Examining the Physiology of Harmful Microalgae During

EXAMINING THE PHYSIOLOGY OF HARMFUL MICROALGAE

DURING ALGICIDAL CONTROL AND DIEL VERTICAL MIGRATION

Charles L. Tilney

A dissertation submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Marine Studies

Summer 2014

UMI Number: 3642365

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.

UMI 3642365 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346

EXAMINING THE PHYSIOLOGY OF HARMFUL MICROALGAE

DURING ALGICIDAL CONTROL AND DIEL VERTICAL MIGRATION

Charles L. Tilney

Approved: ______Mark A. Moline, Ph.D. Director of the School of Marine Science and Policy

Approved: ______Nancy M. Targett, Ph.D. Dean of the College of Earth, Ocean, and Environment

Approved: ______James G. Richards, Ph.D. Vice Provost for Graduate and Professional Education

I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: ______Mark E. Warner, Ph.D. Professor in charge of dissertation

Signed: ______Kathryn J. Coyne, Ph.D. Member of dissertation committee

Signed: ______Jonathan H. Cohen, Ph.D. Member of dissertation committee

Signed: ______Christopher J. Gobler, Ph.D. Member of dissertation committee

ACKNOWLEDGMENTS

Many people contributed to the successful completion of this dissertation, and I am much obliged to every one of them. First and foremost, I must thank Dr. Mark Warner for his support, sage guidance, and patience over the years, and for imparting his knowledge of biology, ecology, and science in general. I too must pay a special thanks to Dr. Kathryn Coyne, for her guidance throughout much of this dissertation, and for patiently training me in the realm of molecular biology. Certainly, everyone in the Warner and Coyne laboratories deserve my thanks for their help and friendship over the years, including: Dr. Sebastian Hennige, Dr. Michael McGinley, Dr. Matthew Aschaffenburg, Kenneth Hoadley, Dr. Tye Pettay, Kaytee Pokrzywinski, Chris Main, Colleen Bianco, Katherine Lee, Michelle Stuart and Dr. Jennifer Stewart. I also wish to thank my committee members, Dr. Jonathan Cohen and Dr. Christopher Gobler, for providing advice and direction. I must thank everyone on the Lewes campus, and should thank in particular Dr. Jonathan Sharp, and Dr. Edward Whereat. The participants of the University of Delaware’s Citizen Monitoring Program also deserve an acknowledgement.

I would like to thank my family in Washington DC for their myriad offerings of support especially upon moving to Delaware. My sisters, parents, and grandparents in Great Britain deserve my eternal thanks for their love and support. Finally, without question, my wife Josée deserves my final words here for her patience and love. Thank you.

iv TABLE OF CONTENTS

LIST OF TABLES ...... ix LIST OF FIGURES ...... x ABSTRACT ...... xviii

Chapter

1 INTRODUCTION ...... 1

1.1 What Are Harmful Algal Blooms? ...... 1 1.2 A Brief Overview of HAB Dynamics ...... 3 1.3 Circadian Rhythms and Diel Vertical Migration ...... 7 1.4 Control of HABs ...... 9 1.5 Dissertation Chapters ...... 11

2 GROWTH, DEATH AND PHOTOBIOLOGY OF DINOFLAGELLATES UNDER BACTERIAL-ALGICIDE CONTROL ...... 13

2.1 Abstract ...... 13 2.2 Introduction ...... 15 2.3 Methods ...... 19

2.3.1 Bacterial Filtrate Preparation ...... 19 2.3.2 Algal Species ...... 19 2.3.3 Species Testing ...... 20 2.3.4 Algal Photochemistry ...... 20 2.3.5 Dose Responses ...... 21 2.3.6 Membrane Permeability ...... 22 2.3.7 Dark Incubation ...... 23 2.3.8 Statistical Analyses ...... 24

2.4 Results ...... 25

2.4.1 Species Comparison ...... 25 2.4.2 Dose Response ...... 26 2.4.3 Membrane Permeability ...... 27 2.4.4 Darkness ...... 28

v 2.5 Discussion ...... 29

2.5.1 General Dynamics of Photobiology & Cell Number ...... 29 2.5.2 Species Specificity & Membrane Exposure ...... 30 2.5.3 Dose Responses ...... 32 2.5.4 Membrane Permeability and Fv/Fm ...... 33 2.5.5 Darkness ...... 34 2.5.6 Conclusion ...... 35

3 EFFECTS OF A BACTERIAL ALGICIDE, IRI-160AA, ON DINOFLAGELLATES AND THE MICROBIAL COMMUNITY IN MICROCOSM EXPERIMENTS ...... 43

3.1 Abstract ...... 43 3.2 Introduction ...... 45 3.3 Methods ...... 51

3.3.1 Bacterial Filtrate Preparation ...... 51 3.3.2 Natural Waters, Incubation & Sampling ...... 51 3.3.3 Chlorophyll-a ...... 53 3.3.4 Active Chlorophyll-a Fluorescence ...... 54 3.3.5 Cell Enumeration ...... 54 3.3.6 DNA Extraction, PCR-DGGE, and Sequencing ...... 55 3.3.7 Quantitative-PCR ...... 57 3.3.8 Statistical Analyses ...... 59

3.4 Results ...... 60

3.4.1 Bloom Experiments ...... 60

3.4.1.1 Photochemistry ...... 60 3.4.1.2 Autotrophic Biomass & Target Cell Number ...... 61 3.4.1.3 Eukaryotic Community Structure ...... 62

3.4.2 G. instriatum Dose Experiment ...... 64

3.4.2.1 Photochemistry ...... 64 3.4.2.2 Eukaryotic and Prokaryotic Community Structure ...... 64 3.4.2.3 Group and Species Abundances ...... 65

3.4.2.3.1 Specificity of Primers For qPCR ...... 67

3.5 Discussion ...... 68

vi 3.5.1 Similarities and Discrepancies From Cultures ...... 68 3.5.2 Shifts in Community Structure ...... 74 3.5.3 Conclusions ...... 77

4 COMPARING THE DIEL VERTICAL MIGRATION OF KARLODINIUM VENEFICUM (DINOPHYCEAE) AND CHATTONELLA SUBSALSA (RAPHIDOPHYCEAE): PSII PHOTOCHEMISTRY, CIRCADIAN CONTROL, AND CARBON ASSIMILATION ...... 92

4.1 Abstract ...... 92 4.2 Introduction ...... 94 4.3 Methods ...... 97

4.3.1 Stock Algal Culture, Column Design, and Lighting ...... 97 4.3.2 Common Experimental Procedures ...... 98 4.3.3 Chlorophyll Fluorescence Measurements ...... 99 4.3.4 DVM Sampling and Calculations ...... 101 4.3.5 Initial Migration & Photobiology Experiments ...... 101 4.3.6 Assessing Circadian Control of Rhythms ...... 102 4.3.7 Comparing DVM and Static Cultures ...... 103 4.3.8 Carbon Assimilation ...... 104 4.3.9 Statistical Analyses ...... 105

4.4 Results ...... 106

4.4.1 Diel Vertical Migration & Diel Photobiology ...... 106 4.4.2 Circadian Control Experiments ...... 110 4.4.3 Comparing Growth Between DVM and Static Cultures ...... 112 4.4.4 Carbon Assimilation Rates ...... 113

4.5 Discussion ...... 113

4.5.1 DVM, Photobiology, and Growth ...... 113 4.5.2 Circadian Rhythms ...... 118 4.5.3 Value of DVM ...... 121 4.5.4 Conclusion ...... 122

5 CONCLUSIONS ...... 134

REFERENCES ...... 141

Appendix

A APPENDED FIGURES ...... 170

vii B RE-PRINT LICENCE ...... 173

viii LIST OF TABLES

Table 2.1: Specific growth rates from Figure 2.1 (algal species tested with 4% IRI-160AA or 4% f/2 medium). IRI-160AA growth rates and percent cell loss were calculated from T0 to the time point with the lowest cell number, except for P. minimum and Rhodomonas sp. which did not show a drop in cell density (N/a) and were calculated over 90 h. Percent cell loss was rounded to the nearest whole percentage with ±1 SD given in parentheses ...... 37

Table 3.1: Environmental conditions at the time of sampling each bloom, as well as incubation conditions used during the experimental period ...... 79

Table 3.2: Descriptions of DGGE and qPCR primers, qPCR thermocycling conditions, assay efficiency and regression coefficient of determination ...... 80

Table 3.3: Sequencing results from DGGE bands, qPCR reactions, and cloned qPCR reactions, in order of appearance in the text ...... 81

Table 4.1: Descriptions of photosynthetic components, and chlorophyll a fluorescence terms used in this study...... 124

ix LIST OF FIGURES

Figure 2.1: Species specific responses in cell density (a-e) and Fv/Fm (f-j) among 4 dinoflagellates: G. instriatum (a,f), K. veneficum (b,g), A. tamarense (c,h), and P.minimum (d,i), and 1 cryptophyte, Rhodomonas sp. (e,j), after addition of 4% IRI-160AA concentration (filled squares) compared to control cultures (open circles) after addition of 4% f/2 medium. The species are ordered from highest to lowest response (left to right) and by relative plasma-membrane exposure, which reflects the change from naked to thecate dinoflagellates and then also by number of plates/valves (see discussion). Asterisks indicate a significant difference as determined by 2-way ANOVA and Bonferroni post tests (n=3, P<0.05). Error bars are ±1 SD of the mean. 38

Figure 2.2: Species specific responses in the QA- re-oxidation rate τ (a), and in PSII connectivity ρ (b), both expressed as relative change from the f/2 control, so that the 0% horizontal lines represent the average f/2 control. Gi = G. instriatum (open bars); Kv = K. veneficum (grey bars); At = A. tamarense (black bars); Pm = P. minimum (striped bars); R sp. = Rhodomonas sp. (fine checkered). Data are from the respective experiments shown in Figure 2.1. Asterisks indicate a significant difference (n=3, P<0.05) determined by 2-way ANOVA, and Bonferroni post tests. Error bars are ±1 SD of the mean...... 39

Figure 2.3: Dose response based on cell density (filled triangles) and Fv/Fm (open circles) in K. veneficum expressed as % inhibition relative to the f/2 control, after 18 h. Variable slope dose response curves are fitted to the data. Error bars are ±1 SD. Percent inhibition based on cell density and Fv/Fm are significantly different at the 4% concentration (2-way ANOVA and Bonferroni post-tests, n=3, P<0.05). The inset figure shows relative inhibition in Fv/Fm (filled bars) relative to the f/2 control and cell density (open bars) relative to initial densities in culture dilutions treated with 4% IRI-160AA. Full = no dilution, 2/3 = 33% dilution, 1/3 = 66% dilution. No significant differences were observed among dilutions for either cellular inhibition, or photochemical inhibition (1-way ANOVA and Tukey test). However, IRI-160AA treatments all differed significantly from f/2 controls (n=3, P<0.05) ...... 40

x Figure 2.4: 24-hour incubations with K. veneficum (left panes) and G. instriatum (right panes). Cultures were incubated with either 8% v/v IRI-160AA (filled squares) or 8% v/v f/2 medium (open circles). Fv/Fm (a,b), cell density (c,d) and SYTOX-fluorescence (e,f; relative to the f/2 control) are presented. Greyed areas behind the data represent the dark period. Asterisks indicate time points with a significant difference, as determined by two-way ANOVA and Bonferroni post-tests (P<0.05, n=3). Error bars represent ±1 SD of the mean...... 41

Figure 2.5: Effects of shifting treated cultures into continuous darkness immediately upon addition of 4% (v/v) IRI-160AA in K. veneficum (a, b) and G. instriatum (c, d). Light treatments are indicated by open symbols, dark treatments are indicated by filled symbols, circles indicate controls and squares are IRI-160AA treatments. Asterisks indicate a significant difference only between IRI-160AA light and IRI-160AA dark treatments (2-way ANOVA and Bonferroni post- tests, P<0.05, n=3). Error bars are ±1 SD of the mean...... 42

Figure 3.1: Photophysiology measured over 2 or 3 days of incubation with IRI- 160AA. Filled squares represent IRI-160AA algicide treatments, and open circles represent f/2 controls. (A-C) maximum quantum yield of photosystem II (Fv/Fm ratio), (D-F) photosystem II reaction center connectivity (unitless), (G-I) maximum effective absorption cross section of photosystem II (Sigma; in nm2). Error bars indicate 1 S.D. and asterisks indicate significance level (*=p<0.05, **=p<0.01, ***=p<0.001) as determined by repeated measures ANOVA and Bonferoni post tests...... 82

Figure 3.2: Measures of autotrophic biomass over 2 or 3 days of incubation with IRI-160AA. Filled squares represent IRI-160AA algicide treatments, and open circles represent f/2 controls. (A-C) in-vivo fluorescence presented in relative fluorescence units, (D-F) extracted chlorophyll-a presented in µg/L, and (G-I) target dinoflagellate cell density in 103 cells mL-1. Error bars indicate 1 S.D., and asterisks indicate significance level (*=p<0.05, **=p<0.01, ***=p<0.001) as determined by repeated measures ANOVA and Bonferoni post tests. ... 83

xi Figure 3.3: MDS plots of the eukaryotic community composition based on DGGE fractionation of partial 18S rDNA PCR amplicons from particles > 3 µm, and determined by Sorensen similarity among samples within 3 separate experiments: P. minimum (A), K. veneficum (B), G. instriatum (C). Contours indicate SIMPROF significance at p<0.05. Bubble size indicates lane-standardized band volume of the band which contributed most to community dissimilarity in Bray-Curtis similarity matrices (see text for details). Bubble scale (% of community) is shown as semi-circles below each plot. The sequence identification of each band is shown in the top left of each plot. In all plots, T0 samples are represented by white bubbles (or triangles in A where band volumes were too small for bubble depiction), final f/2 control samples are dark grey bubbles, and final IRI-160AA samples are light-grey bubbles. Scales are indicated below each plot...... 85

Figure 3.4: MDS bubble plot of eukaryotic community composition in the > 3 µm size fraction of the G. instriatum dose experiment at the final time point (T42 hours). Grey filled bubbles are IRI-160AA treatments, and black filled triangles are f/2 controls, and the percent of IRI-160AA or f/2 used is written above every sample. Bubbles represent lane- standardized band volume of Paraphysomonas sp. where the scale (% of community) is shown below the plot with semi-circles. Symbols are used for f/2 control samples because bubbles were too small. Dotted lines delineating groups of samples indicate a significant difference in community structure based on SIMPROF tests on Sorensen presence/absence similarity matrices (p<0.05)...... 86

Figure 3.5: MDS plot of prokaryotic community composition in the 0.2-3 µm size fractions of the G. instriatum dose experiment. Prokaryotic community composition was assessed by PCR-DGGE, and analyzed by multivariate statistics (see text for details). Symbols shown are: initial community composition (T0 hours, closed triangles), final community composition in IRI-160AA algicide treatments (T42 hours, grey filled squares), and final community composition in controls (T42 hours, black filled circles). The percent application is noted above each symbol. Dotted lines delineating groups of samples indicate a significant difference in community structure based on SIMPROF tests on Sorensen presence/absence similarity matrices (p<0.05)...... 87

xii Figure 3.6: Relative qPCR abundance as fold change from control microcosms, of dinoflagellates (A), diatoms (B), ciliates (C), G. instriatum (D), Leptocylindrus sp. (E, striped bars), Cyclotella sp. (E, filled bars), and Paraphysomonas sp. (F) in the G. instriatum dose experiment. Error bars indicate ±1 SD, and asterisks indicate significant differences between control microcosms (not displayed) and IRI-160AA treatments (*=p<0.05, **=p<0.01, ***=p<0.001), determined by RM- ANOVA (n=3) and Bonferoni post tests...... 88

Figure 3.7: Confirming the bioactivity of IRI-160AA filtrates harvested for use in this study, at 10 % (v/v) against uni-algal cultures after 1 day of incubation following the methods in Pokrzyiwinski et al. (2012) and Tilney et al. (2014). Inihibition in Fv/Fm is presented in panel A, where open symbols represent f/2 control cultures, and closed symbols represent IRI-160AA treated cultures. Filled circles show IRI-160AA used in the K. veneficum bloom experiment, tested against K. veneficum; filled diamonds show IRI-160AA used in the P. minimum bloom experiment tested against G. instriatum; and filled triangles show IRI-160AA used in the G. instriatum bloom experiment and dose experiment, tested against G. instriatum. Algicidal activity (in %) is shown in panel B, and originates from the same experiments described above for panel A. The dinoflagellate noted below the bars represents the filtrate batch from that species’ bloom experiment in the current study. Algicidal activity in IRI-160AA treatments was tested by RM-ANOVA and Bonferroni post-tests, and revealed that significant algicidal activity was induced by all 3 batches of algicide used in this study after just 1 day of incubation (p<0.001, n=3)...... 89

Figure 3.8: Correlation of relative DGGE band volume with relative qPCR abundance of Leptocylindrus sp., Paraphysomonas sp., G. instriatum, and Cyclotella sp. tested in the G. instriatum dose experiment (data from Fig. 3.4 and Fig. 3.6). Pearson product moment correlation r = 0.95, R2=0.91, p<0.0001, n=36. Relative abundance refers to IRI- 160AA treatment as a percentage of the average f/2 controls, and only IRI-160AA samples are presented...... 90

xiii Figure 3.9: Mixed culture experiment with K. veneficum and Chattonella subsalsa (Raphidophyceae) showing the effect of eliminating K. veneficum from a 2-specie mixed culture on the maximum effective absorption cross section of PSII (SigmaPSII, nm2). We tested 3 algal cultures, each in triplicate with 10% IRI-160AA or 10% f/2 medium added as a control, which were 1) unialgal K. veneficum, 2) unialgal C. subsalsa, and 3) a 1:1 mix of K. veneficum with C. subsalsa. Panel A shows percent algicidal activity (filled bars) where Kv = K. veneficum, Mixture = 1:1 mixed culture, and Cs = C. subsalsa. Paenl B shows maximal effective absorption cross-section of PSII (SigmaPSII, nm2), where closed symbols are IRI-160AA incubations, and open symbols are controls. Circles represent unialgal K. veneficum, squares represent uni-algal C. subsalsa, and triangles represent the 1:1 mixed culture. Only the mixed culture shows a decline in SimgaPSII due to the shift in the proportion of each alga’s photochemical signatures...... 91

Figure 4.1: DVM and PSII photochemistry in K. veneficum (A, C, E, G, I) and C. subsalsa (B, D, F, H, J) measured every 3 hours over 39 hours, with columns kept in 14:10 LD cycles. Measurement began on the third day after inoculation. Filled grey bars underneath the data represent the dark period, and white areas represent the light period. Light intensity is presented above panels A, B. Presented for each species 2 are: DVM (A, B); Fv/Fm (C, D); σPSII (nm , E, F); τ (µs, G, H); ρ (I, J). DVM is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measures from the bottom of the columns (filled circles) and at the surface of the columns (open cicles) are presented. Note the difference in scale between panels E and F. Error bars represent ±1 SEM where n=2 for K. veneficum, and n=3 for C. subsalsa...... 126

xiv Figure 4.2: DVM and PSII photochemistry in K. veneficum (A, C, E, G, I) and C. subsalsa (B, D, F, H, J) measured an 1 hour before dawn (05:00), at midday (13:00), and 1 hour before dark (19:00) every day for 6 days, with columns kept under 14:10 LD cycles. Measurement began on the second day after inoculation. Filled grey bars underneath the data represent the dark period, and white areas represent the light period. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, B); Fv/Fm (C, D); PSII light harvesting center quenching (E, F); τ (µs, G, H); ρ (I, J). DVM is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measurements at the bottom (filled circles), at the surface (open circles), and light acclimated measurements at the surface (open inverted triangles) are presented. Light acclimated measurements at the bottom were obscuring dark measurements, and so were excluded for clarity. Asterisks in C, D represent significant differences (p<0.05) between dark acclimated measurements at the surface and bottom samples as determined by RM-ANOVA and Bonferroni post tests (n=3). Error bars represent ±1 SEM...... 128

Figure 4.3: DVM and Fv/Fm in K. veneficum (A, C) and C. subsalsa (B, D) measured at 08:00, 14:00, 23:00 in K. veneficum, and at 08:00, 14:00, 17:00 in C. subsalsa, in columns maintained for two days in 14:10 LD cycles (same data as Fig. 4.1), and then maintained in constant darkness from 06:00, at time = 48 hours. Light intensity during the incubations is presented above panels A and B. Dark filled grey bars underneath the data represent dark periods, white areas represent light periods, and the pale filled grey area represents constant darkness. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, B); Fv/Fm (C, D). Error bars represent ±1 SEM where n=2 for K. veneficum, and n=3 for C. subsalsa...... 129

xv Figure 4.4: DVM and PSII photochemistry in K. veneficum (A,C,E,G,I) and C. subsalsa (B,D,F,H,J) measured at 1 hour before dawn (05:00), midday (13:00), and 1 hour before dark (19:00) every day for 6 days, in the continuous light experiment (LL). Columns were kept under 14:10 LD cycles until day 3, when columns were shifted at midday (13:00) into continuous midday light for 2 days, before being returned to 14:10 LD cycles at midday (13:00) for a further 24 hours. Light intensity during the incubations is presented above panels A and B. Dark filled grey bars underneath the data represent dark periods, white areas represent light periods, and pale filled grey bars represent ‘night-time’ during the constant light period. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, B); Fv/Fm (C, D); τ (µs, E, 2 F); σPSII (nm , G, H); ρ (I, J). DVM (A, B) is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measurements at the bottom (filled circles), and at the surface (open circles) are presented. Note the difference in scale between panels G and H. Error bars represent ±1 SEM...... 131

Figure 4.5: Specific growth rate µ (d-1) in cells forced to grow in surface light intensities (High), bottom light intensities (Low), or in cells free to move within the column (Column). In each case growth was calculated by regression over 6 to 9 days (see methods). Error bars represent ±1 SEM, and statistical differences (2-way and 1-way ANOVA followed by Bonferroni post-tests, P<0.05, n=3), are denoted by capital letters at the top of each bar...... 132

Figure 4.6: Carbon assimilation rates as pg C cell-1 h-1, in K.veneficum (A), and C. subsalsa (B) calculated from 55 min incubations using sub-samples from 5 depths in the columns. Statistical differences (1-way ANOVA followed by Bonferroni post-tests, P<0.05, n=3) between rates at each depth are denoted by capital letters at the top of each bar. Error bars represent ±1 SEM...... 133

Figure A1: Testing the effects of IRI-160AA concentration on SYTOX-green fluorescence (y-axis) in K. veneficum after 30 minute incubations with 1 µM final concentration SYTOX-green and different concentrations of IRI-160AA (x-axis)...... 170

xvi Figure A2: Testing change in Fv/Fm from the top and bottom of un-mixed test tubes over 66 hours. Four sets of triplicate test tubes contained a final volume of 5 mL of K. veneficum culture, two sets were treated with 4% (v/v) IRI-160AA, and two treated with 4% (v/v) f/2 medium as a control. One set of treatments, and one set of controls were mixed by pipette before measuring a 2 mL aliquot in the FRR fluorometer after dark acclimation. The other set of treatments and controls remained unmixed, and the top 2 mL of culture carefully removed for measurement, with the remaining 3 mL from these tubes being mixed before measuring a 2mL aliquot, also after dark acclimation...... 171

Figure A3: Testing for circadian rhythmicity in multiple turnover (MT) Fv/Fm in C. subsalsa during the same experiment as that presented in Fig. 4.4. The single turnover (ST) Fv/Fm from Fig. 4.4 is also presented...... 172

xvii ABSTRACT

The frequency of harmful algal blooms (HABs) has increased, and so too has the societal and environmental costs associated with them. Anthropogenic eutrophication has played some part in this, but is not always implicated, as HABs and their drivers are diverse. Continued investigation into HAB biology and ecology may help to identify the underlying factors controlling algal blooms in general, and factors implicated in their recent increase in particular. Investigation of the various methods for the prevention, control, and mitigation of HABs is also warranted, to identify ‘solutions’ to reduce or eliminate some or all of the costs imposed by HABs. In this dissertation, both basic and applied work was conducted, initially on the control of dinoflagellates with a sterile bacterially-derived algicidal filtrate that is strongly specific toward dinoflagellates, and then on the diel vertical migration (DVM) of two potentially harmful species of microalgae. Initial investigations of the algicidal filtrate (termed IRI-160AA) assessed the effects in cultured dinoflagellates, and identified that electron transport through the photosynthetic apparatus at photosystem II (PSII) was broadly inhibited in algicide treated dinoflagellates. Moreover, IRI-160AA induced dose-dependent effects on both cell viability and PSII efficiency, and although dinoflagellate responses were highly species specific, one apparent generalization was that thecae reduced overall susceptibility. Next, the algicidal filtrate was tested on field collected water samples containing dinoflagellates. These dinoflagellates were less susceptible than uni-algal cultures because they were not in the most susceptible phase of growth, which corroborated the idea of Pokrzywinski et al. (2012) that

xviii identifying the growth stage of dinoflagellate blooms in the field would be crucial to achieving high levels of bioactivity with the algicide. Using a combination of polymerase chain reaction denaturing gradient gel electrophoresis (PCR-DGGE) and quantitative real-time PCR, I determined what changes occurred in the eukaryotic and prokaryotic microbial communities after applications of IRI-160AA. A common theme among these applications was the rise of bactivorous protists, most notably Paraphysomonas spp., which likely arose as a response to bacterial growth from dissolved organic matter (DOM) within the filtrate and released from dying dinoflagellates. This result highlighted how the microbial loop might transfer dinoflagellate carbon back to higher trophic levels in large-scale applications, and perhaps in HAB control more generally. The results suggest that IRI-160AA could be a potent and species-specific method to control harmful dinoflagellates in nature, but future work is required to identify methods to appropriately and cost-effectively scale- up algicide production, as well as to isolate the active compounds and determine the molecular targets and mode of action.

Finally, a detailed comparison of the DVM of two sympatric harmful algal species (Karlodinium veneficum and Chattonella subsalsa) was conducted in laboratory columns under various sampling regimes and lighting conditions. Both species exhibited markedly different DVM patterns, which were consistent with different photoprotective mechanisms, and carbon assimilation rates. DVM was found to be under the control of a circadian clock in C. subsalsa but not K. veneficum. This work provides a useful comparison between two potentially harmful species from the Delaware Inland Bays, and improves our understanding of the niches that these algae

xix are adapted to. Collectively, this dissertation has raised many interesting questions for future research addressing the control of harmful algae with biologically derived compounds, and laid important foundations for deeper investigations into the role of DVM in two cosmopolitan harmful algae.

xx Chapter 1

INTRODUCTION

This introductory chapter aims to provide a broad overview of topics covered in this dissertation. The literature review in this chapter is not intended to be exhaustive, but instead to provide complementary background material for readers to place the research chapters of this dissertation into the context of the wider HAB field and related areas.

1.1 What Are Harmful Algal Blooms?

Harmful algal blooms (HABs) are biological phenomena that develop from the growth of certain algae, which can result in harm to humans and the environment. HABs are diverse phenomena, and the ‘harmful’ designation usually stems from either excessive growth, so called “high-biomass blooms”, or from toxin production by low biomass but toxic algae, so called “toxic algal blooms”. The biomass discrepancy between these types of HABs can be enormous, for example the high-biomass brown tide species Aureococcus anophagefferens can reach cell densities of >109 cells L-1 (n.b. in biomass terms it’s a small cell), whereas a HAB of Dinophysis spp. can cause harm at <103 cells L-1, a full 6 orders of magnitude difference (Smayda 1997). The biomass discrepancy between these types of HABs also serves as a good metaphor for the large diversity and complexity of the HAB problem, a common theme throughout this dissertation. High-biomass blooms can lead to hypoxia and anoxia at the benthos due primarily to the high attenuation of light, and the high respiratory demands from

1 the decomposition of dead algal organic matter (GEOHAB 2006). These types of blooms are sometimes classified as ‘ecosystem disruptive’ if usual ecosystem functions are inhibited, such as carbon transfer to higher trophic levels (Edvardsen and Paasche 1998). Toxic algal blooms on the other hand are caused by the production of a wide variety of toxins, each varying in their targets, mechanisms, and degree of toxicity. These toxins can affect human health primarily by the bioaccumulation of toxins in filter-feeding shellfish, which upon ingestion can cause one of five toxicity syndromes (FAO 2004). Whether an individual bloom is deemed to be ‘harmful’ may be open to debate (Smayda 1997), as the consequences of HABs can range from lethal toxicity to humans, fish and mammals, to the more innocuous consequences of unattractive odors or water-body aesthetics (Granéli and Turner 2006). The determination is however important, as it could affect estimates of HAB occurrences. Sometimes however, the consequences of HABs are not immediately apparent, and some toxic microalgae can produce harmful effects at a later date. For example, sea lions and humans can experience amnesic shellfish poisoning shortly after consumption of domoic acid toxins, but then experience a latent period of weeks to months before symptoms of domoic acid epileptic disease, such as seizures and unusual behaviors, appear (Ramsedell and Gulland 2014). Although harmful algal blooms are societally undesirable, they do represent natural phenomena that have evolved over time. In this vein, historical records prior to the 18th century cite HAB-related events like the attribution of human deaths to specific shellfish consumption (e.g. see Lewitus et al. 2012 and references therein), and biblical records of ‘red’ waters (Granéli and Turner 2006). Although HABs occur

2 naturally, their frequency is increasing world wide (Smayda 1990, Hallegraeff 1993, Van Dolah 2000), and is accompanied by biogeographic range expansions of certain species (Kooistra et al. 2001, Penna et al. 2005, Bolch & de Salas 2007, McLeod et al. 2012). If this trend is extrapolated, it implies that negative effects will also continue to increase. Moreover, growing demand for protein is driving up aquaculture efforts, which in terms of total fisheries production is expected to match or exceed total wild catches by 2030 (World Bank 2013). This is problematic, first because certain aquaculture methods increase nutrient loading to aquatic environments, which correlates positively with HAB occurrences (see below, and Bouwman et al. 2013; though not all aquaculture has this effect e.g. Smayda 2005), but secondly because the negative effects of HABs can strongly impact classical finfish and shellfish aquaculture systems (e.g. Shumway 1990, Mackenzie 1991, Azanza et al. 2005) potentially forming a positive feedback loop. Thus HABs may have the potential to have larger economic impacts in the future, and possibly risk destabilizing global supplies of protein. Consequently, research efforts to understand the core biology of harmful algae and HAB dynamics (i.e. factors contributing to the formation, persistence, and decline of blooms) are required to address the HAB issue. Indeed, research efforts have made successful inroads in a number of directions, including in determining some important factors driving the recent global HAB expansion, but many more questions remain.

1.2 A Brief Overview of HAB Dynamics Anthropogenic eutrophication is widely believed to have driven some of the observed increases in HABs to date (Anderson et al. 2002, Heisler et al. 2008), however the relationship is also more complex than a simple “more nutrients = more

3 HABs” hypothesis, partly because HABs are diverse, with many culpable species and many forms of ‘harm’. Consequently, the relationship between HAB occurrences and eutrophication may sometimes be simple (i.e. as earlier, ‘more nutrients = more HABs’), but may also be mediated through altered nutrient speciation (e.g. urea, Glibert et al. 2006), shifts in the ratio of organic:inorganic nutrient composition (e.g. in brown tides, Gobler and Sunda 2012), and by modified nutrient ratios which can affect toxin production (e.g. Fehling et al. 2004) and patterns of species dominance (e.g. Gilpin et al. 2004). However, the importance of macronutrient ratios (particularly inorganic nutrients) is now being reconsidered (Flynn 2010, Gowen et al. 2012, Davidson et al. 2012). The reduced importance placed on inorganic nutrient ratios has stemmed in small part from an increasing appreciation of nutritional flexibility among HAB species (e.g. Gobler et al. 2012), and at the extreme end by the prevalence of mixotrophy among flagellates (Stoecker 1999, Lewitus et al. 1999, Tillmann 2003). Moreover, nutrient regulation of HABs can stem from micronutrient controls (e.g. Doblin et al. 2000, Peacock and Kudela 2014) and vitamins, since many HAB species are auxotrophic for certain forms (Tang et al. 2010). In some HAB species, photosynthetic adaptations can provide possible advantages over other non-HAB species (e.g. resistance to high light, Hennige et al. 2013). Release from top-down controls are also likely of importance in HAB dynamics, with HAB species effectively inhibiting grazing through a variety of mechanisms such as poor nutritive value, chemical deterrents, and production of exopolymers (Strom et al. 2003, Tillmann 2004, Stoecker et al. 2008, Waggett et al. 2008). Chemical ecology, specifically allelopathy (inhibitory chemical interactions), among harmful algae and competitor algae is also important in HAB dynamics (Prince et al. 2008, Poulson et al. 2009).

4 Ship-ballast dispersal is also thought to have contributed to the geographic expansion of HAB species (Bolch & de Salas 2007), which can affect allelopathic interactions by bringing together plankton with no shared evolutionary history (Kubanek et al. 2007). Other biotic interactions may be important in regulating HABs, including interactions with bacteria (Doucette 1995, Mayali and Azam 2004), parasites (Velo-Suárez et al. 2013), and viruses (Brussard et al. 2005). The potential role of intraspecific variability among HABs has been known for a long time and continues to be substantiated (e.g. Burkholder and Glibert 2006). Many species of toxic algae are dinoflagellates (Smayda 1997b), and undergo complex sexual life cycles (e.g. Blackburn et al. 1989 and references therein), which are still incompletely characterized (Brosnahan 2011). To be relevant, any new understanding of HAB biology must be considered with constraints imposed by the physical and chemical environment, for example including, for example, hydrographic and meteorological conditions, coastal geomorphology, and climate change. Environmental characteristics such as water residence times, turbulence, stratification, land-sea interactions, and local geomorphology can all interact with the biology and ecology of HABs (Margalef 1978, Smayda 1997b, Cembella et al. 2005, Lai and Yin 2014). Thus, each system presents a different suite of interactions between the physical and chemical environment and the harmful algae that are present. A comparative approach among similar (and dissimilar) systems has been proposed as a way to tease apart the different mechanisms influencing HABs (Anderson et al. 2005, Kudela et al. 2008). Many environmental factors and groups of factors (e.g. those associated with climatic variables like the North Atlantic Oscillation) have been implicated in HAB dynamics at various geographical scales, and remain under investigation. For example, evidence

5 that climate changes can impact HAB occurrences, geographic expansions, and modify ‘bloom periods’ is beginning to emerge, though large geographical heterogeneity in the responses is often reported (Breton et al. 2006, Edwards et al. 2006, Gieskes et al. 2007, Moore et al. 2008, Hallegraeff 2010, Mcleod et al. 2012). The effect of warming and acidification on HABs is not yet known, but evidence to date has shown effects may range from minimal (Rost et al. 2006) to enhancing the competitive ability of some algal species, with increased growth rates due to so-called ‘carbon-fertilization’ (Fu et al. 2008, Hutchins et al. 2009). Furthermore in one HAB species, high-CO2 treatments induced both higher growth and overall toxicity (Fu et al. 2010). Notably, little evidence has been published for suppressive effects of climate change on HABs, though various mechanisms could theoretically provoke an inhibitory effect on HABs (e.g. via competitive interactions, Tortell et al. 2002). The effect of climate change on HAB dynamics may also manifest through altered rainfall patterns and stratification, which may benefit a number of flagellated HABs (Jephson et al. 2011). Consequently, progress in understanding HABs has required and will continue to require enormous multi-interdisciplinary research efforts. For example, conceptual frameworks that cover many diverse systems have been developed to help tease apart commonalities among systems, and help to identify areas where research is lacking (Smayda and Reynolds 2003). Modelling efforts are another example where researchers are bringing together multiple disciplines by incorporating an understanding of the biology into physical forcing models to explore, for example, inter-annual variability in bloom magnitudes and spatial distributions (McGillicuddy et al. 2005). The success and utility of these models for management and researchers

6 is spurring similar developments in other systems, but require a thorough understanding of the local biological, environmental and hydrographical conditions (Anderson et al. 2013). Many species and HAB systems remain poorly characterized, and continued basic research is needed in almost all HAB species and systems where they commonly occur.

1.3 Circadian Rhythms and Diel Vertical Migration Biological rhythms are ubiquitous among life on earth, and the period (i.e. the length of one full cycle) of these rhythms can vary greatly. In the plankton for example, biological rhythms include ultradian rhythms (periods < 24 hours, e.g. cell division in Prochlorococcus spp., Shalapyonok et al. 1998), diel rhythms (periods of ~ 24 hours, e.g. diel vertical migration in dinoflagellates, Eppley et al. 1968), circalunar rhythms (periods of ~30 days e.g. in zooplankton biomass, Hernández-León et al. 2002) and circannual rhythms (periods of ~ 1 year e.g. dinoflagellate excystment, Anderson and Keafer 1987, Matrai et al. 2005). Biological rhythms are controlled either by environmental cues (exogenous control), or by internal cues that stem from a ‘biological clock’ (endogenous control). Both types of control are common in nature and are important for organisms to synchronize biological processes to each other and to the environment over time. Following from this distinction, diel rhythms that are endogenously controlled are called circadian rhythms. Four core features that define circadian clocks are 1) the ability to free-run (i.e. continue cycling in the absence of any external stimuli), 2) temperature compensation (i.e. the circadian period is maintained across a range of temperatures), 3) entrainment (the ability for the clock to be set by external stimuli called zeitgebers, usually light for circadian rhythms), and 4) conditionality (the ability to ‘switch off’ clock control of certain phenotypes under

7 specific conditions) (Vitaterna et al. 2001, Johnson 2001, Xu et al. 2013). Circadian rhythms are adaptive in the evolutionary sense, and enhance photosynthesis, growth, and survival in plants and cyanobacteria when the clock is correctly matched to the light:dark cycle (Sharma 2003, Dodd et al. 2005, Johnson 2005 and references therein). Interestingly, dinoflagellates have been central to the study of circadian rhythms because their rhythm in biolumninescence facilitated automation of clock measurements (Hastings 2001). Moreover, rhythms in dinoflagellates that have been identified to date include bioluminescence, photosynthesis, motility and migration, cell division, protein synthesis, and ultrastructural reorganization (Nassoury et al. 2005, Hastings 2001). Adding complexity to the role of the biological clock is the evidence that multiple clocks exist within a single cell (Roenneberg and Morse 1993). An understanding of the circadian biology of harmful algae (many of which are dinoflagellates) could reveal characteristics that promote their proliferation. Diel vertical migration is a ubiquitous behavior in nature and is used by diverse aquatic organisms such as cyanobacteria (Richardson and Castenholz 1987), dinoflagellates (Eppley et al. 1968), zooplankton (Forward 1988), jellyfish (Moriarty et al. 2012) and fish (Mehner 2012). The behavior also occurs in a wide variety of habitats and covers a large range of spatial scales, from migrations of <1mm within sediments by benthic thermophilic cyanobacteria (Richardson and Castenholz 1987), to migrations of many hundreds of meters into virtual darkness by certain deep sea zooplankton (Van-Haren and Compton 2013). Although vertical migration has important implications for biogeochemical flux in water columns (Villareal et al. 1993) it can also be important ecologically. Migrations require motility, which can limit which organisms can perform DVM. Among the plankton, motility is generally

8 considered a specialty of the flagellates (Smayda 1997b), though buoyancy can also be used (Villareal 1988). There are a number of ultimate drivers of DVM in phytoplankton, including avoidance of predators (e.g. see ‘cascading migrations’ hypothesis of Bollens et al. 2011, and Bollens et al. 2012), nutrient acquisition at depth during water column stratification (e.g. Cullen 1985), avoidance of currents (e.g. Crawford and Purdie 1992), and photosynthetic optimization (e.g. Ault 2000). Mechanistically too, there are a number of possible drivers of DVM in phytoplankton, including phototaxis, chemotaxis, geotaxis, and the circadian clock (Hasle 1950, Byrne et al. 1992, Kamykowski et al. 1998b). The signal transduction pathways for all of these mechanisms remain unknown. Although vertical migrations in phytoplankton have been studied for a long time (e.g. Gran 1915), a full understanding of DVM remains enigmatic in part due to the diversity of species that participate, and the complexity of the behavior itself. Specifically, the behavior is plastic, and may represent the integrated physiological status of the cells, while using multiple unknown signaling pathways. Thus, understanding both the patterns of DVM in different environments, as well as the mechanisms that drive the behavior in harmful algae, may help to reveal greater detail of their realized niche and provide a better prediction of their spatial distribution over time.

1.4 Control of HABs Understanding core HAB biology and ecology is essential for appropriate management direction. Once sufficient perspective is gained on the issue, there are a variety of different options (and combinations of options) that may be available to tackle problems associated with HABs. As noted by Hoagland and Scatasta (2006), if the costs of active management intervention are higher than the costs imposed by the

9 HAB, a viable option would be to do nothing. Otherwise, efforts are either preventative or reactive, and are then either targeted toward remedying the physical entity (i.e. environments that promote HABs, or HABs themselves) or toward alleviating the harmful effects. The main preventative effort to alleviate the harmful effects of HABs is monitoring, which can provide effective direction to management (Rhodes et al. 2001, Trainer and Suddleson 2005). The main preventative effort to alleviate the occurrences of HABs has been to limit nutrient loading, which has also seen successes in certain environments (Moore and Christensen 2009) but is difficult and expensive to implement, and may not be effective in all systems (Zingone and Envoldsen 2000, Heisler et al. 2008, Smith and Schindler 2009). More direct preventative approaches to remedy environments that promote HABs include the ‘fixation’ of macronutrients, such as phosphate, into refractory forms (e.g. Lürling and Oosterhout 2013). There are fewer reactive options available to limit the harmful effects of a HAB, but one effective example is the moving of finfish cages away from a HAB (Rensel and Whyte 2003). However, this action does not address other harm inflicted by a HAB (e.g. toxicity to wild animals). Lastly, reactive efforts to address the actual HAB itself are available (so called HAB ‘control’), and are increasingly investigated. Examples of HAB control include physical methods (e.g. mixing, Jungo et al. 2001), clay flocculation (Sengco and Anderson 2004), chemical methods (e.g. copper sulfate (Rounsefell and Evans 1958) and H2O2 (Matthijs et al. 2012)), biologically derived compounds (Shao et al. 2013), biological methods such as microalgal grazers (Jeong et al. 2008), algicidal bacteria (Kang et al. 2011), macroalgae (Tang et al. 2014), and viruses (Brussaard et al. 2005). All of the above mentioned strategies are viable; and many would be complementary within a holistic

10 approach to management, and so continued research in all directions will be valuable. Methods using biological agents and their derived bioactive compounds are typically very nascent in terms of progression toward implementation, but tend to offer a number of unique and desirable characteristics, most importantly potency and species specificity.

1.5 Dissertation Chapters Chapters two and three of this dissertation investigate the control of dinoflagellates with an allelopathic filtrate (termed IRI-160AA) derived from a bacterium, Shewanella sp. IRI-160. This bacterium was isolated from the Delaware Inland Bays, and was previously found to inhibit dinoflagellate population growth in culture (Hare et al. 2005). Chapter two investigates how IRI-160AA affects dinoflagellate physiology, mortality, and growth. Specifically, this chapter addresses photosystem II (PSII) function with active chlorophyll fluorescence, the timing of PSII inhibition and membrane permeation, the effects of IRI-160AA dose on mortality and PSII function, and the effect of darkness on dinoflagellate susceptibility to the filtrate. Chapter three tests IRI-160AA on natural plankton communities to assess dinoflagellate susceptibility and changes to the microbial community. Molecular techniques were used to quantitatively and qualitatively assess community structure and species/class/phyla abundance. Chapter four departs from the allelopathic filtrate work, describing and comparing the DVM of two harmful algae isolated from the Delaware Inland Bays, Karlodinium veneficum and Chattonella subsalsa, in laboratory columns. DVM and PSII photobiology were compared in short and long term experiments. Detected rhythms were tested in continuous light or dark, for evidence of free-running rhythms.

11 Furthermore, measurements of growth and carbon assimilation were used to help decipher the differences in migration patterns between species.

12 Chapter 2

GROWTH, DEATH AND PHOTOBIOLOGY OF DINOFLAGELLATES UNDER BACTERIAL-ALGICIDE CONTROL

2.1 Abstract Naturally occurring allelopathic compounds, specific to some phytoplankton, may be a good source of bio-control agents against microalgae responsible for harmful algal blooms (HABs). Global expansion of HABs has invigorated research into different approaches to control these algae, including the search for naturally derived algicidal compounds. Here we investigated the effects of a filtrate from the algicidal marine bacterium Shewanella sp. IRI-160 on photochemical function of four cultured dinoflagellates, Karlodinium veneficum, Gyrodinium instriatum, Prorocentrum minimum, and Alexandrium tamarense. The filtrate (designated IRI-160AA) contains bioactive compound(s), which were recently shown to inhibit growth of several dinoflagellate species. Results of this study show that all dinoflagellates but P. minimum exhibited photosystem II (PSII) inhibition, loss of photosynthetic electron transport, and varying degrees of cellular mortality. Exposure assays over 24 h showed that PSII inhibition and loss of cell membrane integrity occurred simultaneously in G. instriatum, but not in K. veneficum, where PSII activity declined prior to losing outer membrane integrity. In addition, PSII inhibition and population growth inhibition were dose-dependent in K. veneficum, with an average EC-50 of 7.9% (v/v) IRI-160AA. Application of IRI-160AA induced significantly higher PSII inhibition and cell mortality in K. veneficum subjected to continuous darkness as compared to cells

13 maintained with 12:12 h light:dark cycles, while no such dark effect was noted for G. instriatum. The marked differences in the rate and impact of this algicide suggest that multiple cellular targets and different cascades of cellular dysfunction occur across these dinoflagellates.

14 2.2 Introduction It is widely accepted that harmful algal blooms (HABs) are increasing (Hallegraffe 1993, Van Dolah 2000, Anderson et al. 2002), and that certain HAB species are expanding their biogeographic ranges (e.g. Mcleod et al. 2012). Anthropogenic eutrophication (Anderson et al. 2002, Anderson et al. 2008, Heisler et al. 2008), ballast water transportation (Hallegraeff 1993, Bolch and de Salas 2007, Smayda 2007) and potentially climate change (Moore et al. 2008, Hallegraeff 2010, Mcleod et al. 2012) are all thought to contribute to a continued global expansion of HABs. Irrespective of the proximal causes, these blooms will continue to increase environmental and economic costs, which are currently estimated at $82 M and $813 M annually in the US and European Union, respectively (Hoagland and Scatasta 2006), thus highlighting the need for effective management solutions. HAB management is broadly segregated by whether methods target solutions for the present or the future, and then by whether the actions are direct or indirect. Actions that aim to prevent or reduce HAB occurrences in the future are termed ‘prevention’ approaches. The most common form of prevention aims to restrict nutrient loads to aquatic systems to reduce anthropogenic eutrophication. However, in estuarine and marine waters, successful prevention of HABs by nutrient management may be less straightforward than for freshwater HABs because of a wider variety of HAB species and environmental conditions (Smith and Schindler 2009). Furthermore, Heisler et al. (2008) note that enacting effective policies for nutrient restrictions is quite challenging and costly and would require adaptive management to account for imperfect understanding of HAB dynamics (see also Zingone and Enevoldsen 2000). Likewise, some HABs are not linked to anthropogenic eutrophication and therefore would not respond to restricted nutrient releases (e.g. Bachmann et al. 2003, Anderson et al.

15 2008). Recognizing these difficulties, it is imperative to pursue more immediate strategies as well. In this regard, actions that aim to minimize the impacts of HABs (e.g. aquaculture losses, human illnesses) in the present or immediate future, are termed ‘mitigation’ approaches, and are not mutually exclusive from yielding ‘prevention’. Mitigation approaches are further sub-divided into either direct actions that reduce or contain a HAB population, termed “control”, or indirect/passive actions to ameliorate HAB impacts, which we term “non-control mitigation”. Non-control mitigation, such as moving finfish cages to avoid fish kills (Rensel and Whyte 2003), and monitoring to prevent the harvesting of toxic organisms (Zingone and Enevoldsen 2000, Trainer and Suddleson 2005) has proved successful. However, non-control mitigation is sometimes impractical and costly, or too passive to ameliorate certain impacts while a bloom still exists (e.g. aerosolized toxins or toxicity to wild animals). Consequently, under these circumstances control mitigation may be more effective. Control mitigation encompasses many different chemical, physical and biological strategies (Kim 2006, Sengco 2009). Physical controls, such as mixing (accomplished by bubbling) alleviated Microcystis sp. blooms in a lake (Jungo et al. 2001), and clay flocculation has been successful at removing a wide variety of HAB species (Sengco and Anderson 2004). Chemical controls, such as copper sulphate additions (Rounesefell and Evans 1958), and hydrogen peroxide (Matthijs et al. 2012) have also been effective. Often however, broad toxicity is cited as an obstacle to real- world use (Boylan and Morris 2003, HAB RDDTT 2008). Recent discoveries of less broadly-toxic chemicals have shown promise, such as L-Lyseine which has been used against Microcystis spp. (Takamura et al. 2004) and Thiazolidinediones (Kim et al. 2010). In addition, a number of chemicals effective against HAB species have been

16 derived from biological origins, such as unknown compounds in barley straw extracts (Terlizzi et al. 2002), polyphenols from tea leaf extracts (Lu et al. 2013), rhamnolipid biosurfactants from Pseudomonas aeruginosa (Wang et al. 2005) and prodigiosin from the bacterium Hahella chejuensis (Jeong et al. 2005). Lastly, biological control techniques have been successful in controlling certain HAB species in controlled incubations, primarily via algicidal bacteria (Kim et al. 2008, Kang et al. 2011, Paul and Pohnert 2012) and micrograzers (Jeong et al. 2008). To date, no single approach has been truly championed, and the unique advantages of these many strategies have warranted continued examination. Hare et al. (2005) described a bacterium, Shewanella sp. IRI-160, that had a growth inhibiting effect on three dinoflagellates (Prorocentrum minimum, Pfiesteria piscicida, and Gyrodinium uncatenaum), but had no growth inhibiting effects on four non-dinoflagellate species. A recent investigation by Pokrzywinski et al. (2012) found that the algicidal activity of Shewanella sp. IRI 160 was due to a thermally stable, polar and water soluble compound or compounds secreted by the bacterium. The agent(s) accumulated in the medium, such that cell-free bacterial filtrates induced algicidal and growth inhibitory effects, thereby negating the need to apply live bacteria. Furthermore, algicidal activity was significantly greater when applied to dinoflagellates in logarithmic growth, as compared to stationary or lag-phase cultures. The specific cellular targets, and mode of action of this secreted algicide is currently under investigation. Testing of naturally derived algicides and allelochemicals in-vitro has revealed that many act on specific components of the photosynthetic apparatus of susceptible algal cells. For example, free fatty acids disrupt membranes, resulting in the

17 dissociation of phycobilins (but not integral chlorophylls) from the thylakoid membranes of cyanobacteria (Wu et al. 2006). Polyphenolic algicides produced by a macrophyte inhibit photosystem II (PSII hereafter) in Anabaena sp. by interfering with electron transport between the primary and secondary quinones QA and QB in the PSII reaction center (Leu et al. 2002). Likewise, the compounds Fischerellin A and Fischerellin B, from the cyanobacterium Fischerella muscicola target multiple sites on the electron transport chain surrounding PSII (Smith and Doan 1999), and

Cyanobacterin, derived from Scytonema hofmannii likely targets QA on the acceptor side of PSII (Gleason and Case 1986). Lastly, rhamnolipid biosurfactants from the bacterium Pseudomonas aeruginosa inactivated PSII in a range of HAB taxa (Gustafsson et al. 2009). Furthermore, Shi et al. (2009) found that algicidal agents produced in-vivo were also capable of inhibiting the photosynthetic apparatus, showing that the algicidal bacteria Pseudomonas mendocina caused PSII inactivation in Aphanizomenon flos-aquae. In the current study, we explored whether the algicide IRI-160AA affects PSII function in a range of cultured dinoflagellates and report on the timing of membrane permeability in affected cells.

2.3 Methods

2.3.1 Bacterial Filtrate Preparation

Bacterial filtrates were prepared following Hare et al. (2005) and modifications detailed in Pokrzywinski et al. (2012). Briefly, Shewanella sp. IRI-160 was plated onto LM media agar without antibiotics. Plates were incubated at room temperature until colony formation, whereupon a colony was transferred to 100 ml of liquid LM media without antibiotics and incubated at 25 °C on an orbital shaker at 100 RPM for 18 h. Cultures were split into two, centrifuged, and washed twice in f/2 seawater algal medium (Guillard and Ryther 1962). The bacteria were resuspended in 80 ml fresh f/2, and incubated at 30°C for one week. Finally, the culture was centrifuged, and the supernatant was filtered (0.2 µm) and stored at -80°C until later use. The filtrate was thawed prior to use and then stored at 4°C during subsequent applications. Following Pokrzywinski et al. (2012), we refer to this filtrate as IRI-160AA.

2.3.2 Algal Species Five algal species were tested for photochemical inhibition with the algicide: 4 dinoflagellates [Karlodinium veneficum (CCMP 2936), Gyrodinium instriatum (CCMP 2935), Prorocentrum minimum (CCMP 2233), Alexandrium tamarense (CCMP 1493)], and 1 cryptophyte, [Rhodomonas sp. (CCMP 757)]. Non-axenic batch cultures were grown in autoclaved 20 PSU f/2 medium at 24°C, under 150 µmol photons m-2 s-1 on a 12:12 light:dark cycle provided by cool-white fluorescent lights. For all experiments, algal cultures were sampled during logarithmic growth, as this

19 growth phase shows the greatest susceptibility to the algicidal and growth-inhibitory effects of IRI-160AA (Pokrzywinski et al. 2012).

2.3.3 Species Testing Tests comparing five species were carried out aseptically in autoclaved 15 ml glass test tubes, with independent replication. Treatments and controls were run in triplicate. Treatment cultures included addition of 4% (v/v) IRI-160AA while controls included 4% (v/v) f/2 medium. Logarithmic-stage cultures were inoculated mid light- cycle (of a 12:12 L:D) and were sampled for both photochemistry (see below) and cell density after 18 h of exposure and then in three subsequent 24 h intervals (= 42, 66, 90 h). The 18 h sampling point was chosen because this was previously noted as a time of peak inhibition in other physiological variables (Pokrzywinski et al. submitted). Cells were fixed with Lugols Iodine solution, and later enumerated by light microscopy in a Neubaur Haemocytometer or Sedgwick-Rafter cell. Percent cell loss was calculated as (number of cells lost / initial abundance) x100.

2.3.4 Algal Photochemistry

PSII function was assessed by fast repetition rate fluorometry (Kolber et al.

1998; FASTtracka II and FASTact system, Chelsea Instruments, UK). A series of 100 subsaturating LED ‘flashlets’ at ~1 µs intervals were applied to generate a single- turnover fluorescence transient. Fluorescence transients were used to evaluate three primary parameters: 1) the maximum quantum yield of PSII (Fv/Fm = Fm-Fo/Fm), 2) the rate of primary quinone re-oxidation and electron transport out of PSII (τ; QA- à QA), and 3) the photochemical connectivity between PSII reaction centers (ρ) (Kolber et al.

20 1998, Cosgrove and Borowitzka 2011). The Fv/Fm ratio describes the maximum efficiency of PSII for converting absorbed light into a photochemical charge separation, and lower Fv/Fm values can be used to infer PSII inactivation (i.e. damage) or down-regulation of PSII reaction centers. Tau (τ) describes the time constant of electron transfer on the acceptor side of PSII, and specifically describes the time required to re-oxidize the reduced primary quinone (QA-) measured in µs. PSII connectivity (ρ) describes the relative capacity to re-direct excitation energy from closed reaction centers, to nearby open PSII reaction centers. Prior to fluorescence measurement, samples were dark acclimated at room temperature for ≥20 minutes to re-open PSII reaction centers and relax non- photochemical quenching. The FRR sample chamber was maintained at 24 °C. The fluorescence measurement protocol consisted of 3 consecutive acquisitions, with each acquisition consisting of 40 repetitions of the following sequence: 100 flashlets at 1 µs intervals to reach saturation, and 50 flashlets at 49 µs intervals to record relaxation kinetics. Curve fitting software provided with the instrument (FASTpro v3.0, Chelsea

Instruments, UK) was used to derive Fv/Fm, τ, and ρ. All curve fits and fluorescence transients were manually inspected in real time.

2.3.5 Dose Responses

The dose response of K. veneficum to IRI-160AA was assessed in triplicate at final concentrations of 2, 4, 6, 8, 12 and 16% (v/v), alongside an 8% (v/v) f/2 control.

Fv/Fm and cell density were measured after 18 h of incubation. The control was limited to an 8% concentration because a dilution series of f/2 concentrations was expected to yield identical results. Fv/Fm and cell density were converted to percent inhibition of the f/2 control ((treatment / average-control) x100)). Prior to the conversion of cell

21 density to percent inhibition, the average control cell densities were corrected to account for the differing dilutions used relative to the control using:

% !"#$!%#&! − 8 !"#$%&# 8% !"#$%"& !"## !"#$%&' − × !"#$%"& !"## !"#$%&' 100

The Fv/Fm ratio is unaffected by cell density and does not require this dilution correction. In order to test whether IRI-160AA activity was affected by dinoflagellate cell density, a second experiment was conducted which compared IRI-160AA activity against dilutions of K. veneficum. IRI-160AA was added at a 4% (v/v) final concentration, in triplicate to 3 dilutions of K. veneficum culture (no dilution, 1:3 dilution and 2:3 dilution in f/2 medium). Control cultures included addition of 4%

(v/v) f/2 medium to undiluted culture. Measurements of cell density and Fv/Fm were made after 18 h.

2.3.6 Membrane Permeability Outer membrane permeability in relation to IRI-160AA activity was investigated in K. veneficum and G. instriatum with the fluorescent dye, SYTOX Green (Invitrogen, CA, USA; hereafter referred to as SYTOX). SYTOX fluoresces brightly when bound to DNA but cannot pass through intact membranes, and can therefore be used to indicate the presence of dead cells in which the membrane has been compromised. Treatments included 8% (v/v) IRI-160AA additions, and controls included 8% (v/v) f/2. Subsamples were collected every 3 h for 24 h. SYTOX was added to subsamples in duplicate at a 1 µM final concentration and incubated for 90

22 min in the dark, and fluorescence was measured on a microplate reader with excitation and emission at 485nm and 520nm respectively (FLUOstar Omega, BMG Labtech). The cellular origin of SYTOX fluorescence was verified using an epi-fluorescence microscope (Microphot-FXA, Nikon, Japan). Total SYTOX fluorescence values were blank corrected (f/2 medium + 1 µM SYTOX) and converted to SYTOX fluorescence per cell. The SYTOX fluorescence per cell of IRI-160AA treatments were converted to a percent of the average f/2 control fluorescence per cell (which did not vary significantly over 24h, 1-way ANOVA, n=3, P>0.05).

2.3.7 Dark Incubation To test how light might influence the susceptibility of K. veneficum, and G. instriatum to IRI-160AA, we compared the effect of IRI-160AA incubations in the dark to IRI-160AA incubations in light:dark (L:D) cycles. Using cultures of K. veneficum and G. instriatum that were grown under 12:12 L:D cycles, two 4% IRI- 160AA treatments were tested. One of these treatments was transferred to continuous (24h) darkness, and the other treatment was kept in the regular 12:12 light:dark cycle. Controls included addition of 4% f/2, with one control transferred to continuous darkness, and the other kept in 12:12 light:dark cycles. Both treatments and both controls were performed in triplicate and transferred to their respective light/dark treatments immediately after IRI-160AA or f/2 addition, at approximately mid light- cycle. Cell density and photochemistry were measured at 0, 18, 42, 66 and 90 h after addition of IRI-160AA.

23 2.3.8 Statistical Analyses Statistical analyses were performed in Prism 5 (GraphPad Software Inc., CA, USA). All experimental treatments tested were independent with small balanced sample sizes (n=3). Different treatments were analyzed by 2-way ANOVA with time and treatment as independent variables, and were followed by Bonferroni post hoc tests. The culture dilution experiment was tested by one-way ANOVA and the Tukey’s HSD post hoc test.

24 2.4 Results

2.4.1 Species Comparison

All 4 species of dinoflagellates showed significant growth inhibition with IRI- 160AA, whereas the cryptophyte, Rhodomonas sp., responded with an increased growth rate and a higher maximal abundance relative to f/2 controls (Table 2.1 and Fig. 2.1a–e). Despite all dinoflagellates showing susceptibility to IRI-160AA by growth inhibition, there were species-specific responses in the relative magnitude and timing of cell mortality, growth inhibition and recovery (Fig. 2.1a-d, Table 2.1), and photochemical inhibition and recovery (Fig. 2.1f-i, Fig. 2.2). Gyrodinium instriatum and K. veneficum showed similar levels of susceptibility to the algicide (54 ±19% and 50±20% cell loss respectively), with some recovery or stabilization in cell density occurring after 42 h and 66 h respectively (Fig. 2.1, Table 2.1). Alexandrium tamarense exhibited a continuous decline in cell density before reaching a maximum cell loss of 7% (±10%), at 90 h (Fig. 2.1c, Table 2.1). P. minimum did not decline in abundance during the experiment, but did exhibit growth inhibition relative to the control (Fig. 2.1d, Table 2.1). Gyrodinium instriatum, K. veneficum, and A. tamarense exhibited similar levels of PSII inactivation after exposure to IRI-160AA. Maximal inhibition occurred at 18 h for algicide-treated cultures of both K. veneficum and G. instriatum, and at 66 h in A. tamarense (Fig. 2.1f, g, h). Recovery from PSII inactivation in remaining cells in the treatment cultures occurred by 42 h in G. instriatum and by 66 h in K. veneficum (Fig. 2.1f,g), while A. tamarense showed negligible recovery between 66 h and 90 h (Fig. 2.1h). Prorocentrum minimum showed no significant PSII inactivation (2-way ANOVA, P>0.05, Fig. 2.1i) after exposure to IRI-160AA. Rhodomonas sp. treated

25 with IRI-160AA exhibited no significant difference in Fv/Fm compared to the control (Fig. 2.1j). Following the species’ responses in Fv/Fm, PSII re-oxidation (τ) in the algicide-treated samples took significantly longer than the f/2 controls in G. instriatum, K. veneficum, and A. tamarense (2-way ANOVA, P<0.05), and was also observed in P. minimum (Figure 2.2a). Temporally, τ was altered in G. instriatum and K. veneficum only after 18 h of exposure, and recovered to control levels (0% horizontal line) by 42 h. In A. tamarense, τ remained significantly slower as compared to the controls until the end of the experiment (Fig. 2.2a), and in P. minimum τ was inhibited after 90 h only. Connectivity (ρ) was significantly lower only in K. veneficum and A. tamarense (2-way ANOVA, P<0.05), and recovered by 42 h in K. veneficum, and did not recover in A. tamarense (Fig. 2.2b). Conversely, ρ was significantly higher relative to the control in P. minimum after 90 h when exposed to the algicide (2-way ANOVA, P<0.05, Fig. 2.2b). In Rhodomonas sp. treated with IRI- 160AA τ was significantly faster at 42 and 90 h, and ρ was significantly lower than the conrol at 90 h.

2.4.2 Dose Response K. veneficum exhibited a typical dose response in photochemical inhibition and cell density with increasing IRI-160AA concentrations from 2% to 16% (Fig. 2.3).

The percent inhibition using Fv/Fm values was significantly lower than for cell density only at a 4% concentration (2-way ANOVA, n=3, P<0.05). The average EC-50 calculated from the fitted dose-response curves was 7.0% and 8.7% (v/v) IRI-160AA when calculated using cell density and Fv/Fm respectively. Final IRI-160AA concentrations of 12 to 16% resulted in near to 100% photochemical and cellular

26 inhibition relative to the control after 18h. The time required for recovery in growth and Fv/Fm was also dose dependent (i.e. longer at higher doses), as was relative membrane permeability as determined by SYTOX fluorescence (higher SYTOX fluorescence was noted at higher doses; Fig. A1). In contrast to the effects of IRI- 160AA concentration, modifying algal cell density by dilution had no effect on the relative inhibition in photochemistry or cell density after 18 h exposure to IRI-160AA at 4% (v/v) concentration (Fig. 2.3 inset). Cell densities and Fv/Fm in all IRI-160AA treatments were, however, significantly lower than the control in the algal dilution experiment (n=3, 1-way ANOVA & Tukey test, P<0.05).

2.4.3 Membrane Permeability When photochemistry was followed with greater temporal resolution, K. veneficum exhibited moderate PSII inactivation after 18 h exposure to the algicide (Fig. 2.4a), whereas G. instriatum showed large and immediate (< 3 h) PSII inactivation (Fig. 2.4b). Growth for K. veneficum appeared to slow by 6 h after addition of IRI-160AA and cell density was significantly lower than the controls between 18 and 24 h (Fig. 2.4c), whereas a large immediate decline in cell density occurred within the first 6 h of exposure to the algicide for G. instriatum (Fig. 2.4d).

By 9 h, there was a significant increase in cell membrane permeability (detected by SYTOX green) in algicide-treated K. veneficum, which peaked abruptly after 24 h (Fig. 2.4e). In comparison, significant cell permeability was detected in G. instriatum immediately after IRI-160AA addition, which then peaked earlier near 12 h (Fig. 2.4f). For G. instriatum, the timing and relative changes in SYTOX fluorescence correlated well with the relative changes in Fv/Fm and cell density. In contrast, the

27 main decline in Fv/Fm in K. veneficum at 18 h preceded the maximum permeability by 6 h, but did coincide with the commencing decline in cell density between 15 and 24 h.

2.4.4 Darkness Karlodinium veneficum and G. instriatum showed different responses to darkness both in the presence and absence of IRI-160AA. Compared to light controls, dark K. veneficum controls showed photochemical inhibition and stunted growth by 42 h, and precipitous declines in both photochemistry and cell number by 90h (Fig 2.5a,b). Contrarily, dark G. instriatum controls showed a small but consistent downregulation of PSII photochemistry after 18h and showed inhibited growth from 42h, but neither photochemistry nor cell density showed the precipitous declines by 90 h as in K. veneficum. Dark IRI-160AA treatment of K. veneficum induced greater PSII inactivation after 18 h (Fig. 2.5a), and a larger loss in cell density after 42 h (Fig. 2.5b) as compared to light IRI-160AA treatments. Neither cell density nor PSII inactivation recovered in K. veneficum treated with IRI-160AA and kept in continuous darkness. Contrary to the dark response of K. veneficum, G. instriatum showed no significant increase in cellular mortality in dark algicide applications relative to those kept in light:dark cycles (Fig. 2.5d). However, G. instriatum did show significantly more PSII inactivation by 42 h in dark IRI-160AA treatments, and like K. veneficum failed to fully recover from the initial PSII inactivation (Fig. 2.5c) in the dark IRI-160AA treatments.

28 2.5 Discussion A large proportion of algicides and allelochemicals derived from marine and freshwater microalgae appear to target competitor phytoplankton cell membranes, often simultaneously inactivating PSII and photosynthetic electron transport, and inhibiting growth (Smith and Doan 1999, Legrande et al. 2003 and references therein). Here we have employed analysis of single turnover chlorophyll fluorescence to evaluate PSII inhibition in dinoflagellates under algicidal treatment, using a previously described filtrate from the bacteria Shewanella sp. IRI-160 (Hare et al. 2005, Pokrzywinski et al. 2012).

2.5.1 General Dynamics of Photobiology & Cell Number

With the exception of P. minimum, we found a significant decline in the maximum quantum yield of PSII (Fv/Fm) during algicide incubations, and observed declines in cell density and inhibition of growth in all dinoflagellates tested.

Significant loss in PSII photochemistry was accompanied by inhibition in τ (QA- re- oxidation) and ρ (PSII connectivity). Broadly, these results describe the disruption of photosynthetic electron transport (PET) through PSII. While we cannot ascertain temporal loss in carbon fixation from this data, such reduction in PET would most likely lead to a substantial loss in ATP and NADPH necessary for carboxylation and other cellular processes. Although the relative magnitude of PSII inactivation showed some variability among experiments, we are confident in our species-specific differences because repeated experiments returned the same trends on average. We believe some of this variability may stem from natural variations in the production and accumulation of the bioactive compound(s) in each batch of Shewanella sp. IRI-160.

29 The recovery of Fv/Fm back to control levels after incubation with 4% IRI- 160AA occurred within 90 h in all but A. tamarense, and was considered to originate from a small population of less effected cells, while the majority were lysed and were no longer contributing to the active chlorophyll a fluorescence signal. In addition to the photochemical recovery seen in the residual population that avoided mortality at lower concentration algicide treatments, these cells were also capable of cell division. Clearly the dynamics of any population re-growth would be complicated further if these experiments were carried out in natural parcels of water that contained other competing phytoplankton as well as micro and mesograzers.

2.5.2 Species Specificity & Membrane Exposure Pokrzywinski et al. (2012) concluded that thecate dinoflagellates exhibited reduced levels of algicidal activity compared with naked dinoflagellates. The results presented here corroborate this pattern, with K. veneficum and G. instriatum exhibiting similar declines in cell density and photochemistry, whereas the thecate dinoflagellate A. tamarense exhibited reduced susceptibility to IRI-160AA with a lag in the photochemical response. However, the lack of any photochemical response with algicide treatment yet significant reduction in cell growth noted for P. minimum provides evidence that the algicidal mode of action across these different dinoflagellates is not ubiquitous. Furthermore, the thecal morphology of A. tamarense and P. minimum are very different, with A. tamarense having many similarly sized thecal plates (Fukuyo 1985), and P. minimum comprised primarily of 2 large valves (Faust et al. 1999). It is possible that plasma membrane access is required for the algicidal compound(s) to be effective against chloroplast function or that P. minimum is only susceptible to the algicide by an as yet unknown cellular pathway. Several

30 examples in the literature of algicide resistance in organisms with physical barriers lend support to the thecae protection hypothesis. For example, reduced cytotoxicity from fatty acids in Anabaena P-9 heterocysts compared to vegetative cells implicated a protection mechanism of the heterocyst’s thick cell wall (Wu et al. 2006). Secondly, the cellulose plates of the dinoflagellates, Cochlodinium polykrikoides, P. minimum, and Prorocentrum micans, were thought to reduce the speed and magnitude of algicidal effects by peptides, compared with the high activity on the relatively exposed plasma membranes of raphidophytes (Park et al. 2011). Thirdly, gram-positive bacteria were susceptible to a rhamnolipid biosurfactant, but a gram-negative bacterium was not, leading Sotirova et al. (2008) to suggest that lipopolysaccharides in the outer-membrane formed a barrier to surfactant activity. Further, the results of Sotirova et al. (2008) led Gustafsson et al. (2009) to propose that the complex cell wall of Microcystis aeruginosa (a gram negative harmful cyanobacteria) may contribute to this species’ resistance to a rhamnolipid biosurfactant as well. Another consideration is the different plastid origins of the dinoflagellates tested in this study, with a view to identifying possible links driving the species- specific susceptibilities and photosynthetic responses observed. Gyrodinium instriatum, P. minimum and A. tamarense share the putatively ancestral type plastid, with the photosynthetic pigment peridinin, a form II RuBisCo, and a highly reduced mini-circle genome, while K. veneficum has a more derived plastid with fucoxanthin, form I RuBisCo, and a full circular plastid genome (Daugbjerg et al. 2000, Yoon et al. 2005, Wisecaver and Hackett 2011). In consequence, the fewer plastid encoded genes, and possibly less redox regulation of remaining plastid genes in peridinin-containing dinoflagellates, require a higher degree of nuclear encoded plastid-bound protein

31 translocation and are therefore possibly more susceptible to plastid damage if inter- organelle signaling and plastid protein targeting is impaired by IRI-160AA. However, this relationship between plastid genome reduction and algicide susceptibility is likely complex, as evidence suggests that K. veneficum has a chimeric proteome derived from nuclear encoded secondary (peridinin-based) and tertiary (fucoxanthin-based) endosymbioses (Patron et al. 2006). If there is an influence of plastid origin, then the importance of the plastid would likely depend upon the extent of endosymbiotic gene transfer, and in particular, how many and which genes have been transferred to the nucleus.

2.5.3 Dose Responses The small but significantly greater inhibition based on cell density as compared to Fv/Fm at 4% (Fig. 2.3) is difficult to interpret, but may be due to highly susceptible cells lysing quickly, with little to no time spent in an inhibited-state, such that photoinhibitory signals are not readily observed. The broad similarity between the dose response curves based on photochemistry and cell density however suggests that the increased inhibition based on Fv/Fm values is likely due to a larger number of affected cells, rather than from increased inhibition in a subset of affected cells. In further support of this, we observed that under a low IRI-160AA concentration (4%), a proportion of the population (found at the surface of the culture medium) retains a significantly higher photosynthetic competency than the integrated population as a whole (Fig. A2). Likewise, this also explains why photochemistry appears to recover by 42 h following treatment with 4% IRI-160AA (Fig. 2.1). It is important to stress here, however, that this recovery represents a substantially smaller population of cells than at the start of the experiment, and the number of residual photochemically active

32 cells is dose dependent (Fig. 2.3). Further testing is needed in order to discern if multiple applications of a low concentration of IRI-160AA may eliminate these residual cells as well. In a second dose response experiment, serial dilutions of algal culture were treated with 4% IRI-160AA to examine the possibility that the algicidal activity is dependent on cell density. Here, the degree of mortality and the decline in Fv/Fm did not vary between culture dilutions (Fig. 2.3 inset). This response differs from the results described in Mu et al. (2009), in which enhanced algicidal activity was observed by the algicidal bacterium, Ochrobactrum sp. FDT5, when inoculated into cultures with lower chlorophyll-a concentrations (Mu et al. 2009). Given the dose response elicited by increasing IRI-160AA concentration as described above, where higher concentrations induced higher mortalities and larger decreases in PSII efficiency, we conclude that IRI-160AA activity is independent of cell density within the range tested here.

2.5.4 Membrane Permeability and Fv/Fm When photochemistry and membrane permeability were compared in tandem, the two dinoflagellate species tested responded differently to IRI-160AA addition. G. instriatum had a simultaneous decline in Fv/Fm and cell viability, while K. veneficum lost chloroplast function before plasma membrane integrity, with peak PSII inactivation 6 h prior to the peak loss in cell viability. Two hypotheses could explain this discrepancy between the species. Firstly, that the location, concentration, and susceptibility of algicide target sites differ between the two dinoflagellates (Pokrzywinski et al. submitted). Alternatively, Pokrzywinski et al. (2012) identified 3 mass ions within the bacterial filtrate (with masses of 239.2, 301.2, and 470.1 daltons),

3 3 which may each contribute to algal mortality and/or growth inhibition. Given that multiple targets and multiple compounds are a common feature of allelopathy (Einhellig 1999), an alternative hypothesis is that differing susceptibility to each component may explain the differences observed between the two dinoflagellates. Confirmation of such hypotheses cannot be resolved without the isolation and identification of the bioactive compound(s) and subsequent identification of the specific cellular target sites.

2.5.5 Darkness There is a range in dark tolerance across phytoplankton taxa, including species known to cause harmful algal blooms (Smayda and Mitchell-Innes 1974, Peters and Thomas 1996, Furusato et al. 2004, Popels et al. 2007), thus it is not surprising that G. instriatum displayed a much higher degree of general dark tolerance than K. veneficum. However, it is important to note the significant difference between the responses of these two species when treated with the algicide under prolonged darkness, and how their response correlates to what appears to be their basal tolerance to prolonged dark exposure. While this area of research is limited, Mayali et al. (2007) have noted that the bacterial exposure while in prolonged darkness increased the rate of ecdysis in the dinoflagellate Lingulodinium polyedrum. Mayali et al. (2007) suggested that darkness may reduce the ability of L. polyedrum to produce antibiotics, and although not directly applicable here since live bacteria were not used, an analogous mechanism involving antagonistic/protective molecule production could be a possibility for K. veneficum as well. The more rapid loss of PSII photochemistry noted for IRI-160AA-treated K. veneficum in the dark could be due to either a direct amplification of chloroplast dysfunction or may point to an accelerated loss of cellular

34 function at another location. In this regard, disruption in mitochondrial activity and/or synthesis of chloroplast-targeted proteins in the cytoplasm could result in a similar decline in PSII activity as noted here. It is well known that homeostatic chloroplast protein repair requires some light and complete darkness can lead to greater inactivation (Bergo et al. 2003), however, our data may point to indirect disruption of chloroplast function as a better explanation. Notably, Pokrzywinksi et al. (submitted) have shown that IRI-160AA exposure in dinoflagellates induces several cellular markers indicative of programmed cell death (PCD), including extracellular release of hydrogen peroxide, caspase-like activity, and phosphatidylserine inversion of cellular membranes. Likewise, Segovia and Berges (2005) have noted that PCD in the chlorophyte Dunaliella tertiolecta was accelerated when cells were placed in prolonged darkness in the presence of the cytoplasmic protein synthesis inhibitor cycloheximide. Given that maintenance respiration in dinoflagellates can represent over half of the total dark respiration rate in some species, due in part to high protein turnover (Jauzein et al. 2011), and that disruption in mitochondrial activity may lead to down-stream loss in photochemistry (Saradadevi and Raghavendra 1992), it appears that IRI-160AA may affect K. veneficum mitochondrial activity to a greater degree than G. instriatum, and this leads to an even faster disruption in chloroplast electron transport activity in K. veneficum.

2.5.6 Conclusion Dinoflagellate responses to IRI-160AA were found to be species-specific, but included PSII inactivation, loss in membrane integrity, inhibition of growth, and mortality. In agreement with Pokrzywinski et al. (2012), thecae appeared to provide protection from algicidal attack, and also appeared to limit PSII inactivation. Due to

35 differences in susceptibility between A. tamarense and P. minimum, we suggest that the thecae-armoring hypothesis could be extended to include relative plasma- membrane exposure. In comparing the effects of IRI-160AA on K. veneficum and G. instriatum we found that peak PSII inactivation was decoupled from peak SYTOX permeability in K. veneficum but not in G. instriatum, and that darkness affected K. veneficum substantially more than G. instriatum. These discrepancies led us to speculate that physiological differences can substantially alter susceptibility to IRI- 160AA. Ultimately, the specificity of this algicide for certain dinoflagellates remains promising from an applied standpoint and further work on the mode of action, the efficacy of the algicide in-situ, and identification of the bioactive compounds are currently underway.

36 Table 2.1: Specific growth rates from Figure 2.1 (algal species tested with 4% IRI- 160AA or 4% f/2 medium). IRI-160AA growth rates and percent cell loss were calculated from T0 to the time point with the lowest cell number, except for P. minimum and Rhodomonas sp. which did not show a drop in cell density (N/a) and were calculated over 90 h. Percent cell loss was rounded to the nearest whole percentage with ±1 SD given in parentheses

Growth Rates µ (d-1) Cell Loss (%) Control IRI-160AA

G. instriatum 0.27 -0.26 54 (±19) K. veneficum 0.39 -0.40 50 (±20) A. tamarense 0.27 -0.01 7 (±10) P. minimum 0.21 0.08 N/a Rhodomonas sp. 1.00 1.06 N/a

Figure 2.4: 24-hour incubations with K. veneficum (left panes) and G. instriatum (right panes). Cultures were incubated with either 8% v/v IRI-160AA (filled squares) or 8% v/v f/2 medium (open circles). Fv/Fm (a,b), cell density (c,d) and SYTOX-fluorescence (e,f; relative to the f/2 control) are presented. Greyed areas behind the data represent the dark period. Asterisks indicate time points with a significant difference, as determined by two-way ANOVA and Bonferroni post-tests (P<0.05, n=3). Error bars represent ±1 SD of the mean.

Figure 2.5: Effects of shifting treated cultures into continuous darkness immediately upon addition of 4% (v/v) IRI-160AA in K. veneficum (a, b) and G. instriatum (c, d). Light treatments are indicated by open symbols, dark treatments are indicated by filled symbols, circles indicate controls and squares are IRI-160AA treatments. Asterisks indicate a significant difference only between IRI-160AA light and IRI-160AA dark treatments (2-way ANOVA and Bonferroni post-tests, P<0.05, n=3). Error bars are ±1 SD of the mean.

42 Chapter 3

EFFECTS OF A BACTERIAL ALGICIDE, IRI-160AA, ON DINOFLAGELLATES AND THE MICROBIAL COMMUNITY IN MICROCOSM EXPERIMENTS

3.1 Abstract

Many biological agents and biologically derived compounds show promise for use in controlling harmful algal blooms (HABs). Allelopathic filtrates from the bacterium Shewanella sp. IRI-160 (termed IRI-160AA) have been shown to inhibit population growth and kill a variety of dinoflagellates grown in culture. Here we tested the efficacy of IRI-160AA in microcosms initiated using three natural dinoflagellate blooms (Prorocentrum minimum, Karlodinium veneficum and Gyrodinium instriatum). We measured target dinoflagellate abundance, total chlorophyll-a, photosystem II (PSII) photochemistry, and changes to the prokaryotic and eukaryotic community composition over 2-3 days of IRI-160AA incubation. Naked dinoflagellates (K. veneficum and G. instriatum) were impacted more, while abundance of the thecate P. minimum was not affected. However, dinoflagellate growth inhibition was generally lower than that observed in uni-algal cultures, and took longer to occur. Eukaryotic community composition in IRI-160AA treated microcosms was significantly different from control incubations, and was driven predominantly by increases in heterotrophic protists (e.g. Euplotes sp. and Paraphysomonas sp.). Similarly, significant changes to the prokaryotic community structure were evident. Microcosms with G. instriatum and higher algicide

43 concentrations indicated that algicidal activity was enhanced in a dose dependent manner. Furthermore, total ciliate abundance as well as a bactivorous chrysophyte (Paraphysomonas sp.) increased in a dose dependent manner. Total diatom abundance increased at lower IRI-160AA concentrations, but increased less with increasing dose. Overall, the bio-activity of IRI-160AA on naturally occurring dinoflagellates in mixed natural microbial communities is encouraging from the applied perspective of using the active compound(s) in IRI-160AA as natural agent(s) to manage harmful dinoflagellate blooms.

44 3.2 Introduction The recognition that harmful algal blooms (HABs) are increasing in frequency, extent and geographic range (Hallegraeff 1993, Anderson et al. 2002) highlights the need to assess possible methods for the prevention, control, and mitigation of HABs or their negative effects. To date, no one approach has emerged as a universally applicable approach to managing HABs in nature. Of those mentioned, prevention is an instinctively attractive option, but it is considered a larger challenge than establishing effective control efforts, in part due to the complexity of blooms in nature (Jewett et al. 2008). Non-control mitigation such as monitoring toxin levels, and moving aquaculture pens is an important aspect for holistic HAB management, but may not be universally applicable, and may not ameliorate all impacts of a HAB (e.g. toxicity to wild animals). Thus, the need for cost effective, practical HAB control is expected to grow with the continued expansion of HABs and their associated impacts (Jewett et al. 2008, Anderson 2009). Indeed, despite the early “reticence” noted in the late 1990’s for HAB control (Anderson 1997, Anderson 2004) developments in this area have grown substantially in recent years (e.g. Kang et al. 2007, Kim et al. 2008, Kim et al. 2008b, Pan et al. 2011, Baek et al. 2013). Proposed strategies for HAB control include physical, chemical, and biologically based techniques (Kim 2006, Anderson 2009, Sengco 2009). A theoretically perfect control measure fulfills the four criteria of control efficiency (i.e. how effective population control is), price/economics (i.e. full costs of implementation), feasibility (i.e. can the control technique scale appropriately), and specificity (i.e. how many non-target organisms may be affected). The last of these criteria, species specificity, is important because community diversity maintains ecosystem services (Hooper et al. 2005, Isbell et al. 2011) and imparts resilience to

45 ecosystems (Folke et al. 2004) at both macro and micro scales of biological organization. Consequently, the specificity of a control measure could impact community function by changing how microbial communities re-structure after an implementation. To date, the most commonly tested control measures, such as clay flocculation and chemical controls, have fit the first three criteria well, but are deficient in the last criterion, the desired specificity (Kim 2010). For example, chemical controls, such as copper sulphate, are poor in this criterion due to taxonomically broad toxicity among the plankton (Jančula and Maršálek 2011). However, recent reports of reduced collateral losses through dose optimization of

H2O2 (Matthijs et al. 2012, Burson et al. 2014) shows that it is possible to optimize the specificity of certain chemical controls. Tests of clay applications on non-target organisms have been described as broadly positive, although evidence of species specific inhibition does exist and the extent of this inhibition depends on the species present, final loading levels and the mixing regime (Pan et al. 2011, Shumway et al. 2003, Archambault et al. 2004, Van Oosterhout and Lürling 2013). Sublethal stresses from clay applications have also been noted, which may increase macrofaunal biological oxygen demands, and will be important in comparing the finer aspects of different control measures (Orizar et al. 2013). In contrast to physical and chemical strategies, biological (or biologically derived) control strategies are often associated with high levels of species specificity, where proposed agents have included the use of parasites (Taylor 1968, Velo-Suárez et al. 2013), viruses (Baudoux and Brussaard 2005, Brussaard et al. 2005, Gobler et al. 2007), protozoan grazers (Jeong et al. 2008), and algicidal bacteria (Jung et al. 2008, Kim et al. 2008, Kim et al. 2009, Takamura et al. 2004). Additionally, a growing

46 number of biologically derived compounds (often allelochemicals) have been proposed for HAB control, e.g. stilbenes from plants (Mizuno et al. 2008), alkaloids such as nostocarboline from a cyanobacterium (Blom et al. 2006), polyphenols such as those derived from tea leaf extracts (Lu et al. 2013), and rhamnolipid biosurfactants from the bacterium Pseudomonas aeruginosa (Wang et al. 2005). To date however, biological or biologically derived control strategies are commonly associated with poor fulfillment of the first three criteria mentioned above (Anderson 1997, Shao et al. 2013). For example, in a review of available chemical compounds for use in cyanobacterial bloom management, Jančula and Maršálek (2011) suggest that the scarcity of field studies with biologically derived compounds is likely related to the high cost of intricate extraction or synthesis that is needed to acquire many of these compounds. Despite the current limitations, the potential of these organisms and natural products to deliver safe and specific control of target organisms, encourages continued assessment of biologically-based HAB control strategies. Continued assessment is especially important given the growing environmental safety requirements of developed and developing countries, which may curb the use of traditional algicides (Jančula and Maršálek 2011). The bacterium Shewanella sp. IRI-160 was originally found to be algicidal, inhibiting dinoflagellate population growth, and showing little effect on other microalgal classes tested (Hare et al. 2005). More recently, Pokryziwinski et al. (2012) corroborated this result, and showed that naked dinoflagellates were more susceptible than thecate species, and that susceptibility to the algicide was highest during active algal growth. Furthermore, the compound(s) responsible for bio-activity were secreted into the medium and were highly thermally stable. The use of a sterile bio-active

47 filtrate, instead of live bacteria to control dinoflagellates is useful from an applied perspective, as applications of exotic bacteria to non-indigenous waters are likely to meet regulatory hurdles (Anderson 2004). Tilney et al. (2014) corroborated the lower efficacy of IRI-160AA against thecate species, and found that algicidal activity was associated with disruption in photosystem II (PSII) photochemical activity, which along with population growth inhibition, was found to be dose dependent. However, susceptibility of the heterotrophic dinoflagellate Oxyrrhis marina to IRI-160AA (Pokryziwinski et al. 2012) indicates that PSII is not the only target site, and suggests that effects on PSII function may be a secondary effect in photosynthetic dinoflagellates. Although the mechanisms underlying cell susceptibility are still unknown, Pokryziwinski et al. (submitted) have observed cellular and biochemical changes that are consistent with programmed cell death in dinoflagellates treated with IRI-160AA. This implies that the mode of action is more complex than simple lytic activity, which may represent an advantage over other control techniques if the autocatalytic death process involved intracellular toxin transformation/degradation before the release of cellular contents during the final cell lysis. Uni-algal dinoflagellate cultures were a useful model for initial algicide characterization, but cannot reveal the effects of an algicide (or an algicidal bacteria) in complex natural communities. Other studies that assessed the efficacy of algicidal bacteria in mixed species assemblages have had mixed results, such that algicidal activity was dependent on the combinations of species present. For example, Kim et al. (2007) noted that both the ciliate, Stentor roeselii, and the algicidal bacteria, Streptomyces neyagawensis, were able to control the cyanobacterium Microcystis aeruginosa individually, but when applied together, they acted antagonistically and M.

48 aeruginosa density increased. In contrast, enhanced control of M. aeruginosa was observed when a different algicidal bacteria was applied with the same ciliate (Kim et al. 2007). In another example, Mayali and Doucette (2002) found that resistance to the algicidal bacterium, Cytophaga sp. 41-DBG2, was transmissible to susceptible algal cultures by transfer of the free-living bacteria from ‘resistant’ algal cultures. Moreover, bacteria may also positively or negatively affect algicidal activity through physiological effects on the target algae (Green et al. 2010, Mayali and Azam 2004). These examples highlight the complex ecological responses that can occur in natural waters following applications of biological control measures. The outcome of control measures using biologically derived compounds, rather than the living organisms, could also be affected by the microbial community. For example, in terrestrial plant allelopathy, soil bacteria mineralize allelopathic compounds (phenolic acids), thereby reducing their inhibitory effect on cucumber plant seedlings (Blum et al. 2000), and in marine waters, bacteria are known to transform paralytic shellfish toxins (Smith et al. 2001, Donovan et al. 2009). Eukaryotic cells too, are potent mediators of ecologically relevant chemical interactions in marine environments. For example, the allelopathic effects of a marine macrophyte on a harmful cyanobacterium were reversed by co- culture with a green microalga (Chang et al. 2012). In another example, Redshaw et al. (2011) showed that the diatom Skeletonema grethae effectively inhibited the toxicity of brevetoxin (PbTx-2) toward brine shrimp. In addition to biotic interactions, abiotic interactions such as oxidation may reduce the efficacy of allelopathic compounds (Weidenhamer and Romeo 2004). In the current study, we examined the bio-activity of IRI-160AA against naturally occurring dinoflagellates in their sympatric microbial communities, and

49 characterized the effects of IRI-160AA on the eukaryotic and prokaryotic community structures. We found that naturally occurring dinoflagellates were susceptible to IRI- 160AA, and that the structure of eukaryotic and prokaryotic communities were altered by IRI-160AA applications in a dose-dependent manner. The results continue to show promise from an applied perspective of using IRI-160AA as a control agent for harmful dinoflagellate blooms in nature.

50 3.3 Methods

3.3.1 Bacterial Filtrate Preparation

Glycerol storage stocks of Shewanella sp. IRI-160 held at -80 °C were used to aseptically inoculate modified LM agar plates (Sambrook et al. 1989, Luria Bertani medium, 20 g L−1 NaCl, 10 g L−1 tryptone, 5 g L−1 yeast extract, and 15 g L−1 agar) and were then incubated at room temperature. A bacterium colony was transferred to liquid LM medium, which was incubated at 24 °C on an orbital shaker (Lab Rotator; Barnstead Lab-Line, IL, USA). After reaching an optical density of 1.49 (measured with a NanoDrop 2000 Spectrophotometer; Thermo Scientific, DE, USA), cultures were centrifuged for 5 min at 5,000 RPM (200 mL volumes in Sorvall RC-5B Refrigerated Superspeed Centrifuge; DuPont Instruments, CT, USA) and re-suspended in 20 PSU f/2 medium to a final volume 80% that of the original. This wash was repeated and the culture was re-combined and incubated for 1 week in f/2 medium at 24 °C with daily re-suspension. Subsequently, cultures were centrifuged at 6,000 RPM for 10 min, and the supernatants were combined and autoclaved (20 min at 121 °C). The sterile supernatant, referred to as IRI-160AA after Pokryziwinski et al. (2012), was filtered through 0.2 µm nylon filters (Whatman POLYCAP AS or Fisherbrand 25 mm syringe filters) and stored at 4 °C. The autoclaved bacterial filtrates were tested on cultures of the sensitive dinoflagellates, G. instriatum and K. veneficum to ensure that they had significant algicidal activity.

3.3.2 Natural Waters, Incubation & Sampling Two experiments were conducted to examine the effectiveness of IRI-160AA on dinoflagellates in natural water samples. The first, referred to as “bloom experiments”, monitored the plankton daily over a period of 2 to 3 days after

51 inoculation with IRI-160AA at one concentration. The second, referred to as the “dose experiment”, monitored the plankton only once, after incubating with varying concentrations of IRI-160AA. Short experimental periods were used to minimize enclosure effects on the microbial communities. All experiments were conducted in the laboratory in triplicate or quadruplicate as described below. Control microcosms were prepared by addition of f/2 medium at volumes equivalent to the volumes of algicide added to the treatment microcosms. The use of f/2 nutrients in the algicidal filtrate (and for control incubations) facilitated a more direct comparison to previous work with dinoflagellate cultures, and increased the likelihood of stimulating dinoflagellate growth and hence susceptibility (Pokryziwinski et al. 2012). For the bloom experiments, three separate natural blooms were sampled. Sampling site information, environmental conditions (measured with a YSI sonde, 556 MPS) at the time of sample collection, and conditions provided for laboratory incubations are presented in Table 1. The 3 separate bloom experiments consisted of blooms of Prorocentrum minimum (600,000 cells mL-1), Karlodinium veneficum (20,000 cells mL-1), and Gyrodinium instriatum (6,000 cells mL-1). Due to the extremely high density of the P. minimum bloom, the water sample was diluted by half with 0.2 µm filtered sea water (collected from non-bloom water and not amended with nutrients) with a salinity of 25 (matching the bloom water). Environmental samples were prefiltered through a 209 µm mesh to remove meso-grazers and 350 mL or 1 L were incubated in glass Erylenmer flasks. Algicidal filtrates (for treatments) and f/2 medium (for controls) were warmed to room temperature and added to flasks immediately following a pre-addition sampling (“T0 samples”), to final concentrations (v/v) of 8 % (K. veneficum) and 10 % (P. minimum and G. instriatum). Flasks were sampled daily

52 for biomass (target cell counts, total chlorophyll-a, and in-vivo chlorophyll-a fluorescence), photophysiology (Fast Repetition Rate chlorophyll fluorescence), and molecular analysis (size fractionated into > 3 µm and 0.2-3 µm) as described below. For the dose experiment, three concentrations of IRI-160AA (v/v; 8 %, 16 % and 32 %) were used in triplicate 40 mL (final volume) experiments from a single environmental sample collected during a bloom of G. instriatum. The environmental sample used for this experiment was freshly collected one day following collection for the previously described G. instriatum bloom experiment and was also prefiltered through a 209 µm mesh. All treatment (IRI-160AA added) and control (f/2 medium added) incubations in this experiment were sampled after 42 hours for photophysiology, in-vivo chlorophyll-a fluorescence, and molecular analysis (size fractionated as above), described below.

3.3.3 Chlorophyll-a For samples containing the naked dinoflagellates K. veneficum and G. instriatum, known volumes of sample were filtered onto GF/F filters (Whatman, USA) and stored at -15 °C until extraction. Filters were extracted in 90% acetone for 24 hours in the dark at -15 °C and chlorophyll-a was then measured using a Turner AU- 10 Fluorometer (Turner Designs, USA). Samples of the thecate dinoflagellate P. minimum were extracted by centrifuging 20 mL volumes at 3,500 RPM for 4 min, discarding the supernatant and re-suspending the pellet in 5 mL of 90% acetone, transferred to darkness at -15°C for 24 h. Samples were then sonicated for 60 s (Misonix Sonicator 3000; Misonix Inc., USA) and centrifuged as above before measuring chlorophyll-a content of the supernatant (modified from Berden-Zrimec et al. 2008).

53 3.3.4 Active Chlorophyll-a Fluorescence Community photochemistry was monitored by active chlorophyll fluorescence with a FASTtracka II Fast Repetition Rate Fluorometer and associated laboratory FASTact system (Chelsea Instruments, UK). Samples were dark acclimated for > 30 min to allow all closed PSII reaction centers to open. Each fluorescence transient was inspected manually at the time of acquisition, and maximum photochemical efficiency of PSII (Fv/Fm), the connectivity among PSII reaction centers (PSII connectivity), and the maximum effective absorption cross-section (Sigma, nm2) were calculated from curve fits to the active fluorescence transient in real time with FASTpro software (V3.0, Chelsea Instruments, UK). Two to three acquisitions were performed on each sample, with each acquisition consisting of 40 sequences, of 100 saturation flashlets at 1µs interval, followed by 50 relaxation flashlets at 49µs intervals. Cuvettes were maintained at the same temperature used for incubation (18, 24, 29°C) by a water bath connected to the FASTact, preventing phytoplankton shock and changes to fluorescence.

3.3.5 Cell Enumeration

In the bloom experiments, target cells were enumerated microscopically using Lugol’s Iodine fixed samples (1% final concentration) within 6 weeks of preservation.

The smaller cells (K. veneficum, and P. minimum) were counted on a Neubaur Haemocytomer in quadruplicate, while the larger G. instriatum was counted in 30 µl aliquots on a Sedgewick-Rafter chamber in triplicate. In the dose experiment, total dinoflagellates, total diatoms, total ciliates, G. instriatum, Leptocylindrus sp., Cyclotella sp., and Paraphysomonas sp. were all quantified by qPCR (described below).

54 3.3.6 DNA Extraction, PCR-DGGE, and Sequencing Between 40 and 80 mL of sample was gently vacuum filtered (< 10 mmHg) sequentially onto 3 µm and 0.2 µm polycarbonate filters, which were then frozen at - 80 °C in CTAB extraction buffer (Dempster et al. 1999) that was previously amended with 20 ng/mL of a pGEM plasmid as an internal standard for extraction efficiency (Coyne et al. 2005). Following Dempster et al. (1999) and Coyne et al. (2005), frozen vials were heated to 65 °C for 10 min, and the DNA was extracted from the CTAB buffer via chloroform:isoamyl-alcohol (24:1), followed by a high salt, isopropanol / ethanol precipitation. Recovered DNA was re-suspended in 10 µL LOTE [3 mM Tris- HCl (pH 7.5), 0.2 mM EDTA] and quantified spectrophotometrically. Changes in eukaryotic and prokaryotic community composition were assessed by denaturing gradient gel electrophoresis (DGGE) of universally primed eukaryotic and prokaryotic PCR products respectively. The DGGE method fractionates PCR products based on sequence by electrophoresing the PCR products on an acrylamide gel with a gradient of denaturant, such that products with low melting temperatures migrate slower than products with high melting temperatures and hence over time become separated. The method typically detects DNA sequences with a copy number >1% of the total copy number present (Green et al. 2009). Each 50 µL PCR reaction contained, 1X PCR buffer, 0.2 mM dNTPs, 2.5 mM MgCl2, 0.05 µg/µL BSA, 0.2 µM each primer (see Table 2), 0.625 U JumpStart Taq (Sigma-Alrich, USA), 28.25 µL water, and 50 ng of template DNA. Thermal cycler settings followed a touch-down protocol consisting of an initial denaturation at 94 °C for 2 min; 20 cycles of 94 °C for 30 s, 65 °C for 30 s (-0.5 °C/cycle) and extension at 72 °C for 1 min; followed by 10 cycles as above but keeping a constant annealing temperature at 56 °C; and a final elongation step at 72 °C for 5 min. PCR amplifications (including negative controls)

55 were verified on 1 or 2 % (w/v) agarose gels for eukaryotic and prokaryotic reactions respectively. For eukaryotic DGGE, between 20 and 40 µL of PCR product (keeping identical volumes within experiments) was loaded onto 6% acrylamide gels with denaturing gradients between 20 and 50% (100% denaturant = 7M urea and 40% formamide), which were electrophoresed overnight at 100V at 60 °C on a D-code Universal Mutation Detection System (Bio-Rad, CA, USA). The same protocol was followed for prokaryotic reactions (only assessed in the dose experiment), except that gels were composed of 8% acrylamide, with a denaturing gradient from 20 to 60 %, and samples were electrophoresed for 6 h at 130 V at 60 °C. For G. instriatum and P. minimum eukaryotic bloom experiments, a target dinoflagellate marker lane was added, using PCR products from samples known to contain only DNA from the target algae. After electrophoresis, gels were stained for ~15 min in 5% (v/v) EtBr in DI- water, and briefly rinsed for several minutes in DI-water before imaging on a UV transilluminator (Alpha Innotech Corp., CA, USA). After immediate inspection of the DGGE banding patterns, bands of interest were stabbed with a sterile pipette, and used to inoculate 20 µL of PCR-water for sequencing. To sequence bands of interest, 5 µL of stab-inoculated water were amplified as above and supplementing extra cycles for amplifications with low product. These PCR products were then assessed by DGGE to ensure a single primary band was amplified. The PCR amplicons were purified (PCR Clean-up Kit, Qiagen or Sigma) and sequenced using the Big Dye Sequencing Ready Reaction Kit V 2.1 (Invitrogen, CA, USA) on an ABI PRISM 310 Genetic Analyzer. The taxanomic identity of retrieved sequences was determined with a BLAST search of GenBank (Altschul et al. 1990).

56 DGGE banding patterns (representing microbial community structure) were analyzed with Phoretix 1D software (TotalLab, UK), using a rolling ball background subtraction and by matching bands manually among samples. Band volumes (calculated as band area * intensity) were exported into PRIMER V6.0 (PRIMER-E Ltd, Plymouth, UK) and were used as a proxy for relative abundance across matched bands. DGGE band density has been used successfully as a proxy of species abundance (e.g. Nikolcheva et al. 2003, Pascoal et al. 2010) although amplification biases can sometimes skew this correlation (Neilson et al. 2013), warranting careful interpretation. Consequently we applied the most conservative transformation (to presence/absence) for non-parametric multi-dimensional scaling analyses (MDS), and included band density in separate analyses. Sorensen similarity matrices were calculated based on the presence/absence data. Cluster-analysis and MDS plots were generated based on 1000 permutations, with clusters determined by SIMPROF (p<0.05; Clarke et al. 2008), which were overlaid onto MDS plots and can be used to compare all samples without a priori assumptions about the differences between samples. Supplementing the conservative MDS plots, lane-standardized band volume data was used to generate Bray-Curtis percent similarity matrices and assessed by SIMPER, which identifies the bands (species) that contributed most to the community dis-similarity among samples. Lane-standardized volume of the band that contributed most to community dissimilarity is presented in the MDS plots as bubbles. Symbols are used where band volume was lower than the smallest bubble size.

3.3.7 Quantitative-PCR Primers used for qPCR are included in Table 2. Three new primer pairs were designed for species of interest from DGGE analyses and sequence identification.

57 These primers were designed in PrimerBLAST (Ye et al. 2012; http://www.ncbi.nlm.nih.gov/), using the default settings. Primers for Leptocylindrus sp. were designed to amplify 149 bp of the 18S rDNA. Primers for Paraphysomonas sp. were designed to amplify 185 bp of the 18S rDNA. Primers for Cyclotella sp. were designed to amplify 223 bp of the 18S rDNA. The specificity of Cyclotella sp., Leptocylindrus sp., and Paraphysomonas sp. primers were tested in-silico by Primer- BLAST, by agarose gel electrophoresis, melt-curve analysis, and by directly sequencing the products from qPCR reactions. All qPCR assays were carried out on an ABI Prism 7500 Sequence Detection System (Applied Biosystems) using 10 µL reaction volumes with 1 µL (5–11 ng) of template DNA. All primer sequences and concentrations used, as well as thermocycling conditions, qPCR efficiency, and standard curve R2 are included in Table 2. The total dinoflagellate, ciliate and diatom abundances in each sample were assayed by qPCR using primers specific to each class. Assays for dinoflagellates, diatoms, Cyclotella sp., Leptocylindrus sp. and Paraphysomonas sp. were performed in 10 µL reactions with SYBR Select Master Mix (Applied Biosystems). The total ciliate community was assayed in 10 µL reactions containing 0.25 U JumpStart Taq DNA Polymerase (Sigma-Aldrich, USA), 1x PCR buffer (Sigma-Aldrich, USA), 0.4 mM dNTPs, 5 mM MgCl2, 50 ng/µl Bovine Serum Albumin, and 0.2x SYBR Green (Invitrogen, CA, USA). The two TaqMan assays (pGEM internal plasmid standard and G. instriatum) were performed in 10 µL reactions with TaqMan Environmental Master Mix 2.0 (Applied Biosystems), as described in Handy et al. (2008). Relative quantification of each species or class was determined by linear regression against a standard curve generated with 10-fold dilutions of one sample of whole community

58 DNA extract. Assay results were then normalized to the pGEM internal standard before calculating percent abundance relative to the average control at the same time point.

3.3.8 Statistical Analyses Statistical analyses were performed in Prism 5 (GraphPad Software, Inc.) using an alpha of 0.05. All experiments had small balanced sample sizes (n=3 or 4). Changes in measured variables over time (e.g. cell density and chlorophyll-a) were analyzed by repeated measures ANOVA and where significant (p<0.05), followed by Bonferroni post-tests between treatments and time points. The dose response experiment was tested by RM-ANOVA and Bonferroni post-tests.

59 3.4 Results In testing batches of harvested algicide against pure dinoflagellate cultures, significant declines in Fv/Fm comparable to that presented previously (Fig. 3.7A, 1way-ANOVA, p<0.001, n=3, Tilney et al. 2014), and induced significant algicidal activity comparable to that presented previously (Fig. 3.7B, 1-way ANOVA, p<0.001, n=3, Pokryziwinski et al. 2012).

3.4.1 Bloom Experiments

3.4.1.1 Photochemistry The photochemical responses of the three bloom communities to IRI-160AA were small and variable (Fig. 3.1). Both P. minimum and K. veneficum communities exhibited small (+0.04 and +0.03 respectively), but statistically significant increases in the maximum quantum yield of PSII photochemistry (Fv/Fm) after 2 and 3 days of algicide exposure respectively (Fig. 3.1A, B). Contrastingly, the community Fv/Fm in the G. instriatum bloom experiment did not differ from the control (Fig. 3.1C), nor did PSII connectivity or the maximum effective absorption cross section of PSII (sigma) vary from the control in this experiment (Fig. 3.1F, I). Despite similar increases in Fv/Fm in P. minimum and K. veneficum bloom samples, the two experiments exhibited opposite patterns in PSII connectivity (Fig. 3.1D, E) and in functional absorption cross section of PSII (Fig. 3.1G, H) compared to controls. PSII connectivity increased significantly (p<0.01, RM-ANOVA, n=3) in P. minimum algicide treatments (+0.05) and decreased significantly (p<0.01, RM-ANOVA, n=3) in K. veneficum algicide treatments (-0.04) relative to controls after 2 days (Fig. 3.1D,E). Sigma did not deviate significantly from the control in P. minimum (Fig. 3.1G) whereas sigma declined rapidly in K. veneficum algicide treatments, and was

60 significantly lower than T0 after 1 day. Sigma continued to decrease to 81% (0.44 nm2) that of the control after 3 days (Fig. 3.1H).

3.4.1.2 Autotrophic Biomass & Target Cell Number In-vivo fluorescence (Fig. 3.2A-C) and extracted chlorophyll-a (Fig. 3.2D-F) were consistent with each other in all 3 bloom experiments, and statistical differences (between the control and IRI-160AA samples) between the two metrics occurred in the same samples and times. In the P. minimum experiment, total autotrophic biomass remained steady in both the IRI-160AA treatments and the controls, and no significant difference was evident between controls and treatments (Fig. 3.2A and D; p>0.05, RM-ANOVA, n=3). Total autotrophic biomass increased significantly in both the IRI- 160AA treatment and the control in the K. veneficum and G. instriatum experiments (p<0.05, RM-ANOVA, n=4 and n=3 respectively). However, total autotrophic biomass in the IRI-160AA treatment was significantly higher than the control only in the K. veneficum experiment (p<0.05), and not in the G. instriatum experiment (p>0.05). Target dinoflagellate cell density in each of the bloom communities are presented in Figure 3.2 (G-I). In all three bloom communities, the target dinoflagellates in the control microcosms remained constant or increased over the duration of the experiment. At the final time point in IRI-160AA treatments; P. minimum density was 119 % (+50,277 cells mL-1) of the control density, K. veneficum density was 64 % (-14,861 cells mL-1) of the control density, and G. instriatum density was 13 % (-685 cells mL-1) of the control density. However, differences in target dinoflagellate cell density between control and treatment microcosms were only significant for K. veneficum and G. instriatum bloom experiments (p<0.05, RM-

61 ANOVA, n=4 and n=3 respectively). Additionally, the lower relative abundance of K. veneficum in IRI-160AA treatments relative to control microcosms was not associated with a population decline over time since K. veneficum cell density increased 23 % ± 11 (p<0.05, t-test, n=3) in IRI-160AA treatments. Contrastingly, the lower relative abundance of G. instriatum in IRI-160AA treatments relative to controls was associated with a population decline of 84% (±17, p<0.01, t-test, n=3) in the IRI- 160AA treatment relative to initial abundance. The majority of G. instriatum population decline occurred on day 2 and 3.

3.4.1.3 Eukaryotic Community Structure Non-metric multi-dimensional scaling (MDS) comparisons of the dominant eukaryotic community structure (based on presence-absence, and excluding organisms <3µm) are presented for each bloom experiment, P. minimum (Fig. 3.3A), K. veneficum (Fig. 3.3B), and G. instriatum (Fig. 3.3C). In all three bloom experiments, IRI-160AA and control microcosms shared a common community composition prior to inoculation (p>0.05, SIMPROF, indicated by unique clustering in Fig. 3.3). Final community composition of IRI-160AA treatments and control microcosms were different in two of the three experiments (in P. mimium and K. veneficum, Fig. 3.3A,B, p<0.05, SIMPROF), but not in the G. instriatum experiment (Fig. 3.3C, p>0.05,

SIMPROF). We confirmed that standardized relative DGGE band volumes (of four ‘bands’) were significantly correlated with relative qPCR abundance using all data available from the dose experiment (Fig. 3.8, Pearson-Product Moment Correlation, Pearson r = 0.95, R2 = 0.91, p<0.0001, n=36), lending support to the use of band volume as a proxy of abundance in the current study. Consequently, SIMPER analysis on Bray-Curtis similarity matrices (using band volume as a proxy of abundance)

62 revealed that Paraphysomonas sp., K. veneficum, and Euplotes sp. (see Table 3.3 for sequence results) contributed most to community dissimilarity in the P. minimum, K. veneficum, and G. instriatum bloom experiments, respectively. SIMPER analysis indicated that Paraphysomonas sp. contributed 31% (±4) to community dissimilarity in the P. minimum bloom experiment; K. veneficum contributed 46% (±13) to community dissimilarity in the K. veneficum bloom experiment; and Euplotes sp. contributed 55% (±19) to community dissimilarity in the G. instriatum bloom experiment. Two additional bands in the DGGE of the P. minimum bloom experiment showed declines in the IRI-160AA treatments relative to the control, and were both identified by sequence analysis in two independent DNA samples to be dinoflagellates most closely related to Gymnodinium aureolum and Symbiodinium sp. (Table 3.3; the search revealing Symbiodinium sp. excluded uncultured/environmental sample sequences). According to the DGGE band volumes, the band resembling G. aureolum declined by 86% (±5.2, p<0.01, n=3, t-test) in IRI-160AA treatments relative to the controls and the band resembling Symbiodinium sp. declined by 78% (±6.1, p<0.001, n=3, t-test) in IRI-160AA treatments relative to the controls. In the DGGE community analysis of the K. veneficum bloom experiment, one additional band identified in two separate samples as Heterosigma akashiwo (Table 3.3), had a significantly lower band volume in control samples compared to IRI- 160AA samples at the final time sampling time (p<0.001, t-test, n=3).

63 3.4.2 G. instriatum Dose Experiment

3.4.2.1 Photochemistry

Statistically, Fv/Fm did not show a dose dependent decline with IRI-160AA concentration after 42 hours of incubation (p>0.05, RM-ANOVA, n=3). However, there were significant differences (p<0.05, RM-ANOVA, n=3) in Fv/Fm between treatment and control microcosms, and Bonferroni post tests indicated that only Fv/Fm in the 32% IRI-160AA treatment was significantly different from the corresponding control microcosms (32%), and declined to 79 % of the Fv/Fm in the control microcosms (p<0.05, RM-ANOVA, n=3).

3.4.2.2 Eukaryotic and Prokaryotic Community Structure MDS plots of the resultant eukaryotic and prokaryotic community compositions in the G. instriatum dose experiment are presented in Figs. 3.4 and 3.5 respectively. Comparisons of the eukaryotic community composition between samples at the final time point, revealed that all of the control microcosms were not significantly different from one another, and that the 8% IRI-160AA samples were not significantly different from the control samples (p>0.05, SIMPROF test, 100 permutations). In contrast, the 16% IRI-160AA samples and the 32% IRI-160AA samples were all significantly different from the ‘control + 8% IRI-160AA’ samples (p<0.05, SIMPROF). Additionally, community composition in the 32% IRI-160AA samples was significantly different from 2 of 3 16% IRI-160AA samples. This latter group containing 2 of 3 16% IRI-160AA samples was more similar to the ‘controls + 8% IRI-160AA’ samples than the group containing the ‘32% IRI-160AA + 1 of 3 16% IRI-160AA’ samples (determined by distance in MDS plot). SIMPER analysis revealed that Paraphysomonas sp. (Table 3.3) was the most important species

64 contributing to community dissimilarity in the G. instriatum dose experiment. Relative band volume of this species is overlain in Fig. 3.4, and shows the predominance of this species in IRI-160AA treatments compared to the controls. A dose dependent increase in band volume was observed with increasing IRI-160AA concentration for this species (p<0.05, RM-ANOVA, n=3). The initial (T0 hours) prokaryotic community composition was significantly different from the final (T42 hours) community composition of both the control and IRI-160AA treated microcosms (Fig. 3.5, p<0.05, SIMPROF). Furthermore, the prokaryotic community structure in the final IRI-160AA treated microcosms was significantly different from that in the control microcosms (Fig. 3.5, p<0.05, SIMPROF). No SIMPER or sequencing was conducted for the prokaryotic analysis.

3.4.2.3 Group and Species Abundances In assessing the eukaryotic community structure of the G. instriatum dose experiment, 4 DGGE bands (in addition to G. instriatum – sequence not shown) were sequenced due to interesting dynamics between IRI-160AA treatments and controls. The four sequences were revealed by BLAST to be most similar to Paraphysomonas sp., Leptocylindrus sp., Cyclotella sp, and Heterocapsa triquetra (see Table 3.3). Primers developed for Cyclotella sp., Leptocylindrus sp., and Paraphysomonas sp. all yielded a single product of approximately the correct size on agarose gel electrophoresis. The fourth band, Heterocapsa triquetra, was only observed in control samples, and to a minor extent in the 8% algicide treatment (data not shown).

All group-level and species-level abundance data for the dose experiment is presented as a fold change relative to control abundance after 42 hours, as measured

65 by qPCR, and is presented in Figure 3.6. Group abundance of dinoflagellates, diatoms, and ciliates are presented in Fig. 3.6A, B and C respectively, and were determined using ‘universal’ group-specific primers (see Table 3.3 for references). Total dinoflagellate abundance declined in a dose dependent manner with IRI-160AA concentration (Fig. 3.6A, RM-ANOVA, p<0.05, n=3). Dinoflagellate abundance in IRI-160AA treatments was 0.92 (8 % treatment, n.s.), 0.58 (16 % treatment, p<0.01), and 0.12 (32 % treatment, p<0.001) fold that of the total dinoflagellate abundance in control microcosms. Total diatom abundance was affected differently depending on IRI-160AA dose (Fig. 3.6B, RM-ANOVA, p<0.05, n=3), where total diatom abundance in IRI-160AA treatments was 1.49 (8 % treatment, p<0.001), 1.27 (16 % treatment, p<0.05), and 1.10 (32 % treatment, n.s.) fold that of the diatom abundance in control microcosms. Ciliate abundance increased in a dose dependent manner with IRI-160AA concentration (Fig. 3.6C, 2way ANOVA, p<0.05, n=3), with ciliate abundance in IRI-160AA treatments 1.45 (8 % treatment, n.s.), 2.02 (16 % treatment, p<0.05), and 2.90 (32 % treatment, p<0.001) fold that of the ciliate abundance in control microcosms. Species-level abundance data for G. instriatum, Leptocylindrus sp. and Cyclotella sp., and Paraphysomonas sp. are presented in Fig. 3.6D, E, F respectively, and were quantified by qPCR with species-specific primers (see Table 3.3). G. instriatum abundance declined in a dose dependent manner with IRI-160AA concentration (Fig. 3.6D, RM-ANOVA, p<0.05, n=3) with abundance in IRI-160AA treatments 0.89 (8 % treatment, n.s.), 0.63 (16 % treatment, p<0.05), and 0.09 (32 % treatment, p<0.001) fold that of G. instriatum abundance in control microcosms. Leptocylindrus sp. abundance did not vary in a dose dependent manner with IRI-

66 160AA (Fig. 3.6E, p>0.05, RM-ANOVA, p<0.05, n=3), despite a significant difference in its abundance between the treatments. Leptocylindrus sp. abundance in IRI-160AA microcosms were 1.03 (8 % treatment, n.s.), 0.77 (16 % treatment, n.s.), and 0.32 (32 % treatment, p<0.05) fold that of the Leptocylindrus sp. abundance in the control microcosms. Cyclotella sp. abundance increased at all concentrations of IRI- 160AA in a dose dependent manner (Fig. 3.6E, RM-ANOVA, p<0.05, n=3), with Cyclotella sp. abundance in IRI-160AA treatments 3.10 (8 % treatment, n.s.), 5.75 (16 % treatment, p<0.001), and 3.37 (32 % treatment, p<0.05) fold that of the Cyclotella sp. abundance in control microcosms. Paraphysomonas sp. abundance also increased in a dose dependent manner (Fig. 3.6F, RM-ANOVA, p<0.001, n=3) and to the largest extent of all the species quantified. Paraphysomonas sp. abundance in IRI-160AA microcosms was 9.80 (8 % treatment, p<0.001), 47.93 (16 % treatment, p<0.001), and 152.29 (32 % treatment, p<0.001) fold that of the Paraphysomonas sp. abundance in control microcosms.

3.4.2.3.1 Specificity of Primers For qPCR A number of the qPCR reactions from each assay were pooled (within the same assay) for sequencing. Direct sequencing from qPCR confirmed the correct amplicon lengths and identities of Cyclotella sp., Leptocylindrus sp., and

Paraphysomonas sp. (Table 3.3). Single peaks in the melting curves for Cyclotella sp. and Leptocylindrus sp. after qPCR amplification suggested the likelihood that the products contained only one sequence. However, the melt curve from Paraphysomonas sp. indicated a two-stage melting process, which may be due to discrepancies in length, GC/AT ratio, or sequence (Ririe et al. 1997). Consequently, the specificity of these primers was further investigated by cloning (TOPO TA

67 Cloning Kit using pCR 4-TOPO Vector) and directly sequencing from colony PCR products. Two clones were sequenced, and both yielded similar (2-stage) melt curves as were observed in the environmental samples. Cloned sequences were the correct size with high similarity to Paraphysomonas sp. sequences (Table 3.3). The sequences revealed a GC-rich region of 42 bp (75.9% GC content, bases 47-88), immediately adjacent to a 51 bp length with low GC content (31.4% GC content, bases 89-139). All 3 primer sets were considered species specific, and the 2-peak melt curve of Paraphysomonas sp. was most likely due to adjacent lengths of the aforementioned GC-rich and GC-poor regions.

3.5 Discussion Previous work has demonstrated the algicidal and growth inhibitory activities of IRI-160AA derived from Shewanella sp. IRI-160 on dinoflagellates grown in culture (Hare et al. 2005, Pokryziwinski et al. 2012, Tilney et al. 2014). In the current study, we show that IRI-160AA also causes algicidal and growth inhibitory activity against natural dinoflagellate blooms found in complex environmental water samples. In the context of utilizing IRI-160AA as a control agent for HABs, we discuss both the relative potency of the algicide on dinoflagellates in natural communities compared to cultured dinoflagellates, as well as qualitative and quantitative changes to the microbial community and potential risk of the algicide to fish health.

3.5.1 Similarities and Discrepancies From Cultures

The pattern of dinoflagellate species susceptibility in natural waters was consistent with the pattern of susceptibility in previous uni-algal culture experiments (Pokryziwinski et al. 2012, Tilney et al 2014); specifically that P. minimum was least

68 susceptible and G. instriatum was the most susceptible of the three dinoflagellates tested. However, three observations were noted when comparing results from the current study with results obtained previously in laboratory culture experiments and from the bioactivity tests of IRI-160AA harvested for use in the current study. The first difference was the minor and inconsistent patterns of photochemical responses. The second difference was the stability (or increase) in total autotrophic biomass. The third difference noted between responses of natural communities and laboratory cultures was the smaller and/or delayed decline in target dinoflagellate abundance. The first two of these differences are most simply accounted for because photochemistry and total autotrophic biomass are community metrics defined by signals from the whole autotrophic community and not only the target dinoflagellate population. In regards to the photochemistry, Suggett et al. (2009) highlighted the importance of interpreting the photo-physiological component of FRR fluorescence transients in the context of community composition. This is because community FRR signals represent an average of unique species-specific photobiological signatures, each overlain with species- and individual-specific photo-physiological stress signals. In the current study, the minor changes in photochemistry were likely due to signals from the unaffected algal community, which were able to mask any photoinactivation signatures supplied by affected dinoflagellates. However, results from the dose response experiment with G. instriatum provided evidence that photochemical disruption is observable within the community signal given sufficiently high doses of the algicide. Despite masking of the photoinactivation signal, the photochemical data do contain valuable information on community composition. For example, an increase of the raphidophyte Heterosigma akashiwo in the algicide treatment was confirmed by

69 DGGE analysis, and changes in the maximum effective absorption cross section (sigma) were consistent with this change in phytoplankton composition. Specifically, sigma values measured in cultured individuals of K. veneficum are high (~2.6 nm2, pers. obs.), while sigma values in H. akashiwo are low (~1.5 nm2, Hennige et al. 2013). Thus, the photochemical signal measured here corresponds to a reduction in “high-sigma” signals (K. veneficum) and a gain in “low-sigma” signals (H. akashiwo). This same effect of species composition on the photochemical signal can be seen in IRI-160AA applications on mixed cultures of Chattonella subsalsa and K. veneficum in Fig. 3.9. Thus, the FRR signals are consistent with the shift observed in community composition by DGGE. Although not identical (since biomass is absolute), a similar effect occurs in the measurements of total autotrophic biomass. Specifically, following previous work showing minimal effects on non-dinoflagellate algae, it is not surprising that total autotrophic biomass remained high, since non-dinoflagellate cells were unaffected by IRI-160AA and continued to contribute to total autotrophic biomass. However, given the high densities (and proportions) of dinoflagellates in the blooms sampled, the stability of total autotrophic biomass corroborates the lower efficiency at removing target dinoflagellates. As noted above, there was also a difference in the relative decline in dinoflagellate abundance over time in the algicide-treated microcosms compared to laboratory cultures. For example, P. minimum cell density in natural community microcosms treated with the algicide was not significantly different from the control, or over time. K. veneficum increased in abundance over time in the algicide treated microcosm experiments and only exhibited growth inhibition relative to the control by day 3. G. instriatum still exhibited significant declines in cell density during

70 microcosm experiments, although the majority of cell losses were delayed until the 3rd day as opposed to the immediate effects normally observed in G. instriatum culture experiments (Tilney et al. 2014, Fig. 3.7). Overall, the results follow a trend of reduced or delayed susceptibility in natural water applications relative to results using laboratory cultures. Importantly, a comparison of absolute susceptibility between lab and field applications cannot be made, and instead, comparisons of susceptibility can only be made for the specific incubations we have presented. Therefore the comparison noted above does not represent the absolute, maximal, or even the potential range of susceptibilities that may be achieved in field dinoflagellates more generally. While the factors driving this trend of reduced or delayed susceptibility are unknown, several possibilities seem likely. Firstly, changes in cellular physiology have previously been cited as a potential driver of changes in susceptibility to algicidal bacteria (Manage et al. 2000, Mayali and Azam 2004). Further, a broader range of cellular physiologies among the field populations, compared to synchronously entrained cultures, could prevent temporally-synchronized mass susceptibility and therefore mask observations of population level effects (Mayali and Azam 2004). Secondly, slower population growth by dinoflagellates in natural bloom waters could limit the proportion of cells with a susceptible physiology/phenotype that is known to occur most frequently during active growth (Pokryziwinski et al. 2012). In the current study, dinoflagellate growth was negligible in control incubations, potentially due to sampling the blooms near stationary phase. Additionally, we cannot rule out if partial inactivation of the bioactive compound(s) by microbial transformation may have reduced algicide potency and also contributed to the trend of reduced or delayed susceptibility observed in field-collected dinoflagellates. Lastly, it is notable that

71 Shewanella sp. IRI-160 has been found to be widespread throughout the Delaware Inland Bays (KJ Coyne, unpublished data), and consequently adaptation to Shewanella’s allelopathy by dinoflagellates may be occurring, as has been observed in the allelopathy of other marine plankton and in plants (Vivanco et al. 2004, Kubanek et al. 2007). However, since our laboratory cultures were isolated <10 years ago from the Delaware Inland Bays, the extent of genetic adaptation of field specimens may be limited, though the mortal nature of the interaction may be a potent pressure selecting for resistant genotypes. Regardless, if adaptation were to occur in some form, this could imply that natural product development for HAB control may need to be conducted continuously, similar to the issues of antibiotic resistance in the biomedical sciences and of pesticide resistance in agronomy, and may ultimately undermine the viability of this approach to HAB management. Other authors have noted reduced efficacy of algicidal bacteria, and other algicides, on algae in environmental samples. For example, Kim et al. (2008) observed a significant increase in their target cyanobacteria, Microcystis aeruginosa, at low concentrations of the algicidal bacteria Xanthobacter autotrophicus HYS0201-SM02, consistent with the effect seen on P. minimum in the current study. Kang et al. (2011) also found lower effects of the algicidal bacteria Pseudomonas fluorescens SK09 in a natural water mesocosm compared to uni-algal in-vitro assays. It is important to note that these two studies (Kim et al. 2008, Kang et al. 2011) used live bacteria and are thus not directly comparable to our study, as these studies must also consider effects on in situ algicide production. More comparably however, Schrader et al. (2000) found that trans-ferulic acid was ineffective at controlling the cyanobacterium Oscillatoria perornata in the field despite potency in laboratory studies, which they attributed to

72 rapid microbial degradation or chemical transformation in the field. It is important to note that all of these experimental results represent severe application scenarios since microcosm, mesocosm, and pond studies, as enclosed bodies, prevent advection and dilution into adjacent waters (Mayali et al. 2007), which modeling efforts indicate is crucial in mediating the importance of allelopathy and zooplankton grazing on HAB dynamics (Solé et al. 2006). These results raise the question of how to improve bioactivity for natural water applications. One option may be to employ higher concentrations of the control agent, which is supported by evidence of enhanced activity at higher concentrations presented in the current study (Fig. 3.6) and also observed in other studies (e.g. Kange et al. 2007, Kim et al. 2008). Increasing algicide concentrations for field applications might be achieved at the application step, during the production/harvest steps, or post harvest. Identification of the bio-active compound(s) and biosynthetic pathways as a means to mass produce algicidal compounds (Kim et al. 2008) may be the most labor intensive option in the short run, but may be the most practical and economical route in the long run, and would be an appropriate strategy for future application of IRI-160AA in treatment of dinoflagellate blooms. Another option to improve the bio-activity in natural water applications, as identified by Pokrzywinski et al. (2012), is to use the algicide at an early stage of bloom formation, because bioactivity is highest during active algal growth phases (Pokrzywinski et al. 2012). Consequently, monitoring the dinoflagellate growth phase in situ could improve the level of bio-activity achieved by allowing applications to be made at times of optimal dinoflagellate susceptibility. Considering potential sources of physical accumulations may also be necessary in this regard since physical forcing would not alter dinoflagellate susceptibility like during biological growth, despite both

73 yielding increased biomass (Smayda 1997, Hall et al. 2008). Identifying the bioactive molecules(s), as well as understanding the root of dinoflagellate susceptibility will be crucial to optimizing IRI-160AA efficiency in the field.

3.5.2 Shifts in Community Structure One common theme among the algicide treatments was the dramatic increase in heterotrophic/bactivorous protists (i.e. Paraphysomonas spp. and Euplotes sp.) compared to controls in P. minimum and G. instriatum bloom experiments, and highlighted in the dose experiment by dose-dependent increases in ciliates (Fig. 3.5C) and Paraphysomonas sp. (Fig. 3.5F). It is difficult to ascertain why these groups in particular increased, but an increase in bacterial prey-availability or quality appears likely. Although we did not measure bacterial abundance, a significant increase in total DNA was recovered from the higher concentration (16% and 32%) 0.2-3 µm fraction collected from the algicide treated samples compared to the control samples in the dose experiment (RM-ANOVA, p<0.05, n=3) which indirectly supports this hypothesis. Although this is only a tentative proxy of increased bacterial abundance (this fraction includes picoeukaryotes), this increase is expected due to the dissolved organic matter (DOM) release by dying dinoflagellates (Mayali and Doucette 2002; not measured in the current study) and/or from other organic constituents within the

Shewanella exudate. Furthermore, the shift in bacterial community composition (Fig. 3.5) is likely to have been associated with a shift in nutritional quality, which may have opened a niche for bacteriovores with specific prey preferences (Turley et al. 1986, Dopheide et al. 2011). Another study that assessed the effect of immobilized algicidal bacteria (Pseudomonas fluorescens HYK0210-SK09) on natural blooms of a diatom (Stephanodiscus hantzschii) observed an increase in heterotrophic ciliates and

74 flagellates, and concluded this was due to predation on added algicidal bacteria (Jung et al. 2013). Similarly, Kang et al. (2011) found significant increases in ciliate abundance in algicidal-bacteria treated mesocosms relative to controls. That we observed a similar effect without adding live bacteria highlights that DOM from the Shewanella exudate or from cell death in the dinoflagellate community are also important factors in enhancing net community heterotrophy in algal control studies. Consequently, it appears that HAB control strategies that induce cellular destruction in the water column are likely to mimic the events following DOM addition to the plankton. For example, growth of bacterial biomass, increases in heterotrophic nano- flagellates, a decline in primary productivity per chlorophyll-a and broadly, a shift toward a heterotrophic, bacterial production-based food web (Forsström et al. 2013). This rise in DOM could be disadvantageous if harmful algae capable of utilizing these enhanced organic nutrients were present (e.g. brown tide species, Gobler and Sunda 2012). Changes to community structure were expected in all incubations due to containment and the effects of inorganic nutrient enrichment (Chen et al. 1997, Gattuso et al. 2002), which ultimately defined the short incubation times used in this study. Differences in community structure between controls and treatments however, show that IRI-160AA contributed to community re-structuring. This effect could be mediated by a combination of direct and indirect factors. Direct control may be exerted by the addition of organic components in the IRI-160AA filtrate (potentially including the bio-active component) that can be used by heterotrophic or mixotrophic organisms stimulating their growth. Indirect control may be exerted by affecting the physiology or viability of dinoflagellates, but could also be driven by effects from

75 non-target microbes. Others have highlighted the difficulty and importance of defining the direct and indirect effects of allelochemicals on microbial communities (Weissbach et al. 2011). Some of these questions could be resolved through isotopic tracer studies, but are probably more efficiently assessed after the bioactive compounds are known and can be assayed or even synthesized in the laboratory. Although the dinoflagellates in the blooms sampled here appeared to be mono- specific, DGGE analysis revealed that non-target dinoflagellates were present, and were affected by IRI-160AA. In the P. minimum experiment, two bands most closely related to the dinoflagellates, Gymnodinium aureolum and Symbiodinium sp. (see results section), both declined in the algicidal treatment based on DGGE band volumes. Secondly, in the G. instriatum dose experiment, the increase in a dinoflagellate species most closely related to Heterocapsa triquetra was limited to final control samples, and conspicuously absent from final IRI-160AA samples (except for a faint presence in 8% IRI-160AA samples). Consequently, multiple dinoflagellates are simultaneously affected, to differing degrees of susceptibility. This may be useful for targeting multi-species harmful dinoflagellate blooms, but may also represent a source of collateral damage if non-target or beneficial dinoflagellates are present, for example as symbionts. Surprisingly, Paraphysomonas sp. did not dominate the G. instriatum bloom experiment while it did in all concentrations of the G. instriatum dose experiment. One explanation is that Euplotes sp. as a bacteriovore (Turley et al. 1986) effectively filled the niche occupied by Paraphysomonas sp. (e.g. Smayda and Reynolds 2001). Weissbach et al. (2011) also drew attention to the importance of starting microbial community composition in determining community outcomes due to allelochemical

76 additions. Stochastic re-structuring of the microbial community could affect the efficiency of carbon transfer out of the microbial loop to higher trophic levels between each IRI-160AA application. For example, VanHannen et al (1999) observed growth in rotifer populations after a mass viral lysis event, and posited that this should be successful at channeling carbon out of the microbial loop. Notably, community re- structuring in situ will likely differ from that observed here in vitro because of photochemical modifications of DOM and POM by sunlight (Gobler et al. 1997), and by the inclusion of meso-plankton > 209 µm, which we excluded. The exclusion of mesozooplankton facilitated the clear observation that protist abundances increased, and would act to fuel the food chain in a whole community setting.

3.5.3 Conclusions The lower and/or delayed algicide efficacy seen in the current study, along with evidence of dose dependent effects, suggests that higher concentrations of the algicide may be needed to fully control dinoflagellate blooms that are detected at the height of the bloom in situ. Maintenance of high autotrophic biomass and PSII photochemistry at lower doses of IRI-160AA corroborates the lower algicidal efficacy, and also suggests collateral mortalities and/or inhibition were minimal. We hypothesize that reduced and/or delayed efficacy in microcosms may be due to differences in algal physiology in the field collected samples, supported by the slow growth observed in control samples. Our study therefore supports the suggestion by Pokryziwinski et al. (2012) that the timing of IRI-160AA applications will be crucial to achieving high bio-activity in the field, by ensuring that dinoflagellates are actively growing and thus express the most susceptible physiology. The shift towards communities dominated by heterotrophic and bactivorous organisms is a common

77 theme among algicidal bacteria studies, and one which we also observed, despite inoculating with bacteria-free filtrate. We speculate that organic compounds in the filtrate, in addition to DOM excretion by lysed target algae, may be important factors determining the observed community restructuring. Lastly, the confirmation that the algicide was effective in mixed assemblages supports further work to deduce the responsible compound(s), which could lead to compound concentration, synthetic synthesis, the potential to minimize direct affects to community structure, and the ability to enhance efficacy in the field.

78 Table 3.1: Environmental conditions at the time of sampling each bloom, as well as incubation conditions used during the experimental period

P. minimum K. veneficum G. instriatum

In Situ Latitude 38°34'13.75"N 38°32'27.82"N 38°34'13.75"N Longitude 75° 5'3.90"W 75°3'27.59"W 75° 5'3.90"W Sampling Date 18 Apr 2012 8 May 2012 17 Jul 2012 Water Temperature (°C) 18.3 23.8 31.0 Salinity (PSU) 28.3 25.5 28.5 Total Dissolved Solids - 25.9 28.8 (g/L) pH - 8.11 7.54 DO (mg/L) - 10.2 6.2

Laboratory Temperature (°C) 18.0 24.0 29.0 Irradiance (µmol m-2 s-1) 54 200 75 Container 0.5L Flask 2L Flask 0.5L Flask

79 Table 3.2: Descriptions of DGGE and qPCR primers, qPCR thermocycling conditions, assay efficiency and regression coefficient of determination

(q)PCR Thermocycling: Organism(s) Targeted Primer Sequence (5'-->3') Primer Conc. Efficiency (%) r2 ** Reference Denature Anneal Extend

Eukaryotes Euk29F GTC TCA AAG ATT AAG CCA TGC 0.2 µM - - Coyne et al. 2005

Euk517R * GGA CCA GAC TTG CCC TC 0.2 µM Coyne et al. 2005

Prokaryotes 338F * ACT CCT ACG GGA GGC AGC AG 0.2 µM - - Lane 1991

519RC ATT ACC GCG GCT GCT GG 0.2 µM Muyzer et al. (1993)

Dinoflagellates Dino06F CCG ATT GAG TGW TCC GGT GAA TAA 0.9 µM 95.12 0.999 Handy et al. 2008 2 95° - 15s 56° - 30s 72° - 60s

EukB GAT CCW TCT GCA GGT TCA CCT AC 0.9 µM Handy et al. 2008 3

Diatoms 1256F TAG TGA GGA TTG ACA GAT TGA 0.3 µM 83.48 0.998 Lee 2012 95° - 15s 53° - 30s 72° - 60s

1536R CAA TAA TCT ATC CCT ATC ACG ATG 0.9 µM Lee 2012

Ciliates 384F YTB GAT GGT AGT GTA TTG GA 0.5 µM 79.52 0.996 Dopheide et al. 2008 94° - 45s 55° - 60s 72° - 90s

1147R GAC GGT ATC TRA TCG TCT TT 0.5 µM Dopheide et al. 2008

G. instriatum Gi166F GCA CAA ATT CCC AAC TTC GCG G 0.3 µM 94.71 0.999 Handy et al. 2008 95° - 15s 56° - 30s 70° - 60s

Gi274R GCT CGA ATG ATT CAT CGC CAG CA 0.3 µM Handy et al. 2008

Paraphysomonas sp. ParaF ATT GGA GGG CAA GTC TGG TG 0.9 µM 90.59 0.999 This study 95° - 30s 60° - 30s 72° - 45s

ParaR GAC AAC TGA ATG CCA GAC GC 0.9 µM This study

Cyclotella sp. CycloF TGCATCAATACCCGACTTCTG 0.9 µM 85.65 0.994 This study 95° - 30s 60° - 30s 72° - 35s

CycloR AGGCTCCCTCTCCGAAATCT 0.9 µM This study

Leptocylindrus sp. LeptoF AATTTAGGGATTGATTCCGGAGAGG 0.9 µM 85.47 0.998 This study 95° - 30s 60° - 30s 72° - 35s

LeptoR ATTCCAAGTGACAAACCTGAAGA 0.9 µM This study

pGEM Standard M13F CCCAGTCACGACGTTGTAAAACG 0.9 µM 97.32 1 Coyne et al. 2005 95° - 15s 60° - 30s 72° - 60s

pGEM R TGTGTGGAATTGTGAGCGGA 0.9 µM Coyne et al. 2005

* GC-clamp added

2 Modified from Oldach et al. (2000)

3 Modified from Medlin et al. (1988)

80 Table 3.3: Sequencing results from DGGE bands, qPCR reactions, and cloned qPCR reactions, in order of appearance in the text

Sequencing Results Percent Experiment Species / Genus Taxa No. Seqs. bp Queried Q-Coverage Identity Accession Match

DGGE sequences P. mimimum Paraphysomonas sp. Chrysophyceae 1 371 bp 100% 100 JQ967325.1 K. veneficum K. veneficum Dinophyceae 3 284 bp 100% 99 EF492506.1

G. instriatum Euplotes sp. Spirotrichea 1 288 bp 100% 93 EF094967.1

P. mimimum Gymnodinium aureolum Dinophyceae 2 478 bp 100% 96 FN392226.1

P. mimimum Symbiodinium sp. Dinophyceae 2 472 bp 100% 92 EF419282.1

K. veneficum Heterosigma akashiwo Raphidophyceae 2 154 bp 100% 100 JQ250796.1

G. instriatum Dose Paraphysomonas sp. Chrysophyceae 2 463 bp 100% 99 JQ967325.1

G. instriatum Dose Leptocylindrus sp. Bacillariophyceae 1 465 bp 100% 99 KC814809.1

G. instriatum Dose Cyclotella sp. Bacillariophyceae 2 470 bp 100% 100 JF708166.1

G. instriatum Dose Heterocapsa triquetra Dinophyceae 1 458 bp 100% 88 AJ415514.1

qPCR sequences G. instriatum Dose Cyclotella sp. Bacillariophyceae 1 180 bp 100% 100 KF918178.1 G. instriatum Dose Leptocylindrus sp. Bacillariophyceae 1 101 bp 100% 99 KC894150.1

G. instriatum Dose Paraphysomonas sp. Chrysophyceae 1 145 bp 100% 99 JQ967324.1

Cloned sequences G. instriatum Dose Paraphysomonas sp. Chrysophyceae 2 145 bp 100% 99 JQ967324.1

Figure 3.2: Measures of autotrophic biomass over 2 or 3 days of incubation with IRI- 160AA. Filled squares represent IRI-160AA algicide treatments, and open circles represent f/2 controls. (A-C) in-vivo fluorescence presented in relative fluorescence units, (D-F) extracted chlorophyll-a presented in µg/L, and (G-I) target dinoflagellate cell density in 103 cells mL-1. Error bars indicate 1 S.D., and asterisks indicate significance level (*=p<0.05, **=p<0.01, ***=p<0.001) as determined by repeated measures ANOVA and Bonferoni post tests.

84 Figure 3.3: MDS plots of the eukaryotic community composition based on DGGE fractionation of partial 18S rDNA PCR amplicons from particles > 3 µm, and determined by Sorensen similarity among samples within 3 separate experiments: P. minimum (A), K. veneficum (B), G. instriatum (C). Contours indicate SIMPROF significance at p<0.05. Bubble size indicates lane-standardized band volume of the band which contributed most to community dissimilarity in Bray-Curtis similarity matrices (see text for details). Bubble scale (% of community) is shown as semi-circles below each plot. The sequence identification of each band is shown in the top left of each plot. In all plots, T0 samples are represented by white bubbles (or triangles in A where band volumes were too small for bubble depiction), final f/2 control samples are dark grey bubbles, and final IRI- 160AA samples are light-grey bubbles. Scales are indicated below each plot.

Figure 3.4: MDS bubble plot of eukaryotic community composition in the > 3 µm size fraction of the G. instriatum dose experiment at the final time point (T42 hours). Grey filled bubbles are IRI-160AA treatments, and black filled triangles are f/2 controls, and the percent of IRI-160AA or f/2 used is written above every sample. Bubbles represent lane-standardized band volume of Paraphysomonas sp. where the scale (% of community) is shown below the plot with semi-circles. Symbols are used for f/2 control samples because bubbles were too small. Dotted lines delineating groups of samples indicate a significant difference in community structure based on SIMPROF tests on Sorensen presence/absence similarity matrices (p<0.05).

Figure 3.6: Relative qPCR abundance as fold change from control microcosms, of dinoflagellates (A), diatoms (B), ciliates (C), G. instriatum (D), Leptocylindrus sp. (E, striped bars), Cyclotella sp. (E, filled bars), and Paraphysomonas sp. (F) in the G. instriatum dose experiment. Error bars indicate ±1 SD, and asterisks indicate significant differences between control microcosms (not displayed) and IRI-160AA treatments (*=p<0.05, **=p<0.01, ***=p<0.001), determined by RM-ANOVA (n=3) and Bonferoni post tests.

Figure 3.7: Confirming the bioactivity of IRI-160AA filtrates harvested for use in this study, at 10 % (v/v) against uni-algal cultures after 1 day of incubation following the methods in Pokrzyiwinski et al. (2012) and Tilney et al. (2014). Inihibition in Fv/Fm is presented in panel A, where open symbols represent f/2 control cultures, and closed symbols represent IRI-160AA treated cultures. Filled circles show IRI-160AA used in the K. veneficum bloom experiment, tested against K. veneficum; filled diamonds show IRI-160AA used in the P. minimum bloom experiment tested against G. instriatum; and filled triangles show IRI-160AA used in the G. instriatum bloom experiment and dose experiment, tested against G. instriatum. Algicidal activity (in %) is shown in panel B, and originates from the same experiments described above for panel A. The dinoflagellate noted below the bars represents the filtrate batch from that species’ bloom experiment in the current study. Algicidal activity in IRI- 160AA treatments was tested by RM-ANOVA and Bonferroni post-tests, and revealed that significant algicidal activity was induced by all 3 batches of algicide used in this study after just 1 day of incubation (p<0.001, n=3).

Figure 3.8: Correlation of relative DGGE band volume with relative qPCR abundance of Leptocylindrus sp., Paraphysomonas sp., G. instriatum, and Cyclotella sp. tested in the G. instriatum dose experiment (data from Fig. 3.4 and Fig. 3.6). Pearson product moment correlation r = 0.95, R2=0.91, p<0.0001, n=36. Relative abundance refers to IRI-160AA treatment as a percentage of the average f/2 controls, and only IRI- 160AA samples are presented.

Figure 3.9: Mixed culture experiment with K. veneficum and Chattonella subsalsa (Raphidophyceae) showing the effect of eliminating K. veneficum from a 2-specie mixed culture on the maximum effective absorption cross section of PSII (SigmaPSII, nm2). We tested 3 algal cultures, each in triplicate with 10% IRI-160AA or 10% f/2 medium added as a control, which were 1) unialgal K. veneficum, 2) unialgal C. subsalsa, and 3) a 1:1 mix of K. veneficum with C. subsalsa. Panel A shows percent algicidal activity (filled bars) where Kv = K. veneficum, Mixture = 1:1 mixed culture, and Cs = C. subsalsa. Paenl B shows maximal effective absorption cross-section of PSII (SigmaPSII, nm2), where closed symbols are IRI-160AA incubations, and open symbols are controls. Circles represent unialgal K. veneficum, squares represent uni-algal C. subsalsa, and triangles represent the 1:1 mixed culture. Only the mixed culture shows a decline in SimgaPSII due to the shift in the proportion of each alga’s photochemical signatures.

91 Chapter 4

COMPARING THE DIEL VERTICAL MIGRATION OF KARLODINIUM VENEFICUM (DINOPHYCEAE) AND CHATTONELLA SUBSALSA (RAPHIDOPHYCEAE): PSII PHOTOCHEMISTRY, CIRCADIAN CONTROL, AND CARBON ASSIMILATION

4.1 Abstract Diel vertical migration (DVM) is thought to provide an adaptive advantage and may help to determine the ecological niche of certain harmful algae that use it. Here we separately compared DVM patterns between two mid-Atlantic sympatric species of harmful algae, Karlodinium veneficum and Chattonella subsalsa, in laboratory columns. We interpreted the DVM patterns of each species with Photosystem II (PSII) photochemistry, rates of carbon assimilation, and specific growth rates. Each species migrated differently, wherein K. veneficum migrated closer to the surface each day with high population synchrony, while C. subsalsa migrated with lower synchrony, and migrated near to the surface from the first day of measurements. Both species appeared to downregulate PSII in high light at the surface, but used different mechanisms for this. C. subsalsa grew slower than K. veneficum in very low light intensities (equivalent to the bottom of columns), and exhibited maximal rates of C- assimilation (Pmax) at surface light intensities, suggesting this species may prefer high light, potentially explaining this species’ rapid surface migration. Contrastingly, K. veneficum showed declines in carbon assimilation at surface light intensities, and exhibited a smaller reduction in growth at low (bottom) light intensities (compared to C. subsalsa), suggesting that this species’ step-wise migration was photoacclimative

92 and determined its daily migration depth. We also found differences in the circadian regulation of DVM between the species, and investigated circadian control of PSII photochemistry. Migration conformed to each species’ physiology and is useful in describing each alga’s realized niche in the environment.

93 4.2 Introduction It is widely accepted that the frequency of harmful algal blooms (HABs) is increasing and that some of these species are expanding their biogeographic ranges (Hallegraeff 1993, Van Dolah 2000, Mcleod et al. 2012). In seeking to understand why this is occurring, a thorough understanding of the biotic and abiotic controls on HAB dynamics is required. This is daunting given the sheer number and diversity of species responsible for HABs and the variety of environments in which they occur, but nevertheless much progress has been made (Anderson et al. 2008, Heisler et al. 2008, Anderson 2009, Lewitus et al. 2012). One particularly complex factor that has long been studied and implicated in bloom initiation and persistence is diel vertical migration (DVM). DVM is a cyclical behaviour in which the vertical distribution of organisms inverts between day and night, such that the distribution cycle (starting and ending in the same place), happens over approximately 24-hour intervals (i.e. has a circadian period). DVM is particularly complex because it integrates a wide variety of cellular physiologies into a plastic behaviour (Cullen and MacIntyre 1998), via the use of a variety of sensory perceptions (Kamykowski et al. 1998b), which operate by incompletely characterized signal transduction pathways (Allan and Fisher, 2011). In this regard, understanding how DVM is employed by different algal species could be useful to better characterize their ecological niches (Cullen and MacIntyre 1998).

Two hypotheses have been postulated to explain how DVM confers an adaptive advantage to phytoplankton. The first is that DVM provides access to deep nutrients when surface nutrients become depleted (Eppley et al. 1968, Cullen and Horrigan 1981, MacIntyre et al. 1997, Doblin et al. 2006). The second is that DVM provides relief from grazers, many of which show DVM patterns in an opposite phase to phytoplankton (Pedrós-Alió et al. 1995, Bollens et al. 2012). These hypotheses are

94 not mutually exclusive and both factors were used to explain the dominance of a raphidophyte in a stratified lake (Salonen and Rosenberg 2000). Other advantages conferred by DVM include avoidance of flushing into unfavourable waters (Anderson and Stolzenbach 1985, Crawford and Purdie 1992), and photosynthetic optimisation (Ault 2000). The nutrient hypothesis has received greater attention than the other possibilities. There are a variety of mechanisms that algae use to sense their immediate environment for perception and orientation, and a number of these are used to direct movement and DVM. These include phototaxis (movement in respect to light; e.g. Moldrup et al. 2013), chemotaxis (movement in respect to chemicals; e.g. Byrne et al. 1992, Ucko et al. 1994) and geotaxis/gravitaxis (movement in respect to gravity; e.g. Kamykowski et al. 1998b, Hader et al. 2003). Furthermore, DVM appears to be driven by an endogenous circadian clock in some species (Weiler and Karl 1979, Roenneberg et al. 1989, Shikata et al. 2013), which is thought to operate (in part) by controlling the sensitivity of the alga to the above taxes (e.g. sensitivity of phototaxis depends on the endogenous clock’s rhythm; Forward 1975). The clock itself must be regulated however, and this is almost always controlled primarily by light, but can also receive inputs from pH and certain chemicals (Roenneberg and Rehman 1996, Eisensamer and Roenneberg 2004). The role of the circadian clock has been studied extensively in one particular dinoflagellate, Lingulodinium polyedrum, which shows clock controlled rhythms in movement as well as bioluminescence, photosynthesis, cell division, protein synthesis, and ultrastructure (Fritz et al. 1990, Mitag 2001 and references therein, Nassoury et al. 2005). Determining whether a biological rhythm such as DVM and photosynthesis in microalgae is under clock control is ecologically relevant because clock-controlled

95 rhythms provide an adaptive advantage by synchronising specific cellular physiologies to the optimal time of day, without reliance on external factors (Ouyang et al. 1998, Woelfe et al. 2004, Dodd et al. 2005). Most studies that have noted the existence of endogenous control of algal DVM have cited movement prior to the light-dark or dark-light transitions (e.g. Eppley et al. 1968, Doblin et al. 2006). However, observation of a free-running rhythm in constant conditions should help to validate that these rhythms are indeed circadian (Johnson 2001), but such measures have rarely been assessed (Handy et al. 2005, Shikata et al. 2013). Both the proximate (i.e. mechanistic) and ultimate (i.e. adaptive significance) drivers regulating DVM appear to vary among different species (Kamykowski et al. 1998b, Doblin et al. 2006), and even within a species under different environmental conditions (e.g. Eppley et al. 1968). Despite an imperfect understanding of the DVM behaviour, from an ecological point of view DVM can be empirically tested and compared among species to generate interesting and useful knowledge about important species such as those responsible for harmful algal blooms. Few studies have compared detailed migratory behaviour of two or more species in a single study. In the current study we describe and compare the DVM patterns of two locally isolated HAB forming species from the Delaware Inland Bays in the mid Atlantic region of the United States, a dinoflagellate, Karlodinium veneficum, and a raphidophyte, Chattonella subsalsa. A few studies have assessed the DVM of Chattonella spp., including Chattonella antiqua (Watanabe et al. 1991, Watanabe et al. 1995), and the same strain of C. subsalsa used in the current study (Handy et al. 2005). In contrast, almost nothing is known of the vertical migrations of K. veneficum, with only one study reporting DVM in this species (a field

96 study assessing DVM in phytoflagellates in response to water column structure; Hall and Paerl 2011). Here, physiological measurements, including photochemistry, specific growth rates, and carbon assimilation rates, were used to help explain different DVM behaviours. Additionally, the potential role of a circadian clock was assessed by testing if rhythms in DVM and PSII photochemistry would free-run in constant light or constant dark conditions.

4.3 Methods

4.3.1 Stock Algal Culture, Column Design, and Lighting

The dinoflagellate, K. veneficum (CCMP 2936) and the raphidophyte, C. subsalsa (CCMP 2191), were both isolated from the Delaware Inland Bays ~ 8 years before this study. Prior to this study, these cultures were grown in f/2 seawater medium under static 12:12 light:dark (LD) cycles. Two months before this study began, stocks were grown in 20 PSU filtered sea water supplemented with f/2 nutrient concentrations (“f/2 medium”, Guillard and Ryther 1962, Guillard 1975), at 24 °C, and grown under static 14:10 light:dark (LD) cycles at ~50 µmol photons m−2 s−1 from above with cool white fluorescence lights. In stock cultures and column experiments detailed below, the light period commenced at 6 a.m. EST (06:00), and ended at 8 p.m.

(20:00).

Three columns were constructed from extruded transparent acrylic piping, which was bonded to square acrylic bases. Columns were 1.82 m tall, had an internal diameter of 14.6 cm and held approximately 31 L of culture. Thirty nylon barbed-style sampling ports were threaded into the walls of the column, 3 each at 10 depths. The

9 7 bottom sampling port was 2.8 cm above the bottom of the column, and ports above this were placed every 19 cm, with the final port placed at a depth of 9 cm from the top of the column. The tops of the columns were covered with a square of thin acrylic, which was removed for surface sampling (~ 1 cm depth). Ports were sampled via 5 mL syringes, which were attached to the barbed ports by 3 cm of silicone tubing that remained clamped while not in use. The columns were kept in a temperature controlled room at 24 °C, and each illuminated from above by a set of 9 fan-cooled white LEDs (Cool White XLamp XP-G, Cree Inc., CA, USA), which were attached to a digital controller (AquaController-Apex Jr., Neptune Systems, CA). The standard lighting regime in the columns was a 14:10 LD cycle with a surface light intensity at dawn and dusk of 230 µmol photons m−2 s−1, which ramped up to 1000 µmol photons m−2 s−1 at the mid day peak. Irradiance (PAR, 400-700 nm) at the top and bottom of the columns was measured in air, with a Li-Cor 4π quantum sensor. A two point calculation of the downwelling attenuation coefficient (Kd) was made using the equation from Kirk (1983):

1 !!! !! = ×!" (!2 − !1) !!!

Where z is the depth and E is the light intensity. The resultant Kd was used to model the light field throughout the columns, which was used for the carbon assimilation experiments (see below).

4.3.2 Common Experimental Procedures

As with the stock cultures, all experiments were conducted in 20 PSU seawater amended with f/2 nutrients as per stock cultures. Initially, 30 L of f/2 medium was added to each column, and then 1 L of culture was added from the top at

98 approximately 19:00 to avoid a high light exposure. Columns were cleaned between inoculations with liquid detergent and sponge, followed by bleach and multiple rinses with tap water and a final rinse with DI water. Sampling under darkness was always performed under dim green light (< 0.5 µmol photons m−2 s−1), which was verified not to induce any photochemical activity in the algae studied.

4.3.3 Chlorophyll Fluorescence Measurements In-vivo chlorophyll-a fluorescence of dark acclimated samples was used as a proxy of cellular biomass (Eppley et al. 1968, Wood et al. 2005, Gustavs et al. 2009). Use of this proxy provided an efficient way to measure algal distribution, and facilitated high-resolution spatio-temporal sampling strategies. Single turnover fluorescence of Photosystem II (PSII) was used to probe algal photochemistry, using a FASTtracka II fast repetition rate fluorometer (FRRf; Chelsea Instruments, UK), and a FASTact laboratory system (Chelsea Instruments, UK). The FRRf induces a single turnover of the PSII reaction center by cumulatively saturating PSII with a series of short (~ 1 µs) blue flashlets until a first fluorescence plateau is reached (Kolber et al. 1998). Our protocol consisted of 100, 1 µs flashlets each separated by 1 µs, followed by 50, 1 µs flashlets separated by 49 µs to probe relaxation kinetics. Fluorescence transients and the derived parameters were calculated with

FASTpro V3.0 (Chelsea Instruments, UK) in real time and immediately inspected for integrity. Fluorescence parameters derived from the transients are presented in Table 4.1, and brief descriptions of the most prominent terms are noted here (see also

Cosgrove and Borowitzka 2011). Fv/Fm is a robust measure of the maximum potential efficiency that captured photons will drive photochemistry through PSII (i.e. the reduction of components in the linear electron transport chain), which is measured

99 after dark acclimation to release the photosynthetic units from excitation pressure.

’ ’ Fq /Fm is a measure of the operating efficiency of PSII photochemistry under actinic light. The PSII functional absorption cross section (σPSII or σPSII’) quantifies the area of PSII where light absorption is directed to photochemistry. The connectivity among PSII reaction centers (ρ or ρ’) is the probability that an exciton reaches an open reaction center after initially encountering a closed PSII reaction center. The time required for relaxation of PSII fluorescence (τ or τ’), describes the efficiency of e-

- transport from Qa to Qb. Except where noted, fluorescence measurements were made on dark acclimated samples, in order to relax possible non-photochemical quenching. A standard dark acclimation was used for K. veneficum by placing cells removed from the column into low light (< 5 µmol photons m−2 s−1) for approximately 30 min and into darkness for 15 min. Testing dark acclimation intervals in Chattonella subsalsa revealed that, similar to other raphidophytes (Hennige et al. 2013), low light acclimation (< 5 µmol photons m−2 s−1) yielded higher PSII quantum yields than complete darkness, and thus, samples collected in the light were acclimated only under low light prior to fluorescence measurement. However, samples collected in the dark phase of growth were maintained in complete darkness for measurement. In one experiment (the 6 d experiment), fluorescence measurements were also collected from light acclimated samples to assess the type and degree of photochemical and non- photochemical fluorescence quenching used by each alga at the surface and at the bottom of the columns. White LEDs in the FRR sample chamber were used to match the light environment of the column location where the cells were removed (i.e.

100 varying with depth and time), and light acclimated fluorescence transients were recorded after 4 minutes of actinic light exposure.

4.3.4 DVM Sampling and Calculations In all experiments, two samples were withdrawn and combined from each depth, except at the surface where only one sample was withdrawn (21 samples per column per sampling time). Subsequently, 2 mL from each depth was dark acclimated and measured for active chlorophyll fluorescence (simultaneous calculations of PSII photochemistry and chlorophyll-a concentration). To calculate whole population DVM, chlorophyll-a concentrations from each depth were standardised to a percentage of the total column chlorophyll-a concentration. This was used to calculate a cumulative percent distribution from the bottom to the top of the column, such that 0% of the population was accounted for at 183 cm depth (the bottom), and 100% of the population are accounted for at 1 cm depth. Multiple pairwise linear regressions (i.e. between each two sampling ports) were used to interpolate the depth (in cm) of the lower (25%), median (50 %) and upper (75%) quartiles of the population. Changes in depth over time of the median, upper, and lower quartiles of the population revealed whole-population patterns in DVM, and the interquartile range (IQR) provided a quantitative, comparable measure of dispersion (i.e. synchrony). A description of this method and its benefits are described in Pennak (1943).

4.3.5 Initial Migration & Photobiology Experiments

To characterise DVM and diel periodicity of PSII photochemistry of each species, two experiments were conducted on different time scales under standard lighting conditions (i.e. 14:10 ramping LD cycles, in nutrient replete f/2 media). In the

101 first experiment, 3 days after inoculation, migration and photochemistry were measured every 3 h for 39 h commencing 1 hour before ‘lights-on’ (05:00). Measurements made at the lights-off transition (20:00) were made 5 minutes past the hour (i.e. during darkness). Column light intensity in this experiment was lower compared to later experiments because these were initial experiments where algal response to very high light was not yet known (peak surface intensity 650 µmol photons m−2 s−1). In the second experiment measurements were made 3 times per day for 6 days, starting on day 2, at 1 h before ‘lights-on’ (05:00), at peak midday irradiance (13:00), and 1 h before dark (19:00). Light acclimated photochemistry was also measured in this experiment (see chlorophyll fluorescence measurements above for protocol). In the longer term K. veneficum experiment, 7 measurements were excluded from surface photochemistry data (6 from the 05:00 samples and 1 from a 13:00 sample) when chlorophyll-a concentration was < 0.8, because fluorescence transients in these low cell density samples consistently overestimated variable fluorescence (pers. obs.).

4.3.6 Assessing Circadian Control of Rhythms Two experiments were conducted to test if diel rhythms in PSII photochemistry or DVM persist under constant light or dark conditions (i.e. were under circadian control) in either alga. Columns under standard conditions (i.e. entrained to 14:10 LD cycles) were shifted into either continuous darkness (abbreviated as the DD experiment hereafter) or continuous light (abbreviated as the LL experiment hereafter). The DD experiment was conducted immediately following completion of the previously described 39 h photobiology/DVM experiments. Columns were shifted into darkness at 06:00 on day 5 after inoculation. Sampling with

102 K. veneficum in DD was performed at 08:00 and 14:00 for two days and once in between at 23:00. Sampling with C. subsalsa in DD was performed at 08:00, 14:00, and 17:00 for two days. The LL experiments were conducted on freshly inoculated columns (new media new culture), and on day 2 after inoculation DVM and photobiology were measured for 2.5 or 3.5 d (for K. veneficum and C. subsalsa respectively) before shifting the columns at 13:00 to continuous peak light intensity. After 2 d under LL, the columns were shifted back to standard LD conditions and monitored for 24 h. The columns were sampled at 05:00, 13:00, and 19:00 throughout the experiment. CircWave V3.3 was used to test for significant circadian rhythms (R.A. Hut; available at http://www.EUclock.org) by fitting a sine wave with linear harmonic regression in order to test the data fits against the fit of a horizontal linear regression (Oster et al. 2006, Keller et al. 2009). Sine wave fitting was chosen over other methods due to the short time series of data, relatively coarse and un-even sampling strategy, and because the period was expected to remain circadian (Leise et al. 2011).

4.3.7 Comparing DVM and Static Cultures In order to determine whether DVM provided an advantage under nutrient replete conditions in either alga, we compared growth rates in the columns to growth rates of algae kept at fixed surface and bottom light intensities. Triplicate 600 mL polystyrene tissue culture flasks were microwave sterilised (Sanborn et al. 1982) and inoculated with a 1:30 (v:v) ratio of culture:f/2 as used in columns. Flasks were maintained under lighting equivalent to that in columns (equivalent LED spectral output, and ramping on 14:10 LD cycles) but maintained at surface intensities (peak 1000 µmol photons m−2 s−1), or at bottom intensities (peak 30 µmol photons m−2 s−1).

103 Dark acclimated in-vivo fluorescence in the flasks was monitored at 13:00 for 9 days and specific growth rate µ (d-1) was calculated by regression of the natural log of in- vivo fluorescence against time. Growth rate in columns was calculated in the same way but used total column in-vivo fluorescence at midday (in C. subsalsa) and at 05:00, 13:00, and 19:00 (in K. veneficum). Three measurements per day were used to calculate K. veneficum growth rates to account for lower coefficients of determination.

4.3.8 Carbon Assimilation To compare how DVM and column location relates to carbon fixation, 14C incorporation rates were measured in samples withdrawn from the columns and placed in a photosynthetron following Fu et al. (2007) with modifications. Fresh columns were inoculated as above, and were grown for 6 or 7 d (K. veneficum and C. subsalsa respectively). DVM and photochemistry were measured 1 day before, 1 day after, and on the day of the 14C experiment at 05:00, 13:00, and 19:00. At 13:00 on the day of the experiment, samples were withdrawn from each column at 5 depths (180, 123, 66, 28,

14 and 1 cm) for ex situ NaH CO3 incubation, dissolved inorganic carbon (DIC), and cell enumeration. For 14C assimilation, 1 mL sub-samples (5 depths, 3 columns) were

14 distributed into 7 mL glass vials and 0.3 µCi NaH CO3 was added to each vial. Vials were incubated for 55 min in a photosynthetron at 24 °C, and supplied with white

LED light at the intensity that each sample would expect to receive at their sampled column depth (intensity determined using Kd). Two additional samples were incubated

14 with NaH CO3 in the dark. After incubation, all samples were fixed with 2% glutaraldehyde. Each sample was acidified with 250 µL of 6N HCL for >8 hours to

14 off-gas unincorporated NaH CO3. Samples were then read in a scintillation counter (LS-6500, Beckman Coulter, CA, USA) and values were corrected for minor dark

104 uptake with the dark samples. Samples for cell enumeration were fixed in 1% Lugols Iodine and counted within 5 d (C. subsalsa) and 20 d (K. veneficum). K. veneficum was enumerated by nuebauer haemocytometer and C. subsalsa was enumerated by counting all cells in 25 µL aliquots (diluted as necessary) on a sedgewick rafter grid. For DIC samples, ~20 mL of culture was poured into glass scintillation vials with 200

µL of 5% (w/v) HgCl2 and stored at 4 °C until analysis. DIC samples were analysed by acidifying with 5% (v/v) phosphoric acid with a DIC Analyzer (AS-C3, Apollo

SciTech Inc., GA, USA) and sent to an infra-red CO2 analyzer (LI-7000, LI-COR, NB,

14 12 USA) using CO2 free air. C assimilation was then calculated using the measured C DIC concentrations in each sample.

4.3.9 Statistical Analyses Statistical analyses were performed in Prism 5 (GraphPad Software Inc., USA) with small balanced sample sizes throughout (n=3 or 4). Two-way repeated measures ANOVA (RM-ANOVA) was used to compare the top and bottom photochemical parameters (n=3 or 4) over time in the short-term and long-term migration and photochemistry experiments. In the short term DVM and photobiology experiment, 1 column became contaminated before measurements began and was discarded, statistical tests were therefore carried out by combining both days (n=4) and testing between 05:00 to 20:00 (as no replicate measures were made at 23:00 and 02:00). To measure changes over time, a one-way RM-ANOVA was used and where significant, followed by Tukey post-hoc tests. Statistics were not performed in the K. veneficum DD experiment due to the loss of 1 column to contamination.

105 4.4 Results

4.4.1 Diel Vertical Migration & Diel Photobiology

Distributions of K. veneficum showed a clear pattern of DVM, rising toward the surface near midday, and descending to the bottom at night (Fig. 4.1A). Synchrony was highest at the bottom of the columns at night (i.e. >75% of K. veneficum were at or below 180 cm). Distributions of C. subsalsa also showed a clear pattern of DVM, rising to the surface near midday and descending to depth at night (Fig. 4.1B). Synchrony in C. subsalsa DVM was low throughout the diel period, and showed significantly less population synchrony in DVM compared to K. veneficum (comparing IQR; P<0.05, RM-ANOVA, n=3 and 4), a flatter mid-day peak, and descended more slowly to depth at night compared to K. veneficum (comparing slope of descent from 14:00 to 02:00, in the upper quartile; F1,36 = 62.7, P<0.0001, n=24 and 16). Both species exhibited diel changes in PSII photochemistry, which were unique to each species. In K. veneficum Fv/Fm was significantly different between the bottom and top of the columns (Fig. 4.1C; 05:00 to 20:00; RM-ANOVA, n=4,

P<0.05). At the bottom, Fv/Fm exhibited a small decline throughout the day, a larger decline of ~ 0.1 through the night, and a significant large rapid return to the morning maxima after illumination at dawn (P<0.05, 05:00 to 20:00; RM-ANOVA, n=4). At the surface, K. veneficum also had a maxima in the morning and a slight non- significant decrease during the night, but also exhibited a large significant decrease of ~0.07 during the day (P<0.05, 05:00 to 20:00; RM-ANOVA, n=4) with a variable degree of recovery when light intensity declined at the end of the day. As with

Karlodinium, Fv/Fm differed significantly between the bottom and top of the columns

106 in C. subsalsa (Fig. 4.1D; P<0.05, RM-ANOVA, n=3). At the bottom, Fv/Fm declined significantly by ~0.05 over the day (P<0.05, RM-ANOVA, n=3), and returned steadily through the night to a morning maximum (at 08:00; but was not significant P>0.05,

RM-ANOVA, n=3). For C. subsalsa at the surface, Fv/Fm declined significantly (P<0.05, RM-ANOVA, n=3) to a minimum around mid-afternoon (14:00 to 17:00) with some recovery evident toward the end of the day as light intensity ramped down (Fig. 4.1D; like in K. veneficum). The maximum effective absorption cross section of

PSII (σPSII) in K. veneficum was similar at the top and bottom of the columns, and showed minima (~2.2 nm2) during the night and maxima (~2.6 nm2) late in the day

(Fig. 4.1E). There was a slight, yet non-significant difference in σPSII in C. subsalsa between the top and bottom of the columns (Fig. 4.1F; P>0.05, RM-ANOVA, n=3). At the top and bottom of the column, maxima (~1.2 nm2) occurred just before dawn, and decreased significantly throughout the day (P<0.05, RM-ANOVA, p=3), before rising back to the maxima through the second half of the night. Photosystem II reoxidation time (τ) in K. veneficum at the top and bottom of the columns was not significantly different (Fig. 4.1G; 05:00 to 20:00; P>0.05, RM-ANOVA, n=4), and remained constant throughout the day at ~ 490 µs, before rising gradually throughout the night to a maximum of ~600 - 670 µs toward the end of the night. Tau (τ) in C. subsalsa was similar at the top and bottom throughout the experiment (Fig. 4.1H), showing a midday minima of ~600 µs, which increased steadily toward a maxima during the early night (~700 µs), and then fell steadily back toward the midday minima. PSII connectivity (ρ) differed significantly between the top and bottom of the columns in K. veneficum (Fig. 4.1I; 05:00 to 20:00; P<0.05, RM-ANOVA, n=4). At the bottom, ρ was constant during the day and night, but remained at a slightly higher

107 level during the day (0.41) than during the night (0.37). At the surface, ρ dropped significantly by ~0.15 (P<0.05, RM-ANOVA, n=4) to a minimum at midday under peak irradiance, then returned to the level of the bottom cells by the end of the day, and rose slightly to a maxima in the early morning. PSII connectivity (ρ) in C. subsalsa also exhibited differences between the top and bottom of the column (Fig. 4.1J; P<0.05, RM-ANOVA, n=3). At the bottom, ρ dropped significantly by ~0.06 (05:00) from the morning to the end of the day (P<0.05, RM-ANOVA, n=3) before rising back to the morning maxima through the night. At the top, the drop in ρ from the morning maxima to the end of the day was also significant, and more abrupt than at the bottom (P<0.05, RM-ANOVA, n=3). The main differences between the diel photochemical patterns between K. veneficum and C. subsalsa were 1) the reduction of

PSII in the dark in K. veneficum (decrease in Fv/Fm and increased tau), and 2) the larger drop in PSII connectivity at the top (compared to the bottom) in K. veneficum, and 3) the inverted timing of maximal sigma value (late day phase in K. veneficum and early day phase in C. subsalsa). When assessed over 6 days, the migratory patterns of K. veneficum and C. subsalsa were broadly similar to that observed in Fig. 4.1 (Fig. 4.2A, B). Specifically, both maintained similar inter-quartile ranges and migration amplitudes as Fig. 4.1. However, for K. veneficum, the population migrated closer to the surface with each proceeding day. Specifically, median migration amplitude, calculated as the difference between 13:00 and the following morning (05:00), increased significantly in K. veneficum when comparing day 2 to day 5 (+99 cm ± 5 SEM; P<0.001, 1way- ANOVA, n=3). Conversely, in C. subsalsa, the population immediately distributed with an upper quartile close to the surface, but as growth progressed, C. subsalsa

108 began to migrate with greater synchrony toward the bottom, although median migration amplitude was not significant when comparing day 2 to day 6 (+63 cm ± 28 SEM; not significant, 1way-ANOVA, n=3). Similarities and differences in photochemistry and photo-acclimation between K. veneficium and C. subsalsa were evident after monitoring the columns for 6 days. Prominent differences were highlighted in the measurements of light acclimated surface samples compared to dark acclimated samples. The operating quantum yield

’ ’ Fq /Fm was significantly lower at the surface than at the bottom every day in K. veneficum (Fig. 4.2C; P<0.05; see legend) and C. subsalsa (Fig. 4.2D; P<0.05; see

’ ’ legend). Midday surface Fq /Fm followed a similar trend in both species, increasing for the first 3 days and gradually decreasing for the last 3 days. Recovery in Fq’/Fm’ from midday (13:00) to the end of the day (19:00) decreased in both species by day 6.

Surface Fv/Fm values were significantly different from bottom Fv/Fm values in both K. veneficum (Fig. 4.2B; P<0.05), and C. subsalsa (Fig. 4.2D; P<0.05), although C. subsalsa showed larger, and gradually increasing declines. Non-photochemical quenching from the PSII antenna bed was significantly higher in the light acclimated surface cells of K. veneficum (~0.5; Fig. 4.2E) compared to C. subsalsa (~0.2; Fig. 4.2F) (P<0.05, RM-ANOVA, n=3). Compared to K. veneficum, there appeared to be greater diel variability in light-acclimated τ in C. subsalsa at 19:00 in surface samples in the final 3 days (Fig. 4.2H) though this was not significant (P>0.05, RM-ANOVA, n=3). At midday at the surface, PSII connectivity in the light activated state (ρ’) dropped to a larger extent compared to dark-acclimated ρ in K. veneficum as compared to C. subsalsa (Fig. 4.2I, J; P<0.05, RM-ANOVA, n=3). However, a downward trend

109 in ρ’ in C. subsalsa led to ρ’ values that were equivalent to those in K. veneficum by the final day.

4.4.2 Circadian Control Experiments After shifting K. veneficum into continuous darkness (DD), the median and interquartile range at 08:00 appeared to be higher and wider compared to the previous days in LD (Fig. 4.3A). Under DD, changes to the distribution of K. veneficum over time continued to follow the up/down directions that were observed under LD (i.e. rising from 08:00 to 14:00 and falling from 14:00 to 23:00) but was not found to be a significant rhythm (Fig. 4.3A; P>0.05, CircWave, n=3). After shifting C. subsalsa into DD, it maintained a constant depth and no circadian rhythm was detected (Fig. 4.3B;

P>0.05, CircWave, n=3). After shifting K. veneficum into DD, Fv/Fm did not rise at 08:00 in either the surface or bottom samples as was noted in the LD growth regime, but, instead remained low, at a level equivalent to that usually observed at the end of the dark phase and showed no evidence of rhythmicity (Fig. 4.3C). After shifting C. subsalsa into DD, both the surface and bottom Fv/Fm values consistently declined and there was no evidence of circadian regulation of Fv/Fm in C. subsalsa (Fig. 4.3D). Before shifting K. veneficum into continuous light (LL), it migrated in the same manner as observed in Fig. 4.2A, including the trend to more shallow midday depth distributions over time. After shifting K. veneficum into continuous light (LL), the depth distribution continued to rise almost linearly, without any evidence of DVM (Fig. 4.4A; P>0.05, CircWave, n=3). After returning K. veneficum back to LD, DVM returned immediately with similar phase and amplitude. Before shifting C. subsalsa into continuous light (LL), cells migrated in the same manner as observed in Fig. 4.2B, showing the same migration amplitude and characteristically wide inter quartile

110 range observed in Figs. 4.1B and 4.2B. After shifting C. subsalsa into continuous light (LL), it continued to migrate with significant diel periodicity (P<0.01, CircWave, n=3), with the same apparent phase as before the shift, but with diminished amplitude. After returning C. subsalsa back to LD, the amplitude of DVM was immediately regained.

Before shifting K. veneficum into LL, Fv/Fm, τ, and σPSII showed distinct diel fluctuations consistent with that depicted in Figs. 4.2C, 2G, and 1E respectively. Fv/Fm fluctuated by ~ 0.1, driven predominantly by a fall during the night (at the bottom) and to midday declines (at the top) (Fig. 4.4C). Tau (τ) decreased by ~150 µs during the day before rising again at night (Fig. 4.4E), and σPSII tended to peak at midday and fell from midday through to the next morning (Fig. 4G). After shifting to LL, K. veneficum lost the diel fluctuations in Fv/Fm, τ, and σPSII. Fv/Fm remained at day-time levels at the bottom, while cells at the surface had significantly lower Fv/Fm values at 13:00 compared to values at 13:00 during the first 2 days under LD (Fig. 4C; P<0.05, RM- ANOVA, n=3); τ remained at ~590 µs at the bottom, while steadily increasing from

~530 µs to ~670 µs at the surface; and there was no day-time increase in σPSII at the bottom or the surface, and remained significantly lower (Fig. 4G; P<0.05, RM-

ANOVA, n=3) at the surface compared to midday values under LD. Similar to Fv/Fm and sigma, ρ remained low and had no continued diel periodicity (Fig. 4I). Upon returning K. veneficum to LD, Fv/Fm, τ, σPSII, and ρ returned to pre-LL patterns. Before shifting C. subsalsa into LL, Fv/Fm, tau, and sigma showed diel fluctuations consistent with Figs. 4.2D, 2H, and 1F respectively; Fv/Fm at the surface was reduced at midday (13:00) and before dark (19:00) (Fig. 4.4D). Tau (τ) was highest before dark (19:00) and lowest around midday (13:00) (Fig. 4.4F), and sigma was highest just before dawn

111 and decreased until the end of the day (Fig. 4.4H). After shifting C. subsalsa into LL,

Fv/Fm at the bottom was similar to the trends noted under LD conditions until the second night, while shifts in Fv/Fm at the surface did not continue as under LD conditions and generally remained low (Fig. 4.4D). Tau (τ) at the bottom and top maintained a LD pattern only for 1 day (Fig. 4.4F), and σPSII at the bottom maintained a similar pattern to LD at reduced amplitude and with a general trend to lower values, but surface σPSII did not maintain the pattern observed under LD conditions (Fig. 4.4H). PSII connectivity (ρ) at the bottom continued as in LD until the end of the first day, increasing at 13:00 and decreasing to 19:00, and the amplitude was larger than under LD. At the surface, connectivity remained low and showed no rhythmicity (Fig. 4.4J).

4.4.3 Comparing Growth Between DVM and Static Cultures Growth rates of each species held under ramping 14:10 LD cycles at either surface or bottom light intensities, and in the columns, are presented in Fig. 4.5. In surface high-light flasks, K. veneficum and C. subsalsa both grew at approximately 0.5 d-1, and were not significantly different from each other (P>0.05, 2-way ANOVA, n=3). However, in bottom low-light flasks K. veneficum growth was 0.36 d-1 while C. subsalsa grew significantly slower at 0.22 d-1 (P<0.01, 2-way ANOVA, n=3). C. subsalsa growth in the columns was 0.47 d-1 and was significantly faster than low- light acclimated cells (P<0.05, 1-way ANOVA, n=3), but not significantly different from high-light acclimated cells (P>0.05, 1-way ANOVA, n=3). K. veneficum growth in the columns (0.17 d-1) was significantly slower than high-light acclimated cells (P<0.05, 1-way ANOVA, n=3) but not significantly different from low-light acclimated cells (P>0.05, 1-way ANOVA, n=3).

112 4.4.4 Carbon Assimilation Rates

-1 -1 Maximal carbon fixation (Pmax) in K. veneficum was 9.0 pg C cell h and occurred at intermediate column depths (statistically at 28 and 66 cm; Fig.4.6A). For

-1 -1 C. subsalsa Pmax was 88 pg C cell h and was not significantly different (P>0.05, 1- way ANOVA, n=3) between samples collected at shallow to intermediate column depths (1, 28, 66 cm; Fig. 4.6B). Significant declines in carbon uptake were observed in K. veneficum at 28 cm and 1 cm depth, while no similar trend was noted for C. subsalsa, which instead maintained Pmax at the surface.

4.5 Discussion

4.5.1 DVM, Photobiology, and Growth K. veneficum and C. subsalsa both showed clear patterns of DVM, but migrated differently within each day. The main differences in DVM patterns between the two species were 1) a rapid descent in K. veneficum compared to a slow descent in C. subsalsa, and 2) less synchrony in C. subsalsa’s migration compared to K. veneficum. Regarding the descent, it may be that K. veneficum descends actively (e.g. via negative phototaxis and/or positive geotaxis) (Eggersdorfer and Häder 1991, Kamykowski et al. 1998b), whereas C. subsalsa descends passively via a ‘dispersal’ mechanism like that of Karenia brevis as noted by Van Dolah et al. (2008). The lower migration synchrony in C. subsalsa compared to K. veneficum is difficult to explain, but could be due to individual-specific migrations (to different depths) resulting from phased/asynchronous cell division, where biochemically unequal cell division and/or cell division gating, limits each cells’ timing and extent of migration (Kamykowski et al. 1998, Van Dolah et al. 2008). A similar pattern has been noted in the dinoflagellate K. brevis that divides unevenly, such that the ‘poorer’ daughter cells migrate higher

113 and earlier than the parent cells (Schaeffer et al. 2009). Alternatively, the wide C. subsalsa distribution may be due to bioconvection streams (‘clumps’ of cells moving rapidly downwards creating a stream of cells; pers. obs.). Bioconvection streams are initiated by dense swarms of cells at the surface, which causes the surface water to become denser than the subsurface water, creating an overturning instability (Pedley and Kessler 1992). However, it is not clear how this affects C. subsalsa specifically because K. veneficum also forms such ‘streams’, but contrasting cell size and density could mediate such differences. Although we have discussed this difference from the perspective of C. subsalsa, the converse effects from K. veneficum may be equally important (e.g. more synchronous division, less efficient streaming). Regardless of the mechanisms, a wide vertical distribution in situ may be advantageous by limiting the probability of catastrophic population loss from an area due to currents or grazing (Anderson and Stolzenbach 1985, Crawford and Purdie 1992, Bollens et al. 2012), or increasing the probability of partial transfer to a better environment and hedging survival through wider distributions (Tyler and Seliger 1978). The appearance of cells at the surface despite evidence for the majority of cells occurring at lower depths, highlights that sub-populations can migrate differently from the population as a whole. Individual level assessments of swimming behaviour (e.g. via video analysis; Tobin et al. 2013) and physiology would be very complementary to further characterize the patterns in DVM. In the longer experiment (Fig. 4.2), K. veneficum increased its migration amplitude significantly by migrating closer to the surface each day. Surprisingly, K. veneficum continued to migrate closer to the surface despite the highest Pmax occurring at an intermediate depth (Fig. 4.6). Importantly, cell growth in the columns by day 6

114 would have increased light attenuation, so the light levels used to determine Pmax were probably overestimated, and the true (day 6) Pmax could be shallower. Consequently, K. veneficum appears to migrate to a depth optimal for photosynthesis as described in Prorocentrum triestinum (Ault 2000). The high growth (Fig. 4.5) and maximal photosynthesis (Fig. 4.6) at low to intermediate light intensities may facilitate the migration strategy of K. veneficum. Although the majority of K. veneficum cells did not require an immediate use of fast photoprotective mechanisms (see below), a small population of cells at the surface did use these mechanisms. Use of fast

’ ’ photoprotective mechanisms was evidenced by the midday reductions in Fq /Fm at the surface, which recovered toward the end of the day and were not evident in dark acclimated samples. These photoprotective mechanisms included the use of both PSII connectivity and non-photochemical quenching in the antennae complex. The use of these photoprotective mechanisms at the surface did not change over time in K. veneficum. Although not directly measured, antennae-based non-photochemical quenching is consistent with the operation of a xanthophyll cycle (Long et al. 1994, Brown et al. 1999, Gorbunov et al. 2001, Hennige et al. 2009), as previous reported in both dinoflagellates (Brown et al. 1999) and raphidophytes (Hennige et al. 2013). Reports of PSII connectivity as it relates to photoprotection are rare (Gorbunov et al. 2001, Ihnken et al. 2011, Trimborn et al. 2014). Trimborn et al. (2014) suggest that high ρ at the start of the day can act as a photoprotective mechanism by enhancing electron transport rates to rapidly set-up a thylakoid pH gradient and initiate NPQ mechanisms. Correspondingly, Ihnken et al. (2011) showed a positive relationship between NPQ and ρ in the first few minutes of illumination from darkness (high ρ correlating with high NPQ). Our results also show morning maxima in ρ, such that

115 NPQ could be initiated quickly in both species. However ρ, and ρ’ particularly, decrease during exposure to high-light throughout the day, which would not enhance photochemical e- sinks during this period. Instead ρ’ may correspond to other NPQ mechanisms, including LHC-quenching and RCII downregulation. In particular, by preventing absorbed excitation energy from entering closed RCIIs, LHC-quenching should by definition reduce ρ, by increasing the likelihood that excitation energy transiting between PSII units via LHCIIs would be quenched. Indeed, Gorbunov et al.

(2001) showed a positive correlation with ρ and σPSII’ in Symbiodinium in agreement with our data. Consequently, reductions in ρ and ρ’ may simply indicate specific photoprotective mechanisms, rather than being an active mechanism itself. Although unlikely to be the case here, reductions in ρ may contribute to photoprotection by allowing continued quenching by photoinactived RCIIs, rather than adding excitation pressure to remaining functional RCIIs. Similar to K. veneficum, C. subsalsa exhibited

’ ’ similar reductions in surface Fq /Fm that recovered slightly by the end of the day and were not evident in surface Fv/Fm values during the first 2 days, suggesting non- photochemical photoprotective mechanisms were implemented and effective at preventing photoinhibtion. From the first day onwards, > 25% of the C. subsalsa population migrated near to the surface each day, thus a larger proportion of the population (compared to K. veneficum) required rapid photoprotection. Mechanistically, C. subsalsa differed from K. veneficum, in being much less reliant on LHC-quenching and somewhat less reliant on PSII connectivity (ρ). The lower reliance on LHC-quenching, along with the smaller drop in ρ (Fig. 4.1J, 2J) and larger increases in τ and τ’ (time for electron flow out of PSII) toward the end of the day in C. subsalsa (Fig. 4.1H, 2H) compared to K. veneficum, strongly suggest that different

116 mechanisms of photoprotection are used in this species. C. subsalsa may instead be more reliant on one or more other mechanisms such as; cyclic electron flow around PSII (Falkowski et al. 1986) or PSI (Asada 1999, Endo et al. 1999); the use of alternative electron sinks such as O2 at RuBisCo (i.e. photorespiration, Kozaki and

Takeba 1996) O2 at PSI (i.e. water-water cycle, Asada 1999) and nitrate by nitrate- reductases via NADPH (Guerrero et al. 1981, Long et al. 1994); the use of constitutive RCII NPQ mechanisms (so-called “RCII downregulation”, Larkum 2003, Ivanov et al. 2008); or enhanced PSII repair cycles (Jeans et al. 2013). It is not possible to determine which mechanisms are primarily employed by C. subsalsa, but given the coincident reductions in surface Fv/Fm in tandem with increases in τ late in the day, RCII downregulation appears likely (Ivanov et al. 2008). An even larger difference in τ was observed between the species in LL at the surface (constant high light, Fig. 4.4).

After 3 days, Fv/Fm in C. subsalsa at the surface declined further (Fig. 4.3D), and this may be due to longer-term photoprotective mechanisms that were not relaxed within

45 min of dark acclimation. This loss in Fv/Fm is consistent with the finding of Warner and Madden (2007) that C. subsalsa cannot maintain high maximal rates of electron transport for prolonged periods (>1 week), when shifted into high light, though a decrease in Fv/Fm does not always imply reductions in overall electron transport or photosynthesis (Behrenfeld et al. 1998) and indeed Pmax is maintained at surface levels after 7 days (Fig. 4.6). Ultimately, the low growth rate (Fig. 4.5B) and Pmax (Fig. 4.6) in low-light grown C. subsalsa, provides further evidence that this alga maximises photosynthesis and growth under high light. Thus, C. subsalsa and K. veneficum migrate, photoacclimate, and photoprotect in different ways, which will affect their ecological niche in the Delaware Inland Bays and elsewhere, but the importance of

117 these differences will depend on the stratification and mixing conditions. C. subsalsa relies on rapid photoprotective mechanisms that over 3-4 days in a static column, resulted in reduced Fv/Fm but did not immediately affect Pmax (Fig 4.6). C. subsalsa is poised to exploit periods (or seasons) with high light, and can acclimate quickly to rapid changes in light intensity. Interestingly, this is consistent with another harmful alga, Cochlodinium cf. polykrikoides, which is also a vertical migrator, with strong phototactic responses, that is minimally photoinhibited in high light (1600 µmol photons m-2 s-1), perhaps suggesting a common niche of these species (Kudela et al. 2008). K. veneficum migrated to an optimal light intensity in the column (directed by its prior photoacclimation state; and potentially self shading in uni-algal culture,

Roenneberg and Mittag 1996) which enabled this species to maintain high Fv/Fm values and C-assimilation rates near to Pmax. As a result, K. veneficum uses migration to photoacclimate and prevent photoinhibition, and is better suited to stable light conditions, and may outcompete C. subsalsa in low light conditions.

4.5.2 Circadian Rhythms Circadian rhythms in DVM were not observed in constant darkness in either species, but evidence for endogenous cues for migration and orientation were observed in both species. Specifically, K. veneficum’s ascent at 08:00 to 14:00 in DD suggests that the morning ascent is at least partially endogenous and likely geotaxis based (Kamykowski et al. 1998b). C. subsalsa did not ascend in DD, but did appear to ascend prior to dawn, suggesting an endogenous cue in this species (Handy et al. 2005, Shikata et al. 2013). In C. subsalsa, the lack of DVM rhythm in DD but presence of a rhythm in LL may indicate that phototaxis is more strongly under clock control than geotaxis. Phototaxis was not under clock control in K. veneficum under constant light

118 (Fig. 4.4A) and downward migrations were also not initiated. It is not clear why this species does not entrain its phototactic behaviour to a circadian oscillator like C. subsalsa (this study) and other dinoflagellates (Forward 1975). Perhaps entraining phototaxis to the circadian clock is inhibitory to an opportunistic predator like K. veneficum (Place et al. 2012), where the optimal depth may be independent of light (i.e. K. veneficum may migrate towards its prey, e.g. see Bollens et al. 2011). In contrast, although C. subsalsa is likely bactivorous (Jeong 2011), a greater reliance on photosynthesis may make phototactic circadian rhythms more advantageous. Differences between clock entrainment of DVM in K. veneficum and C. subsalsa are not unexpected, since photo and geotactic responses vary even among different dinoflagellate species (Kamykowski et al. 1998b). Previous reports of DVM in constant light in other dinoflagellates and in Chattonella antiqua show rhythms with high amplitude lasting >3 days (Weiler and Karl 1979, Roenneberg et al. 1989, Shikata et al. 2013). Thus, the lack of a robust circadian rhythm in K. veneficum, and to a lesser extent C. subsalsa, was surprising. These results may reflect the true extent of circadian rhymicity that these particular species exhibit in nature, i.e. zero entrainment (K. veneficum) or low levels of entrainment (C. subsalsa). Alternatively, a number of other factors may have limited rhythmicity observed in this study. Firstly, light spectra used for entrainment may not have been optimal, since entrainment to white light yields weaker DVM rhythms in C. antiqua compared to blue light (Shikata et al. 2013). Spectral changes experienced in the environment may entrain stronger rhythms of DVM (see also Figueroa et al. 1998). Secondly, culture artefacts may have developed in these cultures over time, since circadian rhythms in photosynthesis have been lost in older cultures of G. polyedra

119 (Sweeney 1986), and phototaxis was lost in older cultures of Kryptoperidinium foliaceum and linked to eyspot degeneration (Moldrup et al. 2013). However, the cultures used in this study have only been in culture for ~8 years, hence mutation may be less likely. Lastly, deprivation of non-photic clock inputs (e.g. nitrate and pH) may have repressed circadian rhythmicity in DVM (Roenneberg and Rehman 1996, Eisensamer and Roenneberg 2004). Photosynthesis and PSII activity are known to show circadian rhythms in a number of phytoplankton (Brand 1982) including many dinoflagellates (Hastings et al. 1961, Prezelin and Sweeney 1977, Prezelin et al. 1977, Sorek et al. 2013) and can continue with high amplitude in excess of 3 days in constant conditions (Brand 1982, Prezelin et al. 1977). In particular, the maximum quantum yield of PSII follows a circadian rhythm (e.g. Samuelson et al. 1983, MacKenzie and Morse 2011, Sorek et al. 2013), and was initially thought to stem from PSII rather than downstream from the PQ pool or PSI (Samuelsson et al. 1983). Additionally, Sorek et al. (2013) noted that along with rhythms in Fv/Fm, transcription of the oxygen-evolving enhancer gene showed circadian rhythmicity in Symbiodinium sp. implying that PSII rhythmicity is linked to the donor side of PSII. Consequently it was surprising to observe such low circadian variability in PSII fluorescence measures in both species in constant conditions here. Most prior results showing circadian regulation of Fv/Fm were determined with multiple turnover fluorescence induction, whereby the plastoquinone- pool becomes fully reduced, but this had no effect on the circadian rhythm of Fv/Fm in C. subsalsa here (Fig. A3). MacKenzie and Morse (2011) found no circadian rhythm in oxygen evolution or Fv/Fm in extracted chloroplasts from L. polyedrum, and proposed that CO2 availability drives the circadian rhythms of photosynthesis and

120 electron transport in L. polyedrum in vivo. However, components localised at PSII may affect Fv/Fm more than oscillations mediated by the dark reactions. For example, psbA synthesis and degradation is controlled by light rather than a circadian oscillator

(Wang et al. 2005b) and could severely affect PSII function and Fv/Fm. In K. veneficum, the majority of the diel periodicity in Fv/Fm stems from the night-time drop which may be caused by chlororespiration that reduces the PQ pool (Jones and Hoegh-

Guldberg 2001, Peltier and Cournac 2002). Large decreases in Fv/Fm by chlororespiration do not occur in the light because accumulation of PQH2 is prevented by linear electron transport through the Z-scheme. Thus, a large proportion of the diel variability in Fv/Fm in K. veneficum cannot occur in constant conditions, as no shift between chlororespiration and photosynthesis occurs. If the majority of the diel periodicity were due to light-dependent downregulation at PSII in K. veneficum and C. subsalsa at the surface, then diel periodicity is unlikely to continue in the absence of changes in light intensity. In at least one diatom, both photoprotective and photosynthetic pigment synthesis is under some degree of circadian control (Ragni and D’alcalà 2007). In this regard, if lower light intensities had been used for continuous light, which prevented substantial PSII downregulation, PSII photochemistry at the surface may have shown rhythmicity due to diel periodicity in pigment synthesis and the arrangement into thylakoid membranes.

4.5.3 Value of DVM In K. veneficum, a large reproductive cost was associated with DVM, but due to the low coefficient of determination of the regression to calculate growth (0.38 ± 0.27 S.D.), this result was not considered further. In C. subsalsa, the growth of migrating cultures was compared against a population prevented from migrating.

121 Contrastingly, DVM in C. subsalsa, did not show a reproductive advantage, or cost, under nutrient replete LD conditions compared to cells held at surface light intensities over 6 to 9 days (Fig. 4.5; ‘col.’ vs. ‘high’). Although reliant upon estimated growth rates without mixing the column, the coefficients of determination for these rates were high (0.95 ± 0.04 S.D.), and provide support for our results. This suggests that any adaptive value from DVM stems from conditions not tested here, such as nutrient depletion (Cullen 1985, MacIntyre et al. 1997), or high grazing pressure (Bollens et al. 2012). Only two studies in cyanobacteria have directly measured an adaptive advantage of circadian rhythms (i.e. reproductive fitness, Johnson et al. 2005, Woelfe et al. 2004, Xu et al. 2013), so the unexpected lack of a circadian rhythm in DVM in our K. veneficum isolate may present a unique strain to study the adaptive value of circadian rhythms if new isolates from the field do show high levels of circadian rhythmicity in DVM. Since dinoflagellates have not yet been amenable to laboratory transformation, this could be particularly useful (Lapointe and Morse 2008).

4.5.4 Conclusion We present the first detailed assessment of DVM in K. veneficum and found notably distinct patterns of DVM between K. veneficum and C. subsalsa. Each species’ photosynthetic physiology appeared to define their migration patterns in the conditions used in the current study. C. subsalsa appeared to be a high-light specialist, and should be competitive during peak midday light intensities in the field. C. subsalsa tended to have a wide vertical distribution, which may influence lateral population dispersal and losses in the field. K. veneficum’s migration was consistent with this species’ mixotrophic and generalist life style, and we suggest that its lack of circadian control of DVM is consistent with the potential for prey-dependent depth

122 orientation. To more fully determine the migratory preferences of these algae, DVM will need to be assessed under more conditions, such as nutriclines and with predator/prey additions, which will further help to characterize the possible ecological and biogeochemical roles that such migratory patterns play in nature.

123

Table 4.1: Descriptions of photosynthetic components, and chlorophyll a fluorescence terms used in this study.

Components & Definition (units where applicable) Terms - Qa, Qa Oxidized and reduced forms of the primary quinone at PSII

Qb Oxidized form of the secondary quinone PSI, PSII Photosystem I and Photosystem II RCII Reaction Center of PSII LHCII Light harvesting complex of PSII

Fo, Fm Minimum and maximum fluorescence in the dark, after dark aclimation

Fv Variable fluorescence between Fo and Fm (= Fm-Fo)

Fv/Fm Maximum quantum yield (efficiency) of PSII photochemistry in the dark

F’, Fm’ Steady-state and maximum fluorecence in actinc light

Fq’ Varibale fluorescence between F’ and Fm’ (= Fm’-F’)

Fq’/Fm’ Effective quantum yield (efficiency) of PSII photochemistry in actinc light 2 σPSII Maximum effective absorption cross section (nm ) of PSII in the dark 2 σPSII’ Steady-state effective absorption cross section (nm ) of PSII in actinic light

LHC-quenching Non-photochemical quenching of PSII fluorecence from the LHCII/antennae bed (= 1- σPSII’/σPSII) τ Time (µs) for re-oxidation/re-opening of PSII reaction centers in the dark (i.e. Qa- to Qa via Qb reduction) τ’ Time (µs) for re-oxidation/re-opening of PSII reaction centers in actinc light ρ Connectivity among PSII units in the dark (probability of absorbed excitation migrating from closed to open RCIIs) ρ’ Connectivity among PSII units in actinc light

124

125 Figure 4.1: DVM and PSII photochemistry in K. veneficum (A, C, E, G, I) and C. subsalsa (B, D, F, H, J) measured every 3 hours over 39 hours, with columns kept in 14:10 LD cycles. Measurement began on the third day after inoculation. Filled grey bars underneath the data represent the dark period, and white areas represent the light period. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, 2 B); Fv/Fm (C, D); σPSII (nm , E, F); τ (µs, G, H); ρ (I, J). DVM is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measures from the bottom of the columns (filled circles) and at the surface of the columns (open cicles) are presented. Note the difference in scale between panels E and F. Error bars represent ±1 SEM where n=2 for K. veneficum, and n=3 for C. subsalsa.

126

127 Figure 4.2: DVM and PSII photochemistry in K. veneficum (A, C, E, G, I) and C. subsalsa (B, D, F, H, J) measured an 1 hour before dawn (05:00), at midday (13:00), and 1 hour before dark (19:00) every day for 6 days, with columns kept under 14:10 LD cycles. Measurement began on the second day after inoculation. Filled grey bars underneath the data represent the dark period, and white areas represent the light period. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, B); Fv/Fm (C, D); PSII light harvesting center quenching (E, F); τ (µs, G, H); ρ (I, J). DVM is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measurements at the bottom (filled circles), at the surface (open circles), and light acclimated measurements at the surface (open inverted triangles) are presented. Light acclimated measurements at the bottom were obscuring dark measurements, and so were excluded for clarity. Asterisks in C, D represent significant differences (p<0.05) between dark acclimated measurements at the surface and bottom samples as determined by RM- ANOVA and Bonferroni post tests (n=3). Error bars represent ±1 SEM.

128

129

130 Figure 4.4: DVM and PSII photochemistry in K. veneficum (A,C,E,G,I) and C. subsalsa (B,D,F,H,J) measured at 1 hour before dawn (05:00), midday (13:00), and 1 hour before dark (19:00) every day for 6 days, in the continuous light experiment (LL). Columns were kept under 14:10 LD cycles until day 3, when columns were shifted at midday (13:00) into continuous midday light for 2 days, before being returned to 14:10 LD cycles at midday (13:00) for a further 24 hours. Light intensity during the incubations is presented above panels A and B. Dark filled grey bars underneath the data represent dark periods, white areas represent light periods, and pale filled grey bars represent ‘night-time’ during the constant light period. Light intensity is presented above panels A, B. Presented for each species are: DVM (A, B); Fv/Fm (C, D); τ (µs, E, F); 2 σPSII (nm , G, H); ρ (I, J). DVM (A, B) is presented as the median (filled squares), upper (open triangles), and lower (open inverted triangles) quartiles. In photochemistry figures (C-J), dark acclimated measurements at the bottom (filled circles), and at the surface (open circles) are presented. Note the difference in scale between panels G and H. Error bars represent ±1 SEM.

131

132

133 Chapter 5

CONCLUSIONS

The increasing frequency of HABs worldwide portends increasing societal and environmental costs for the future. The problem can be tackled simultaneously from two fronts. First, by gaining a better understanding of the core biology and ecology of these organisms to determine the underlying drivers of these trends (i.e. a ‘bottom-up’ approach). Second, the economic and ecological damage by HABs may be ameliorated by seeking methods to control the blooms or their symptoms without concern for the underlying drivers of the HAB increase (‘top-down’). Both of these approaches have advantages and disadvantages, and this dissertation adds new findings to both fronts. Identification of the underlying drivers of the HAB problem is what distinguishes the two approaches, where the former strives to identify possible underlying drivers, while the latter does not require this. As a bottom-up approach, the former also requires a comprehensive understanding of the multitudes of components (and their interactions) that make up the problem as a whole, whereas the ‘top-down’ approach does not require this knowledge and research can proceed quickly and independently of that information. However, both approaches are complementary, with the former in particular offering direction for the latter. Thus, the two approaches to the HAB problem are reminiscent of the direction that climate change research has taken, where some focus is given to basic blue skies research, and other focus is given to considering geo-engineering options (Lampitt et al. 2008). Thus, as in climate change research, both approaches should be considered viable and deserving avenues for continued research, and this view is advocated in the HAB literature (e.g.

134 Anderson 2009). This dissertation presents new research in both avenues: firstly in regards to investigating a bacterially derived exudate in the context of its potential use as a biologically-derived dinoflagellate control agent, and secondly as a comparison of behaviour and physiology between two harmful algae. Chapter two investigated how bacterial filtrates of Shewanella sp. IRI-160 (Hare et al. 2005) affected photosynthetic electron flow through photosystem II (PSII) in susceptible dinoflagellates. Alongside growth inhibition, functional electron flow through PSII was inhibited in all four tested dinoflagellates, and both variables were found to be dose dependent with filtrate concentration. Although many natural algicides and allelochemicals affect PSII function (e.g. Smith and Doan 1999 and references therein), it is not yet known how PSII is affected by the algicide in the current study. It was however determined that inhibition at PSII was extensive, as

- evidenced by inhibition in Fv/Fm, Qa re-oxidation and PSII connectivity. Moreover, susceptibility of the heterotrophic dinoflagellate Oxyrrhis marina (noted by Pokrzywinski et al. 2012) implies that PSII cannot be the sole target of the algicide. The overarching trend in these experiments was that susceptibility was species specific, but that thecae in particular provided some protection, corroborating a finding in Pokryzwinski et al. (2012). Moreover, the naked dinoflagellates differed in their temporal susceptibility, with cells of G. instriatum losing viability sooner than K. veneficum. A novel outcome from this study was that darkness expedited and enhanced PSII dysfunction and cell mortality in K. veneficum, but not in G. instriatum. This type of interaction has only been noted once before (Mayali et al. 2007), and it was speculated that in this case, physiological differences in the form of higher basal respiratory demands in K. veneficum, led to this enhanced susceptibility in the dark.

135 Overall, chapter two revealed important baseline information on how the filtrate affected dinoflagellate growth and physiology in optimum culture conditions, and was valuable for the subsequent study on effects in natural water microcosms. The results of chapter two are complementary to the results and conclusions presented in Pokrzyziwinski (2014), where it is shown that the algicidal filtrate induces a form of programmed cell death in affected dinoflagellates, and that nuclear degradation is a crucially important component of this process and to the dismantling of affected cells. These results too, provide important context for the results in chapter 3 of this dissertation. Chapter three studied the effect of the algicidal filtrate in natural water microcosms with the aid of molecular fingerprinting (PCR-DGGE) and quantitative PCR (qPCR). Dinoflagellate growth was inhibited in almost all experiments, but generally to a lesser degree than in culture. This result was explained by low growth measured in these dinoflagellates, and underlined the necessity of a monitoring campaign if this algicide (in its current form) were to be operationalized. This is a notable limitation of the algicide, and more broadly highlights how high target specificity among biologically derived agents, can become disadvantageous beyond a certain specificity threshold and may hamper its utility (e.g. extreme virus-host specificity, Nagasaki et al. 1999). The growth of heterotrophic protists was a common theme among microcosms, and increased in a dose dependent manner with algicide concentration. These heterotrophic protists were thought to have arisen due to bacterial growth from dying dinoflagellates and from organic impurities in the filtrate. This appeared to be one route transferring harmful algae carbon toward non-harmful organisms. Since this effect was noted in other algal control studies (e.g. Kang et al.

136 2011, Jung et al. 2013), it may be a common process among HAB control techniques that cause mortality in the water column. That dinoflagellates could be controlled, that the total algal biomass remained high, and that the microbial community continued to function (PSII function, and growth of certain organisms) in algicide treated microcosms, suggested that the algicide holds promise as a dinoflagellate control agent. One possible limitation of the current work is that large applications in the field may show greater variability than that observed in small controlled in-vitro microcosms. This may for example be due to the addition of in-situ factors such as, different light quality and quantity, dilution with adjacent water, addition of meso and macrograzers, and greater microbial diversity. Given that aquatic ecosystems harbor enormous microbial diversity, the effort in the current research to quantify changes to the microbial community following algicide addition, can only be described as a superficial assessment at best. Consequently, continued work will be required to better quantify the prokaryotic and eukaryotic community dynamics following algicide addition. Work in this direction should prove interesting, particularly because the planktonic community present at the time of algicidal bloom termination may be mis- matched to that usually present during natural bloom termination, altering the natural succession. More research will also be required to identify whether potency can be feasibly increased in lieu of dinoflagellate growth, and whether presumed bacterial growth can be minimized, both of which may be achieved via purification of the active compounds. Thus, another important caveat of the research presented here on the effects of IRI-160AA, is that the biological responses to raw sterilized bioactive filtrates, may not be identical to those using purified compounds. Despite the importance of citing this issue, the results and interpretations can still be valuable, and

137 conclusions on the mode of action may be less likely to be confounded by this, than conclusions on changes to the microbial community. Finally, evidence in the literature points to a great degree of success in regards to bioprospecting for biologically- derived agents that are effective inhibitors of harmful algae. However, it will be important to fully understand the mechanisms underlying the observed allelopathic effects of these agents. The investigations here, and in Pokryziwinski (2014), suggest that these mechanisms may sometimes be complex and multi-faceted, and that a combination of many and varied approaches will be necessary to fully evaluate these agents for their supposed purpose. Despite years of ‘hesitance’ and slow growth (Anderson 2004), the field of bio-control techniques for HABs appears poised for exciting leaps in development in the near future. Chapter four compared the patterns of diel vertical migration of two potentially harmful algae, K. veneficum (a dinoflagellate) and C. subsalsa (a raphidophyte). Detailed comparisons on the vertical migrations of different algae are rare, so this study provides an opportunity for comparison between two different, and important harmful flagellates. Although both species exhibited diel vertical migration by migrating to the surface during the day and to depth at night, other aspects of the migration were found to be distinct. One of these differences was in the degree of population synchrony that could pose advantages or disadvantages in situ by controlling population flushing, and represents an interesting avenue for further research, especially given the biogeographic range expansions observed in some harmful algae. A difference in the rate of descent was also observed between the species, and suggested that different sensory or motor responses were being used by each alga. An investigation to confirm the suspected sensory perceptions in these algae

138 would prove useful for future work. Differences in migration patterns were interpreted in the context of 1) coincident measurements of PSII photochemistry, 2) by further experiments that determined carbon acquisition rates in cells obtained from different depths in the column, and 3) by specific growth rates determined at surface and bottom light levels. Each species employed different photoacclimation and photoprotection mechanisms. Specifically, K. veneficum migrated in a step-wise pattern, initially avoiding high light, whereas C. subsalsa migrated immediately near to the surface. Moreover, K. veneficum exhibited much larger light harvesting center (LHCII) quenching than C. subsalsa, which instead appeared to photoprotect PSII by reaction centre II (RCII) downregulation. In K. veneficum, cells retrieved from intermediate depths supported the maximal rate of carbon assimilation, and is consistent with the hypothesis that this species migrates to lower depths that are photosynthetically optimal for this species. C. subsalsa showed maximal carbon assimilation rates at the surface, which was also consistent with this species’ migration into surface waters. Specific growth rates measured in cells maintained statically at surface and bottom light intensities indicated that C. subsalsa had lower growth rates at bottom light intensities compared to K. veneficum, which is consonant with C. subsalsa’s immediate migration into surface light intensities. Migration was controlled by an endogenous circadian rhythm in C. subsalsa, but not in K. veneficum, which corresponds with these species’ trophic flexibility, namely, a greater dependence on photosynthesis in C. subsalsa, and greater trophic flexibility in K. veneficum that may warrant prey-dependent depth orientation rather than light-dependent orientation. Overall, this study helped to identify the spatial niche that these species occupy in quiescent dead-end canals where these algae frequently bloom in the Delaware Inland

139 Bays. Importantly, migration patterns are species specific, and appear to correspond well with the photosynthetic and physiological capabilities that distinguish these organisms. The hypothesis I cited at the start of this chapter, that basic HAB science can often be directly utilized to the benefit of ‘control-orientated’ research, can be exemplified with the studies presented in this dissertation. For example, the migration work presented in chapter 4 will be helpful to define how monitoring programs for dinoflagellates could be best implemented (both spatially and temporally), and more specifically, it identifies how algicide applications should consider depth dependent applications to maximize contact time and to synchronize with times of highest physiological susceptibility. HABs are a major and growing problem worldwide, and theoretically, the environmental and societal costs are unlikely to abate in the short term. Research in this thesis adds to two bodies of literature, the first regarding the potential control of harmful algal blooms in nature, and the second, to the general understanding of harmful algae biology, focused on comparing behaviour and physiology between two harmful algae. I advocate that continued research in both realms is likely to be necessary to adequately solve the growing HAB problem in a timely manner.

140 REFERENCES

Allan CY, Fisher PR (2011) Phototaxis: Microbial. In: Encyclopedia of life sciences. DOI: 10.1002/9780470015902.a0000399.pub2.

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol, 215(3), 403-410.

Anderson DM (1997) Turning back the harmful red tide. Nature, 388, 513-514.

Anderson DM (2004) Prevention, control and mitigation of harmful algal blooms: Multiple approaches to HAB management. In: Hall S, Etheridge S, Anderson D, Kleindinst J, Zhu M, Zou Y (eds) Harmful algae management and mitigation. Asia-Pacific Economic Coooperation (Singapore): APEC Publication #204-MR-04.2.

Anderson DM (2009) Approaches to monitoring, control and management of harmful algal blooms (HABs). Ocean Coast Manag, 52(7), 342-347.

Anderson DM, Stolzenbach KD (1985) Selective retention of two dinoflagellates in a well-mixed estuarine embayment: the importance of diel vertical migration and surface avoidance. Mar Ecol Prog Ser, 25, 39-50.

Anderson DM, Keafer BA (1987) An endogenous annual clock in the toxic marine dinoflagellate Gonyaulax tamarensis. Nature, 325, 616-617.

Anderson DM, Glibert PM, Burkholder JM (2002) Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries, 25, 704-726.

Anderson DM, Pitcher GC, Estrada M (2005) The comparative “systems” approach to HAB research. Oceanography, 18, 148–157.

Anderson DM, Burkholder JM, Cochlan W, Glibert PM, Gobler CJ, Heil CA, Kudela RM, Parsons ML, Rensel JEJ, Townsend DW, Trainer VL, Vargo GA (2008) Harmful algal blooms and eutrophication: Examining linkages from selected coastal regions of the United States. Harmful Algae, 8, 39–53.

141 Anderson DM, McGillicuddy DJ, Keafer BA, He R, Townsend DW (2013) Population dynamics of Alexandrium fundyense in the Gulf of Maine: outlook for improved management and forecasting. In: Pagou P, Hallegraeff G (eds.) Proceedings of the 14th International Conference on Harmful Algae. ISSHA IOC, ISBN 978-87-990827-3-5.

Archambault MC, Bricelj VM, Grant J, Anderson DM (2004) Effects on juvenile hard clams, Mercenaria mercenaria, from the application of phosphatic clays to mitigate harmful algal blooms. Mar Biol, 144, 553–565.

Asada K (1999) The water-water cycle in chloroplasts: scavenging of active oxygens and dissipation of excess photons. Annu Rev Plant Physiol Plant Mol Biol, 50, 601-639.

Ault TR (2000) Vertical migration by the marine dinoflagellate Prorocentrum triestinum maximizes photosynthetic yield. Oecologia, 125, 466-475.

Azanza RV, Fukuyo Y, Yap LG, Takayama H (2005) Prorocentrum minimum bloom and its possible link to a massive fish kill in Bolinao, Pangasinan, Northern Philippines. Harmful Algae, 4(3), 519-524.

Bachmann RW, Hoyer MV, Fernandez C, Canfield DE (2003) An alternative to proposed phosphorus TMDLs for the management of Lake Okeechobee. Lake Resev Manage, 19, 251-264.

Baek SH, Son M, Bae SW, Shin K, Na DH, Cho H, Yamaguchi M, Kim YO, Kim SW (2013) Algicidal activity of the thiazolidinedione derivative TD49 against the harmful dinoflagellate Heterocapsa circularisquama in a mesocosm enclosure. J Appl Phycol, 25, 1555-1565.

Baudoux A-C, Brussaard CPD (2005) Characterization of different viruses infecting the marine harmful algal bloom species Phaeocystis globosa. Virology, 341(1), 80-90.

Behrenfeld MJ, Prasil O, Kolber ZS, Babin M, Falkowski PG (1998) Compensatory changes in Photosystem II electron turnover rates protect photosynthesis from photoinhibition. Photosynth Res, 58, 259-268.

Berden-Zrimec M, Flander-Putrle V, Drinovec L, Zrimec A, Monti M (2008) Growth, delayed fluorescence and pigment composition of four Prorocentrum minimum strains growing at two salinities. Biol Res, 41, 11-23.

142 Bergholtz T, Daugbjerg N, Moestrup Ø, Fernandez-Tejedor M (2005) On the identity of Karlodinium veneficum and description of Karlodinium armiger sp. nov. (dinophyceae), based on light and electron microscopy, nuclear-encoded LSU rDNA, and pigment composition. J Phycol, 42, 170-193.

Bergo E, Segalla A, Giacometti GM, Tarantino D, Soave C, Andreucci F, Barbato R (2003) Role of visible light in the recovery of photosystem II structure and function from ultraviolet-B stress in higher plants. J Exp Bot, 54(388), 1665- 1673.

Blackburn SI, Hallegraeff GM, Bolch CJ (1989) Vegetative reproduction and sexual life cycle of the toxic dinoflagellate Gymnodinium catenatum from Tasmania, Australia. J Phycol, 25, 577-590.

Blom JF, Brutsch T, Barbaras D, Bethuel Y, Locher HH, Hubschwerlen C, Gademann K (2006) Potent algicides based on the cyanobacterial alkaloid nostocarboline. Org Lett, 8, 737–740.

Blum U, Staman KL, Flint LJ, Shafer SR (2000) Induction and/or selection of phenolic acid-utilizing bulk soil and rhizosphere bacteria and their influence on phenolic acid phytotoxicity. J Chem Ecol, 26(9), 2059-2078.

Bolch CJS, de Salas MF (2007) A review of the molecular evidence for ballast water introduction of the toxic dinoflagellates Gymnodinium catenatum and the Alexandrium “tamarensis complex” to Australasia. Harmful Algae, 6, 465–85.

Bollens SM, Quenette JA, Rollwagen-Bollens G (2012) Predator enhanced diel vertical migration in a planktonic dinoflagellate. Mar Ecol Prog Ser, 447, 49- 54.

Bollens SM, Rollwagen-Bollens G, Quenette JA, Bochdansky AB (2011) Cascading migrations and implications for vertical fluxes in pelagic ecosystems. J Plankton Res, 33(3), 349-355.

Bouwman L, Beusen A, Glibert PM, Overbeek C, Pawlowski M, Herrera J, Mulsow S, Yu R, Zhou M (2013) Mariculture: significant and expanding cause of coastal nutrient enrichment. Environ Res Lett, 8, 044026.

Boylan JD, Morris JE (2003) Limited effects of barley straw on algae and zooplankton in a midwestern pond. Lake Reserv Manage, 19, 265–271.

Brand LE (1982) Persistent diel rhythms in the chlorophyll fluorescence of marine phytoplankton species. Mar Biol, 69(3), 253-262.

143 Breton E, Rousseau V, Parent JY, Ozer J, Lancelot C (2006) Hydroclimatic modulation of diatom/Phaeocystis blooms in nutrient-enriched Belgian coastal waters (North Sea). Limnol Oceanogr, 51(3), 1401-1409.

Brosnahan ML (2011) Life cycle studies of the red tide dinoflagellate species complex Alexandrium tamarense. Ph.D. Thesis, Massachusetts Institute of Technology, and Woods Hole Oceanographic Institution.

Brown BE, Ambarsari I, Warner ME, Fitt WK, Dunne RP, Gibb SW, Cummings DG (1999) Diurnal changes in photochemical efficiency and xanthophyll concentrations in shallow water reef corals: Evidence for photoinhibition and photoprotection. Coral Reefs, 18, 19-105.

Brussaard CPD, Kuipers B, Veldhuis MJW (2005) A mesocosm study of Phaeocystis globosa population dynamics: I. Regulatory role of viruses in bloom control. Harmful Algae, 4(5), 859-874.

Burkholder JM, Glibert PM (2006) Intraspecific variability: An important consideration in forming generalisations about toxigenic algal species. S Africa J Mar Sci, 28(2), 177-180.

Burson A, Matthijs HCP, Bruijne WD, Talens R, Hoogenboom R, Gerssen A, Visser PM, Stomp M, Steur K, Scheppingen YV, Huisman J (2014) Termination of a toxic Alexandrium bloom with hydrogen peroxide. Harmful Algae, 31, 125- 135.

Byrne TE, Wells MR, Johnson CH (1992) Circadian rhythms of chemotaxis to ammonium uptake in Chlamydomonas. Plant Physiol, 98, 879–86.

Cembella AD, Ibarra DA, Diogene J, Dahl E (2005) Harmful algal blooms and their assessment in fjords and coastal embayments. Oceanography, 18(2), 158-171.

Chang X, Eigemann F, Hilt S (2012) Do macrophytes support harmful cyanobacteria? Interactions with a green alga reverse the inhibiting effects of macrophyte allelochemicals on Microcystis aeruginosa. Harmful Algae, 19, 76-84.

Chen CC, Petersen JE, Kemp WM (1997) Spatial and temporal scaling of periphyton growth on walls of estuarine mesocosms. Mar Ecol Prog Ser, 155, 1–15.

Clarke KR, Somerfield PJ, Gorley RN (2008) Testing of null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J Exp Mar Biol Ecol, 366, 56-69.

144 Cosgrove J, Borowitzka MA (2011) Chlorophyll Fluorescence Terminology: An Introduction. In: Suggett DJ, Borowitzka MA, Prasil OJ (eds) Chlorophyll a Fluorescence in Aquatic Sciences: Methods and Applications. Springer, Dordrecht Heidelberg London New York, pp 1-18.

Coyne KJ, Handy SM, Demir E, Whereat EB, Hutchins DA, Portune KJ, Doblin MA, Cary SC (2005) Improved quantitative real-time PCR assays for enumeration of harmful algal species in field samples using an exogenous DNA reference standard. Limnol Oceanogr-Meth, 3, 381–391.

Crawford DW, Purdie DA (1992) Evidence for avoidance of flushing from an estuary by a planktonic, phototrophic ciliate. Mar Ecol Prog Ser, 79, 259-265.

Cullen JJ (1985) Diel vertical migration by dinoflagellates: roles of carbohydrate metabolism and behavioral flexibility. Contr Mar Sci, 27, 135-152.

Cullen JJ, Horrigan SG (1981) Effects of nitrate on the diurnal vertical migration, carbon to nitrogen ratio, and the photosynthetic capacity of the dinoflagellate Gymnodinium splendens. Mar Biol, 62, 81-89.

Cullen JJ, MacIntyre JG (1998) Behavior, physiology and the niche of depth- regulating phytoplankton. In: Anderson DM, Cembella AD, Hallegraeff GM (eds.) Physiological ecology of harmful algal blooms. NATO ASI Series Vol G 41. Springer-Verlag, Berlin, pp. 21.

Daugbjerg N, Hansen G, Larsen J, Moestrup Ø (2000) Phylogeny of some of the major genera of dinoflagellates based on ultrastructure and partial LSU rDNA sequence data, including the erection of three new genera of unarmoured dinoflagellates. Phycologia, 39, 302-317.

Davidson K, Gowen RJ, Tett P, Bresnan E, Harrison PJ, McKinney A, Milligan S, Mills DK, Silke J, Crooks A-M (2012) Harmful algal blooms: How strong is the evidence that nutrient ratios and forms influence their occurrence? Estuar Coast Shelf S, 115, 399-413.

Dempster EL, Pryor KV, Francis D, Young JE, Rogers HJ (1999) Rapid DNA extraction from ferns for PCR-based analyses. Biotechniques, 27, 66-68.

Doblin MA, Blackburn SI, Hallegraeff GM (2000) Intraspecific variation in the selenium requirement of different geographic strains of the toxic dinoflagellate Gymnodinium catenatum. J Plankton Res, 22, 421–432.

145 Doblin MA, Thompson PA, Revill AT, Butler ECV, Blackburn SI, Hallegraeff GM (2006) Vertical migration of the toxic dinoflagellate Gymnodinium catenatum under different concentrations of nutrients and humic substances in culture. Harmful Algae, 5, 665-677.

Dodd AN, Salathia N, Hall A, Kévei E, Tóth R, Nagy F, Hibberd JM, Millar AJ, Webb AAR (2005) Plant circadian clocks increase photosynthesis, growth, survival and competitive advantage. Science, 309, 630-633.

Donovan CJ, Garduno RA, Kalmokoff M, Ku JC, Quilliam MA, Gill TA (2009) Pseudoalteromonas bacteria are capable of degrading paralytic shellfish toxins. Appl Environ Microbiol, 75, 6919–6923.

Dopheide A, Lear G, Stott R, Lewis G (2011) Preferential feeding by the ciliates Chilodonella and Tetrahymena spp. and effects of these protozoa on bacterial biofilm structure and composition. Appl Env Microb, 77(13), 4564–4572.

Doucette GJ (1995) Interaction between bacteria and harmful algae: a review. Nat Toxins, 3(2), 65-74.

Edvardsen B, Paasche E (1998) Bloom dynamics and physiology of Prymnesium and Chrysochromulina. In: Anderson DM, Cembella AD, Hallegraeff GM (eds) Physiological ecology of harmful algal blooms. NATO ASI Series 41. Springer, Berlin Heidelberg New York, pp 193–208.

Edwards M, Johns DG, Leterme SC, Svendsen E, Richardson AJ (2006) Regional climate change and harmful algal blooms in the northeast Atlantic. Limnol Oceanogr, 51(2), 820-829.

Eggersdorfer B, Hader D-P (1991) Phototaxis, gravitaxis, and vertical migrations in the marine dinoflagellate, Prorocentrum micans. FEMS Microb Ecol, 85, 319- 326.

Einhellig FA (1999) An integrated view of allelochemcials amid multiple stresses. In: Inderjit, Dakshini KMM, Foy CL (eds) Principles and Practices in Plant Ecology: Allelochemical Interactions. CRC Press, Boca Raton, pp 479-494.

Eisensamer B, Roenneberg T (2004) Extracellular pH is under circadian control in Gonyaulax polyedra and forms a metabolic feedback loop. Chronobiol Int, 21(1), 27-41.

Endo T, Shikanai T, Takabayashi A, Asada K, Sato F (1999) The role of chloroplastic NAD(P)H dehydrogenase in photoprotection. FEBS Lett, 457, 5-8.

146 Eppley RW, Holm-Hansen O, Strickland JDH (1968) Some observations on the vertical migrations of dinoflagellates. J Phycol, 4, 333-340.

Falkowski PG, Fujita Y, Ley AC, Mauzerall DC (1986) Evidence for cyclic electron flow around photosystem II in Chlorella pyrenoidosa. Plant Physiol, 81, 310- 312.

FAO (2004) Marine Biotoxins, FAO Food and Nutrition Paper, 80. Food and Agriculture Organization of the United Nations, Rome, 278 pp.

Faust MA, Larsen J, Moestrup Ø (1999) Potentially toxic phytoplankton. 3. Genus Prorocentrum (Dinophyceae). In: Lindley JA (ed) ICES Identification Leaflets for Plankton. Leaflet no. 184. Copenhagen. pp 1–24.

Fehling J, Davidson K, Bolch CJ, Bates SS (2004) Growth and domoic acid production of Pseudo-nitzschia seriata (P.T. Cleve) H. Peragallo (Bacillariophyceae) under phosphate and silicate limitation. J Phycol, 40, 674- 683.

Figueroa FL, Niell FX, Figueiras FG, Villarino ML (1998) Diel migration of phytoplankton and spectral light field in the Ría de Vigo (NW Spain). Mar Biol, 130, 491-499.

Flynn KJ (2010) Do external resource ratios matter? Implications for modelling eutrophication events and controlling harmful algal blooms. J Mar Syst, 83, 170-180.

Folke C, Carpenter S, Walker B, Scheffer M, Elmqvist T, Gunderson L, Holling CS (2004) Regime shifts, resilience and biodiversity in ecosystem management. Annu Rev Ecol Evol S, 35, 557-581.

Forsström L, Roiha T, Rautio M (2013) Responses of microbial food web to increased allochthonous DOM in an oligotrophic subarctic lake. Aquat Microb Ecol, 68, 171-184.

Forward RB (1988) Diel vertical migration: Zooplankton photobiology and behaviour. Oceanogr Mar Biol Annu Rev, 26, 361-393.

Forward RB (1975) Dinoflagellate phototaxis: pigment system and circadian rhythm as related to diurnal migration. In: Vernberg FJ (ed) Physiological ecology of estuarine organisms. University of South Carolina Press. Columbia. 367-381.

147 Fritz L, Morse D, Hastings JW (1990) The circadian bioluminescence rhythm of Gonyaulax is related to daily variations in the number of light-emitting organelles. J Cell Sci, 195, 321-328.

Fu F-X, Warner ME, Zhang Y, Feng Y, Hutchins DA (2007) Effects of increased temperature and CO2 on photosynthesis, growth and elemental ratios in marine Synechococcus and Prochlorococcus (cyanobacteria). J Phycol, 43, 485-496.

Fu F-X, Zhang Y, Warner ME, Feng Y, Sun J, Hutchins DA (2008) A comparison of future increased CO2 and temperature effects on sympatric Heterosigma akashiwo and Prorocentrum minimum. Harmful Algae, 7(1), 76-90.

Fu F-X, Place AR, Garcia NS, Hutchins DA (2010) CO2 and phosphate availability control the toxicity of the harmful bloom dinoflagellate Karlodinium veneficum. Aquat Microb Ecol, 59, 55-65.

Fukuyo Y (1985) Morphology of Protogonyaulax tamarensis (Lebour) Taylor and Protogonyaulax catenella (Whedon and Kofoid) Taylor from Japanese coastal waters. Bull Mar Sci, 37(2), 529–537.

Furusato E, Asaeda T, Manatunge J (2004) Tolerance for prolonged darkness of three phytoplankton species, Microcystis aeruginosa (Cyanophyceae), Scenedesmus quadricauda (Chlorophyceae), and Melosira ambigua (Bacillariophyceae). Hydrobiologia, 527, 153-162.

Gattuso J-P, Peduzzi S, Pizay M-D, Tonolla M (2002) Changes in freshwater bacterial community composition during measurements of microbial and community respiration. J Plankton Res, 24(11), 1197-1206.

GEOHAB (2006) Global Ecology and Oceanography of Harmful Algal Blooms: HABs in Eutrophic Systems. Glibert P M ed. IOC and SCOR, Paris and Baltimore. 74p.

Gieskes WWC, Leterme SC, Peletier H, Edwards M, Reid PC (2007) Phaeocystis colony distribution in the North Atlantic since 1948 and interpretation of long- term changes in the Phaeocystis hotspot in the North Sea. Biogeochemistry, 83, 49e60.

Gilpin LC, Davidson K, Roberts EC (2004) The influence of changes in nitrogen:silicon ratios on diatom growth dynamics. J Sea Res, 51, 21-35.

Gleason FK, Case DE (1986) Activity of the natural algicide, cyanobacterin, on angiosperms. Plant Physiol, 80(4), 834-837.

148 Glibert PM, Harrison J, Heil C, Seitzinger S (2006) Escalating worldwide use of urea – a global change contributing to coastal eutrophication. Biogeochemistry, 77, 441-463.

Gobler CJ, Hutchins DA, Fisher NS, Cosper EM, Sañudo-Wilhelmy SA (1997) Release and bioavailability of C, N, P, Se, and Fe following viral lysis of a marine chrysophyte. Limnol Oceanogr, 42(7), 1492-1504.

Gobler CJ,Anderson OR, Gastrich MD, Wilhelm SW (2007) Ecological aspects of viral infection and lysis in the harmful brown tide algal Aureococcus anophagefferens. Aquat Microb Ecol, 47, 25-36.

Gobler CJ, Sunda WG (2012) Ecosystem disruptive algal blooms of the brown tide species, Aureococcus anophagefferens and Aureoumbra lagunensis. Harmful Algae, 14, 36-45.

Gobler CJ, Burson A, Koch F, Tang Y, Mulholland MR (2012) The role of nitrogenous nutrients in the occurrence of harmful algal blooms caused by Cochlodinium polykrikoides in New York estuaries (USA). Harmful Algae, 17, 64-74.

Gorbunov MY, Kolber ZS, Lesser MP, Falkowski PG (2001) Photosynthesis and photoprotection in symbiotic corals. Limnol Oceanogr, 46(1), 75-85.

Gowen RJ, Tett P, Bresnan E, Davidson K, McKinney A, Milligan S, Mills DK, Silke J, Harrison P, Crooks AM (2012) Anthropogenic nutrient enrichment and blooms of harmful micro-algae. Oceanogr Mar Biol, 50, 65-126.

Gran HH (1915) The plankton production of north European waters in the spring of 1912. Bulletin Planktonique pour l’année 1912.

Granéli E, Turner JT (2006) Ecology of Harmful Algae. Ecological Studies, Vol. 189 Springer-Verlag, Berlin Heidelberg.

Green DH, Hart MC, Blackburn SI, Bolch CJS (2010) Bacterial diversity of Gymnodinium catenatum and its relationship to dinoflagellate toxicity. Aquat Microb Ecol, 61, 73-87.

Green SJ, Leigh MB, Neufeld JD (2009) Denaturing gradient gel electrophoresis (DGGE) for microbial community analysis. In: Timmis KN (ed) Handbook of hydrocarbon and lipid microbiology. Springer, Heidelberg, Germany, pp 4137- 4158.

149 Guerrero MG, Vega JM, Losada M (1981) The assimilatory nitrate-reducing system and its regulation. Annu Rev Plant Physiol, 32, 169-202.

Guillard RRL (1975) Culture of phytoplankton for feeding marine invertebrates. In: Smith WL, Chanley MH (eds.) Culture of Marine Invertebrate Animals. Plenum Press, NY, USA, pp. 26-60.

Guillard RRL, Ryther JH (1962) Studies of marine planktonic diatoms. I. Cyclotella nana Hustedt and Detonula confervacea Cleve. Can J Microbiol, 8, 229-239.

Gustafsson S, Hultberg M, Figueroa RI, Rengefors K (2009) On the control of HAB species using low biosurfactant concentrations. Harmful Algae, 8, 857-863.

Gustavs L, Schumann R, Eggert A, Karsten K (2009) In vivo growth fluorometry: accuracy and limits of microalgal growth rate measurements in ecophysiological investigations. Aquat Microb Ecol, 55, 95-104.

HAB RDDTT (2008) Harmful Algal Bloom Research, Development, Demonstration, and Technology Transfer National Workshop Report. Dortch, Q, Anderson DM, Ayres DL, Glibert PM (eds) Woods Hole, Massachusetts.

Hader DP, Lebert M, Richter P, Ntefidou M (2003) Gravitaxis and graviperception in flagellates. Adv Space Res, 31(10), 2181-2186.

Hall NS, Litaker RW, Fensin E, Adolf JE, Bowers HA, Place AR, Paerl HW (2008) Environmental factors contributing to the development and demise of a toxic dinoflagellate (Karlodinium veneficum) bloom in a shallow, eutrophic, lagoonal estuary. Estuar Coast, 31, 402-418.

Hall NS, Paerl HW (2011) Vertical migration patterns of phytoflagellates in correlation to light and nutrient availability in a shallow microtidal estuary. Mar Ecol Prog Ser, 425, 1-19.

Hallegraeff GM (1993) A review of harmful algal blooms and their apparent global increase. Phycologia, 32, 79–99.

Hallegraeff GM (2010) Ocean climate change, phytoplankton community responses, and harmful algal blooms: a formidable predictive challenge. J Phycol, 46, 220–235.

Handy SM, Coyne KJ, Portune KJ, Demir E, Doblin MA, Hare CE, Cary SC, Hutchins DA (2005) Evaluating vertical migration behavior of harmful raphidophytes in the Delaware Inland Bays utilizing quantitative real-time PCR. Aquat Microb Ecol, 40, 121-132.

150 Handy SM, Demir E, Hutchins DA, Portune KJ, Whereat EB, Hare CE, Rose JM, Warner M, Farestad M, Cary SC, Coyne KJ (2008) Using quantitative real- time PCR to study competition and community dynamics among Delaware Inland Bays harmful algae in field and laboratory studies. Harmful Algae, 7, 599-613.

Hare CE, Demir ED, Coyne KJ, Cary SC, Kirchman DL, Hutchins D (2005) A bacterium that inhibits the growth of Pfiesteria piscicida and other dinoflagellates. Harmful Algae, 4, 221-234.

Van-Haren H, Compton TJ (2013) Diel vertical migrations in deep sea plankton is finely tuned to latitudinal and seasonal day length. PLoS ONE, 8(5), e64435.

Hasle GR (1950) Phototactic vertical migration in marine dinoflagellates. Oikos, 2, 2.

Hastings JW (2001) Fifty years of fun. J Biol Rhythms, 16(1), 5-18.

Hastings JW, Astrachan L, Sweeney BM (1961) A persistent daily rhythm in photosynthesis. J Gen Physiol, 45, 69–76.

Heisler J, Glibert P, Burkholder J, Anderson D, Cochlan W, Dennison W, Dortch Q, Gobler C, Heil C, Humphries E, Lewitus A, Magnien R, Marshall HG, Sellner K, Stockwell D, Stoecker DK, Suddleson M (2008) Eutrophication and harmful algal blooms: a scientific consensus. Harmful Algae, 8, 3-13.

Hennige SJ, Suggett DJ, Warner ME, McDougall KE, Smith DJ (2009) Photobiology of Symbiodinium revisited: bio-physical and bio-optical signatures. Coral Reefs, 28, 179-195.

Hennige SJ, Coyne KJ, MacIntyre H, Liefer J, Warner ME (2013) The photobiology of Heterosigma akashiwo. Photoacclimation, diurnal periodicity, and its ability to rapidly exploit exposure to high light. J Phycol, 49(2), 349-360.

Hernández-León S, Almeida C, Yebra L, Arístegui J (2002) Lunar cycle of zooplankton biomass in subtropical waters: Biogeochemical implications. J Plankton Res, 24(9), 935-939.

Hoagland P, Scatasta S (2006) The economic effects of harmful algal blooms. In: Granéli E, Turner JT (eds) Ecology of Harmful Algae. Springer, Verlag Berlin Heidelberg, pp 391-402.

151 Hooper DU, Chapin FS, Ewel JJ, Hector A, Inchausti P, Lavorel S, Lawton JH, Lodge DM, Loreau M, Naeem S, Schmid B, Setälä H, Symstad AJ, Vandermeer J, and Wardle DA (2005) Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol Monogr, 75, 3–35.

Hutchins DA, Mulholland MR, Fu F (2009) Nutrient cycles and marine microbes in a CO2 enriched world. Oceanography, 22(4), 128-145.

Ihnken S, Kromkamp JC, Beardall J (2011) Photoacclimation in Dunaliella tertiolecta reveals a unique NPQ pattern upon exposure to irradiance. Photosynth Res, 110, 123-137.

Isbell F, Calcagno V, Hector A, Connolly J, Harpole WS, Reich PB, Scherer-Lorenzen M, Schmid B, Tilman D, van Ruijven J, Weigelt A, Wilsey BJ, Zavaleta ES, Loreau M (2011) High plant diversity is needed to maintain ecosystem services. Nature, 477, 199-202.

Ivanov AG, Sane PV, Hurry V, Öquist G, Huner NPA (2008) Photosystem II reaction centre quenching: mechanisms and physiological role. Photosynth Res, 98, 565-574.

Jančula D, Maršálek B (2011) Critical review of actually available chemical compounds for prevention and management of cyanobacterial blooms. Chemosphere, 85, 1415–1422.

Jauzein C, Collos Y, Laabir M, Vaquer A (2011) Dark metabolism and carbon- nitrogen uncoupling in the toxic dinoflagellate Alexandrium catenella (Dinophyceae). Harmful Algae, 11, 73-80.

Jeans J, Campbell DA, Hoogenboom MO (2013) Increased reliance on photosystem II repair following acclimation to high-light by coral-dinoflagellate symbioses. Photosynth Res, 118, 219-229.

Jeong H, Yim JH, Lee C, Choi SH, Park YK, Yoon SH, Hur CG, Kang HY, Kim D, Lee HH, Park KH, Park SH, Park HS, Lee HK, Oh TK, Kim JF (2005) Genomic blueprint of Hahella chejuensis, a marine microbe producing an algicidal agent. Nucleic Acids Res, 33, 7066-7073.

Jeong HJ (2011) Mixotrophy in the red tide algae raphidophytes. J Eukaryot Microbiol, 58, 215–222.

152 Jeong HJ, Kim JS, Yoo YD, Kim ST, Song JY, Kim TH, Seong KA, Kang NS, Kim MS, Kim JH, Kim S, Ryu J, Lee HM, Yih WH (2008) Control of the harmful alga Cochlodinium polykrikoides by the naked ciliate Strombidinopsis jeokjo in mesocosm enclosures. Harmful Algae, 7, 368–377.

Jephson T, Fagerberg T, Carlsson P (2011) Dependency of dinoflagellate vertical migration on salinity stratification. Aquat Microb Ecol, 63, 255-264.

Jewett EB, Lopez CB, Dortch Q, Etheridge SM, Backer LC (2008) Harmful algal bloom management and response: assessment and plan. Interagency Working Group on Harmful Algal Blooms, Hypoxia, and Human Health of the Joint Subcommittee on Ocean Science and Technology. Washington, DC.

Johnson CH (2001) Endogenous timekeepers in photosynthetic organisms. Ann Rev Physiol, 63, 695-728.

Johnson CH (2005) Testing adaptive value of circadian systems. Method Enzymol, 393(43), 818-837.

Jones RJ, Hoegh-Guldberg O (2001) Diurnal changes in the photochemical efficiency of the symbiotic dinoflagellates (Dinophyceae) of corals: photoprotection, photinactivation and the relationship to coral bleaching. Plant Cell Environ, 24(1), 89-99.

Jung SW, Kim BH, Katano T, Kong DS, Han MS (2008) Pseudomonas fluorescens HYK0210-SK09 offers species-specific biological control of winter algal blooms caused by freshwater diatom Stephanodiscus hantzschii. J Appl Microbiol, 105(1), 186-195.

Jung SW, Kang Y-H, Baek SH, Lim D, Han M-S (2013) Biological control of Stephanodiscus hantzschii (Bacillariophyceae) blooms in a field mesocosm by the immobilized algicidal bacterium Pseudomonas fluorescens HYK0210- SK09. J Appl Phycol, 25(1), 41-50.

Jungo E, Visser PM, Stroom J, Mur LR (2001) Artificial mixing to reduce growth of the blue-green alga Microcystis in Lake Nieuwe Meer, Amsterdam: An evaluation of 7 years of experience. Water Sci Technol: Water Supply, 1, 17- 23.

Kamykowski D, Milligan EJ, Reed RE (1998) Biochemical relationships with the orientation of the autotrophic dinoflagellate Gymnodinium breve under nutrient replete conditions. Mar Ecol Prog Ser, 167, 105-117.

153 Kamykowski D, Milligan EJ, Reed RE (1998b) Relationships between geotaxis/phototaxis and diel vertical migration in autotrophic dinoflagellates. J Plankton Res, 20(9), 1781-1796.

Kang Y-H, Kim BR, Choi HJ, Seo JG, Kim BH, Han MS (2007) Enhancement of algicidal activity by immobilization of algicidal bacteria antagonistic to Stephanodiscus hantzschii (Bacillariophyceae). J Appl Microbiol, 103(5), 1983-1994.

Kang Y-H, Jung SW, Jo S-H, Han M-S (2011) Field assessment of the potential of algicidal bacteria against diatom blooms. Biocontrol Sci Techn, 21(8), 969- 984.

Keller M, Mazuch J, Abraham U, Eom GD, Herzog ED, Volk H-D, Kramer A, Maier B (2009) A circadian clock in macrophages controls inflammatory immune responses. Proc Nat Acad Sci USA, 106(50), 21407-21412.

Kim HG (2006) Mitigation and controls of HABs. In: Granéli E, Turner JT (eds) Ecology of harmful algae. Springer, Verlag Berlin Heidelberg, pp 327-338.

Kim B-H, Hwang SJ, Kim YO, Hwang SO, Takamura N, and Han MS (2007) Effects of biological control agents on nuisance cyanobacterial and diatom blooms in freshwater systems. Microbes Environ, 22, 52–58.

Kim B-H, Sang M, Hwang S-J, Han M-S (2008) In situ bacterial mitigation of the toxic cyanobacterium Microcystis aeruginosa: implications for biological bloom control. Limnol Oceanog-Meth 6:513-522.

Kim D, Kim JF, Yim JH, Kwon S-K, Lee CH, Lee HK (2008b) Red to red - the marine bacterium Hahella chejuensis and its product Prodigiosin for mitigation of harmful algal blooms. J Microbiol Biotechn, 18(10), 1621-1629.

Kim JD, Kim JY, Park JK, Lee CG (2009) Selective control of the Prorocentrum minimum harmful algal blooms by a novel algal-lytic bacterium Pseudoalteromonas haloplanktis. Mar Biotechnol, 11(4), 463-472.

Kim, HG (2010) An overview on the occurrences of harmful algal blooms (HABs) and mitigation strategies in Korean coastal waters. In: Ishimatsu A, Lie H-J (eds) Coastal environmental and ecosystem issues of the East China Sea, pp. 121–131.

Kim Y-M, Wu Y, Duong TU, Ghodake GS, Kim SW, Jin ES, Cho H (2010) Thiazolidinediones as a novel class of algicides against red tide harmful algal species. Appl Biochem Biotechnol, 162, 2273–2283.

154 Kirk JTO (1983) Light and photosynthesis in aquatic ecosystems. Cambridge Univ Press, Cambridge New York. USA, 401 p.

Kolber ZS, Prasil O, Falkowski PG (1998) Measurements of variable chlorophyll fluorescence using fast repetition rate techniques: Defining methodology and experimental protocols. Biochim Biophys Acta, 1367, 88-106.

Kooistra WHCF, de Boer MK, Vrieling EG, Connell LB, Gieskes WWC (2001) Variation along ITS markers across strains of Fibrocapsa japonica (Raphidophyceae) suggests hybridisation events and recent range expansion. J Sea Res, 46(3-4), 213-222.

Kozaki A, Takeba G (1996) Photorespiration protects C3 plants from photoinhibition. Nature, 384, 557-560.

Kubanek J, Snell TW, Pirkle C (2007) Chemical defense of the red tide dinoflagellate Karenia brevis against rotifer grazing. Limnol Oceanogr, 52, 1026-1035.

Kudela RM, Ryan JP, Blakely MD, Lane JQ, Peterson TD (2008) Linking the physiology and ecology of Cochlodinium to better understand harmful algal bloom events: A comparative approach. Harmful Algae, 7(3), 278-292.

Lai Z, Yin K (2014) Physical–biological coupling induced aggregation mechanism for the formation of high biomass red tides in low nutrient waters. Harmful Algae, 31, 66-75.

Lampitt RS, Achterberg EP, Anderson TR, Hughes JA, Iglesias-Rodriguez MD, Kelly-Gerreyn BA, Lucas M, Popova EE, Sanders R, Shepherd JG, Smythe- Wright D, Yool A (2008) Ocean fertilization: A potential means of geoengineering? Phil Trans R Soc A, 366, 3919-3945.

Lane DJ (1991) 16S/23S rRNA sequencing. In: Stackebrandt E, Goodfellow M (eds) Nucleic Acid Techniques in Bacterial Systematics. Wiley, New York, USA. pp. 115-175.

Lapointe M, Morse D (2008) Reassessing the role of a 3’-UTR-binding translational inhibitor in the regulation of circadian bioluminescence rhythm in the dinoflagellate Gonyaulax. Biol Chem, 389, 13-19.

Larkum AWD (2003) Light-harvesting systems in algae, In: Larkum AWD, Douglas SE, Raven JA (eds), Photosynthesis in algae, Kluwer Academic Publishers, The Netherlands, pp. 277-304.

155 Lee K (2012) Molecular assessment of benthic diatom assemblages in Delaware’s inland bays. MS Thesis, University of Delaware.

Legrande C, Rengefors K, Fistarol GO, Granéli E (2003) Allelopathy in phytoplankton - biochemical, ecological and evolutionary aspects. Phycologia, 42, 406-419.

Leu E, Krieger-Liszkay A, Goussias C, Gross EM (2002) Polyphenolic allelochemicals from the aquatic angiosperm Myriophyllum spicatum inhibit photosystem II. Plant Physiol, 130, 2011-2018.

Lewitus A, Horner R, Caron D, Garcia-Mendoza E, Hickey B, Hunter M, Huppert D, Kudela R, Langlois G, Largier J, Lessard E, RaLonde R, Rensel J, Strutton P, Trainer V, Tweddle J (2012) Harmful algal blooms along the North American west coast region: history, trends, causes and impacts. Harmful Algae, 19, 133–159.

Lewitus AJ, Willis BM, Hayes KC, Burkholder JM, Glasgow HB, Glibert PM, Burke MK (1999) Mixotrophy and nitrogen uptake by Pfiesteria piscicida (Dinophyceae). J Phycol, 35, 1430-1437.

Long SP, Humphries S, Falkowski PG (1994) Photoinhibition of photosynthesis in nature. Annu Rev Plant Physiol Plant Mol Biol, 45, 633-662.

Lu Y, Wang J, Yu Y, Su W, Kong F (2013) Inhibition of Camellia sinensis (L.) O. Kuntze on Microcystis aeruginosa and isolation of the inhibition factors. Biotechnol Lett, doi:10.1007/s10529-013-1188-4.

Lürling M, Oosterhout FV (2013) Controlling eutrophication by combined bloom precipitation and sediment phosphorous inactivation. Water Res, 47(17), 6527- 6537.

MacIntyre JG, Cullen JJ, Cembella AD (1997) Vertical migration, nutrition and toxicity in the dinoflagellate Alexandrium tamarense. Mar Ecol Prog Ser, 148, 201-216.

MacKenzie TDB, Morse D (2011) Circadian photosynthetic reductant flow in the dinoflagellate Lingulodinium is limited by carbon availability. Plant Cell Environ, 34, 669-680.

Mackenzie L (1991) Toxic and noxious phytoplankton in Big Glory Bay, Stewart Island, New Zealand. J Appl Phycol, 3, 19–34.

156 Manage PM, Kawabata Z’i, Nakano S-I (2000) Algicidal effect of the bacterium Alcaligenes denitrificans on Microcystis spp. Aquat Microb Ecol, 22, 111-117.

Margalef R (1978) Life-forms of phytoplankton as survival alternatives in an unstable environment. Oceanol Acta, 1, 493- 509.

Matrai P, Thompson B, Keller M (2005) Alexandrium spp. from eastern Gulf of Maine populations: Circannual excystment of resting cysts. Deep Sea Res II, 52(19- 21), 2560-2568.

Matthijs HCP, Visser PM, Reeze B, Meeuse J, Slot PC, Wijn G, Talens R, Huisman J (2012) Selective suppression of harmful cyanobacteria in an entire lake with hydrogen peroxide. Water Res, 46, 1460–1472.

Mayali X, Azam F (2004) Algicidal bacteria in the sea and their impact on algal blooms. J Eukaryot Microbiol, 51, 139–144.

Mayali X, Doucette GJ (2002) Microbial community interactions and population dynamics of an algicidal bacterium active against Karenia brevis (Dinophyceae). Harmful Algae, 1, 277–293.

Mayali X, Franks PJS, Azam F (2007) Bacterial induction of temporary cyst formation by the dinoflagellate Lingulodinium polyedrum. Aquat Microb Ecol, 50, 51-62.

McGillicuddy DJ, Anderson DM, Lynch DR, Townsend DW (2005) Mechanisms regulating the large-scale seasonal fluctuations in Alexandrium fundyense populations in the Gulf of Maine: Results from a physical–biological model. Deep Sea Res II, 52, 2698–2714.

Mcleod DJ, Hallegraeff GM, Hosie GW, Richardson AJ (2012) Climate-driven range expansion of the red-tide dinoflagellate Noctiluca scintillans into the Southern Ocean. J Plankton Res 34: 332-337

Medlin L, Elwood HJ, Stickel S, Sogen ML (1988) The characterization of enzymatically amplified eukaryotic 16S-like rRNA-coding regions. Gene, 71, 491–499.

Mehner T (2012) Diel vertical migration of freshwater fishes – proximate triggers, ultimate causes and research perspectives. Freshwater Biol, 57(7), 1342-1359.

Mittag M (2001) Circadian rhythms in microalgae. Int Rev Cytol, 206, 213-247.

Mizuno CS, Schrader KK, Rimando AA (2008) Algicidal activity of stilbene analogues. J Agric Food Chem, 56, 9140–9145.

157 Moldrup M, Moestrup Ø, Hansen PJ (2013) Loss of Phototaxis and Degeneration of an Eyespot in Long-term Algal Cultures: Evidence from Ultrastructure and Behaviour in the Dinoflagellate Kryptoperidinium foliaceum. J Eukaryot Microbiol, 60, 327–334.

Moore BC, Christensen DR (2009) Newman Lake restoration: A case study. Part 1. Chemical and biological responses to phosphorous control. Lake Reserv Manage, 25, 337-350.

Moore SK, Trainer VL, Mantua NJ, Parker MS, Laws EA, Backer LC, Fleming LE (2008) Impacts of climate variability and future climate change on harmful algal blooms and human health. Environ Health doi:10.1186/1476-069X-7-S2- S4.

Moriarty PE, Andrews KS, Harvey CJ, Kawase M (2012) Vertical and horizontal movement patterns of a scyphozoan jellyfish in a fjord-like estuary. Mar Ecol Prog Ser, 455, 1-12.

Mu R, He Y, Liu S, Wang X, Fan Z (2009) The algicidal characteristics of one algae- lysing FDT5 bacterium on Microcystis aeruginosa. Geomicrobiol J, 26, 516- 521.

Muyzer G, De Waal ED, Uitterlinden AG (1993) Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl Environ Microbiol, 59, 695–700.

Nagasaki K, Tarutani K, Yamaguchi M (1999) Growth characteristics of Heterosigma akashiwo virus and its possible use as a microbiological agent for red tide control. Appl Environ Microbiol, 65(3), 898-902.

Nassoury N, Wang Y, Morse D (2005) Brefeldin a inhibits circadian remodeling of chloroplast structure in the dinoflagellate gonyaulax. Traffic, 6, 548–561.

Neilson JW, Jordan FL, Maier RM (2013) Analysis of artifacts suggests DGGE should not be used for quantitative diversity analysis. J Microbiol Meth, 92, 256–263.

Nikolcheva LG, Cockshutt AM, Bärlocher F (2003) Determining diversity of freshwater fungi on decaying leaves: Comparison of traditional and molecular approaches. Appl Environ Microbiol, 69(5), 2548-2554.

158 Oldach DW, Delwiche CF, Jakobsen KS, Tengs T, Brown EG, Kempton JW, Schaefer EF, Bowers HA, Glasgow HB, Burkholder JM, Steidinger KA, Rublee PA (2000) Heteroduplex mobility assay-guided sequence discovery: elucidation of the small subunit (18S) rDNA sequences of Pfiesteria piscicida and related dinoflagellates from complex algal culture and environmental sample DNA pools. Proc Natl Acad Sci USA, 97, 4303–4308.

Orizar IS, Rivera PPL, Diego-McGlone ML, Azanza RV (2013) Harmful Algal Bloom (HAB) mitigation using ball clay: Effect on non-target organisms. Journal of Env Sci and Management, 1, 36-43.

Oster H, Damerow S, Hut RA, Eichele G (2006) Transcriptional profiling in the adrenal gland reveals circadian regulation of hormone biosynthesis genes and nucleosome assembly genes. J Biol Rhythms, 21, 350-361.

Ouyang Y, Andersson CR, Kondo T, Golden SS, Johnson CH (1998) Resonating circadian clocks enhance fitness in cyanobacteria. Proc Natl Acad Sci USA, 95, 8660-8664.

Pan G, Chen J, Anderson DM (2011) Modified local sands for the mitigation of harmful algal blooms. Harmful Algae, 10, 381-387.

Park SC, Lee JK, Kim SW, Park Y (2011) Selective algicidal action of peptides against harmful algal bloom species. PLoS ONE, 6(10), e26733.

Pascoal C, Cássio F, Nikolcheva L (2010) Realized fungal diversity increases functional stability of leaf litter decomposition under zinc stress. Microb Ecol, 59, 84-93.

Patron NJ, Waller RF, Keeling PJ (2006) A tertiary plastid uses genes from two endosymbionts. J Mol Biol, 357, 1373-1382.

Paul C, Pohnert G (2011) Interactions of the algicidal bacterium Kordia algicida with diatoms: regulated protease excretion for specific algal lysis. PLoS ONE, 6(6), e21032.

Peacock MB, Kudela RM (2014) Evidence for active vertical migration by two dinoflagellates experiencing iron, nitrogen, and phosphorus limitation. Limnol Oceanogr, 59(3), 660-673.

Pedley TJ, Kessler JO (1992) Hydrodynamic phenomena in suspensions of swimming microorganisms. Annu Rev Fluid Mech, 24, 313-358.

159 Pedrós-Alió C, Massana R, Latasa M, García-Cantizano J, Gasol JM (1995) Predation by cilates on a metalimnetic Cryptomonas population: feeding rates, impact and effects of vertical migration. J Plankton Res, 17(11), 2131-2154.

Peltier G, Cournac L (2002) Chlororespiration. Ann Rev Plant Biol, 53, 523-550.

Penna A, Garcés E, Vila M, Giacobbe MG, Fraga S, Lugliè A, Bravo I, Bertozzini E, Vernesi C (2005) Alexandrium catenella (Dinophyceae), a toxic ribotype expanding in the NW Mediterranean Sea. Mar Biol, 148, 12-23.

Pennak RW (1943) An effective method of diagramming diurnal movements of zooplankton organisms. Ecology, 24(3), 405-407.

Peters E, Thomas DN (1996) Prolonged darkness and diatom mortality: I. Marine Antarctic species. J Exp Mar Biol Ecol, 207, 25–41.

Place AR, Bowers HA, Bachvaroff TR, Adolf JE, Deeds JR, Sheng J (2012) Karlodinium veneficum - The little dinoflagellate with a big bite. Harmful Algae, 14, 179-195.

Pokrzywinski KL, Place AR, Warner ME, Coyne KJ (2012) Investigation of the algicidal exudate produced by Shewanella sp. IRI-160 and its effect on dinoflagellates. Harmful Algae, 19, 23-29.

Pokrzywinski KL (2014) Characterization of bacterial algicide IRI-160AA: Insights on programmed death pathways and organelle impacts in harmful dinoflagellates. PhD Thesis, University of Delaware, USA.

Pokrzywinski KL, Tilney CL, Warner ME, Coyne KJ (Submitted) Induction of programmed cell death in harmful dinoflagellates following exposure to bacterial algicide IRI-160AA.

Popels LC, MacIntyre HL, Warner ME, Zhang YH, Hutchins DA (2007) Physiological responses during dark survival and recovery in Aureococcus anophagefferens (Pelagophyceae). J Phycol, 43, 32-42.

Poulson KL, Sieg RD, Kubanek J (2009) Chemical ecology of the marine plankton. Nat Prod Rep, 26, 717-856.

Prézelin BB (1982) Effects of light intensity on aging in the dinoflagellate Gonyaulax polyedra. Mar Biol, 69, 129-135.

160 Prézelin BB, Meeson BW, Sweeney BM (1977) Characterization of photosynthetic rhythms in marine dinoflagellates. I. Pigmentation, photosynthetic capacity and respiration. Plant Physiol, 60, 384-387.

Prézelin BB, Sweeney BM (1977) Characterization of photosynthetic rhythms in marine dinoflagellates. II. Photosynthesis-irradiance curves and in-vivo chloropyll-a fluorescence. Plant Physiol, 60, 388-392.

Prince EK, Myers TL, Naar J, Kubanek J (2008) Competing phytoplankton undermines allelopathy of a bloom-forming dinoflagellate. Proc R Soc B, 275, 2733-2741.

Ragni M, D’alcalà MR (2007) Circadian variability in the photobiology of Phaeodactylum tricornutum: pigment content. J Plankton Res, 29(2), 141-156.

Ramsdell JS, Gulland FM (2014) Domoic acid epileptic disease. Mar Drugs, 12(3), 1185-1207.

Redshaw CH, Sutter DA, Lim-Hing K, Heckman ML, Jakobsson SL, Naar J, Kubanek J (2011) Tracking losses of brevetoxins on exposure to phytoplankton competitors: Ecological impacts. Harmful Algae, 12, 119-124.

Regel RH, Brookes JD, Ganf GG (2004) Vertical migration, entrainment and photosynthesis of the freshwater dinoflagellate Peridinium cinctum in a shallow urban lake. J Plankton Res, 26(2), 143-157.

Rensel JE, Whyte JNC (2003) Finfish mariculture and harmful algal blooms. In: Hallegraeff, GM, Anderson DM, Cembella AD (eds) Manual on harmful marine algae, revised edn. IOC UNESCO, pp 693-722.

Rhodes LL, Mackenzie AL, Kaspar HF, Todd KE (2001) Harmful algae and mariculture in New Zealand. ICES J Mar Sci, 58, 398-403.

Richardson LL, Castenholz RW (1987) Diel vertical movements of the cyanobacterium Oscillatoria terebriformis in a sulfide-rich hot spring microbial mat. Appl Environ Microb, 53(9), 2142-2150.

Ririe KM, Rasmussen RP, Wittwer CT (1997) Product differentiation by analysis of DNA melting curves during the polymerase chain reaction. Analytical Biochemistry, 245, 154-160.

Roenneberg T, Colfax GN, Hastings JW (1989) A circadian rhythm of population behavior in Gonyaulax ployedra. J Biol Rhythms, 4(2), 201-216.

161 Roenneberg T, Morse D (1993) Two circadian oscillators in one cell. Nature, 362, 362-364.

Roenneberg T, Mittag M (1996) The circadian program of algae. Semin Cell Dev Biol, 7, 753-763.

Roenneberg T, Rehman J (1996) Nitrate, a nonphotic signal for the circadian system. FASEB J, 10, 1443-1447.

Rost B, Richter K-U, Riebesell U, Hansen PJ (2006) Inorganic carbon acquisition in red tide dinoflagellates. Plant Cell Environ, 29, 810-822.

Rounsefell GA, Evans JE (1958) Large-scale experimental test of copper sulfate as a control for the Florida red tide. US Fish Wld S Spec Sci Rep 270.

Salonen K, Rosenberg M (2000) Advantages from diel vertical migration can explain the dominance of Gonyostomum semen (Raphidophyceae) in a small, steeply- stratified humic lake. J Plankton Res, 22(10), 1841-1853.

Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: A laboratory manual. 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.

Samuelsson G, Sweeney BM, Matlick HA, Prézelin BB (1983) Changes in photosystem II account for the circadian rhythm in photosynthesis in Gonyaulax polyedra. Plant Physiol, 73, 329-331.

Sanborn MR, Wan SK, Bulard R (1982) Microwave sterilization of plastic tissue culture vessels for reuse. Appl Environ Microbiol, 44(4), 960-964.

Saradadevi K, Raghavendra S (1992) Dark respiration protects photosynthesis against photoinhibition in mesophyll protoplasts of pea (Pisum sativum). Plant Physiol, 99, 1232-1237.

Schaeffer BA, Kamykowski D, Sinclair G, McKay L, Milligan EJ (2009) Diel vertical migration thresholds of Karenia brevis. Harmful Algae, 8, 692-698.

Schrader KK, Duke SO, Kingsbury SK, Tucker CS, Duke MV, Dionigi CP, Millie DF, Zimba PV (2000) Evaluation of ferulic acid for controlling the musty-odor cyanobacterium, Oscillatoria perornata, in aquaculture ponds. J Appl Aquacult, 10(1), 1–16.

162 Segovia M, Berges JA (2005) Effect of inhibitors of protein synthesis and DNA replication on the induction of proteolytic activities, caspase-like activities and cell death in the unicellular chlorophyte Dunaliella tertiolecta. Eur J Phycol, 40, 21-30.

Sengco MR (2009) Prevention and control of Karenia brevis blooms. Harmful Algae, 8, 623-628.

Sengco MR, Anderson DM (2004) Controlling harmful algal blooms through clay flocculation. J Eukaryot Microbiol, 51, 169-172.

Shalapyonok A, Olson RJ, Shalapyonok LS (1998) Ultradian growth in Prochlorococcus spp. Appl Environ Microbiol, 64(3), 1066-1069.

Shao J, Li R, Lepo JE, Gu JD (2013) Potential for control of harmful cyanobacterial blooms using biologically derived substances: problems and prospects. J Environ Manage, 125, 149-55.

Sharma VK (2003) Adaptive significance of circadian clocks. Chonobiol Int, 20(6), 901-919.

Shi S, Tang D, Liu Y (2009) Effects of an algicidal bacterium Pseudomonas mendocina on the growth and antioxidant system of Aphanizomenon flos- aquae. Curr Microbiol, 59, 107-112.

Shikata T, Matsunaga S, Iseki M, Nishide H, Higashi S-I, Kamei Y, Yamaguchi M, Jenkinson IR, Watanabe M (2013) Blue light regulates the rhythm of diurnal vertical migration in the raphidophyte red-tide alga Chattonella antiqua. J Plankton Res, 35(3), 542-552.

Shumway SE (1990) A review of the effects of algal blooms on shellfish and aquaculture. J World Aquacult Soc, 21(2), 65-104.

Shumway SE, Frank DM, Ewart LM, Ward JE (2003) Effect of yellow loess on clearance rate in seven species of benthic, filter-feeding invertebrates. Aquaculture Res, 34, 1391-1402.

Smayda TJ (1990) Novel and nuisance phytoplankton blooms in the sea: Evidence for a global epidemic. In: Toxic marine phytoplankton, Proc. 4th Int. Conf., 29-40, Elsevier.

Smayda TJ (1997) What is a blooom? A commentary. Limnol Oceanogr, 42(5), 1132- 1136.

163 Smayda TJ (1997b) Harmful algal blooms: Their ecophysiology and general relevance to phytoplankton blooms in the sea. Limnol Oceanogr, 42(5), 1137-1153.

Smayda TJ (2007) Reflections on the ballast water dispersal - harmful algal bloom paradigm. Harmful Algae, 6, 601-622.

Smayda TJ, Reynolds CS (2001) Community assembly in marine phytoplankton: Application of recent models to harmful dinoflagellate blooms. J Plankton Res, 23(5), 447-461.

Smayda TJ, Reynolds CS (2003) Strategies of marine dinoflagellate survival and some rules of assembly. J Sea Res, 49, 95–106.

Smayda TJ (2005) Harmful Algal Bloom Communities in Scottish Coastal Waters: Relationship to Fish Farming and Regional Comparisons, A Review. Scottish Executive. Available online at: www.scotland.gov.uk/Publications/2006/02/03095327/0.

Smayda TJ, Mitchell-Innes B (1974) Dark survival of autotrophic, planktonic marine diatoms. Mar Biol, 25, 195–202.

Smith EA, Grant F, Ferguson CMJ, Gallacher S (2001) Bio- transformations of paralytic shellfish toxins by bacteria isolated from bivalve molluscs. Appl Environ Microbiol, 67, 2345–2353.

Smith GD, Doan NT (1999) Cyanobacterial metabolites with bioactivity against photosynthesis in cyanobacteria, algae and higher plants. J Appl Phycol, 11, 337-344.

Smith VH, Schindler DW (2009) Eutrophication science: where do we go from here? Trends Ecol Evol, 24(4), 201-207.

Solé J, Estrada M, Garcia-Ladona E (2006) Biological control of harmful algal blooms: A modelling study. J Mar Syst, 61(3-4), 165-179.

Sorek M, Yacobi YZ, Roopin M, Berman-Frank I, Levy O (2013) Photosynthetic circadian rhythmicity patterns in Symbiodinium, the coral endosymbiotic algae. Proc R Soc B, 280, 20122942.

Sotirova AV, Spasova DI, Galabova DN, Karpenko E, Shulga A (2008) Rhamnolipid- biosurfactant permeabilizing effects on gram-positive and gram-negative bacterial strains. Curr Microbiol, 56(6), 639–647.

164 Stoecker DK, Thessen AE, Gustafson DE (2008) “Windows of opportunity” for dinoflagellate blooms: Reduced microzooplankton net growth coupled to eutrophication. Harmful Algae, 8, 158-166.

Stoecker DK (1999) Mixotrophy among dinoflagellates. J Eukaryot Microbiol, 46, 397-401.

Strom SL, Wolfe GV, Slajer A, Lambert S, Clough J (2003) Chemical defense in the microplankton II: Inhibition of protist feeding by dimethylsulfoniopropionate. Limnol Oceanogr, 48, 230–237.

Suggett DJ, Moore CM, Hickman AE, Geider RJ (2009) Interpretation of Fast Repetition Rate (FRR) fluorescence: signatures of community structure versus physiological state. Mar Ecol Prog Ser, 376, 1-19.

Sweeney BM (1986) The loss of the circadian rhythm in photosynthesis in an old strain of Gonyaulax polyedra. Plant Physiol, 80(4), 978-981.

Takamura Y, Yamada T, Kimoto A, Kanehama N, Tanaka T, Nakadaira S, Yagi O (2004) Growth inhibition of Microcystis cyanobacteria by L-lysine and disappearance of natural Microcystis blooms with spraying. Microbes Environ, 19, 31–39.

Tang YZ, Kang Y, Berry D, Gobler CJ (2014) The ability of the red macroalga, Porphyra purpurea (Rhodophyceae) to inhibit the proliferation of seven common harmful microalgae. J Appl Phycol, DOI: 10.1007/s10811-014-0338- y.

Tang YZ, Koch F, Gobler CJ (2010) Most harmful algal bloom species are vitamin B1 and B12 auxotrophs. Proc Nat Acad Sci USA, DOI:10.1073/pnas.1009566107.

Taylor FJR (1968) Parasitism of the toxin-producing dinoflagellate Gonyaulax catenella by the endoparasitic dinoflagellate Amoebophrya ceratii. J Fish Red Board Can, 25(10), 2241-2245.

Terlizzi DE, Ferrier MD, Armbrester EA, Anlauf KA (2002) Inhibition of dinoflagellate growth by extracts of barley straw (Hordeum vulgare). J Appl Phycol, 14, 275–280.

Tillmann U (2003) Kill and eat your predator: A winning strategy of the planktonic flagellate Prymnesium parvum. Aquat Microb Ecol, 32, 73-84.

Tillmann U (2004) Interactions between planktonic microalgae and protozoan grazers. J Eukaryot Microbiol, 51, 156-168.

165 Tilney CL, Pokrzywinski KL, Coyne KJ, Warner ME (2014) Growth, death, and photobiology of dinoflagellates (Dinophyceae) under bacterial-algicide control. J Appl Phycol, DOI: 10.1007/s10811-014-0248-z .

Tobin ED, Grünbaum D, Patterson J, Cattolico RA (2013) Behavioral and physiological changes during benthic-pelagic transition in the harmful alga, Heterosigma akashiwo: Potential for rapid bloom formation. PLoS ONE 8(10): e76663.

Tortell PD, DiTullio GR, Sigman DM, Morel FMM (2002) CO2 effects on taxonomic composition and nutrient utilization in an Equatorial Pacific phytoplankton assemblage. Mar Ecol Prog Ser, 236, 37–43.

Trainer VM, Suddleson M (2005) Monitoring approaches for early warning of domoic acid events in Washington State. Oceanography, 18(2), 228-237.

Trimborn S, Thoms S, Petrou K, Kranz S, Rost B (2014) Photophysiological responses of Southern Ocean phytoplankton to changes in CO2 concentrations: Short-term versus acclimation effects. J Exp Mar Biol Ecol, 451, 44-54.

Turley CM, Newell RC, Robins DB (1986) Survival strategies of two small marine ciliates and their role in regulating bacterial community structure under experimental conditions. Mar Ecol Prog Ser, 33, 59-70.

Tyler MA, Seliger HH (1978) Annual subsurface transport of a red tide dinoflagellate to its bloom area: Water circulation patterns and organism distributions in the Chesapeake Bay. Limnol Oceanogr, 23(2), 227-246.

Ucko M, Garesh S, Simon-Berkovitch B, Malis SA (1994) Predation by a dinoflagellate on a red microalga with a cell wall modified by sulfate and nitrate starvation. Mar Ecol Prog Ser, 104, 293-298.

Van Dolah FV (2000) Marine algal toxins: origins, health effects, and their increased occurrence. Environ Health Persp, 108, 133–141.

Van Dolah FM, Leighfield TA, Kamykowski D, Kirkpatrick GJ (2008) Cell cycle behavior of laboratory and field populations of the Florida red tide dinoflagellate, Karenia brevis. Cont Shelf Res, 28, 11-23.

Van Hannen EJ, Zwart G, Van Agterveld MP, Gons HJ, Ebert J, Laanbroek HJ (1999) Changes in bacterial and eukaryotic community structure after mass lysis of filamentous cyanobacteria associated with viruses. Appl Env Microbiol, 65(2), 795-801.

166 Van Oosterhout F, Lürling M (2013) The effect of phosphorus binding clay (Phoslock) in mitigating cyanobacterial nuisance: a laboratory study on the effects on water quality variables and plankton. Hydrobiologia, 710, 256-277.

Velo-Suárez L, Brosnahan ML, Anderson DM, McGillicuddy DJ Jr (2013) A Quantitative Assessment of the Role of the Parasite Amoebophrya in the Termination of Alexandrium fundyense Blooms within a Small Coastal Embayment. PLoS ONE, 8(12), e81150.

Villareal TA (1988) Positive buoyancy in the oceanic diatom Rhizosolenia debyana H. Peragallo. Deep Sea Res, 35, 1037-1045.

Villareal TA, Altabet MA, Culver-Rymsza K (1993) Nitrogen transport by vertically migrating diatom mats in the North Pacific Ocean. Nature, 363, 709-712.

Vitaterna MH, Takahashi JS, Turek F (2001) Overview of circadian rhythms. Alcohol Res Health, 25(2), 85-93.

Vivanco JM, Bais HP, Stermitz FR, Thelen GC, Callaway RM (2004) Biogeographical variation in community response to root allelochemistry: novel weapons and exotic invasion. Ecol Lett, 7, 285-292.

Waggett RJ, Tester PA, Place AR (2008) Anti-grazing properties of the toxic dinoflagellate Karlodinium veneficum during predator–prey interactions with the copepod Acartia tonsa. Mar Ecol Prog Ser, 366, 31-42.

Wang X, Gong L, Liang S, Han X, Zhu C, Li Y (2005) Algicidal activity of rhamnolipid biosurfactants produced by Pseudomonas aeruginosa. Harmful Algae, 4, 433-443.

Wang Y, Jensen L, Hojrup P, Morse D (2005b) Synthesis and degradation of dinoflagellate plastid-encoded psbA proteins are light-regulated, not circadian- regulated. Proc Nat Acad Sci USA, 102, 2844–2849.

Warner ME, Madden ML (2007) The impact of shifts to elevated irradiance on the growth and photochemical activity of the harmful algae Chattonella subsalsa and Prorocentrum minimum from Delaware. Harmful Algae, 6, 332-342.

Watanabe M, Kohata K, Watanabe M (1991) Diel vertical migration and nocturnal uptake of nutrients by Chattonella antiqua under stable stratification. Limnol Oceanogr, 36, 593–602.

167 Watanabe M, Kohata K, Kimura T, Takamatsu T, Yamaguchi S, Ioriya T (1995) Generation of a Chattonella antiqua bloom by imposing a shallow nutricline in a mesocosm. Limnol Oceanogr, 40, 1447–1460.

Weidenhamer JD and Romeo JT (2004) Allelochemicals of Polygonella myriophylla: Chemistry and soil degredation. J Chem Ecol, 30(5), 1067-1082.

Weiler CS, Karl DM (1979) Diel changes in phased-dividing cultures of Ceratium furca (dinophyceae): Nucleotide triphosphates, adenylate energy charge, cell carbon, and patterns of vertical migration. J Phycol, 15, 384-391.

Weissbach A, Rudström M, Olofsson M, Béchemin C, Icely J, Newton A, Tillmann U and Legrand C (2011) Phytoplankton allelochemical interactions change microbial food web dynamics. Limnol Oceanogr, 56(3), 899-909.

Wisecaver JH, Hackett JD (2011) Dinoflagellate Genome Evolution. Annu Rev Microbiol, 65, 369-387.

Woelfe MA, Ouyang Y, Phanvijhitsiri K, Johnson CH (2004) The adaptive value of circadian clocks: an experimental assessment in cyanobacteria. Curr Biol, 14, 1481-1486.

Wood AM, Everroad RC, Wingard LM (2005) Measuring growth rates in microalgal cultures. In: Andersen RA (ed) Algal culturing techniques. Elsevier Academic Press, Burlington, MA, p 269–285.

World Bank (2013) Fish to 2030: Prospects for fisheries and aquaculture. World Bank Report Number 83177-GLB.

Wu J-T, Chiang Y-R, Huang W-Y, Jane W-N (2006) Cytotoxic effects of free fatty acids on phytoplankton algae and cyanobacteria. Aquat Toxicol, 80, 338-345.

Xu Y, Ma P, Shah P, Rokas A, Liu Y, Johnson CH (2013) Non-optimal codon usage is a mechanism to achieve circadian clock conditionality. Nature, 495, 116- 120.

Ye J, Coulouris G, Zaretskaya I, Cutcutache I, Rozen S, Madden TL (2012) Primer- BLAST: A tool to design target-specific primers for polymerase chain reaction. BMC Bioinformatics, 13, 134.

Yoon HS, Hackett JD, Van Dolah FM, Nosenko T, Lidie KL, Bhattacharya D (2005) Tertiary endosymbiosis driven genome evolution in dinoflagellate algae. Mol Biol Evol, 22, 1299–1308.

168 Zingone A, Enevoldsen HO (2000) The diversity of harmful algal blooms: a challenge for science and management. Ocean Coast Manage, 43, 725-748.

169 Appendix A

APPENDED FIGURES

If you only have one appendix, change the styles of the previous two paragraphs to Appendix - one and APPENDIX TITLE - one, respectively.

Figure A1: Testing the effects of IRI-160AA concentration on SYTOX-green fluorescence (y-axis) in K. veneficum after 30-minute incubations with 1 µM final concentration SYTOX-green and different concentrations of IRI-160AA (x-axis).

170

Figure A2: Testing change in Fv/Fm from the top and bottom of un-mixed test tubes over 66 hours. Four sets of triplicate test tubes contained a final volume of 5 mL of K. veneficum culture, two sets were treated with 4% (v/v) IRI- 160AA, and two treated with 4% (v/v) f/2 medium as a control. One set of treatments, and one set of controls were mixed by pipette before measuring a 2 mL aliquot in the FRR fluorometer after dark acclimation. The other set of treatments and controls remained unmixed, and the top 2 mL of culture carefully removed for measurement, with the remaining 3 mL from these tubes being mixed before measuring a 2mL aliquot, also after dark acclimation.

171

172 Appendix B

RE-PRINT LICENCE

SPRINGER LICENSE TERMS AND CONDITIONS This is a License Agreement between Charles Tilney ("You") and Springer ("Springer") provided by Copyright Clearance Center ("CCC"). The license consists of your order details, the terms and conditions provided by Springer, and the payment terms and conditions. All payments must be made in full to CCC. For payment instructions, please see information listed at the bottom of this form.

License Number: 3418851259465 License date: Jun 30, 2014 Licensed content publisher: Springer Licensed content publication: Journal of Applied Phycology Licensed content title: Growth, death, and photobiology of dinoflagellates (Dinophyceae) under bacterial-algicide control Licensed content author: Charles L. Tilney Licensed content date: Jan 1, 2014 Type of Use Portion: Thesis/Dissertation Full text Number of copies: 1 Author of this Springer article: Yes and you are the sole author of the new work Order reference number: None Title of your thesis / dissertation: Examining the physiology of harmful microalgae during algicidal control and diel vertical migration Expected completion date: July 2014 Estimated size(pages): 190 pages Total: 0.00 USD Terms and Conditions: Introduction The publisher for this copyrighted material is Springer Science + Business Media. By clicking "accept" in connection with completing this licensing transaction, you agree that the following terms and conditions apply to this transaction (along with the Billing and Payment terms and conditions established by Copyright Clearance Center, Inc. ("CCC"), at the time that you opened your Rightslink account and that are available at any time at http://myaccount.copyright.com). Limited License With reference to your request to reprint in your thesis material on which Springer Science and Business Media control the copyright, permission is granted, free of charge, for the use indicated in your enquiry. Jul 17, 2014 Licenses are for one-time use only with a maximum distribution equal to the

173 number that you identified in the licensing process. This License includes use in an electronic form, provided its password protected or on the university’s intranet or repository, including UMI (according to the definition at the Sherpa website: http://www.sherpa.ac.uk/romeo/). For any other electronic use, please contact Springer at ([email protected] or [email protected]). The material can only be used for the purpose of defending your thesis limited to university- use only. If the thesis is going to be published, permission needs to be re-obtained (selecting "book/textbook" as the type of use). Although Springer holds copyright to the material and is entitled to negotiate on rights, this license is only valid, subject to a courtesy information to the author (address is given with the article/chapter) and provided it concerns original material which does not carry references to other sources (if material in question appears with credit to another source, authorization from that source is required as well). Permission free of charge on this occasion does not prejudice any rights we might have to charge for reproduction of our copyrighted material in the future. Altering/Modifying Material: Not Permitted You may not alter or modify the material in any manner. Abbreviations, additions, deletions and/or any other alterations shall be made only with prior written authorization of the author(s) and/or Springer Science + Business Media. (Please contact Springer at ([email protected] or [email protected]) Reservation of Rights Springer Science + Business Media reserves all rights not specifically granted in the combination of (i) the license details provided by you and accepted in the course of this licensing transaction, (ii) these terms and conditions and (iii) CCC's Billing and Payment terms and conditions. Copyright Notice:Disclaimer You must include the following copyright and permission notice in connection with any reproduction of the licensed material: "Springer and the original publisher /journal title, volume, year of publication, page, chapter/article title, name(s) of author(s), figure number(s), original copyright notice) is given to the publication in which the material was originally published, by adding; with kind permission from Springer Science and Business Media" Warranties: None Example 1: Springer Science + Business Media makes no representations or warranties with respect to the licensed material. Example 2: Springer Science + Business Media makes no representations or warranties with respect to the licensed material and adopts on its own behalf the limitations and disclaimers established by CCC on its behalf in its Billing and Payment terms and conditions for this licensing transaction. Indemnity You hereby indemnify and agree to hold harmless Springer Science +

174 Business Media and CCC, and their respective officers, directors, employees and agents, from and against any and all claims arising out of your use of the licensed material other than as specifically authorized pursuant to this license. No Transfer of License This license is personal to you and may not be sublicensed, assigned, or transferred by you to any other person without Springer Science + Business Media's written permission. No Amendment Except in Writing This license may not be amended except in a writing signed by both parties (or, in the case of Springer Science + Business Media, by CCC on Springer Science + Business Media's behalf). Objection to Contrary Terms Springer Science + Business Media hereby objects to any terms contained in any purchase order, acknowledgment, check endorsement or other writing prepared by you, which terms are inconsistent with these terms and conditions or CCC's Billing and Payment terms and conditions. These terms and conditions, together with CCC's Billing and Payment terms and conditions (which are incorporated herein), comprise the entire agreement between you and Springer Science + Business Media (and CCC) concerning this licensing transaction. In the event of any conflict between your obligations established by these terms and conditions and those established by CCC's Billing and Payment terms and conditions, these terms and conditions shall control. Jurisdiction All disputes that may arise in connection with this present License, or the breach thereof, shall be settled exclusively by arbitration, to be held in The Netherlands, in accordance with Dutch law, and to be conducted under the Rules of the 'Netherlands Arbitrage Instituut' (Netherlands Institute of Arbitration).

OR: All disputes that may arise in connection with this present License, or the breach thereof, shall be settled exclusively by arbitration, to be held in the Federal Republic of Germany, in accordance with German law. Other terms and conditions: v1.3 If you would like to pay for this license now, please remit this license along with your payment made payable to "COPYRIGHT CLEARANCE CENTER" otherwise you will be invoiced within 48 hours of the license date. Payment should be in the form of a check or money order referencing your account number and this invoice number 501340126. Once you receive your invoice for this order, you may pay your invoice by credit card. Please follow instructions provided at that time. Make Payment To: Copyright Clearance Center Dept 001 P.O. Box 843006 Boston, MA 02284-3006 For suggestions or comments regarding this order, contact RightsLink Customer Support: [email protected] or +1-877-622-5543 (toll free in the US) or +1- 978-646-2777. Gratis licenses (referencing $0 in the Total field) are free. Please retain this printable license for your reference. No payment is required.

175