoring Network, International Union for the Conservation of Nature Global Marine and Polar Program, Washington, DC

Johnson, R. and E. Jong. 1983. Ciguatera: Caribbean and Indo-Pacific fish poisoning. Western Journal of Medicine 138: 872-874.

Jones, R.S.1968.Ecological relationships in Hawaiian and Johnston Island Acanthuridae (Surgeonfishes). Micronesica 4:309-361.

Kuno, S., R. Kamikawa, S.Yoshimatsu, T.Sagara, S. Nishio, and Y. Sako. 2010. Genetic diversity of Gambierdiscus spp. (Gonyaulacales, Dinophyceae) in Japanese coastal areas. Phycological Research 58: 44–52.

Lange, W.R. 1987. Ciguatera toxicity. American Family Physician 35: 177.182.

Lartigue, J., J.E. Jester, R.W. Dickey, and T.A. Villareal. 2009. Nitrogen source effects on the growth and toxicity of two strains of the ciguatera-causing dinoflagellate Gambierdiscus toxicus. Harmful Algae 8: 781–791.

Lechat, I., F. Partenski, and E. Chungue. 1985. Gambierdiscus toxicus: culture and toxin production. In: Gabrie, C. and B. Salvat (Eds.), Proceedings of the 5th International Coral Reef Congress, Tahiti, vol. 4. Antenne Museum-Ephe, Moorea, pp.443–448.

Legrand, A,M., P. Cruchet, R. Bagnis, M. Murata, Y. Ishibashi, and T. Yasumoto. 1990. Chromatographic and spectral evidence for the presence of multiple ciguatera toxins. In: Toxic Marine Phytoplankton, Graneli, E., Anderson, D.M., Edler, L., Sundstrom, B.G. (Eds), Elsevier Science Publishing, pp 374-378

113

Lewis, R. and M. Sellin. 1992. Multiple ciguatoxins in the flesh of fish. Toxicon 30:915- 919.

Lewis, R.J. 2001.The changing face of ciguatera. Toxicon 39(8): 97-106.

Lewis, R.J. 2006. Ciguatera: Australian perspectives on a global problem. Toxicon 48: 799-809.

Littler, M., P. Taylor, and D. Littler. 1983. Algal resistance to herbivory on a Caribbean barrier reef. Coral Reefs 2:111-118.

Littler, D.S. and M.M. Littler. 2000. Caribbean Reef Plants: An Identification Guide to the Reef Plants of the Caribbean, Bahamas, Florida and Gulf of Mexico. Offshore Graphics, Inc., Washington, D.C. 542 pp.

Lirman, D. and P. Biber. 2000. Seasonal dynamics of macroalgal communities of the Northern Florida Reef Tract. Botanica Marina 43(4) 305-314.

Litaker, R.W., M.W. Vandersea, M.A. Faust, S.R. Kibler, M. Chinain, M.J. Holmes, W.C. Holland, and P.A. Tester.2009. Taxonomy of Gambierdiscus including four new species, Gambierdiscus caribaeus, Gambierdiscus carolinianus, Gambierdiscus carpenteri and Gambierdiscus ruetzleri (Gonyaulacales, Dinophyceae) Phycologia48:344–390.

Litaker, R.W., M.W. Vandersea, M.A. Faust, S.R. Kibler, A.W. Nau, W.C. Holland, M. Chinain, M.J. Holmes, and P. Tester. 2010. Global distribution of ciguatera causing dinoflagellates in the genus Gambierdiscus. Toxicon 56: 711–730.

Llewellyn, L.E., 2010. Revisiting the association between sea surface temperature and the epidemiology of fish poisoning in the South Pacific: reassessing the link between ciguatera and climate change. Toxicon 56: 691–697.

Lobel, P.S., D.M. Anderson, and M. Durand-Clement.1988. Assessment of ciguatera dinoflagellate populations: sample variability and algal substrate selection. Biological Bulletin 175:94–101.

Mackay, J. 1836. Flora hibernica. Part third. Algae.vol. 2. Dublin, pp. 279.

Maliao, R.J., R.G. Turnigan, and J. Lin. 2008.Phase-shift in coral reef communities in the Florida Keys National Marine Sanctuary (FKNMS), USA. Marine Biology 154: 841-853.

Martyr, P. 1555. De Orbe Novo. “The eight decades of Peter Martyr d‟Anghera.” Quoted in Gudger, E.W. (1930) Poisonous fishes and fishes poisonings, with special

114

reference to ciguatera in the West Indies. American Journal of Tropical Medicine 10:43-55.

McMillan, J.P., P.A. Hoffman, and H.R. Granade, 1986.Gambierdiscus toxicus from the Caribbean: a source of toxins involved in ciguatera. Marine Fisheries Review 48: 48–52.

Meyer, K.D., V.J. Paul, H.R. Sanger and S.G.Nelson.1994. Effects of seaweed extracts and secondary metabolites on feeding by the herbivorous surgeonfish Naso lituratus. Coral Reefs 13: 105-112

Morton, S.L., and M.A. Faust. 1997. Survey of toxic epiphytic dinoflagellates from the Belizean barrier reef ecosystem. Bulletin of Marine Science 61: 899–906.

Morton, S.L., D.R. Norris, and J.W. Bomber. 1992. Effect of temperature, salinity and light intensity on the growth and seasonality of toxic dinoflagellates associated with ciguatera. Journal of Experimental Marine Biology and Ecology 157: 79–90.

Morton, S.L., J.W. Bomber, D.R. Tindall, and K.E. Aikman. 1993. Response of Gambierdiscus toxicus to light: cell physiology and toxicity. In: Smayda, T.J., Shimizu, Y.(Eds.), Toxic Phytoplankton Blooms in the Sea. Elsevier, New York, pp. 541–546.

Murata, M., A.-M.Legrand, Y. Ishibashi, and T. Yasumoto. 1989. Structures of Ciguatoxin and its Congener. Journal of the American Chemical Society111: 8929–8931.

Murata, M., A.-M. Legrand, Y. Ishibashi, M. Fukui, and T. Yasumoto. 1990. Structures and configurations of ciguatoxin from the moray eel Gymnothorax javanicus and its likely precursor from the dinoflagellate Gambierdiscus toxicus. Journal of the American Chemical Society 112: 4380–4386.

Nakahara, H., T. Sakami, M. Chinain, and Y. Ishida.1996. The role of macroalgae in epiphytism of the toxic dinoflagellate Gambierdiscus toxicus (Dinophyceae).Phycological Research 44:113–117.

Nishimura, T., S. Sato, W. Tawong, H. Sakanari, H. Yamaguchi, and M. Adachi. 2014. Morphology of Gambierdiscus scabrosus sp. nov. (Gonyaulacales): a new epiphytic toxic dinoflagellate from coastal areas of Japan. Journal of Phycology, 50: 506–514

Norris, J. N. and W. Fenical. 1982. Chemical defense in tropical marine algae. Pages 417-431 in K. Rutzler and I.G. Macintyre, (Eds) .Atlantic barrier reef ecosystem,

115

Carrie Bow Cay, Belize.1.Structure and communities. Smithsonian Contributions to the Marine Sciences 12.

Norström, A. V., M. Nyström, J. Lokrantz, and C. Folke. 2009. Alternative states on coral reefs: beyond coral-macroalgal phase shifts. Marine Ecology Progress Series 376: 295-306.

Parsons, M.L .and L.B. Preskitt.2007. A survey of epiphytic dinoflagellates from the coastal waters of the island of Hawaii. Harmful Algae 6:658–669.

Parsons, M.L., Settlemier, C.J., and Bienfang, P.K., 2010.A simple model capable of simulating the population dynamics of Gambierdiscus, the benthic dinoflagellate responsible for ciguatera fish poisoning. Harmful Algae 10: 71–80.

Parsons, M.L., C.J. Settlemier, and J.M. Ballauer. 2011. An examination of the epiphytic nature of Gambierdiscus toxicus, a dinoflagellate involved in ciguatera fish poisoning. Harmful Algae 10: 598–605.

Parsons, M.L., K. Aligizaki, M.Y. Dechraoui Bottein, S. Fraga, S.L. Morton, and A. Penna. 2012. Gambierdiscus and Ostreopsis: Reassessment of the state of knowledge of their taxonomy, geography, ecophysiology, and toxicology. Harmful Algae 14: 107-129.

Paul, V. J., and W. Fenical, W. 1986. Chemical defense in tropical green algae, order Caulerpales. Marine Ecology Progress Series 34(1-2): 157-169.

Paul, V. J., and W. Fenical, W. 1987.Natural Products Chemistry and Chemical Defense in Tropical Marine Algae of the Phylum Chlorophyta. Bioorganic Marine Chemistry 1:1-29.

Paul, V. J., and M.E. Hay. 1986. Seaweed susceptibility to herbivory: chemical and morphological correlates. Marine Ecology Progress Series 33: 255-264.

Pottier, I., J.P. Vernoux, A. Jones, and R.J. Lewis. 2002. Characterization of multiple Caribbean ciguatoxins and congeners in individual specimens of horse-eye jack(Caranx latus) by high-performance liquid chromatography/mass spectrometry. Toxicon 40: 929–939.

Radke, E., A. Reich, and J. Morris. 2015. Epidemiology of ciguatera in Florida. The American Journal of Tropical Medicine and Hygiene 14-0400.

Ragan, M. A., C.M. Ragan, and A. Jensen.1980. Natural chelators in sea water: detoxification of Zn 2+ by brown algal polyphenols. Journal of Experimental Marine Biology and Ecology 44(2): 261-267.

116

Ragelis, E. P. (1984). Ciguatera seafood poisoning: overview. In ACS Symposium series American Chemical Society.

Randall, J.E. 1958. A review of ciguatera, tropical fish poisoning with a tentative explanation of its cause. Bulletin of Marine Science in the Gulf and Caribbean 8:236–267.

Rhodes, L., Harwood, T., Smith, K., Argyle, P., Munday, R., 2014. Production of ciguatoxin and maitotoxin by strains of Gambierdiscus australes, G. pacificus and G. polynesiensis (Dinophyceae) isolated from Rarotonga, Cook Islands. Harmful Algae 39, 185-190.

Rhodes, L., K. Smith, R. Munday, A. Selwood, P. McNabb, P. Holland, and M.-Y. Bottein.2010. Toxic dinoflagellates (Dinophyceae) from Rarotonga, Cook Islands. Toxicon565, 751–758.

Richlen, M.L., Lobel, P.S. 2011. Effects of depth, habitat, and water motion on the abundance and distribution of ciguatera dinoflagellates at Johnston Atoll, Pacific Ocean. Marine Ecology Progress Series 421: 51-66.

Rongo, T. and R. van Woesik. 2013. The effects of natural disturbances, reef state, and herbivorous fish densities on ciguatera poisoning in Rarotonga, southern Cook Islands. Toxicon 64: 87-95

Russell, B.D., T.S. Elsdon, B. Gillanders, and S. Duncan. 2005. Nutrients increase epiphyte loads: broad-scale observations and an experimental assessment. Marine Biology 147 (2):551-558

Saint Martin, K., M. Durand-Clement and P. Bourdeau. 1988. Contribution to the study of the relations between macroalgae and Gambierdiscus toxicus (Dinophyceae), responsible for ciguatera. Cryptogamie Algologie 9:195–202.

Sakami, T., H. Nakahara, M. Chinain, and Y. Ishida.1999. Effects of epiphytic bacteria on the growth of the toxic dinoflagellate Gambierdiscus toxicus (Dinophyceae) Journal of Experimental Marine Biology and Ecology 233:231–246.

Sakamoto, B., Y. Hokama, F.D. Horgen, P.J. Scheuer, Y. Kan, and H. Nagai.2000. Isolation of a sulfoquinovosyl monocylglycerol from Bryopsis sp. (Chlorophyta): Identification of a factor causing a possible species-specific ecdysis response in Gambierdiscus toxicus (Dinophyceae) Journal of Phycology 36:924–931.

Sanner, B. M., B. Rawert, B. Henning, and W. Zidek. 1997. Ciguatera fish poisoning following travel to the tropics. Zeitschrift Fur Gastroenterologie 35(5): 327-330.

117

Scheuer, P.J., W. Takahashi, J. Tsutsumi and T. Yoshida. 1967. Ciguatoxin: isolation, and chemical nature. Science 155:1267–1268.

Schutte, V.G.W., E.R. Selig, and J.F. Bruno.2010. Regional spatiotemporal trends in Caribbean coral reef benthic communities. Marine Ecology Progress Series 402:115−122.

Sperr, A.E., and G.J. Doucette. 1996. Variation in growth rate and ciguatera toxin production among geographically distinct isolates of Gambierdiscus toxicus. In: Yasumoto, T., Oshima, Y., and Fukuyo, Y. (Eds.), Harmful and Toxic Algal Blooms. UNESCO, Paris, pp. 309–312.

Steidinger, K.A.1983. A re-evaluation of toxic dinoflagellate biology and ecology. Progress in Phycological Research.2:147–188.

Targett, N. M., and T.M. Arnold. 1998. Mini review - predicting the effects of brown algal phlorotannins on marine herbivores in tropical and temperate oceans. Journal of Phycology 34(2): 195-205.

Taylor, F.J.R.1979. A description of the benthic dinoflagellates associated with ciguatoxin, including observations on Hawaiian material. In: Taylor, D.L., Seliger, H.,(Eds.). Toxic Dinoflagellate Blooms. Elsevier; New York. pp. 71–76

Taylor, F.J.R, and M.S. Gustavson.1983.An underwater survey of the organism chiefly responsible for “ciguatera” fish poisoning in the eastern Caribbean region: the benthic dinoflagellate Gambierdiscus toxicus. In: Stefanon, A., Fleming, N.J., (Eds.) 7th International Diving Science Symposium; Padova, Italy. 15–18, September, 1983. pp. 95–111.

Tester, P.A., R.L. Feldman, A.W. Nau, S.R. Kibler, and R.W. Litaker. 2010. Ciguatera fish poisoning and sea surface temperatures in the Caribbean Sea and the West Indies. Toxicon 56: 698–710.

Thresher, R. 1980.Behavior and Ecology on the Reef and in the Aquariums. St. Petersburg, FL: Palmetto Publishing, 171 pp.

Tosteson, T.R. 2004. Caribbean ciguatera: a changing paradigm. Revista de Biologica Tropical 52:109-113.

Vairappan, C. S. 2003. Potent antibacterial activity of halogenated metabolites from Malaysian red algae, Laurencia majuscula (Rhodomelaceae, Ceramiales). Biomolecular Engineering 20(4): 255-259.

Vandersea, M.W., Kibler, S.R., Holland, W.C., Tester, P.A., Schultz, T.F., Faust, M.A., Holmes, M.J., Chinain, M. and Litaker, R.W.. 2012. Development of semi-

118

quantitative PCR assays for the detection and enumeration of Gambierdiscus species (Gonyaulacales, Dinophyceae). Journal of Phycology 48:902-915.

Villareal, T.A. and S.L. Morton. 2002. Use of cell-specific PAM-fluorometry to characterize host shading in the epiphytic dinoflagellate Gambierdiscus toxicus. Marine Ecology 23:127–140.

Walters, L.J., M.G. Hadfield, C.M. Smith. 1996. Waterborne chemical compounds in tropical macroalgae: positive and negative cues for larval settlement. Marine Biology 126:383-393.

Wilkinson, C. and D. Souter. 2008. Status of Caribbean coral reefs after bleaching and hurricanes in 2005. Global Coral Reef Monitoring Network, and Reef and Rainforest Research Centre, Townsville, 152 p.

Wylie, C. R., and V.J. Paul. 1988. Feeding preferences of the surgeonfish Zebrasoma flavescens in relation to chemical defenses of tropical algae. Marine Ecology Progress Series.Oldendorf45(1), 23-32.

Xu, Y., M.L. Richlen, J. Liefer, A. Robertson, D. Kulis, T. Smith, M.L. Parsons, and D.M. Anderson. In prep. Effects of temperature, salinity, and irradiance on Gambierdiscus spp. (Dinophyceae) growth. To be submitted to Journal of Phycology, fall 2015.

Yasumoto, T., I. Nakajima, R. Bagnis, and R. Adachi.1977. Finding of a dinoflagellate as a likely culprit of ciguatera. Bulletin of the Japanese Society for the Science of Fish43:1021–1026.

Yasumoto, T., A. Inoue, and R. Bagnis, R. 1979.Ecological survey of a toxic dinoflagellate associated with ciguatera. In: Taylor, D.L. and H. Seliger (Eds.), Toxic Dinoflagellate Blooms. Elsevier; New York. pp. 221–224.

Yasumoto, T., Y. Oshima, Y. Murakami, I. Nakajima, R. Bagnis, and Y. Fukuyo.1980. Toxicity of benthic dinoflagellates. Bulletin of the Japanese Society for the Science of Fish46:327–331.

119

APPENDICES

120

Appendix A1.Average abundance and attachment data for G. belizeanus in ambient conditions.

Acanthophora Caulerpa Dasya 800 800 800 600 600 600 400 400 400

# cells cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Derbesia Dictyota Laurencia 800 800 800 600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Polysiphonia Ulva Control 800 800 800 600 600 600 400 400 400 # cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

121

Appendix A2.Average abundance and attachment data for G. caribaeus in ambient conditions.

Acanthophora Caulerpa Dasya 500 500 500 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0

Cell abundance Cell Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Derbesia Dictyota Laurencia 500 500 500 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0

Cell abundance Cell Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Polysiphonia Ulva Control 500 500 500 400 400 400 300 300 300 200 200 200 100 100 100 0 0 0

Cell abundance Cell Day 1 Day 8 Day 15Day 22Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

122

Appendix A3.Average abundance and attachment data for G. carolinianus in ambient conditions.

Acanthophora Caulerpa Dasya 400 400 400 300 300 300 200 200 200

# cells # 100 100 100 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Derbesia Dictyota Laurencia 400 400 400 300 300 300 200 200 200

# cells # 100 100 100 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Polysiphonia Ulva Control

400 400 1000 300 300 800 600 200 200 400

# cells # 100 100 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

123

Appendix A4.Average abundance and attachment data for G. carpenteri in ambient conditions.

Acanthophora Caulerpa Dasya 800 800 800 600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Derbesia Dictyota Laurencia 800 800 800 600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Polysiphonia Ulva Control 800 800 800 600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

124

Appendix A5.Average abundance and attachment data for G. yasumotoi in ambient conditions.

Acanthophora Caulerpa Dasya

600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Derbesia Dictyota Laurencia

600 600 600 400 400 400

# cells # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29

Polysiphonia Ulva Control

600 600 600 400 400 400

# eclls # 200 200 200 0 0 0 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29 Day 1 Day 8 Day 15 Day 22 Day 29