Microalgal-Bacterial Consortia for Biofuel Production and Wastewater Treatment

MICROALGAL-BACTERIAL CONSORTIA FOR BIOFUEL PRODUCTION AND WASTEWATER TREATMENT

MARK LORIA

Submitted in partial fulfillment of the requirements for the

degree of Doctor of Philosophy

Advisor

Dr. Kurt Rhoads

Department of Civil Engineering

CASE WESTERN RESERVE UNIVERSITY

January 2018

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

Mark Loria

candidate for the degree of Doctor of Philosophy*

Committee Chair

Dr. Kurt Rhoads

Committee Members

Dr. George Wells

Dr. Harihara Baskaran

Dr. Aaron Jennings

Date of Defense

23rd August, 2017

* We also certify that written approval has been obtained for any proprietary

material contained therein. TABLE OF CONTENTS

LIST OF TABLES ...... iv LIST OF FIGURES ...... v ACKNOWLEDGEMENTS ...... ix LIST OF ABBREVIATIONS ...... xi ABSTRACT ...... xiii Chapter 1 INTRODUCTION ...... 1 1.1 Background ...... 1 1.2 Crop Protection in Wastewater-fed High Rate Algal Ponds ...... 2 1.3 Community Assembly in Microalgal-Microbial HRAPs ...... 3 1.4 Microalgal Harvesting and Bioflocculation ...... 4 1.5 MAB Floc Cultivation and Wastewater Treatment ...... 5 1.6 Economic Values of Wastewater-fed High Rate Algal Ponds ...... 8 1.7 Organization and Objectives ...... 8 Chapter 2 FEAST-FAMINE REACTOR CYCLING FOR NATURAL SELECTION OF LIPID-ACCUMUATING ALGAL COMMUNITIES ...... 11 2.1 Introduction ...... 11 2.2 Materials and Methods ...... 14 2.3 Results and Discussion: ...... 17 2.3.1 Storage Carbohydrate and Lipid Changes with Cycling ...... 17 2.3.2 Microbial Community Composition in Reactors ...... 22 2.4 Conclusions and Recommendations ...... 27 Chapter 3 MAB FLOCS AS A PLATFORM FOR WASTEWATER TREATMENT AND BIOFUEL PRODUCTION ...... 30 3.1 Introduction ...... 30 3.2 Materials and Methods ...... 32 3.3 Results and Discussion: ...... 36 3.3.1 Bioflocculation Rate of Microalgae with Activated Sludge ...... 36 3.3.2 Inoculated MAB Floc Reactors ...... 40 3.3.3 Polyculture MAB Floc Reactors ...... 53 3.4 Conclusions ...... 66

Chapter 4 COMPETITION BETWEEN NITRIFICATION AND ALGAL GROWTH IN MAB REACTORS TREATING SYNTHETIC WASTEWATER ...... 69 4.1 Introduction ...... 69 4.2 Materials and Methods ...... 71 4.3 Results and Discussion: ...... 76 4.3.1 Defined vs Complex N and P source ...... 76 4.3.2 Algal Growth in Reactors ...... 80 4.3.3 Nutrient Conversion and Utilization Across Reactors and Time ...... 82 4.3.4 Nitrate Removal and Biomass Production Rate in MAB Floc Systems ...... 99 4.4 Conclusions ...... 103 Chapter 5 INVASION IN MICROALGAL COMMUNITIES ...... 106 5.1 Introduction ...... 106 5.2 Material and Methods ...... 107 5.3 Results and Discussion ...... 108 5.4 Conclusions ...... 110 Chapter 6 CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS ...... 111 6.1 Chapter 2: Feast-Famine Reactor Cycling for Natural Selection of Lipid-accumulating Algae Communities ...... 112 6.2 Chapter 3: MAB Flocs as a Platform for Wastewater Treatment and Biofuel Production ...... 112 6.3 Chapter 4: Competition Between Nitrification and Algal Growth in MAB Reactors Treating Synthetic Wastewater ...... 114 6.4 Chapter 5: Invasion in Microalgal Communities ...... 115 APPENDIX ...... 116 A.1 Effect of Centrifuging and Resuspension on Algal Growth ...... 116 A.2 Quality Control on Modifications to Starch Kit Assay ...... 117 A.3 R Code for Microbial Community Analysis...... 118 A.4 OTU Information ...... 123 A.5 Grouping Information for ANOVA ...... 124 A.6 Linearity Check for ChlA Fluorometric Analysis ...... 125 A.7 Linearity Check for Colorimetric SPV Lipid Analysis ...... 126 A.8 Images of Settleability in MAB Floc Reactors ...... 127 A.9 S. dimorphus growth curve ...... 128 A.10 Microbial Community Analyses Methods Written by Jim Griffin ...... 129

REFERENCES ...... 131

iii

LIST OF TABLES

Table 3.1 TSS concentration, chlA concentration, and autotrophic index of duplicate C. vulgaris, S. dimorphus, and C. sorokiniana MAB flocs (flocs washed with PBS to remove non- flocculated algae and then re-suspended)...... 39 Table 3.2 Lipid composition of flocs (% of dry weight) as measured and calculated based on published values assuming w/w ratios of algae to activated sludge remained stable...... 52 Table 3.3 Operating conditions for polyculture MAB reactors (biological duplicate reactors for each configuration)...... 53 Table 3.4 TSS concentrations (g/L) of mixed liquor, effluent after 5 min settling, and effluent after 1 h settling in aerated (SB) and non-aerated (SBAn) duplicate MAB floc reactors...... 55 Table 4.1 Summary of nutrients in different synthetic wastewater formulations...... 74 Table 4.2 Standard deviations of daily nitrification or N uptake as a percentage of total influent N between startup and Day 15 and from Day 15 to 30...... 84 Table 4.3 Parameters used for N mass balance for the low, medium, and high BOD bioreactors...... 88 Table 4.4 Nitrate + nitrite production (mg-N/L-d) and N uptake (mg-N/L-d) normalized to biomass for N uptake and nitrification rate experiments conducted over a 2 h period. .... 91 Table 4.5 Standard deviations of daily nitrification, N uptake, and P uptake as a percentage of total influent N/P from Day 15 to 30...... 97

LIST OF FIGURES

Figure 2.1 Overall lipid content of reactors over the course of nutrient cycles. Error bars represent one standard deviation above and below of three biological triplicates (three different reactors)...... 18 Figure 2.2 Overall starch content of reactors over the course of nutrient cycles. Error bars represent one standard deviation above and below of three biological triplicates (three different reactors)...... 20 Figure 2.3 FAME profiles of pH 9 reactors over study period (average of three replicate reactors)...... 21 Figure 2.4 FAME profiles of pH 7.5 reactors over study period (average of three replicate reactors)...... 21 Figure 2.5 Total 16S copy number normalized to ng DNA for biological triplicate reactors at (a) pH 7.5 and (b) pH 9...... 23 Figure 2.6 TSS concentration (g/L) in pH 7.5 and pH 9 reactors over feast-famine cycles for biological triplicate reactors...... 23 Figure 2.7 18S OTU composition in pH 7.5 reactors (designated H for HEPES buffered), and pH 9 reactors (designated T for TAPS buffered), sampled at the first four resuspensions...... 24 Figure 2.8 Relative abundance (fraction of total reads) of 8 highly-abundant 18S OTUs showing statistically significant changes in abundance and >.5% of total reads in pH 9 reactor. OTU_291 relative abundance plotted on secondary vertical axis due to higher relative abundance...... 26 Figure 2.9 Shannon Index (H) of pH 9 and pH 7.5 reactors over several reactor cycles...... 27 Figure 3.1 ChlA fluorescence of non-flocculated C. vulgaris in duplicate reactors after 5 min settling time...... 37 Figure 3.2 ChlA fluorescence of non-flocculated S. dimorphus in duplicate reactors after 5 min settling time...... 37 Figure 3.3 ChlA fluorescence of non-flocculated C. sorokiniana in duplicate reactors after 5 min settling time...... 38

Figure 3.4 Microscope images of MAB flocs inoculated with different microalgal strains, at the same inoculation ratio (1:1 algae:AS) 78 h after inoculation (A) C. vulgaris (B) N. oleoabundans (C) S. dimorphus (D) C. sorokiniana...... 41 Figure 3.5 Dissolved oxygen concentration in MAB floc reactors on Days 4, 5, and 6. Measurements were made during the daily light period. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs) at 5:1, 2:1, 1:1, or 1:5 algae:AS (w/w) ratio...... 42 Figure 3.6 pH in MAB floc reactors on Days 4, 5, and 6. Measurements were made during the daily light period. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs) at 5:1, 2:1, 1:1, or 1:5 algae:AS (w/w) ratio...... 43 Figure 3.7 TSS in MAB bioreactors when completely mixed (blue) and effluent TSS after settling for 30 min (red). Green line indicates the initial TSS concentration at inoculation. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs)...... 44 Figure 3.8 Microscopic images of (A) initial activated sludge floc and 1:1 C. vulgaris:activated sludge flocs (B) 5 h after inoculation (C) 30 h after inoculation (D) 78 h after inoculation. .. 46

Figure 3.9 BOD5 in reactor effluent after settling and filtration (0.45µm). The BOD5 was measured on Days 4, 5, and 6 of reactor operation. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs)...... 47 Figure 3.10 Total soluble nitrogen removal efficiencies in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 4 and 5. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs)...... 49 Figure 3.11 Total soluble phosphorus removal efficiencies in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 4, 5, and 6. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs)...... 49 Figure 3.12 ChlA florescence of MAB floc biomass normalized by OD600 for a subset of the MAB floc reactors...... 51 Figure 3.13 Lipid content (% of dry weight) of a subset of MAB floc reactors...... 51 Figure 3.14 Suspended cell concentration of reactor effluent after 5 min settling, as quantified by optical density at 750 nm...... 54 Figure 3.15 Total soluble nitrogen in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b)

Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL2 duplicate reactors...... 58 Figure 3.16 Total soluble nitrate + nitrite in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b) Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL duplicate reactors...... 60 Figure 3.17 Total soluble phosphorus in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b) Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL2 duplicate reactors...... 62 Figure 3.18 Phosphorus uptake over time in biomass from non-aerated SBAn1 and SBAn2 duplicate bioreactor MLSS, and suspended algae, after feeding with minimal synthetic wastewater containing phosphate (no BOD source, no N source, no complex P source). .... 63 Figure 3.19 Total nitrogen concentration in reactors from phosphate uptake experiment 0 h, 1 h, and 4 h post-feeding...... 65 Figure 3.20 Total phosphorus concentration in reactors from phosphate uptake experiment 0 h and 3 h post-feeding...... 65 Figure 4.1 Picture of 1.5-L MAB CSTRs in light incubator...... 72 Figure 4.2 Schematic of 1.5-L CSTRs...... 72 Figure 4.3 Settling behavior of biomass in 1.5-L MAB CSTRs. Samples from high BOD (left), med BOD (middle), and low BOD (right) reactors after 30 min of settling...... 73 Figure 4.4 Total soluble nitrogen removal efficiencies, and conversion to nitrate + nitrite via nitrification, as a percentage of influent total N in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 3 and 6...... 76 Figure 4.5 Total soluble phosphorus removal efficiencies as percentage of influent total P in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 3 and 6...... 77 Figure 4.6 SVI of all reactors on Day 6 of reactor operation (at which point partially and fully defined reactors were terminated due to excessive bulking)...... 79 Figure 4.7 Filamentous bulking in partially/fully defined N and P source reactors on Day 6. (A1-2) partially defined (B1-2) fully defined...... 79

vii

Figure 4.8 ChlA concentration in reactor mixed liquor for each reactor at sampling points over 30 days of operation. Days 2 and 6 were sampled but had undetectable concentrations of ChlA...... 81 Figure 4.9 AI of biomass in reactor mixed liquor for each reactor at sampling points over 30 days of operation. Days 2 and 6 were sampled but had undetectable concentrations of ChlA, and thus AIs of 0...... 81 Figure 4.10 Effluent nitrate + nitrite as percentage of influent total N in low, medium, and high BOD reactors after settling and filtration (0.45 µm)...... 82 Figure 4.11 Total soluble N removal efficiencies in effluent as percentage of influent total N in low, medium, and high BOD reactors after settling and filtration (0.45 µm)...... 83 Figure 4.12 Stacked bar plot of percentage of total influent N removed by uptake into biomass or converted to nitrate + nitrite via nitrification...... 83 Figure 4.13 Nitrification and N removal as a percentage of total influent N in settled, filtered (0.45 µm) samples taken 30 min after the start of the light period (AM), midway through the light period (PM), and 1 h prior to termination of light period (night)...... 95 Figure 4.14 Total soluble phosphorus removal efficiencies as percentage of influent total P in low, medium, and high BOD reactors after settling and filtration (0.45 µm)...... 96 Figure 4.15 P removal as a percentage of total influent P in settled, filtered (0.45 µm) samples taken 30 min after the start of the light period (AM), midway through the light period (PM), and 1 h prior to termination of light period (night)...... 97 Figure 4.16 Luxury P uptake over a 24 h period by the high BOD reactor biomass re- suspended in synthetic wastewater with no N source and 0, 0.05, and 0.1g/L glucose...... 98

Figure 4.17 Suspended (non-flocculated) algal concentration (OD750), total N uptake, and nitrate + nitrite concentration in duplicate batch reactors inoculated with biomass from the medium BOD reactor...... 100 Figure 5.1 Filament counts of A. flos-aquae after invasion of microcosms. Dotted line indicates the expected filament washout rate without cell growth due to dilution, starting at the average filament concentration at invasion for the three reactors within each propagule pressure treatment...... 108 Figure 5.2 Chlorophyll A fluorescence in microcosms on Days 0, 2, and 4 after invasion by A. flos-aquae...... 109

viii

ACKNOWLEDGEMENTS

I would like to express my most sincere thanks to my advisor Dr. Kurt Rhoads for getting me

through this. Kurt has offered wisdom, encouragement, humor, friendship, and delicious

cooking over the years since we met back in Ithaca, NY. I am extremely fortunate to have

been able to work with him over the last five years. Thanks so much for giving me this opportunity.

I would also like to thank the committee members: Dr. Harihara Baskaran, Dr. Aaron

Jennings, and Dr. George Wells. Your lessons in the classroom, research insight, collaboration,

and sharing of resources have made it possible for me to complete and assemble this work. I

am so appreciative of your time and effort serving on this committee.

Thanks to the undergrads who have helped me over the years – Nick Merchant-Wells, Hana

Litwin, Calvin Zehnder, Kathryn Lundgren, and Aaron Mann, and to the Civil Engineering

faculty and staff – special note to Nancy for helping me get literally anything figured out.

A huge thanks to Brian Flanagan at Southerly Wastewater Treatment Plant (NEORSD) for

sampling help. This research simply would not have been possible without your help and

willingness to offer us your time. The same is true for Jim Berilla, who is an absolute magician

with the ability to fix anything and everything. Special thanks for the various facilities which

have assisted with my research over the years. This includes Scott Howell at the Visual

Sciencies Research Center for microscopy work, Jim Faulk in the Department of Chemistry

for use of the GCFID, and the CWRU Core Facilities for DNA services.

Finally, on a personal note, I want to thank my friends and family, the single most important

part of my life. Thanks to: Alana, Mom, Dad, Anna; Conor, Toby, Brett, Andrew, Tim (the

boyos); Tim (+ Niki), Bmac, and David JHB; Sage and Emily (aka Malfoi and Malfoi); Caspar,

David, and Kevin; Kyle and Tim, Carolina/Jimmy + Beth/Bobby, Tanya and Paul (+ Ava),

and my awesome officemate Yuan Guo. Much love to the big wonderful Bohon family and

all the Lorias. Special shout out to Aunt Jennilyn for everything, Uncle Joe for summer nights watching sports down in TN, Uncle Paul for the guitar which I have used as an escape from

research so often, Aunt Sharon for TV show recommendations. I cannot wait to be able to be able to see you all again now that I’m allowed to leave the lab… Finally, to Granny and Papa,

Grandma, Uncle Buzz, and Aunt Genny, this is for you.

LIST OF ABBREVIATIONS

AI Autotrophic Index

ANOVA Analysis of Variance

AOB Ammonia Oxidizing Bacteria

AS Activated Sludge

BOD Biochemical Oxygen Demand chlA Chlorophyll A

CSTR Continuously Stirred Tank Reactor

DO Dissolved Oxygen

FAME Fatty Acid Methyl Ester

HABs Harmful Algal Blooms

HEPES 4-(2-hydroxyethyl)piperazin-1-piperazineethanesulfonic acid

HRAPs High Rate Algal Ponds

HRT Hydraulic Residence Time

MAB Microalgal-Bacterial

MLSS Mixed Liquor Suspended Solids

N Nitrogen

NOB Nitrite Oxidizing Bacteria

OECD Organization for Economic Co-operation and Development

OTU Operational Taxonomic Unit

P Phosphorus

PBR Photobioreactor

PBS Phosphate-Buffered Saline

PHB Polyhydroxybutyrate qPCR Quantitative Polymerase Chain Reaction

SRT Solids Residence Time

SBR Sequencing Batch Reactor

TAPS [tris(hydroxymethyl)methylamino]propanesulfonic acid

TSS Total Suspended Solids

UTEX University of Texas Collection of Algae

WWTP Wastewater Treatment Plant

xii

Microalgal-Bacterial Consortia for Biofuel Production and Wastewater Treatment

ABSTRACT

MARK LORIA

Microalgae have shown promise as a platform for biofuel production and for providing low- cost nutrient removal from wastewater. The combination of these potential uses of microalgae could allow for energy recovery from wastewater streams. There are multiple approaches to microalgal cultivation, but based on theoretical considerations and previously published results, this dissertation focuses primarily on polyculture microalgal communities and combined microalgal-bacterial consortia. These mixed microbial communities are predicted to be productive and robust to disturbances, key features of a microalgal cultivation platform.

High value microalgal biomass production from wastewater requires research on crop protection strategies, microbial community assembly, approaches to biomass harvesting, and operation of reactors for efficient nutrient removal from wastewater. Adequate crop protection is necessary for high biomass productivities and minimization of reactor crashes.

Low-cost harvesting strategies will help lower energy inputs into microalgal reactors and increase the economic viability of algae for commercial applications. Finally, it is important that algae cultivated on wastewater produce effluent that is as clean, or cleaner, than traditional activated sludge systems.

xiii

This dissertation addresses aspects of these key research questions for microalgal cultivation

platforms. This includes the investigation of a bioprocess control strategy with the goal of

increasing the lipid yield of algal polycultures through natural selection. Additionally,

interactions between bacteria and microalgae in microalgal-bacterial systems, and specifically

the competition between nitrifying bacteria and microalgae, is examined. The performance of

these cultures for nutrient removal from synthetic wastewater was evaluated, revealing luxury

N and P removal from wastewater by MAB consortia. Settleability of algal biomass is a key

component of wastewater treatment and biofuel production systems, and this parameter was

also assessed for the MAB consortia. Preliminary data on microbial invasion of microalgal cultures is presented, with relevance to bioaugmentation of microbial communities with specific microalgae, and to the problem of harmful algal blooms in Lake Erie. These experiments have helped to improve the understanding of how different microalgae and bacteria interact and function within bioreactors, with implications for wastewater treatment and biofuel production. Results reveal several future research directions relating to microalgal- bacterial reactor operation and the underlying ecology of microalgal-bacterial consortia.

xiv

Chapter 1 INTRODUCTION

1.1 Background

Microalgae are promising for biotechnology applications because of their ability to convert

sunlight and atmospheric carbon dioxide into valuable products. Microalgae can convert

inorganic carbon into lipids which can be used for liquid biofuel production1, carbohydrates which can be used to generate ethanol2, and algal biomass could be digested for biogas

production, used as animal feed, or applied as a fertilizer2,3. For biodiesel production from

lipids, algae have the potential for higher yields per land area than other proposed biodiesel

crops4.

As with any crop, microalgae require water and fertilizer for cultivation. However, using clean

water for microalgal production could compete with other water uses in some regions, and

fertilization (e.g. nitrogen and phosphorus) is a key factor in the energy and cost balance of

microalgal fuels5. Using wastewater to feed algal cultivation ponds is potentially advantageous

for algal biofuel cultivation because wastewater provides both water and fertilization.

Wastewater-fed algal ponds could mitigate the economic and energetic impacts of water use and fertilization for algal cultivation and allow for recovery of energy and nutrients from pre-

existing waste streams6–8.

While wastewater-fed algal cultivation is a promising conceptual framework for biofuel

production, with a growing body of experimental evidence supporting its viability, there are

several key aspects of the platform which require additional research. Crop protection will be

one important aspect of algal cultivation in wastewater ponds, where algae will be subject to

invasion by potentially undesirable microbial species; pathogenic effects of viruses, chytrids,

1 or other organisms; as well as grazing pressures9. Harvesting algal biomass is also an important consideration in algal cultivation and is a key component of the energy balance for microalgal biofuel production10. Finally, optimization of the lipid content and wastewater treatment capabilities of microalgae is important for overall viability of such systems. Higher biomass productivity and lipid content of harvested algal biomass will improve the economic and energetic balance of microalgal biofuels. Additionally, optimizing wastewater treatment performance can both enhance microalgal bioreactor effluent quality and improve the efficiency of energy and nutrient recovery from wastewater.

1.2 Crop Protection in Wastewater-fed High Rate Algal Ponds

As with any agricultural operation, protection of algal crops in high rate algal ponds (HRAPs) will be subject to pressures from invading organisms, pathogens, and grazers which can influence overall biomass productivity. Based on a priori considerations derived from ecological theory and some experimental evidence, there is growing consensus that monocultures of algae will not be robust and reliable at scale compared to polycultures of microalgae and other microorganisms9,11–14. Higher algal diversity in polycultures would be expected to confer higher resistance to grazing12 and to invasion by competing microorganisms9, as well as decreased susceptibility to pathogens9,15, relative to monocultures.

Additionally, monocultures are inherently at odds with a wastewater-fed HRAP, which will be subject to continuous immigration of diverse microorganisms in wastewater.

Lab-scale experiments with algal polycultures suggest that increased diversity could also create higher productivities relative to monocultures. Studies have shown that compared to monoculture, polycultures of algae exhibited higher light use efficiency, higher biomass yields, and higher lipid contents16–19. Polycultures, then, are predicted to be more robust, produce

enhanced lipid and biomass yields relative to monoculture, and generally fit better into the

context of wastewater-fed HRAPs. For these reasons, an improved understanding of microalgal-microbial polyculture function is an important research direction for enhancing the

viability of microalgal biofuel and wastewater treatment systems.

1.3 Community Assembly in Microalgal-Microbial HRAPs

Depending on the goals of algal cultivation, there are a number of potentially desirable

characteristics in algae colonizing polyculture HRAPs, including wastewater treatment performance, lipid or carbohydrate yield, biomass productivity, grazing or pathogen resistance, halotolerance, and thermotolerance. The relative importance of these characteristics would depend on the industrial use of the algal biomass, the waste stream used to feed the HRAP, and the local environment. Assembling a community of microalgae and other microbes with desirable characteristics will be an important aspect of algal biofuel production, as it is in other engineered microbial systems13. Two general strategies that have been identified for assembling

a microbial community with desired function are engineered community assembly20 and

selecting a community from a natural pool of microorganisms21. Both of these strategies are relevant to microalgal biofuel production.

Microalgal-microbial communities could be assembled by choosing microalgal strains with desirable traits for inoculation, such as high lipid content microalgae which resist grazing and have complimentary ecological niches20. Additionally, strains could be chosen using bioprospecting studies to characterize algae or cyanobacteria strains from the local environment with desired traits22. Such a community of microalgae would be predicted to be well adapted to local environmental conditions and would not pose the risk of contamination in the event of accidental release from a HRAP. With enough prior information on selected

strains, a stable microbial community with desired functionality could, in theory, be assembled synthetically20. The challenges with synthetically assembled communities include creating

stable coexistence among selected strains under possibly varying conditions and potential

crashes in system performance should one or more selected strains be outcompeted from the

system.

Alternatively, microalgal-microbial reactor communities could be assembled from a large,

diverse inoculum or with repeated colonization from outside microorganisms, but with

operating conditions designed to enhance specific functional characteristics of the

community21. Rather than rely upon a body of knowledge on specific algal strains, and models

for how multiple strains might interact and coexist within a complex and dynamic system,

natural selection takes advantage of the stability inherent to diverse natural systems. The more

challenging aspect of a natural selection-based strategy to community assembly is modelling

and predicting bioprocess controls to select for one or more required functions, perhaps

simultaneously.

1.4 Microalgal Harvesting and Bioflocculation

Harvesting of microalgal biomass is an important aspect of the energy balance of microalgal

biofuels10. Of the harvesting strategies for separation of algal biomass from water, flocculation of microalgae has emerged as an economical solution to the algal harvesting challenge23.

Flocculation could potentially be accomplished in algal culture using chemicals or

autoflocculation at high pH, but another promising approach to algal harvesting is using other

microorganisms to initiate in situ flocculation in algae through a process referred to as

‘bioflocculation’23,24. This bioflocculation in algae has been initiated through the use of bacteria, fungus, or other autoflocculating algae species23.

Bioflocculation can provide a low-cost, low-energy, streamlined approach to microalgal

harvesting if co-culture of microalgae with flocculating microorganisms produces a settleable

microalgal consortia which still functions for the desired goals of biofuel production and/or

wastewater treatment. Such a strategy of algal growth has been employed with the cultivation

of microalgal bacterial (MAB) flocs, typically using activated sludge bacteria to initiate algal

flocculation. MAB floc research has typically emphasized wastewater treatment with success25–

27, though lipid content of these algal cultures has been measured in some cases for viability as

a biofuel feedstock28–30. MAB flocs have applications purely for wastewater treatment,

independent of biofuel production, by reducing the aeration burden in wastewater treatment

systems26 and enhancing nitrogen (N) and phosphorus (P) removal relative to activated sludge

alone31. MAB flocs combine the settleability and biochemical oxygen demand (BOD) removal

of activated sludge with the nutrient uptake and biotechnology potential of microalgae, and

they fit well into the context of HRAPs fed with wastewater.

1.5 MAB Floc Cultivation and Wastewater Treatment

While growing microalgal biofuel feedstock cultures with wastewater can mitigate the energetic

and economic costs of fertilizer and water use, it is important that wastewater treatment

efficiency not be compromised in these HRAPs. Independent of biofuel production,

wastewater treatment with microalgae also presents an opportunity for the development of

improved, lower-cost wastewater treatment systems. Settleable MAB floc biomass is a

promising platform for combined biofuel production and wastewater treatment. MAB flocs

can be developed from activated sludge through the community assembly approaches

previously discussed: inoculation with pre-selected algal strains or natural selection from a

diverse algal inoculum or from algae already resident in the activated sludge itself. Inoculating

activated sludge with specific algal strains falls into the conceptual framework of engineered

community assembly, but since the algae are inoculated into a pre-existing microbial

consortium, this approach is closely related to a technique referred to as bioaugmentation.

In bioaugmentation, strains with a functional characteristic of interest are inoculated to

enhance or change the function of the overall microbial community toward a specific goal.

Bioaugmentation has been employed in a number of engineering applications with some

success32–35. In conjunction with inoculation of the target strains, conditions can be created

which favor the competitive success of the target strain34, thus preserving its continued

success. In MAB flocs, such conditions may be created by simply increasing the solids

retention time (SRT) and hydraulic retention time (HRT) of the reactor or by decreasing

aeration, simultaneously creating an ecological niche for which algae are uniquely suited and

then filling that niche with the bioaugmenting strain. Higher SRT and HRT would allow for

algal growth through photosynthesis and the utilization of N and P not utilized for growth by

heterotrophic bacteria. Additionally, lower aeration rates would create an oxygen demand

which could be partially or fully met by algal photosynthesis, thus creating a sustained

dependence of bacterial growth on algal growth. Repeated inoculation of the target strain has

also been suggested as a possible method to ensure that the bioaugmented strain of interest

maintains abundance and function within the microbial community33. Bioaugmentation for

the assembly of MAB communities has been demonstrated experimentally26,29,30.

An alternative natural-selection approach to MAB floc assembly is altering the conditions of reactor operation to create a niche for algae, and then either inoculating with a diverse natural consortia of algae, which will self-select into the community, or simply allowing native

activated sludge-associated algae to grow in abundance in response to new niche availability.

The technique of inoculating with a diverse natural pool of microalgae and allowing for

community assembly through natural selection has been applied in several MAB floc systems

treating different types of wastewater at several scales25,27,36–38.

Enhancing the viability of MAB flocs through either approach of engineered strain selection or natural selection for community assembly requires additional research. Several experimental studies have operated MAB floc reactors as sequencing batch reactors (SBRs) with and without aeration30,37, with different feeding regiments designed to combine algal nutrient removal with

nitrification/denitrification39, as continuously stirred tank reactors (CSTRs) at different

scales26, and with different microbial communities. The varying reactor conditions, feed

regiments, and engineering end-points in existing studies make it difficult to evaluate whether

MAB floc are suitable for the specific goal of combined algal growth and wastewater

treatment. Creating a more unified understanding of underlying processes and operational

strategies can help guide future research in optimizing performance. To this end, the current

research also improves the understanding of the basic ecology of interactions between bacteria

and algae in these systems. One example is the roles of nitrifying bacteria and algae in MAB

floc reactors. These organisms have similar ecological roles and thus have been shown to

compete under certain conditions40,41. Nitrate or nitrite release into effluents of MAB reactors

has been recorded42, and thus competition between these types of organisms could have an impact on nutrient removal performance. Additionally, studies of nutrient uptake conversion processes and rates in MAB floc reactors can help pinpoint operational considerations for optimizing performance.

1.6 Economic Values of Wastewater-fed High Rate Algal Ponds

Cultivation of settleable microalgal biomass using wastewater resources has several potential

economic benefits compared to alternative algal biofuel production platforms. These include

the offsetting of fertilization requirements for microalgal biomass production, the benefits of

inexpensive tertiary treatment, and the elimination of chemical flocculation costs. Mitigating

the difficulties of harvesting and fertilization have been identified as key components of

achieving a viable microalgal biofuel platform43.

Several published analyses have examined the costs and benefits associated with key aspects

of microalgal cultivation. First, Lee et al. in 2010 estimated the cost of chemical flocculation

of microalgae at $ 0.13 m-1 of reactor volume44. This cost would be eliminated with the use of

settleable MAB consortia. A National Renewable Energy Laboratory analysis in 2016 assessed

the cost of fertilization of algal biomass for biofuel production (using Ammonia and

Diammonium Phosphate) at a total of $21 per ton of ash free dry weight algal biomass (cost

45 of CO2 additions excluded) . Again, this cost would be eliminated with the use of wastewater for HRAP feeding. Finally, the economic benefits of algal wastewater treatment include decreased aeration requirements due to algal oxygen production during photosynthesis, and

tertiary treatment due to autotrophic N and P uptake by algae26,46. Park et al. report that the

construction and maintenance cost of algal HRAPs are lower than conventional activated

sludge systems, and the economic benefits of tertiary wastewater treatment from HRAPs

would offset construction and operational costs for algal biofuel reactors46.

1.7 Organization and Objectives

Research in this dissertation addresses three key areas in algal biofuel production using MAB:

(1) cultivation of high value biomass, (2) creating settleable cultures, and (3) simultaneously

treating wastewater along with biomass cultivation. This research is organized into four

chapters relating to one or more specific aspects of these overarching goals.

Chapter 1 presents the investigation of an approach to cultivating microalgal polycultures with

enhanced lipid content for biofuel production. This approach relates to the concept of natural

selection for microbial community assembly by using a bioreactor operational regime

consisting of feast-famine cycling designed to create a competitive advantage for a microbial

community with a high lipid content and thus a higher value as a biofuel feedstock.

Chapter 2 is an examination of the growth of MAB flocs originating from activated sludge,

in the context of combined wastewater treatment and high value algal biomass production.

MAB flocs bioaugmented with specific strains known for high lipid content are inoculated with activated sludge to examine interactions with activated sludge bacteria, wastewater treatment performance, and overall lipid content. Additionally, MAB flocs created from natural phytoplankton selected from activated sludge are operated to investigate how reactor operation influences wastewater treatment and what processes influence N and P utilization

dynamics in the reactors. These experiments were designed to better understand how bacteria

and microalgae interact, how strain selection influences bioaugmented MAB flocs, and how

operational considerations can guide future research efforts into MAB floc reactor

optimization.

Chapter 3 addresses the roles of nitrification versus microalgae in MAB wastewater treatment

systems. Reactors operated in this chapter are intended to help examine what factors and

nutrient (N and P) dynamics influence competition between these two groups of

microorganisms and how the relative success of these groups influences overall reactor

performance for wastewater treatment.

Chapter 4 presents an experimental framework and preliminary results from a microalgal

community invasion test. Invasion of microbial communities is a developing field with

relevance to environmental, engineering, and health care problems. Invasion of microalgal

communities is related to the concept of ‘bioaugmentation’ for the development of microalgal-

microbial communities for biofuel production. The success of bioaugmented strains is

analogous to a test of the invasibility of the microbial community. Additionally, invasion in

microalgal communities is relevant to understanding how toxic cyanobacteria dominate in

harmful algal blooms (HABs) in Lake Erie and other locations, where different microalgal communities existed earlier in the season47.

Chapter 2 FEAST-FAMINE REACTOR CYCLING FOR NATURAL SELECTION OF LIPID-ACCUMUATING ALGAL COMMUNITIES

2.1 Introduction

Algae are an attractive source of alternative energy due to a high growth rate and a higher

productivity per land area used, when compared to other biofuel feedstocks2. Despite

favorable aspects of algal biofuel production, remaining challenges include lowering cost,

mitigating fertilizer and water requirements, and ensuring bioreactors remain stable and productive.

One proposal for meeting fertilization and water demands of algal biomass production is growing algae using wastewater-fed open ponds28,48,49. The benefits of this platform include

lower construction costs and simpler operation compared to an enclosed, sterile,

photbiotreactor2. Additionally, waste treatment with mixed algae or microalgal-bacterial

consortia reduce the need for mechanical aeration and can result in higher nitrogen and

phosphorus removals from waste when compared to traditional wastewater treatment

systems25. Operation of open pond reactors, however, is contingent on the development of

adequate crop protection measures, since these ponds will be subject to invading pathogens

including viruses, amoeba and zooplankton grazers, and chytrids, which can have devastating

effects on algal cultures9,11,13,15,21,50.

Ecological principles suggest that mixed cultures of multiple species of microalgae and other microorganisms should be more robust to invading organisms and pathogens. In monocultures ecological niches are left vacant, providing opportunities for organisms with

potentially less desirable properties to take hold13. Additionally, the paradigm described for

algal virus infection is one of viral host specificity at the species level, if not the strain level51.

Similarly, chytrids display narrow host specificity in cyanobacteria52. Thus, where a single virus

or chytrid could decimate a single-species crop, it would only affect a single species, or even

strain, within a polyculture. In polycultures, after the decline of a single species or strain from

predation, complementary microorganisms could quickly fill newly opened ecological niches.

Growth of mixed communities of algae for biofuels may offer additional advantages over

monocultures, as well. Specifically, mixed cultures of algae have shown more efficient use of

sunlight during photosynthesis and increased production of lipids when compared to single

species cultures16,17.

Previous work by Kim et al.28 and Mahapatra et al.48 on the cultivation of uncharacterized algal-

microbial communities in wastewater showed that mixed communities provided a high degree

of nutrient removal from wastewater as well as a reasonable lipid content for biofuel

production purposes (18-28.5% of dry weight). The demonstrated effectiveness in producing

algal biomass from wastewater, in addition to ecological considerations of reactor stability,

make this a promising direction for algal biofuel research. The work in this dissertation

expands upon this previous research by specifically seeking to increase the lipid content of biomass in algal-microbial wastewater treatment reactors through a bioprocess control strategy.

The economic viability of algal biofuels depends on lipid-rich biomass production, and thus optimizing lipid content of mixed microalgal-microbial biomass is a key consideration. Several approaches may address this optimization problem, such as using engineered mixed algal community assembly by selecting strains with complementary and desirable traits to inoculate into bioreactors20. Another approach is creating a selective environment during biomass

cultivation which favors desired traits, thus creating sustained ecological pressures which select

for favorable attributes out of a natural, diverse pool of microorganisms21. The latter approach

has been demonstrated for optimizing PHB production in bacteria based on cycling of nutrient

availability to create an environment where PHB accumulation is advantageous for growth53.

Nutrient deficiencies (among other abiotic stresses) promote lipid biosynthesis in microalgae2.

For this reason, a repeated application of feast-famine nutrient stress to the same natural

phytoplankton community might effectively select for higher lipid content biomass. During a

period of oligotrophy, microalgal species with a higher lipid production capacity will

accumulate lipids in response to nutrient deprivation. When nutrients return to the growth

medium, these species that have accumulated lipids would be able to utilize the stored lipids

to grow more quickly than other species without reduced carbon stores. With repeated feast-

famine nutrient cycling, lipid-accumulating algae might be given a competitive advantage,

shifting microbial community composition and resulting in higher overall lipid production. In

addition to this biotechnology application, feast-famine nutrient regiments are relevant to

natural phytoplankton communities experiencing intermittent pulses of nutrients, notably

nitrogen and phosphorus54–57.

In conjunction with feast-famine reactor cycling, a parameter of interest in algal bioreactor operation is pH. Photosynthesis naturally increases the pH of surrounding media58, and algal ponds treating wastewater have demonstrated high operating pH values, up to 9 or 10 under some cultivation conditions28,59,60. This increased pH could aid in attenuating human pathogens

in wastewater, thus providing enhanced wastewater treatment61. Given the natural inclination

of algal bioreactors to operate at high pH, and the possible benefits of this for wastewater

treatment, the effects of pH on lipid and biomass yields are of interest.

A feast-famine cycling selective environment was created in algal-microbial bioreactors to assess the viability of this cultivation strategy for selective enhancement of overall lipid content in reactor biomass. Reactors were operated using nutrient-modulated cultivation at pH 7.5 and pH 9 to investigate the effects of elevated pH which may occur in algal ponds as a result of photosynthetic activity on lipid yields and biomass productivity. Changes in the overall lipid and starch content of biomass, along with changes in fatty acid methyl ester (FAME) profiles, were monitored in conjunction with microbial community composition over reactor cycles to assess the effects of feast-famine cycling on lipid content of cultures and to determine the possible mechanisms underlying these effects.

2.2 Materials and Methods

Media and Culture Inoculum

Cultures were inoculated with samples obtained from Wade Lagoon near the Case Western

Reserve University campus in Cleveland, OH, in November 2014. Cultures were grown using a modified WC Medium62 with nutrient levels adjusted to approximate wastewater effluent 63

(60.67 mg/L NO3, and 6.19 mg/L PO4) and with the following modifications: 26.5 mg/L sodium acetate trihydrate to provide BOD, 1 mg/L H3BO3, and the exclusion of sodium carbonate (since media were buffered with organic buffers), and Vanadium. Cultures were maintained at pH 7.5 or pH 9. For cultures maintained at pH 7.5, media was buffered with

HEPES at 715 mg/L64, and for cultures maintained at pH 9, media was buffered with TAPS at 730 mg/L.

Reactor Operation

Six 1-L sterile glass reactors were operated in a controlled environment plant growth chamber

under 14 h daily illumination with 400 µmol m-2 s-1 PAR, at a temperature of 25°C. Three of

these reactors were operated in replicate at pH 7.5, and the other three operated in replicate

at pH 9. Reactors were mixed with a magnetic stir bar at approximately 120 rpm and were

bubbled with filtered air at 0.22 L min-1 to facilitate gas exchange. After 7 days of growth, 450

mL of culture was removed, and cells were separated from media by centrifuging at 7443 g for

10 min at 4°C and re-suspended in fresh media to allow for another 7 days of growth. Tests

showed that centrifuging monocultures of eukaryotic algae at this speed did not inhibit

exponential cell growth (Appendix A.1). Cultures were initially inoculated with 50 mL of Wade

Lagoon water into 2 1-L photobioreactors (PBRs) at pH 7.5 and 9 and pre-cultured to create

an acclimated microalgal-microbial community. These starter cultures were used to inoculate

the six experimental PBRs (50 mL mixed liquor suspended solids (MLSS) inoculated) and were

allowed to grow for 4 days. Reactor cycle 0 represents an initial resuspension of the six

experimental PBR biomass into fresh media. Cycles 1, 2, 3, and 4 represent biomass sampled

and re-suspended after a 7-day batch operation period. Total biomass in the reactors over time

was tracked using total suspended solids (TSS) analysis.

Carbohydrates

Storage carbohydrates were analyzed using Megazyme Total Starch Assay Kit (AA/AMG)

(product number K-TSTA – Megazyme, Wicklow, Ireland). Total starch was extracted

according to the manufacturer’s protocol for samples with resistant starch and D-glucose

and/or maltodextrins, with modification of reagent volume proportional to the mass of algal

sample (approximately 5 mg) used for analysis. This kit has been previously shown to be

suitable for starch extraction from microalgae65, but it was retested for repeatability with mixed

microbial-algal material and at downsized reaction volumes (Appendix A.2). Starch content

was normalized to lyophilized weight of algae.

Lipids

Lipid content of algae normalized to dry weight was quantified via fatty acid methyl ester

(FAME) content for samples collected at the end of each 7-day growth cycle. FAMEs were

extracted after in situ transesterification of ~5 mg lyophilized algal mass, based on the National

Renewable Energy Laboratory standard procedures for microalgal biofuels analysis, laboratory

analytical procedure for the determination of total lipids as fatty acid methyl ester 66, with

minor modifications. The transesterification reaction was carried out in a water bath at 85°C

after the addition of 300 µl acidified methanol, 200 μL 2:1 (v/v) chloroform:methanol, and

0.2 mg of tricosanoic acid methyl ester internal standard (product number 35044 – Restek,

Bellefonte, PA). FAMEs were then partitioned into 1 mL hexane, and 200 μL of this hexane phase was mixed with 5 µg pentadecane as an internal standard (product number 76509-5ML

– Sigma-Aldrich, St. Louis, MO) and quantified by Gas Chromatography/Flame Ionization

Detection with 37 FAME mix as a calibration standard (product number 18919-1AMP –

Sigma-Aldrich, St. Loius, MO). GC/FID separation and detection was carried out using a

J&W DB Wax column (product number 122-7032 – Chem-Agilent, Santa Clara, CA), with 1µl

injection (1:10 split ratio), an inlet temperature of 253°C, oven temperature: 100°C for 1 min,

25°C/min to 200°C with 1 min hold, 5°C/min to 250°C with 7 min hold, helium flow rate of

1ml/min, and FID at 280°C, 450 mL/min air, 40 mL/min hydrogen, 30 mL/min helium.

Calculated FAMEs were normalized to algal lyophilized weight for total lipid content

determination. Samples were stored after lyophilization at -80°C, and analyzed for FAMEs in

batches of six samples, chosen randomly from all samples to be analyzed.

Community Sequencing Analysis

Community composition analysis as described by clustered 18S operational taxonomic unit

(OTU) reads for the eukaryotic community and 16S total abundance for the bacterial

community. These analyses were performed by Jim Griffin at Northwestern University in

Chicago, IL. A description of the methods used, written by Jim Griffin, is in Appendix A.10.

Differential abundance analysis was performed on the dataset provided by Jim Griffin at

Northwestern using DESeq2 implemented using Phyloseq67–69. Shannon Index was calculated

using the vegan R package70. R code for implementing these analyses is in Appendix A.3.

OTUs were clustered at 97% similarity, and a description of OTU size can be found in

Appendix A.4. Putative taxonomy of OTUs was determined by NCBI BLAST alignment71.

2.3 Results and Discussion:

2.3.1 Storage Carbohydrate and Lipid Changes with Cycling

Overall Lipid and Starch Content of Microbial Communities

Reactors showed a change in overall lipid content over the course of multiple nutrient cycles.

After the first feeding cycle, the lipid content of pH 9 reactors increased from 12.2%, on

average, to 16.2%, on average. Subsequent cycles then produced a decreasing lipid content of the biomass in the reactors, with a final average lipid content of 5.2% (Fig. 2.1). Reactors

operated at pH 7.5 showed overall less variability in lipid content over nutrient cycles, with

smaller (<1.5%) fluctuations over the course of the weekly feedings, with a net decrease in

average lipid content from 12.9% to 10.3% over four cycles (Fig. 2.1). ANOVA and Tukey

Pairwise Comparisons showed that these changes to the mean lipid content of the pH 9

reactors were statistically significant over subsequent cycles except between cycles 3 and 4

(Grouping Information in Appendix A5). ANOVA revealed no statistically significant changes to overall lipid content for pH 7.5 reactors between cycles, however (P = 0.136). Though initial

average lipid contents were similar between pH 9 and pH 7.5 reactors (12.2% and 12.9%,

respectively), by the end of operation pH 7.5 reactors had a significantly higher average lipid

content: 10.3% compared to 5.2% for pH 9 reactors (two sample t-test, equal variance, two-

tail P = 0.0002) as a result of the relative differences in the effects of cycling between pH

treatments.

Previous studies of mixed microalgal-microbial consortia reactors produced lipid contents in

biomass of 20%-28% of dry weight28,48. Even before the effects of nutrient modulation

pressures, the initial cultures had a lower lipid content than these published values, 12.2% and

12.9% of dry weight on average for pH 9 and pH 7.5 reactors, respectively. Cultures post-

cycling treatment exhibited even lower lipid contents (5.2% and 10.3% of dry weight). Rather

than select for lipid-accumulation as predicted, feast-famine cycling decreased lipid content in

cultures. The net decrease in lipid content was more dramatic and only statistically significant

in pH 9 reactors (7.0% decrease) compared to pH 7.5 reactors (2.6% decrease), despite initial

similarities in lipid content in these two sets of reactors (12.2% versus 12.9%). These results

indicate that feast-famine cycling was not an effective means of selecting for lipid-

accumulation in microalgae in the reactors. The statistically significant differences in lipid

changes between pH treatments show an interaction effect between these two variables which

dictated the magnitude and direction of the effects of nutrient cycling on lipid content.

Starch content of the biomass was measured, because starch can be an alternative energy

storage polymer in algae. Starch content of reactor biomass did not show clear trends or

changes with feast-famine cycling, with the exception of an increase in average starch content

in pH 7.5 reactors from 7.4% to 12.9% after the first feast-famine cycle. Overall, the average

starch content of the pH 7.5 reactors was higher (ranging from 7.4% to 11.7%) than the pH

9 reactors (ranging from 6% to 8%) (Fig. 2.2). Feast-famine cycling was chosen as a 7-day

cycle based on published lipid accumulation patterns measured for mixed-microalgal

communities and was longer than the time required for peak starch accumulation (around 48

h) to avoid selecting for starch accumulation instead of lipid accumulation72. Despite the initial

increase in average starch content in pH 7.5 reactors, decreases in lipid content of reactor

biomass as a result of feast-famine cycling does not appear to be a result of starch accumulation being favored over lipid accumulation.

Figure 2.2 Overall starch content of reactors over the course of nutrient cycles. Error bars represent one standard deviation above and below of three biological triplicates (three different reactors).

Fatty Acid Profiles in Reactors

FAME profiles were examined in addition to overall lipid levels over the course of feast-

famine cycles. Both sets of reactors initially showed a fatty acid profile consisting primarily of

C16:0, C16:1, C18:1, C18:2 and C18:3 (Fig. 2.3 and 2.4). The pH 7.5 reactors showed a

decrease in C16:0 and C16:1 content as a result of nutrient cycling, along with an increase in

C18:0 (Fig. 2.4). Reactors operated at pH 9 showed the largest decreases in C18:1 and C16:0, which accounted for 72% of overall lipid content decrease (Fig. 2.3).

Figure 2.4 FAME profiles of pH 9 reactors over study period (average of three replicate reactors).

Figure 2.4 FAME profiles of pH 7.5 reactors over study period (average of three replicate reactors).

The fatty acid profiles show that changes to overall lipid content of algae in the pH 9 reactor

was mediated disproportionately by a subset of FAMEs, specifically C16:0 and C18:1. The

effects of feast-famine cycling on overall lipid content and FAME profiles in pH 9 reactors

could be a result of selecting community members with both different FAME profiles and

overall lower lipid content or a downregulation of specific lipid biosynthesis pathways in one

or more species of algae present in the reactor.

2.3.2 Microbial Community Composition in Reactors

Biomass Concentration and Total 16S Abundance

Over the course of feast-famine cycles, the overall proportion of bacteria relative to total TSS could have an impact on biomass lipid content. If bacteria were outcompeting a subset of lipid accumulating algae, this could result in a net reduction in lipid content of the community.

Bacteria in these reactors includes cyanobacteria, though the lipid content of cyanobacteria is generally found to be lower than green algae, as well73,74. Total bacterial abundance was measured in the bioreactors over feast-famine cycles by qPCR of total 16S copy number normalized to extracted DNA concentration, to examine changes in the balance between bacteria and eukaryotic organisms in the reactors.

The 16S abundance as a share of total biomass decreased as a result of feast-famine cycling for pH 9 reactors between cycles 0 and 3, and it subsequently increased again between cycles

3 and 4 (ANOVA P < 0.001). For pH 7.5 reactors, no statistically significant changes in 16S

abundance were detected (ANOVA P = 0.617) (Fig. 2.5). The overall decrease in lipid content over time in pH 9 reactors (Section 2.3.1), then, does not appear to be the result of bacteria outcompeting green algae. There is not a direct causal relationship between bacterial abundance and lipid content changes in pH 9 reactors, since 16S copy number decreased

between cycles 0 and 1 while lipid content increased in pH 9 reactors over this period. Overall

biomass productivities were, in general, lower in pH 7.5 reactors than pH 9 reactors over the course of reactor operation, as indicated by TSS concentration (Fig. 2.6).

a b

Figure 2.5 Total 16S copy number normalized to ng DNA for biological triplicate reactors at (a) pH 7.5 and (b) pH 9.

Figure 2.6 TSS concentration (g/L) in pH 7.5 and pH 9 reactors over feast- famine cycles for biological triplicate reactors.

Microbial Community Composition

Eukaryotic microbial community compositions as described by 18S OTU profiles in the pH

7.5 and pH 9 reactors were initially similar, but by the end of the first feast-famine cycle, the

composition in the two sets of reactors began to diverge, while replicates of reactors

maintained close similarity (Fig. 2.7). A small number of highly-represented OTUs aligning to

Scenedesmus and Desmodesmus (OTUs: 291, 200, 242) accounted for a large percentage of total

reads in the samples across cycles for pH 7.5 reactors (33-69%) as well as pH 9 reactors (70-

96%). This large amount of Scenedesmus/Desmodesmus aligned reads is consistent with

microbiome analyses in other photobioreactors75. The Desmodesmus aligned OTU_291 was selected for at pH 9, and the Scenedesmus/Desmodesmus aligned OTU_200 was selected for at pH 7.5, from similar initial community profiles at these two different pH values.

Figure 2.7 18S OTU composition in pH 7.5 reactors (designated H for HEPES buffered), and pH 9 reactors (designated T for TAPS buffered), sampled at the first four resuspensions.

Statistically significant changes in lipid content were detected for pH 9 reactors but not pH

7.5, and the microbial communities were different between these sets of reactors. To target

specific OTUs with differential changes in abundance over time between the two pH

treatments, DESeq2 was used to detect OTUs which showed significant changes in abundance

both over time and between treatments at α=0.05. Twenty-four OTUs shifted in composition

significantly over the course of reactor cycling, but their change was specific to pH treatment.

In pH 9 reactors, eight of these OTUs were present at >0.5% of total reads, and these eight

‘highly-abundant’ OTUs together made up >96% of total reads. The eight highly-abundant

OTUs, then, changed in relative abundance differently in pH 9 reactors over time than in pH

7.5 reactors and also accounted for the majority of total sampling reads (>96%) in pH 9

reactors. For these reasons, these highly-abundant OTUs are assumed to be implicated in the

pH-specific changes to lipid content in the pH 9 reactors described in Section 2.3.1.

Over the first reactor cycle, five of the eight highly-abundant OTUs (OTU_307, OTU_98,

OTU_510, OTU_303, and OTU_270) decreased in abundance in the pH 9 reactors (Fig. 2.8).

These OTUs accounted for 23% of total reads initially, but after the first cycle accounted for

only 5% of total reads. OTU_242 and OTU_200 were relatively stable over the first cycle,

with OTU_242 decreasing from 6.7% to 5.9% abundance, and OTU_200 increasing from

16.8% to 18.8% abundance. Over this same period, OTU_291 increased in abundance to

account for an additional 13.5% of reads and thus appeared to be primarily responsible for

displacing the other decreasing OTUs over this time period. This shift in community

composition corresponded to an increase in average lipid content from 12.2% to 16.2%. Over cycles 2 and 3, however, OTU_291 increased monotonically in conjunction with a decreasing overall lipid content in the reactors, and after cycle 3 accounted for around 90% of total reads of a microbial community which had 5.9% lipids by dry weight. A change in the physiological

25 state of algae as a result of feast-famine cycling to produce a lower lipid content appears necessary to reconcile the shifts in lipid composition with shifts in microbial community composition. OTU_291 displacing other OTUs corresponded to both an increase and decrease in overall lipid content of the cultures, an effect which is at odds with algae maintaining a stable intracellular lipid composition. FAME profile data in Section 2.3.1 is consistent with the hypothesized physiological shift in algae, as pH 9 reactor lipid decreases were disproportionately affected by C16:0 and C18:1 FAMEs, possibly indicating downregulation of specific lipid biosynthesis pathways.

Figure 2.8 Relative abundance (fraction of total reads) of 8 highly-abundant 18S OTUs showing statistically significant changes in abundance and >.5% of total reads in pH 9 reactor. OTU_291 relative abundance plotted on secondary vertical axis due to higher relative abundance.

OTU_291 represented a substantial portion of the microbial community (by read count) and

increased monotonically throughout the study period. Competitive inhibition of other

organisms by OTU_291 could explain decreased bacterial abundances (as described previously in this Section) as well as a decrease in Shannon Index diversity in pH 9 reactors, which was

not seen in pH 7.5 reactors (Fig. 2.9). The decrease in diversity in the pH 9 reactor could have

contributed to physiological shifts and net decreases in lipid content over feast-famine cycles

in these reactors, as lipid content of microalgal communities have been shown to increase with

increasing species richness17.

Figure 2.9 Shannon Index (H) of pH 9 and pH 7.5 reactors over several reactor cycles.

2.4 Conclusions and Recommendations

Reactors operated with feast-famine nutrient cycling did not select for a microbial community

with higher lipid content at pH 7.5 or pH 9, as predicted a priori based on ecological theory.

pH 9 reactors, in fact, showed a statistically significant net decrease in overall lipid content

over the course of four reactor cycles. pH 7.5 reactors showed a small net decrease in lipid

content over the four cycles, but this decrease was not statistically significant. The cause of

the decrease in lipid content in pH 9 reactors was investigated using microbial community data

by assessing differential abundances of OTUs over cycle with changes distinct to pH

treatment. From this analysis, it appears that changes to lipid content in pH 9 reactors is a

result of the interaction effect of a community composition shift, which was distinct from pH

7.5 reactors, and a physiological shift in algae toward lower lipid production.

Reactors operated at pH 9 had higher biomass production as indicated by reactor TSS values,

lower microbial diversity, and lower bacterial abundance than pH 7.5 reactors. The lower

diversity and bacterial abundance seems to be the result of a monotonic increase in a single

large OTU aligning to Scenedesmus/Desmodesmus genus, which by the end of the last reactor

cycle accounted for around 90% of total sequencing reads.

Other experimental design parameters are worth investigating and optimizing to further

examine the possibility of a nutrient-based selection pressure to enhance mixed microalgal communities for biofuel production. The time period of one week may have been too long to

optimally select for oleaginous microalgae. After such an extended period of oligotrophy,

other selection pressures may alter algal physiology in a way which does not enhance overall

lipid content. Additionally, a relatively low nutrient growth media was fed into the reactors to

simulate an industrial process where secondary treated wastewater is ‘polished’, and energy

recovered through the use of an algal pond. The growth media used was a chemically defined

medium with NO3 as the nitrogen source, and it is possible that this was restrictive to the

biodiversity of the system and created a dearth of species with high lipid content which could

be selected for in the reactors. A medium with a nutrient composition closer to full-strength

wastewater would be more appropriate and support a wider range of organisms. A complex

medium consisting of either real wastewater or media components such as peptone and meat

extract, and generally higher nutrient concentrations, may promote a more diverse and robust

community. More diversity initially would increase the pool of candidate organisms with a

high lipid production capacity. pH appears to be an important parameter to investigate in algal

PBRs, as it has the potential to interact with cultivation strategy effects to produce changes in

diversity and function of the microbial community.

Similar algal polyculture reactors operated by Mahapatra et al.48 and Kim et al.28 recorded

higher biomass lipid contents than at any sampling point in this study. Several differences in

these studies could have explained these higher lipid contents. The reactors operated by Kim and Mahapatra were inoculated with sample collected from wastewater fed lakes and pretreated wastewater. This inoculum may have had initially higher diversity than that of the feast-famine reactors operated in this study. Additionally, the configuration of the Kim and

Mahapatra reactors was different, one operated batch reactors and the other a semi-continuous

HRAP. This difference in reactor configuration has implications for physiological regulation

of lipid biosynthesis in algae.

The effects of feast-famine cycling on algal polycultures has implications for natural systems,

especially with respect to the pH 9 reactors. These reactors exhibited a decrease in species

diversity as a result of feast-famine cycling. In environments subject to similar nutrient

fluctuations, pH would also be subject to increases due to algal growth. The combinations of

these variables produced decreasing species diversity in feast-famine reactors, and thus could

produce decreasing species diversity in natural environments. This would have implications for the ecosystem stability, which generally has a positive relationship with diversity12.

Chapter 3 MAB FLOCS AS A PLATFORM FOR WASTEWATER TREATMENT AND BIOFUEL PRODUCTION

3.1 Introduction

MAB flocs are a potential alternative to activated sludge wastewater treatment and microalgal

biofuel production systems. For wastewater treatment, MAB flocs have been shown to

successfully remove BOD and provide enhanced N and P removal compared to the typical

activated sludge process25–27,29. The symbiosis between bacterial heterotrophy and algal

photosynthesis in the flocs facilitates oxygen and carbon dioxide exchange, thus decreasing

the aeration requirement for wastewater treatment26.

The symbiosis between bacteria and algae could also be harnessed for algal biofuel production.

Two of the remaining challenges in producing algal biofuel are separating the microalgae from the water and providing the necessary nutrients for microalgal growth5,10. Growing MAB floc in wastewater can potentially solve both problems. MAB flocs self-separate from water by the

process of in situ bioflocculation under quiescent conditions. During bioflocculation, microalgae associate with other microorganisms, such as bacteria, which bind together the cells, forming a larger, faster-settling particle. The resulting MAB floc can serve as a low-cost,

simple solution to algal harvesting without the requirement of chemical flocculation

processes23,29,76,77. Growing the MAB flocs in wastewater provides nutrients required for

microalgal growth while saving the cost and energy of fertilization6,28,48. Additionally,

wastewater treatment systems using microalgae can result in higher N and P removal

efficiencies through uptake for growth and nutrient storage by algae31,78–81. N and P removal is

often incomplete in current wastewater treatment plants (WWTPs), causing algal blooms and

low dissolved oxygen (DO) zones in receiving waters, and traditional advanced N and P

removal relies upon costly chemical methods and/or additional reactor configurations7,82,83.

MAB flocs may also provide protection against algal grazers in outdoor systems. Outdoor algal

bioreactors will be subject to grazing pressure from zooplankton and protozoa, which could

cause microalgal crop loss or reactor crashes12,14,84. Combined colonies containing algal and

other microbes, like MAB flocs, have been shown to resist grazing better than planktonic algal

biomass76. Xenic algae co-cultured with bacteria and other endogenous algal species found in

activated sludge could theoretically produce a more stable and productive microbial

community than an algal monoculture9. Fitting into a general theme of algal biofuel ecology,

MAB flocs for combined biofuel production and wastewater treatment could be approached with the concept of community assembly, by inoculating flocs with pre-selected algae which accumulate lipids and have high N and P removal efficiencies20, or rather by natural selection

from a naturally diverse pool of algae and bacteria to create the desired functionality21. Either strategy of MAB floc creation and cultivation requires a better understanding of the interactions between microalgae and bacteria as well as the operational considerations for stable wastewater treatment with MAB flocs.

The objective of these experiments was to investigate MAB floc biology and function related to the goal of combined wastewater treatment and high value algal biomass production.

Reactors were inoculated with specific oleaginous microalgal strains to determine how species selection and the ratio of algae to activated sludge influences the rate and extent of bioflocculation as well as the nutrient removal capacity of resultant MAB flocs. Additionally, wastewater treatment with MAB flocs was further investigated in SBRs inoculated with polyculture flocs selected over time from activated sludge treating a synthetic wastewater.

These polyculture MAB floc reactors revealed insights into nutrient utilization dynamics in

MAB flocs and operational considerations which can help guide future research efforts toward

optimizing the wastewater treatment capabilities of MAB flocs.

3.2 Materials and Methods

Microalgal Strain Selection and Pre-culture

Four microalgal strains were tested for their ability to form MAB flocs with activated sludge

(AS) bacteria: Chlorella vulgaris UTEX 395, Chlorella sorokiniana UTEX 1633, Scenedesmus dimorphus UTEX 1237, and Neochloris oleoabundans UTEX 1185. These strains were chosen based on capacity for rapid growth, growth in wastewater, and/or high lipid content.

Specifically, C. vulgaris was the dominant species in MAB flocs grown for wastewater treatment in previous studies 26,29, and strain UTEX 395 was shown to have the highest lipid content

among C. vulgaris strains in the UTEX culture collection 85. C. sorokiniana was chosen for a

comparison to C. vulgaris within the Chlorella genus, which is known for high growth rates and

lipid production 86. S. dimorphus UTEX 1237 is a strain that has been g successfully cultivated

in large (≥400L) outdoor ponds87,88, and members of the Scenedesmus genus have a demonstrated ability to grow in wastewater18,89. N. oleoabundans has demonstrated high lipid

content, lipid productivity, and the capacity for wastewater treatment 2,90,91.

To obtain sufficient biomass for the inoculation of reactors, C. sorokiniana, C. vulgaris, and S.

dimorphus were pre-cultured for 4 days, 4 days, and 7 days, respectively, on TAP medium92. N.

oleoabundans was pre-cultured for 14 days on Bold’s Basal Medium93. Activated sludge was

obtained from Southerly Wastewater Treatment Plant (Cleveland, OH), from the first stage of

a two-stage aeration tank system designed to remove BOD (first stage) and then ammonia

(second stage).

Bioflocculation Rate Experiments

C. vulgaris, C. sorokiniana, and S. dimorphus were centrifuged at low gravity (≤1507g) to concentrate cells and then resuspended in PBS. The continued growth of microalgae after centrifuging and re-suspending is validated in Appendix A.1. TSS of the resultant slurry was

measured and then mixed at the appropriate dilution with activated sludge mixed liquor to

produce a 1:5 (w/w) algae:AS ratio. Uptake of algae into flocs was measured by tracking

chlorophyll A (chlA) (fluorometric detection method) of suspended algae remaining separated from the floc after settling. After 28 h of incubation, floc was separated from suspended algae

by settling and washing twice with PBS to remove residual suspended algae. The TSS of the

floc was then measured, and chlA was extracted by hand grinding and measured using the

colorimetric detection method according to standard methods94.

Inoculated MAB Floc Reactor Inoculation and Operation

Microalgae and activated sludge were inoculated at different ratios into 300-mL Erlenmeyer flasks with a 225-mL effective volume. Prior to inoculation, activated sludge was concentrated by settling and decanting, while algal cultures were concentrated by low-gravity centrifugation.

The TSS concentration of the concentrated biomass slurries was measured to determine inoculation volumes of AS and algae.

Reactors were fed with synthetic wastewater prepared according to the OECD standard95. Tap

water used to make synthetic wastewater was exposed to a UV lamp for a minimum of 14 h

to remove chlorine and chloramines, then stored at 4°C. The reactors were incubated at 25°C

under 14 h of illumination per day, and shaken on orbital shakers at 140 rpm. Initially, AS and

algae were incubated together for 48 h, at which point reactors were switched to operation as

SBRs with HRT = 1.5 d. The SBR settling time initially was 15 min, after which effluent was

decanted and replaced with fresh synthetic wastewater. After 3 days, settling time was lowered

to minimize the persistence of suspended algae, decanting immediately after floc settling (2-5

min).

Polyculture MAB Floc Reactor Inoculation and Operation

Additional MAB floc reactors were inoculated from biomass cultivated in CSTR reactors operated over a 30-day period while treating synthetic wastewater. This biomass was

transferred to 300-mL Erlenmeyer flasks with a 225-mL effective volume containing the same

synthetic wastewater, OECD recipe with the addition of 0.05 g/L glucose for BOD, and with

equimolar N replacement of ammonium chloride for urea. The reactors were operated as

duplicate SBRs (HRT = 1.5 d), with and without bubbling aeration at 0.15 L min-1, or 4.5 d

HRT with bubbling aeration at 0.15 L min-1. The reactors were incubated at 25°C under 14 h

of illumination per day and shaken on orbital shakers at 190 rpm. The SBR settling time was

approximately 5 min.

Nutrient Analysis

SBR effluent samples were filtered through a 0.45-μm filter prior to nutrient analysis. BOD5

tests were carried out according to Standard Methods for the Examination of Water and

Wastewater, with GGA standard as a BOD check standard and a blank to check for

contamination96. Filtered effluent was digested using persulfate to allow for quantification of

total N and total P in reactor effluent, and digested samples were stored at 4°C (<24 h before

analysis)96. After digestion, total P was measured using ascorbic acid reduction according to

Standard Methods96. Nitrate + nitrite-N and total-N were both measured using a single reagent

colorimetric assay with vanadium (III) chloride as per Doane and Horwáth, using undigested

samples for nitrate + nitrite determination and digested samples for total N97. Total N and

total P digestion efficiency was checked using a cyanocobalamin digestion standard, and standard curves were prepared with each batch of analyzed samples. pH was measured using a Accumet pH electrode (Fisher 13-620-285). Dissolved oxygen was measured using an

Accumet BOD probe (Fisher 13-620-SSP).

Biomass Characterization

Sample TSS was measured according to Standard Methods using 47 mm 934-AH glass fiber

filters96. Cultures were drawn through a syringe repeatedly to homogenize the sample and

break up flocs, and then they were analyzed for fluorometric chlA content and optical density

at 600 nm. For all fluorometric chlA determinations, 100 μl of biomass was extracted into 1:1

DMSO and acetone for 20 min in the dark at room temperature (20% final sample volume)

using the extraction method of Mayew et al.98. Fluorescence was then measured with an

excitation wavelength of 430 nm and an emission wavelength of 671 nm98. This method was

validated using dilutions of C. vulgaris and S. dimorphus cultures, which can be found in

Appendix A.6. MAB flocs were observed using an optical microscope (Leica DM2500 with

DMC4500 camera) with phase contrast.

Lipid Analysis

Lipid content was analyzed using a sulfophosphovanillin (SPV) reagent for colorimetric total

lipid determination99. Samples were homogenized by diluting with PBS and drawing through

a syringe to allow for precise pipetting of small quantities of biomass, determined by

quantification of TSS of the biomass/PBS mixture. Approximately 0.5 mg of cells were

digested with concentrated sulfuric acid at 100°C, cooled in an ice bath for 5 min, vortexed to

mix, and then incubated at 35°C and 200 rpm shaking before reading absorbance at 530 nm99.

The linear range of this assay was measured for both activated sludge and C. vulgaris and can be found in Appendix A.6.

3.3 Results and Discussion:

3.3.1 Bioflocculation Rate of Microalgae with Activated Sludge

To examine differences in affinity for activated sludge flocs between different genus/species of algae, and to examine the kinetics of algal bioflocculation, three different strains of algae were inoculated into activated sludge at a ratio of 1:5 (w/w) algae:AS. Figs. 3.1-3.3 show the incorporation of the different microalgae into floc over time using chlA fluorescence of supernatant after floc settling as a surrogate for suspended algae, which remained dissociated from floc. Lower chlA fluorescence in the supernatant indicates that more algae have become attached to the floc and thus were more amenable to separation from solution by settling.

Both C. vulgaris treatments showed a 93% decrease in chlA fluorescence after just 2 h of incubation with activated sludge. The S. dimorphus treatments had 83% and 91% reduction in chlA fluorescence over this same time period and 95% and 96% reduction after 3 h. C. sorokiniana, however, showed only 33% and 35% reduction in chlA fluorescence after 3 h.

After 28 h of incubation, the C. vulgaris treatments had a net reduction of 87% and 86% chlA fluorescence, both S. dimorphus reactors a 99% reduction, and the C. sorokiniana treatments showed a net increase in chlA fluorescence of 0.7-0.8%. After several more hours, the C. vulgaris reactors showed an increase in chlA fluorescence, but the algae responsible were once again taken up into the floc over the following 7 h.

Figure 3.1 ChlA fluorescence of non-flocculated C. vulgaris in duplicate reactors after 5 min settling time.

Figure 3.2 ChlA fluorescence of non-flocculated S. dimorphus in duplicate reactors after 5 min settling time.

Figure 3.3 ChlA fluorescence of non-flocculated C. sorokiniana in duplicate reactors after 5 min settling time.

As chlA in the supernatant decreased, more algae became associated with the floc, increasing

the autotrophic index (AI) of the floc. AI (mg chlA/g TSS), a measure of the amount of algae

per unit biomass of a sample, is obtained by dividing chlA content (mg/L) by TSS (g/L) of the sample. The C. vulgaris MAB flocs had lower AIs (1.1 vs. 1.2 mg chlA/g TSS) than the S. dimorphus flocs (2.0 vs. 1.7 mg chlA/g TSS). However, C. vulgaris and S. dimorphus MAB flocs

had AIs more than three times higher than that of C. sorokiniana (0.30 vs. 0.27 mg chlA/g TSS)

(Table 3.1). This agrees with the chlA fluorescence data, in which C. sorokiniana exhibited less

suspended algae uptake into activated sludge floc than the other species. Both the decreases

in chlA fluorescence in the supernatant and the resultant AI for these flocs show that different

algal strains change the rate and extent of algal bioflocculation with activated sludge bacteria.

This difference appears to be species specific, since C. vulgaris and C. sorokiniana are both in

the Chlorella genus but have different responses to inoculation into activated sludge mixed

liquor. This has implications for strain selection in the development of MAB flocs, indicating

the need to pre-screen candidate algae for bioflocculation efficiency (the rate and extent of

algal uptake into activated sludge flocs) to the species level. Additionally, the rapid rate of algal

bioflocculation with activated sludge means that algae and bacteria may not need to be

cultivated together to achieve the benefits of settleability. Instead, activated sludge could be

used to initiate in situ flocculation of suspended algae cultures as long as a 2 to 3 h contact time could be maintained.

C. vulgaris 1 C. vulgaris 2 S. dimorphus 1 S. dimorphus 2 C. sorokiniana 1 C. sorokiniana 2 TSS (g/L) 1.06 1.21 1.00 1.21 0.90 0.95 ChlA (mg/L) 1.29 1.34 2.00 2.11 0.27 0.25 AI (mg chlA/gTSS) 1.22 1.10 2.00 1.74 0.30 0.27

The differences in algal species interactions with bacteria could be mediated by two different mechanisms. One possibility is that the differences in algal bioflocculation are driven simply

by surface properties of the different cells which create differences in physical adhesion of

microalgae to floc surfaces. However, the precedent for Chlorophyta to respond to bacterial

quorum sensing molecules has recently been established by Zhou et al.100. Quorum sensing is employed by bacteria in many contexts to coordinate activities across different microorganisms living together. In the activated sludge context, this includes signals to coordinate bacterial aggregation into flocs, and these same chemicals triggered aggregation responses in microalgae examined in the Zhou study100. This recent finding suggests that these

differences in bioflocculation between the three algal strains could be behavioral. The three

examined microalgal strains may be responding differently to bacterial quorum sensing

chemicals, creating distinct physiological aggregation responses to the activated sludge floc

signals. The possibility that different algal species have different interactions with bacteria

mediated by quorum sensing chemicals has implications not only in engineered systems, but

also in nature by potentially shaping ecological roles and associations with bacteria.

3.3.2 Inoculated MAB Floc Reactors

To further understand the interactions between microalgae species and activated sludge, four

different species of algae (C. vulgaris, S. dimorphus, C. sorokiniana and N. oleoabundans) were

inoculated into activated sludge at a ratio of 5:1, 2:1, 1:1, or 1:5 algae:AS. These mixtures were

incubated together for 48 h, then operated as SBRs. Inoculation ratios (w/w algae:AS) for C.

vulgaris were 5:1, 2:1, 1:1; for S. dimorphus, and N. oleoabundans were 5:1, 2:1, 1:1, and 1:5; and C.

sorokiniana was inoculated initially at a 5:1, 2:1, 1:1, and 1:5 ratio, but SBR operation was

continued only for the 1:5 ratio because of the demonstrably lower uptake into floc by C.

sorokiniana compared to other strains in the bioflocculation rate experiment. The higher ratios

of algae:activated sludge for C. sorokiniana exhibited high concentrations of suspended algae in

effluent.

C. vulgaris, N. oleoabundans, and S. dimorphus microalgae incorporated extensively into AS flocs,

as demonstrated by floc micrographs which show a floc surface covered almost entirely in

green algae cells (Fig. 3.4-A:C). Consistent with the bioflocculation rate experiment, C.

sorokiniana showed much less incorporation into flocs compared to the other algal strains at

the same algae:AS inoculation ratio, forming only sporadic pockets of biomass within floc and

largely suspended growth outside of floc (Fig. 3.4-D).

A B

C D

Inoculated MAB Floc Reactor Operating Parameters

Dissolved oxygen was not supplied by bubbling to reactors, thus the only sources of oxygen

for BOD oxidation were exchange with atmosphere facilitated by mixing on shaking tables

and the oxygen supplied by algal photosynthetic activity. Dissolved oxygen was measured on

Days 4, 5, and 6 to ensure that algae and atmospheric exchange were supplying sufficient

oxygen and to examine the effect of algae:AS inoculation ratio on oxygenation in the reactor.

Dissolved oxygen levels in all reactors were above 6.4 mg/L (Fig. 3.5). There was no clear trend between algae:AS inoculation ratio and dissolved oxygen levels. However, reactors with more persistent suspended algal growth (all S. dimorphus reactors and the C. sorokiniana reactor) had higher dissolved oxygen concentrations.

Figure 3.5 Dissolved oxygen concentration in MAB floc reactors on Days 4, 5, and 6. Measurements were made during the daily light period. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs) at 5:1, 2:1, 1:1, or 1:5 algae:AS (w/w) ratio.

Since photosynthetic activity increases pH, reactor pH was measured on Days 4, 5, and 6. pH values ranged from 6.4 and 10.6 in all reactors on all days measured. For the N. oleoabundans,

C. sorokiniana, and C. vulgaris reactors, there was a higher pH in reactors inoculated with higher algae:AS ratios, but this trend became less prominent or disappeared as the days progressed

(Fig. 3.6). The pH decreased in these reactors over time (all algae:AS inoculation ratios). The decrease could be the result of increased populations of nitrifying bacteria which decrease pH as a result of ammonia oxidation. The pH in the S. dimorphus reactors did not decrease, and remained higher than other reactors, above 9 for all inoculation ratios and above 10 for the

5:1 and 2:1 algae:AS ratios.

Figure 3.6 pH in MAB floc reactors on Days 4, 5, and 6. Measurements were made during the daily light period. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs) at 5:1, 2:1, 1:1, or 1:5 algae:AS (w/w) ratio.

Inoculated MAB Floc Settleability

After 8 days of reactor operation, the biomass concentration and biomass settleability were

assessed by measuring the TSS of the mixed liquor and the effluent after settling. With the

exception of the N. oleoabundans (1:1) and S. dimorphus (5:1) reactors, the mixed liquor TSS increased in all reactors compared to the initial TSS, indicating net growth of biomass during the experiment. Biomass growth showed no clear trend with algae:AS inoculation ratio or species (Fig. 3.7). Settleability was measured as the TSS of the effluent after settling for 30 min. Higher effluent TSS indicates lower settleability.

Figure 3.7 TSS in MAB bioreactors when completely mixed (blue) and effluent TSS after settling for 30 min (red). Green line indicates the initial TSS concentration at inoculation. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs).

All C. vulgaris and N. oleoabundans reactors fell near or below typical effluent suspended solids limits of 30 mg/L101,102 at all inoculation ratios, indicating good solids removal performance

compared to AS. Settleability was lower in the C. sorokiniana reactor, which can be explained

by the lower bioflocculation efficiencies of this species with activated sludge, described in

Section 3.3.1. S. dimorphus reactors, however, also exhibited lower settleability (all >65mg/L),

especially at the 5:1 and 2:1 inoculation ratios (both >240mg/L). S. dimorphus had a higher bioflocculation efficiency in experiments from Section 3.3.1 than C. vulgaris. S. dimorphus reactors had consistently higher pH values than all other reactors, though, and the 5:1 and 2:1 ratio reactors with the lowest settleability had the highest pH values. It is not clear, then, if lower settleability during SBR operation is a property associated with S. dimorphus itself, or if this result related to the high pH in these reactors. The differences in settleability can be seen in pictures of reactor biomass in Appendix A.8, with higher-settleability reactor biomass showing only a layer of flocs separated from clear effluent and lower-settleability reactor biomass showing a layer of flocs at the bottom with green suspended algal growth in the effluent. These suspended algae account for the higher effluent TSS and lower settleability in the reactors.

Aside from the presence of some suspended algae in several reactors, settling of the aggregated

MAB biomass was very rapid, within 2-5 min. For the S. dimorphus, C. vulgaris, and N. oleoabundans inoculated reactors, development of rapidly settling biomass correlated with the development of dense aggregates, similar to aerobic granules and algal-bacterial granules previously observed by other researchers. The floc density steadily increased for 78 h after algal inoculation, relative to the initial activated sludge flocs, as indicated by microscopy (Fig.

3.8). Aerobic granules are a promising platform for wastewater treatment, exhibiting high settleability because of higher densities than activated sludge103, and algal-bacterial granules

have been examined experimentally for wastewater treatment30,42,104.

A B

C D

Figure 3.8 Microscopic images of (A) initial activated sludge floc and 1:1 C. vulgaris:activated sludge flocs (B) 5 h after inoculation (C) 30 h after inoculation (D) 78 h after inoculation.

Nutrient Removal in Inoculated MAB Flocs

BOD, total dissolved nitrogen, and total dissolved phosphorus were measured for filtered reactor effluent between Days 4 and 6. BOD effluent concentrations were <20 mg/L for all reactors, which is within the appropriate range for municipal wastewater effluent discharge

(Fig. 3.9). In contrast to MAB flocs examined by Su et al.36, there was no clear relationship

between algae:AS inoculation ratio and BOD removal. However, Su et al. inoculated MAB

flocs with natural microalgal consortia, and thus much of the algal community found in the

floc may have exhibited less capacity for mixotrophic growth than the microalgae used in this

study. Consequently, the initial bacterial biomass may have had less impact on overall organic

carbon oxidation rate.

Nitrogen removal efficiencies were 35-50% for most of the MAB floc reactors and inoculation

ratios, slightly lower than similar SBRs operated by Liu et al.30, which ranged from 36-66%

total nitrogen removal depending on aeration rate. The exceptions were the two poorly settling

S. dimorphus 5:1 and 2:1 reactors. These reactors had removal efficiencies of around 60% on

Day 4 and 80% on Day 5 (Fig. 3.10). The more settleable 1:1 and 1:5 reactors for S. dimorphus

showed lower N removals of 45%-50%. There was no clear relationship between algae:AS

inoculation ratio and N removal for C. vulgaris, but there was a significant decreasing

relationship between decreasing algae:AS ratio and nitrogen uptake for N. oleoabundans reactors

(Pearson correlation, P<0.001) (averaging the 2 consecutive days measured).

The phosphorus removal efficiency by MAB flocs was highest and most consistent in the S.

dimorphus 5:1 and 2:1 reactors with dispersed growth, with average P removals of 95% and

92%, respectively, on Days 4, 5, and 6. (Fig. 3.11). The other S. dimorphus reactors had lower

phosphorus removals, 35%-53% for the 1:1 algae:AS and 35%-74% for the 1:5 algae:AS

reactor. Reactors with other algal species had lower initial phosphorus removals compared to

S. dimorphus, and removal rates decreased on subsequent days. On Day 6, biomass in all but the S. dimorphus reactors and the N. oleoabundans 5:1 reactor had either zero or negative phosphorus removal, indicating that phosphorus was released into the reactor effluent. These

P removal results are unexpected, because many algae scavenge more N and P from wastewater compared to AS bacteria36. The release of phosphorus into the effluent may be caused by AS bacterial lysis (as the community shifts in response to the new symbiosis with algae), luxury P uptake and subsequent loss by algae, or some combination of both effects.

The effects of microalgal strain selection on wastewater nutrient removal performance in MAB flocs is somewhat unclear from these results. It appears that S. dimorphus had generally enhanced N and P removal efficiencies compared to other strains and that these removal efficiencies increased with increasing algae:AS inoculation ratios. However, the contribution of suspended growth to this result with respect to N removal is still in question, with even the more settleable 1:1 and 1:5 S. dimorphus reactors supporting double the suspended algal population than the N. oleoabundans and C. vulgaris reactors. P removal showed a more

exaggerated distinction between species, with S. dimorphus providing substantially more

consistent and higher P removal efficiencies. These nutrient removal results suggest strain

selection may have an effect on nutrient removal efficiency, more so than ratio of algae to

bacteria in the system, especially with respect to P removal. The effects of settleability (and its

dependence on pH) on these removal efficiencies are not clear, but they appear to be influenced by strain selection.

Figure 3.10 Total soluble nitrogen removal efficiencies in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 4 and 5. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs).

Figure 3.11 Total soluble phosphorus removal efficiencies in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 4, 5, and 6. Reactors inoculated with C. vulgaris (Cv), N. oleoabundans (No), S. dimorphus (Sd), and C. sorokiniana (Cs).

Inoculated MAB Floc Composition and Lipid Content

To confirm that additional microalgae increases the autotrophic index of the resultant flocs,

as opposed to additional microalgal biomass simply being decanted with reactor effluent, the

chlA content of a subset of the better-settling MAB flocs was analyzed on Day 8. In the nine

reactors measured, the chlA content normalized to optical cell density increased with

increasing algae:AS inoculation ratio across all reactors, supporting the hypothesis that more

initial algae results in more algae in MAB flocs (Fig. 3.12).

MAB flocs for N. oleoabundans, S. dimorphus, and C. vulgaris were also extracted and analyzed for

total lipids. Lipid content as percent of dry weight between strains was highest for C. vulgaris

(10.4%), then for S. dimorphus (7.8%), and lowest in N. oleoabundans (6.2%) (Fig. 3.13). N. oleoabundans did not consistently show a higher lipid content with higher algal inoculation ratio, but the lipid content of the 5:1 (7.9%), and 2:1 (9.7%) were both higher than the 1:1 (6.2%) and 1:5 (6.4%) reactors.

In published analyses of the microalgal strains used in this study, replete/deplete culture conditions produced lipid contents of 14%/57% for C. vulgaris105, 13%/44% for N.

oleoabundans,105 and 2.2/28% for S. dimorphus106. Algae often accumulate lipid in response to

nutrient deprivation, thus deplete culture conditions result in higher lipid contents than in

replete conditions. The relative lipid contents of MAB flocs in the reactors at the same

inoculation was not consistent with the relative deplete lipid compositions published for

monocultures of the strains, which would predict N. oleoabundans flocs to have a higher lipid

content than S. dimorphus flocs.

Figure 3.12 ChlA florescence of MAB floc biomass normalized by OD600 for a subset of the MAB floc reactors.

Figure 3.13 Lipid content (% of dry weight) of a subset of MAB floc reactors.

If the w/w ratio of algae to activated sludge is assumed to have remained stable after 8 days

as at inoculation, the experimental percent lipid values can be compared to calculated percent

lipid values assuming a 5:1, 2:1, 1:1, or 1:5 algae:AS w/w ratio, a 13.2% lipid content for AS

(Appendix A7), and published lipid contents for the algae strains under replete or deplete

conditions (Table 3.2). This comparison shows that the percent lipids in the flocs is closer to

values expected for algae under replete conditions than deplete conditions (measured lipid

contents of 6.2%-14.1%, compared to 7.7%-13.6% calculated for replete conditions,

compared to 15.7%-42.4% calculated for deplete conditions). Though the assumption that

algae/activated sludge ratio stayed consistent over reactor operation could not be validated

experimentally, additional algae initially inoculated resulted in higher chlA contents in flocs

after 8 days of operation (Fig. 3.12). Still, this suggests that for optimization of lipid content in MAB flocs for biofuel production, strategies for causing stress-associated lipid accumulation

could raise lipid productivities at higher algae to activated sludge ratios by up to 25%.

Table 3.2 Lipid composition of flocs (% of dry weight) as measured and calculated based on published values assuming w/w ratios of algae to activated sludge remained stable.

Cv 2:1 Cv 1:1 No 5:1 No 2:1 No 1:1 No 1:5 Sd 1:1 Sd 1:5 Measured (%) 14.1 10.4 7.9 9.7 6.2 6.4 7.8 6.1 Calculated deplete (%) 42.4 35.1 38.9 33.7 28.6 18.3 20.6 15.7 Calculated replete (%) 13.7 13.6 13.0 13.1 13.1 13.2 7.7 11.4

3.3.3 Polyculture MAB Floc Reactors

N and P removal was lower than published values and/or inconsistent in the inoculated MAB

floc SBRs (Section 3.3.2). To investigate wastewater treatment in MAB flocs further, a series

of SBRs was inoculated with different biomass and altered operating conditions to achieve

more stable performance. These SBRs were inoculated with polyculture MAB floc biomass

from a CSTR treating a synthetic wastewater for over 30 days. This synthetic wastewater was

similar to the OECD recipe used for the MAB flocs inoculated with specific strains, but with

additional BOD via 0.05 g/L glucose, and ammonium chloride as an equimolar N replacement

for urea. Duplicate SBRs were operated with bubbling aeration at 0.15 L min-1 with a 1.5 d

HRT (SB1 and SB2) and a 4.5 d HRT (SBL1 and SBL2) and also without bubbling aeration at a 1.5-d HRT (SBAn1, SBAn2). The stirring rate was also increased from 140 to 190 rpm in an attempt to increase nutrient exchange with flocs. No solids were removed during the operating period for these reactors. The reactor configurations are summarized in Table 3.3.

Table 3.3 Operating conditions for polyculture MAB reactors (biological duplicate reactors for each configuration).

Settling in Polyculture MAB Floc Reactors

Initially, biomass settleability at the time of SBR inoculation was high (>95% of biomass was settled in 5 min). However, over the following week of SBR operation, settleability of biomass decreased, and more suspended algal growth was seen in the effluent, as shown in Fig. 3.14.

Suspended algal growth was measured using daily absorbance readings (optical density at 750 nm) of the effluent.

Figure 3.14 Suspended cell concentration of reactor effluent after 5 min settling, as quantified by optical density at 750 nm.

The suspended algal concentrations in SBL1 and SBL2 were higher than other reactors by Day

4, consistent with a decreased selection pressure against poorly settling biomass at higher HRT

(Fig. 3.14). SBL1 and SBL2 were terminated after Day 4 because of high concentrations of

algae in the effluent and thus a deterioration of floc growth relative to suspended growth

compared to the other reactors.

On Day 5 of reactor operation, biomass concentration and settleability after 5 min and 1 h

were measured in SB1, SB2, SBAn1, and SBAn2. TSS values in the SBR effluents ranged from

116 to 184 mg/L after 5 min of settling, and 40 mg/L to 128 mg/L after 1 h of settling, above typical effluent discharge limits of 30 mg/L101,102 (Table 3.4).

Table 3.4 TSS concentrations (g/L) of mixed liquor, effluent after 5 min settling, and effluent after 1 h settling in aerated (SB) and non-aerated (SBAn) duplicate MAB floc reactors.

However, pH also remained high in the reactors, consistently between 10 and 11 at daily

readings. Given the relationship between settleability and pH in the inoculated S. dimorphus 5:1

and 2:1 reactors, it is possible that floc strength/stability may have be influenced by pH, and

that this causes algal cells to slough into the effluent over the course of each SBR cycle. Floc

strength or floc stability is a parameter that is challenging to measure because of the lack of a

predefined method107. However, floc stability in the context of activated sludge is discussed

via an experimental study in Chapter 19 of Flocculation in Natural and Engineered Environmental

Systems107. The ‘surface free energy of bioflocculation’ is measured via mathematical

relationships to liquid surface tension and contact angle of the fluid with activated sludge. This

surface free energy was significantly positively correlated with effluent suspended solids

concentration; in other words, as surface free energy increased, effluent suspended solids concentration also increased. The authors also examined the effect of SRT (up to 20 d) on surface free energy of bioflocculation and found that surface free energy, and effluent solids concentration, decreased with increasing SRT. The sludge age of MAB polyculture flocs at

SBR startup was 24 days. Based on the effects of SRT on floc surface free energy, these flocs would be expected to have better settleability than activated sludge, given the floc age and its effect on floc surface free energy. If settleability issues in the MAB floc SBRs was related to floc strength and stability, then, the effect is likely mediated by the algae in the floc itself, rather than the longer SRT required to support an algal population in flocs. The pH-mediated floc instability over pH 10 was documented by Sheng et al.108. The pH in the polyculture MAB floc

reactors remained consistently above 10, presumably the result photosynthesis by the algae,

which increases pH.

Instability in MAB flocs resulting from increased pH associated with photosynthetic activity

could be a reactor failure mechanism which would require control in an MAB floc wastewater

treatment system. Algae will generally induce pH increases in the surrounding environment,

and if this indeed results in MAB floc instability, it could have an impact on the ability of MAB

flocs to meet effluent TSS limits and effectively harvest algal biomass. pH-mediated MAB floc

instability should be examined in future studies to validate this putative effect of pH on

effluent TSS, and establish the operational pH limits which would facilitate proper MAB floc

reactor function.

Dissolved Oxygen and BOD in Polyculture MAB Floc Reactors

Dissolved oxygen was regularly checked during reactor operation, near the middle of the light

portion of the illumination cycle. Dissolved oxygen in the reactors was above 7.5 mg/L at

every reading. Decreasing aeration did not alter midday dissolved oxygen, but this may not

have been the case overnight when photosynthesis was inhibited.

On the final day (Day 5) of reactor operation, BOD was measured in the SB1, SB2, SBAn1,

and SBAn2 reactors. All four reactors had undetectable drawdowns in dissolved oxygen in a

15-mL sample of reactor effluent. This confirms that the high pH of the reactors did not

inhibit BOD consumption and also that increased aeration was not necessary to ensure

adequate BOD removal.

Nitrogen Metabolism in Polyculture MAB Floc Reactors

Nitrate + nitrite production and nitrogen uptake in filtered effluent was measured in the SBRs

over several days of operation.

At all but one sampling point (Day 1 in SBAn2), total N removal was higher in the aerated

reactors than the non-aerated reactors (Fig. 3.15). However, overall effective mixing in the aerated reactors was higher due to the mixing effect of bubbling, thus it is not clear whether it is aeration itself, or more turbulent mixing, which increased N removal in general in the aerated compared to non-aerated reactors. Higher nitrogen removal with additional aeration is consistent with published results of MAB flocs30, though both cases may be a result of shear rather than oxygen concentration itself. For the purposes of operation of MAB floc reactors for wastewater treatment, however, this is evidence that increasing either aeration or mixing can result in higher N removal efficiencies during wastewater treatment.

Gardner et al. reported that the Reynolds number for mixing using shaking tables could be calculated using the rotational speed, kinematic viscosity, and a characteristic length scale of the reactor109, using

= (3.1) 2 𝑁𝑁𝑇𝑇 𝑅𝑅𝑅𝑅 𝜈𝜈 where Re is the Reynolds number, N is the rotational speed, is the kinematic viscosity, and

T is the characteristic length scale calculated as V0.333 with reactor𝜈𝜈 volume V.

Figure 3.15 Total soluble nitrogen in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b) Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL2 duplicate reactors.

Using Equation 3.1, the Reynolds number for the SBRs without aeration (SBAn1 and SBAn2) would be 10.7. However, the Reynolds numbers on the order of 106 are reported to be typical

for typical activated sludge systems110. Based on Reynolds number comparisons, then, the non- aerated SBRs were operating at relatively low turbulence compared to typical values for

activated sludge reactors. Increasing turbulence from mixing and shear on the flocs could have

created smaller MAB flocs in aerated reactors30, which would facilitate more nutrient exchange

with the bulk and could explain the enhanced nutrient removal compared to non-aerated

reactors. Finally, the illuminated DO levels in the afternoon were between 7.5-9 mg/L, ±0.5

mg/L between aerated and non-aerated reactors, over 3 measurement days. Though DO may

have deviated more during the non-illuminated portion of diel cycles, taken together this

suggests that the differences in nutrient removal in aerated versus non-aerated reactors was

most likely a result of mixing and not oxygenation of the fluid.

In experiments with polyculture MAB flocs from Section 4.3.4, nitrate uptake by algae was low, a result consistent with other published work27. In previous studies with MAB floc reactors, lower aeration rates resulted in lower nitrate levels in effluent26 or DO-dependent

inhibition of nitrification39. Indeed, the non-aerated SBRs SBAn1 and SBAn2 had low

concentrations of nitrate + nitrite in the effluent (<1 mg/L-N by Day 5) (Fig. 3.16). However,

nitrate + nitrite in the effluent of the aerated SB2 reactor was comparable to these non-aerated

SBRs. Additionally, nitrate + nitrite in the effluent was consistently higher in SB1 than its

duplicate reactor SB2, ten times higher on Day 4 in SB1 (11 mg/L-N) than SB2 (1.1 mg/L-

N), despite both reactors receiving 0.15 L min-1 aeration. Nitrate + nitrite in the effluent

decreased over subsequent days of reactor operation in all reactors, though afternoon

dissolved oxygen readings were stable. Differences in nitrate + nitrite uptake across reactors

and samples points (excluding data points below the nitrate + nitrite detection limit) were

significantly correlated with settleability as indicated by OD750 of the effluent (Pearson

correlation, P = 0.002). Settleability did not, however, correlate with total N uptake (Pearson

correlation, P = 0.802). The direction of the correlation between nitrate + nitrite in the effluent

and settleability is decreasing nitrate + nitrite in the effluent with increasing suspended algal

59 growth, as indicated by OD750 of the effluent. This suggests the possibility that nitrate/nitrite uptake by algae is mediated disproportionately by suspended algal growth than growth of algae in MAB flocs, an effect that could be explained by nitrate and/or nitrite export out of floc by nitrifying bacteria, for which nitrate and nitrite are a waste product. The correlation between settleability and nitrate + nitrite effluent concentrations could explain the lower levels of nitrate + nitrite in SB2 compared to SB1, since SB2 had lower turbidity readings on all but

Day 3 of operation and lower final 5 min and 1 h settleability on Day 5. This could, however, simply be a spurious correlation to another latent variable influencing nitrate and nitrite production/uptake, potentially an uncharacterized stochastic biological effect which caused deviation in behavior between duplicate SB1 and SB2 reactors.

Figure 3.16 Total soluble nitrate + nitrite in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b) Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL duplicate reactors.

Though the differences between duplicate reactors SB1 and SB2 suggest other possible explanations, controlling aeration may indeed be a viable strategy to decrease nitrification and competition for reduced N uptake in MAB floc wastewater treatment systems. The non- aerated SBRs showed equivalent BOD removals as aerated SBRs and consistently low nitrate

+ nitrite concentrations. If the effect of aeration on overall N uptake is indeed because of turbulent mixing rather than aeration itself, a lesser aerated system with high mixing rate may provide best results. Future research in optimizing MAB flocs for wastewater treatment should compare overall N removal, and nitrate and nitrite concentrations, at both different aeration rates and turbulent mixing rates to allow for deconvolution of the relative effects of mixing versus aeration on N removal and the simultaneous effect on nitrification. A better understanding of the effects of mixing versus aeration on overall N uptake would help to optimize wastewater treatment performance versus economic and energetic cost.

Phosphorus Uptake by Polyculture MAB Flocs

Phosphorus removal was >87% in all reactors on Days 2 and 3 of operation, and by Day 4 removal was >93%, with effluent P concentrations of 0.35-0.53 mg/L-P (Fig. 3.17). To assess the role of luxury P removal on this P removal capacity, and to examine how P uptake could affect suspended algae growth persistence, phosphate uptake was recorded over several hours by biomass from SBAn1 and SBAn2. Biomass was fed using the typical SBR dilution rate, but on Day 5, this feed synthetic wastewater had no N source or BOD source and had only phosphate as a P source. After feeding with this minimal plus phosphate synthetic wastewater,

40 mL of MLSS was transferred from each reactor to a 125-mL Erlenmeyer flask, then the reactors were allowed to settle for 5 min to separate floc from suspended algae, and 40 mL of the suspended algae was also transferred to a 125 mL Erlenmeyer flask. These samples were

incubated under illumination at 25°C and 190 rpm shaking for 4 h. Phosphate uptake was measured at 30 min, 1 h, 2 h, 3 h, and 4 h.

Figure 3.17 Total soluble phosphorus in MAB floc reactor effluent after settling and filtration (0.45 µm) measured on Days 2, 3, 4, and 5 for a) Aerated SB1 and SB2 duplicate reactors, b) Non-aerated SBAn1 and SBAn2 duplicate reactors, and c) Extended HRT SBL1 and SBL2 duplicate reactors.

MAB flocs removed phosphate to below 0.1 mg/L-P within 1 h, demonstrating rapid luxury

P uptake (Fig 3.18). Over 1 h, P removal by suspended algal biomass was minimal (less than

0.12 mg/L-P removal). The TSS of the suspended algae was 64 mg/L for SBAn1, and 116 mg/L for SBAn2, whereas the TSS of SBAn1 mixed liquor was 1.73 g/L and 1.47 g/L. At the beginning of a SBR cycle, then, the MAB flocs dominate suspended algal growth by mass, and rapidly scavenge P, leaving little for suspended algae growth. This is additional evidence that settleability issues may be more related to pH-dependent floc strength than growth of algae in the reactor effluent which compete for nutrients consistently over the course of SBR cycles.

Figure 3.18 Phosphorus uptake over time in biomass from non-aerated SBAn1 and SBAn2 duplicate bioreactor MLSS, and suspended algae, after feeding with minimal synthetic wastewater containing phosphate (no BOD source, no N source, no complex P source).

To control for any unexpected non-biomass effects on P removal, to check the contribution of soluble P to total P availability, and to confirm the low N availability/uptake over the incubation period, samples were digested for total N and P determination at several intervals over the 4 h incubation period. The P uptake experiment was run in conjunction with two non-biomass control samples which were filtered through a 0.45 µm filter, incubated, and digested for total P change over a 3 h period.

The initial N/P ratio in the uptake test reactors was low relative to overall microbial biomass composition, <1:1 N:P by weight (Fig. 3.19). Additionally, P uptake in by SBAn1 and SBAn2 biomass over 1 h was 6.8-7.8 mg/L-P, while N uptake was around 2 mg/L-N over the same time period. Digested total P uptake was comparable to the phosphate uptake measured, 96% total P uptake compared to >98% PO4 uptake for the MLSS over 3 h, and 50%/20% total P removal compared to 52%/22% PO4 uptake for the suspended algae over 3 h (Fig. 3.20).

Additionally, total P was not removed in the two filtered no-biomass controls (6.9 mg/L-P and 7.9 mg/L-P initial, 7.0 mg/L-P and 7.5 mg/L-P final after 3 h). These analyses confirm that P was taken up beyond stoichiometric growth requirements, that all P uptake was biomass-related, and that almost all available P was in the form of phosphate for the uptake rate experiment.

These rate experiments show rapid luxury phosphate-P uptake by MAB flocs. This is a useful property of MAB flocs for wastewater treatment, increasing net P uptake into biomass over a shorter time period than biomass growing without luxury P scavenging. Thus, luxury P uptake could allow for the operation of smaller, more efficient, wastewater treatment systems. This rapid luxury P uptake was a property of the flocs explicitly, since the contribution of suspended algal growth was controlled for in separate reactors. Ensuring that MAB floc wastewater

64 treatment systems operate in a manner that takes advantage of this ecological P scavenging strategy will be important in optimizing wastewater treatment performance.

Figure 3.20 Total nitrogen concentration in reactors from phosphate uptake experiment 0 h, 1 h, and 4 h post-feeding.

Figure 3.19 Total phosphorus concentration in reactors from phosphate uptake experiment 0 h and 3 h post-feeding.

3.4 Conclusions

Reactors operated with MAB flocs using algae inoculated with pre-selected strains, and selected from natural resident algal populations, have revealed insights into the function of

MAB floc treatment systems, as well as general interactions between bacteria and activated sludge with relevance to natural ecological roles. These results can help guide future research efforts in designing MAB flocs and MAB floc reactors with the goal of combined wastewater treatment and high value algal biomass production.

Microalgae associate with activated sludge flocs to bioflocculate quickly, within 2 h of inoculation. However, the extent of bioflocculation between microalgae and flocs varies between species of microalgae. This result was seen in batch reactors inoculated with algae and activated sludge, both by tracking non-flocculated algal populations over time, and by measuring the resultant autotrophic index of flocs after incubation with activated sludge bacteria. Differences in the extent of bioflocculation could be attributed to either purely surface adhesion properties or differences in responses to quorum sensing chemicals, but the differences in floc association has implications for natural ecological roles of microalgae in addition to engineering applications for wastewater treatment and algal biomass harvesting.

Differences in how microalgae associate with activated sludge flocs were further investigated in SBRs inoculated with species of algae at different inoculation ratios. These SBRs showed the effect of algal selection on wastewater treatment parameters with respect to P removal. P removal was inconsistent over time in C. vulgaris and N. oleoabundans flocs but was stable and higher in S. dimorphus reactors. Flocs inoculated with different microalgae at the same ratio had different resultant lipid content, though were not consistent with their relative peak lipid contents in monoculture. Based on calculations using published lipid contents in deplete

media, overall lipid contents could be increased substantially (by up to 25%) using strategies

to trigger stress-dependent lipid accumulation in MAB floc reactors.

The results of reactor operation using MAB flocs inoculated with specific algal strains indicate

that bioaugmentation of pre-selected algal strains to create settleable MAB flocs may be viable, but is contingent on pre-screening of algal strains to ensure high bioflocculation efficiency and wastewater treatment capability within floc, and to choose an appropriate initial inoculation ratio. Optimization of cultivation conditions for stress-dependent lipid accumulation will be

necessary to more fully reap the benefits of inoculating with high lipid producing strains.

The capacity of MAB flocs for wastewater treatment was investigated further in SBRs

inoculated using polycultures of algae acclimated to synthetic wastewater over 30 days in

CSTR. Polycultures exhibited high (>90%) P removal over each SBR cycle, which was shown

to be the result of rapid luxury P uptake (>90% of P removed within 30 min of incubation in

a phosphate minimal medium). Luxury P is typically stored in microalgal cells in the form of

polyphosphate80. This property of the flocs will be an important aspect of achieving optimal nutrient removal during wastewater treatment, allowing for smaller treatment systems to

achieve the same nutrient removals. N removal was higher overall in aerated reactors compared to non-aerated. It is unclear if this result is related to oxygen concentrations or

overall mixing rate. Analysis of the Reynolds number in the non-aerated reactors, along with

the theoretical effects of increased turbulence on floc particle size and comparisons of

illuminated DO levels between aerated and non-aerated reactors, suggests the effect of

aeration on nutrient removal is likely due to differences in overall turbulence from mixing rather than oxygen concentrations in the reactors. It is possible that a higher mixing rate with reduced aeration could simultaneously increase N uptake and reduce the competition of algae

with nitrifying bacteria for reduced N. Further research on the effects of turbulence versus

aeration on overall N uptake may help to optimize nutrient removal efficiency versus energy

input into the system.

Settleability in polycultures was well above typical discharge limits for TSS (30 mg/L) in all

reactors101. Degradation of settleability in activated sludge flocs as a result of high pH has been

documented in the literature108. This effect could explain degradation of settleability of the

polyculture MAB flocs over time, as well as the apparent degradation of bioflocculation

efficiency in S. dimorphus SBRs compared to bioflocculation experiments conducted in

activated sludge mixed liquor over a shorter timescale. Given the effects of photosynthesis on increasing pH, this could be an important operational control for achieving high settleability with MAB floc wastewater treatment, and the effects and operating limits for pH in MAB floc systems should be investigated further.

Despite some possible floc instability at high pH, bioaugmented MAB floc cultures revealed that the flocculated cells themselves settled very rapidly, within approximately 2 min.

Improved settleability of MAB biomass relative to activated sludge alone could be a key component of the viability of MAB cultivation. Currently, relatively poor settling by activated sludge leads to lower treatment flow rates and large reactor sizes111. For these reasons,

improving settleability with MAB flocs could help offset the disadvantages of the lower growth

rate of algae relative to heterotrophic bacteria for wastewater treatment.

Chapter 4 COMPETITION BETWEEN NITRIFICATION AND ALGAL GROWTH IN MAB REACTORS TREATING SYNTHETIC WASTEWATER

4.1 Introduction

Several studies have recently examined wastewater treatment using combinations of activated sludge bacterial communities and microalgae. These microalgal-bacterial (MAB) consortia take advantage of the natural symbiotic exchange of oxygen and carbon dioxide between algae and bacteria to reduce aeration26 and utilize the ability of algae to take up additional N and P from wastewater to enhance treatment efficiency31,78,79. As a byproduct of wastewater treatment,

MAB reactors can produce settleable algal biomass, which may be valuable as a biofuel feedstock or for other industrial applications.

One unknown interaction in MAB reactors is the role of nitrifying bacteria (ammonia oxidizing bacteria and nitrite oxidizing bacteria, or AOBs and NOBs) versus algae in the system. Algae and nitrifying bacteria occupy similar ecological niches in the wastewater treatment context, specifically autotrophic growth at slower growth rates than heterotrophic bacterial biomass

(thus requiring longer SRTs)112. In several studies, competitive inhibition of nitrifying bacteria by algae mediated by competition for nutrients has been shown experimentally40,41,113. Half-

+ saturation constants for ammonia have been reported as 1.0 mg NH4 -N/L at 20°C for

112 AOBs , versus 31.5 NH4+-N/L for the microalgae C. vulgaris growing in a synthetic wastewater medium114, indicating a potential competitive advantage for algae over AOBs for reduced nitrogen uptake. Whereas nitrification releases nitrate and nitrite as byproducts of reduced N oxidation for energy, algae take up nitrogen for biomass production and harvest energy through photosynthesis (or organic carbon oxidation during mixotrophic growth).

Reduced N is a preferred and energetically favorable N source for algal growth, and nitrate uptake has been shown to be low in studies of some MAB reactors27,115. Furthermore, algae have a demonstrated ability for luxury nutrient uptake beyond growth requirements80,116. For

these reasons, competition between algae and nitrifying bacteria may have implications on

overall nutrient removal efficiency in MAB wastewater treatment systems.

In MAB wastewater treatment, nitrifying bacteria are reported to compete and persist in

reactors in a manner which appears dependent in part on aeration and dissolved oxygen levels.

In one study, continuously aerated SBRs had nitrite concentrations in effluent accounting for

28-44% of total influent N, whereas partially aerated SBRs had nitrite concentrations

decreasing with decreasing duration of aeration; and overall N removal in the reactors was 36-

66%30. In another experiment, MAB SBRs with DO levels remaining between 7-9 mg/L had

approximately 60% of influent N leaving in effluent as nitrite or nitrate, and overall total N

removal was around 40%42. Fed-batch reactors operated without aeration showed nitrate in

the effluent increasing only after photosynthesis had increased DO levels, and by the end of a

cycle nitrate accounted for 17-20% of total influent N, and overall N removal was 57-79% of influent N25. Another non-aerated lab-scale MAB CSTR with low (0.2-1.5 mg/L) DO showed

almost no effluent nitrate and 69-87% of total influent N was removed26. Despite previous results showing competitive exclusion of AOB/NOB by algae, nitrite and nitrate concentrations persisted in effluents of several of these published MAB wastewater treatment reactors. Effluent nitrite and nitrate concentrations appear to be influenced by both mechanical aeration and generation of oxygen from photosynthesis through DO-dependent

inhibition of nitrification and/or denitrification removing any oxidized nitrogen produced39.

To help better understand the interactions between algae and nitrifying populations in MAB

wastewater treatment systems, aerated 1.5-L CSTR reactors were inoculated with activated

sludge. These reactors were fed with synthetic wastewaters with different compositions,

particularly BOD:N and BOD:P ratios. It was hypothesized that algae in the wastewater could

exhibit mixotrophic growth and that this mixotrophic growth could give a competitive

advantage over nitrifying bacteria. The reactors provided additional understanding of the

competition between nitrifying bacteria and algae in MAB treatment systems and the

subsequent effect on wastewater treatment efficiency. Additionally, studying nutrient

utilization dynamics in these reactors will help design MAB reactors for different waste stream

compositions.

4.2 Materials and Methods

CSTR Design and Operation

CSTR reactors were fabricated to allow for continuous feeding and decanting while retaining

biomass by creating an effluent port where the velocity of flow was less than the settling

velocity of flocs (Figures 4.1 and 4.2). This selected for a settleable MAB consortia, which was retained inside of the reactors (Figure 4.3).

Figure 4.1 Picture of 1.5-L MAB CSTRs in light incubator.

Influent

Effluent

Figure 4.2 Schematic of 1.5-L CSTRs.

Figure 4.3 Settling behavior of biomass in 1.5-L MAB CSTRs. Samples from high BOD (left), med BOD (middle), and low BOD (right) reactors after 30 min of settling.

CSTRs were designed out of polycarbonate tanks for approximately 1.5-L effective volume with constant feeding at a rate of 1 L/d (HRT = 1.5 d) using a peristaltic pump (Wiz Peristaltic

Pump, Insco, Inc.). Solids removal was accomplished by daily MLSS removal of 60-65 mL, resulting in an SRT of 24 d. Reactors were stirred using a modified Phipps and Bird jar test stirrer at 85-90 rpm. Aeration was provided by bubbling air at 0.15 L min-1. The CSTRs were enclosed in a PBR operated at 25°C with 400 μmol m-2 s-1 PAR illumination for 14 h per day.

Reactors were cleaned once mid-way through the operating period due to occlusion of light by algal biofilms on the reactor surface. The pH in the reactors was adjusted by additions of sodium bicarbonate into the influent synthetic wastewater. The reactors were inoculated at startup with MLSS from Southerly Wastewater Treatment Plant (Cleveland, OH).

Synthetic Wastewater Composition

The synthetic wastewater compositions were modified from the OECD standard95, per (L):

0.11 g meat extract, 0.16 g peptone, 53.5 mg NH4Cl, 28 mg KH2PO4, 13.2 mg CaCl2•2H2O,

30 51.1 mg MgSO4•7 H2O, and trace elements , with 0 g/L glucose (low BOD reactor), 0.05 g/L glucose (medium BOD reactor), and 0.1 g/L glucose (high BOD reactor). The partially

defined reactor composition contained (per L): 0.18 g glucose, 0.055 g meat extract, 0.08 g

peptone, 126 mg NH4Cl, and 31.6 mg KH2PO4. The fully defined reactor had no meat extract

or peptone and: 0.275 mg glucose, 0.126 mg NH4Cl, and 35 mg KH2PO4. The resultant nutrient compositions and BOD of the formulations are summarized in Table 4.1.

Table 4.1 Summary of nutrients in different synthetic wastewater formulations.

low BOD med BOD high BOD partially defined fully defined BOD mg/L 165 200 224 182 159 Total P mg/L-P 7.7 7.7 7.7 7.6 7.7 Total N mg/L-N 45.0 45.0 45.0 42.3 41.4

Nutrient and BOD Analysis

Nutrients and BOD in reactor effluent were analyzed after filtering through a 0.45-μm syringe filter. BOD was measured according to standard methods with GGA standard and blanks to control for test efficacy and contamination96. Total N and P were analyzed after

digesting samples with potassium persulfate96. Each set of samples was digested alongside a check standard (cyanocobalamin), in addition to a re-digestion of influent synthetic

wastewater to check for consistency. Total N and P recovery from cyanocobalamin was

above 90% at all digestions, and the standard deviation of total N and P in repeated

digestions of influent synthetic wastewater was 0.24 mg/L-P and 0.93 mg/L-N.

Additionally, N and P standard curves (of nitrate and phosphate) were created for each

batch of samples. Post digestion, phosphate was measured by the ascorbic acid reduction

method96. Nitrate + nitrite and total N were both measured using a single reagent colorimetric assay with vanadium (III) chloride97, using undigested samples for nitrate +

nitrite-N determination. pH was measured using a Accumet pH electrode (Fisher 13-620-

285). Dissolved oxygen was measured using an Accumet BOD probe (Fisher 13-620-SSP).

Chlorophyll A, TSS, and Autotrophic Index

Biomass was filtered onto 47 mm 934-AH glass fiber filters for TSS and chlorophyll A

determination. Chlorophyll A was extracted into 90% acetone by hand grinding filters using a

Tenbroeck Tissue Grinder (Kimble Kontes) and extracting overnight at 4°C. Chlorophyll A

content of the extract was measured colorimetrically96 and normalized to biomass TSS for

determination of autotrophic index.

Sludge Volume Index and Microscopy

Sludge volume index was determined by removing 60 mL of MLSS and settling for 30 min in

a 100-mL graduated cylinder. Sludge volume after the 30 min was recorded, and then the sample was re-mixed and TSS concentration measured. SVI was calculated based on Standard

Methods96 using

( )/ ( ) 1000 = (4.1) 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑚𝑚𝑚𝑚 𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝐿𝐿 𝑥𝑥 𝑆𝑆𝑆𝑆𝑆𝑆 𝑚𝑚𝑚𝑚 𝑇𝑇𝑇𝑇𝑇𝑇 � 𝐿𝐿 � Reactor biomass was observed using an optical microscope (Leica DM2500 with DMC4500

camera) with phase contrast.

4.3 Results and Discussion:

4.3.1 Defined vs Complex N and P source

Nutrient Removal and Nitrification

Percent nitrogen and phosphorus removal, as well as the percent of influent nitrogen converted to nitrate or nitrite by nitrification, was monitored starting on Day 3 of reactor

operation.

Nitrification increased by 25-60% in all reactors between Days 3 and 6 of operation, and nitrogen uptake by biomass decreased by 5-20% over the same period (Fig. 4.4). Algae was

not present at a detectable level during this period of reactor operation. The initial microbial

community in the original activated sludge system was operated for BOD removal prior to a secondary nitrification tank with a short solids residence time (2-4 days), which is expected to

select for heterotrophic bacteria. The increase in nitrification and corresponding decrease in

Figure 4.4 Total soluble nitrogen removal efficiencies, and conversion to nitrate + nitrite via nitrification, as a percentage of influent total N in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 3 and 6.

N removal in all three reactors between Days 3 and 6 could be explained by a shift from primarily heterotrophic bacteria to nitrifying bacteria. Compared to heterotrophic bacteria, nitrifying bacteria have a slower growth rate and lower biomass yield112, explaining both the 6- day delay in higher rates of nitrification in the reactors as well as the accompanying decrease in N removal from the system. P uptake decreased over the same time period by 25-30%, consistent with the idea of a transition from higher-yield heterotrophic bacteria to lower-yield nitrifying bacteria (Fig. 4.5).

Figure 4.5 Total soluble phosphorus removal efficiencies as percentage of influent total P in MAB floc reactors after settling and filtration (0.45 µm) measured on Days 3 and 6.

On Day 3, combined N uptake and nitrification accounted for 55%-83% of total available influent N, whereas by Day 6, these removal and transformation processes accounted for 88%-

103% of influent N. Between Days 3 and 6, the biomass was likely assimilating to the new synthetic wastewater and growth conditions with respect to overall N utilization. Over the first 6 days of reactor operation, there is an establishment of a nitrifying population of bacteria,

77 and an adjustment to fully utilizing available N in reactor influent. However, the influence of defined vs. complex N and P sources on competition between nitrifying bacteria and algae could not be investigated further because of filamentous bulking interfering with reactor operation in the partially and fully defined reactors.

Filamentous Bulking of Defined N and P Source Reactors

By Day 6, there had not been sufficient time to allow for establishment of a detectable algal population in any of the reactors, and filamentous bulking in the partially and fully defined reactors prevented the continued operation of the CSTRs due to cell washout. SVI is a quantitative measure of sludge density and bulking properties. Lower SVI is better for sludge performance. Primarily, complex N and P source reactors (low, medium, and high BOD) all had SVIs consistent with normal activated sludge systems (<200), but the partially and fully defined N and P source reactors had very high SVIs of more than 400 and more than 800, respectively (Fig. 4.6). Microscopic images confirmed this was due to filamentous bulking in these reactors, showing small, sparse floc with a large amount of long bacterial filaments exuding from the surface and preventing proper sludge compaction (Fig. 4.7).

Figure 4.6 SVI of all reactors on Day 6 of reactor operation (at which point partially and fully defined reactors were terminated due to excessive bulking).

A1 A2

B1 B2

Figure 4.7 Filamentous bulking in partially/fully defined N and P source reactors on Day 6. (A1-2) partially defined (B1-2) fully defined.

4.3.2 Algal Growth in Reactors

After filamentous bulking occurred in the partially and fully defined N and P source reactors,

operation of the high, medium, and low BOD reactors was continued to examine the

relationship between algal growth and nitrification and resultant effects on reactor N and P

removal efficiencies. ChlA of the reactor biomass was undetectable (indicating very small algal population densities) on Days 2 and 6 of reactor operation, but it was detectable in all reactors by Day 10 (Fig. 4.8). Over 30 days of reactor operation, chlA content and AI increased steadily once detectable, indicating more photosynthetic activity within the reactors as time progressed

(Figs. 4.8 and 4.9).

For the first 21 days of reactor operation, chlA content and AI were both higher with higher influent BOD at every sampling, with the exception of Day 16, when the low BOD reactor had a higher AI than the medium BOD reactor. However, this trend was not seen on the final sampling point on Day 30. On Day 30, AI showed an opposite trend, lower with higher influent BOD. Because AI did not correlate with influent BOD, it is unclear to what extent

BOD influenced algal growth in the MAB reactors. It is possible that higher BOD allowed for earlier succession of algae in the system by creating a niche for faster-growing mixotrophic algae early in reactor operation. As purely autotrophic algae communities in the reactors were given more time to compete and grow, they could have minimized the effect of BOD on overall chlorophyll A content. This would explain the inversion of the relationship between

AI and BOD by Day 30. Van den Hende et al.27 found a negative correlation between sucrose

concentration in the influent and dominance of algae in their MAB reactors. This result is

consistent with the final Day 30 sampling point and agrees with the theory that less available

BOD should favor photosynthetic autotrophs over heterotrophic organisms.

Figure 4.9 ChlA concentration in reactor mixed liquor for each reactor at sampling points over 30 days of operation. Days 2 and 6 were sampled but had undetectable concentrations of ChlA.

Figure 4.8 AI of biomass in reactor mixed liquor for each reactor at sampling points over 30 days of operation. Days 2 and 6 were sampled but had undetectable concentrations of ChlA, and thus AIs of 0.

4.3.3 Nutrient Conversion and Utilization Across Reactors and Time

Nitrogen metabolism

Beginning on Day 3 of reactor operation, and continuing for 30 days, effluent N, P and nitrate

+ nitrite were measured for each reactor. The total percent N removal and percent of N converted to nitrate + nitrite are nearly mirror images of one other (Figs. 4.10 and 4.11), because close to 100% of influent N was either converted to nitrate or nitrite or incorporated into biomass (Fig. 4.12).

Figure 4.10 Effluent nitrate + nitrite as percentage of influent total N in low, medium, and high BOD reactors after settling and filtration (0.45 µm).

Figure 4.11 Total soluble N removal efficiencies in effluent as percentage of influent total N in low, medium, and high BOD reactors after settling and filtration (0.45 µm).

Figure 4.12 Stacked bar plot of percentage of total influent N removed by uptake into biomass or converted to nitrate + nitrite via nitrification.

Nitrification in the all reactors reached a peak value on Day 6 of operation, with 88%, 87%, and 81% influent N conversion to nitrate + nitrite for the low, medium, and high BOD reactors, respectively. By Day 15 of operation, nitrate + nitrite in the effluent had decreased, and the rates of nitrification and N-removal stabilized, as shown by the decreased standard deviations in day-to-day readings (11.7-15% nitrification and 8.3-13.7% N removal for Days

0-15 versus 3.7-6.4% nitrification and 3.5-5.3% N removal for Days 15-30) (Table 4.2). After stabilization of reactor performance, mean nitrification percentages were 49-64% of influent

N in the reactors, and mean N removal percentages ranged from 32-50% of influent N.

Table 4.2 Standard deviations of daily nitrification or N uptake as a percentage of total influent N between startup and Day 15 and from Day 15 to 30.

Nitrification N Uptake low BOD med BOD high BOD low BOD med BOD high BOD Days 0-15 11.7% 13.9% 15.0% 8.3% 9.8% 13.7% Days 15-30 3.7% 4.4% 6.3% 3.5% 4.5% 5.3%

After the initial spike in nitrification and drop in N removal between Days 3 and 6, nitrification rates decreased across all reactors, accompanied by an increase in N removal rates, before these values stabilized. The increase in N removal rates coincided with the appearance and growth of algae in the reactors, as indicated by detectable chlA concentrations at Day 10. Such a functional change in reactor performance was expected in conjunction with competition between newly colonized algae and the resident nitrifying bacterial population, based on experiments establishing a competitive relationship between algae and nitrifying bacteria40,41,113.

ChlA concentration and AI of the biomass continued to increase throughout the 30 days of reactor operation, but rather than a corresponding decrease in nitrification associated with

competition, nitrate + nitrite levels in the effluent remained stable over this time period. This

indicated stable coexistence of these populations instead of competitive exclusion.

Algae can utilize nitrate for growth, however nitrate uptake by reactor biomass (measured by

re-suspending reactor biomass in synthetic wastewater with sodium nitrate as the sole N

source) was undetectable on Days 14, 16, and 23 of reactor operation, consistent with other

studies27,115. Additionally, if coexistence between algae and nitrifying bacteria was mediated by

algal utilization of byproduct nitrate or nitrite from nitrification as an N source, a decrease in

effluent nitrate + nitrite concentrations would still be expected as chlA content and AI of the

biomass increased. Rather than demonstrate competition between algae and nitrifying bacteria

for resources, the reactors showed a stable coexistence of algae and AOBs/NOBs (indicated

by nitrate + nitrite levels in effluent), supporting an increasingly large contingent of resident

algae without eroding the nitrifying population. Given the low uptake rate of nitrate displayed

by biomass in the 1.5-L CSTR reactors, the nitrifying bacteria persistence in MAB systems

may have led to decrease overall N treatment efficacy, since nitrate produced by nitrification

was not readily taken up by algal biomass.

Across all reactors, nitrification (as quantified by nitrate + nitrite in effluent) did not decrease

in relation to the increase in algal chlorophyll A content throughout steady-state operation. To

understand the factors leading to coexistence of these two populations, N uptake and

transformation dynamics were investigated further.

Investigation of Nitrification and Nitrogen Uptake Dynamics

Over the course of reactor operation, temporal changes in N metabolism were correlated

between reactors for both nitrification (Spearman pairwise comparisons, P = 0.002, 0.002,

<0.001) and N removal (Spearman pairwise comparisons,𝜌𝜌 P = 0.001, 0.002, <0.001). This

𝜌𝜌 85

indicates that the biological or abiotic factors, which influenced changes and fluctuations in N

uptake and transformation over time, were shared between the reactors. The decrease in day- to-day variation between N removal and nitrification rates from Days 15-30 of operation

compared to Days 0-15 (Table 4.2) suggests initial changes in performance were primarily

associated with shifts in microbial community composition. The shifts were shared among the

reactors while adjusting to new growth conditions until quasi-steady-state operation was

achieved after Day 15. However, these shifts are not attributable to a simple direct competition

between algae and nitrification, since algae continued to grow over the course of the

experiment, but nitrification did not decrease accordingly (as discussed previously in this

Section).

To examine the factors associated with nitrification and N uptake in the reactors, the

relationship of nitrate + nitrite conversion and total N removal (percent of influent N) to both

pH and influent BOD were analyzed using multiple regression analysis with N uptake and

nitrate + nitrite in effluent as response variables, pH as a continuous predictor and

low/medium/high BOD feed as a discrete predictor for each. The regressions were applied

over six sampling days, where both pH and N uptake/transformation were measured together.

Over these sampling dates, pH was not a statistically significant predictor of nitrification (P =

0.07) or N uptake (P = 0.122) in the reactors. This indicates that pH effects of nitrification

were mitigated by bicarbonate addition and didn’t influence N uptake. Influent BOD,

however, was a significant predictor of nitrification and N uptake when examined across the

entire period of reactor operation (P = 0.004, P < 0.001), as well as the quasi-steady-state

period between Days 15 and 30 of operation (P = 0.027, P = 0.003). In fact, N uptake was

lower and nitrification higher with higher influent BOD at every sampling taken over the 30

86 days of reactor performance, with the exception of one reading for the high BOD reactor on

Day 9.

Averaging readings over the steady state period of operation on N removal and nitrification for the three reactors, pairwise two sample t-tests assuming equal variances were applied (equal variance assumption validated with pairwise F tests). T-test comparisons (two tail) showed a significant difference in nitrification between the high BOD reactor and the low (P = 0.002) and medium (P = 0.009) BOD reactors, but not between the low and medium BOD reactors

(P = 0.186). Similarly, N removals were significantly different between the high BOD reactor and the low (P = 0.0002) and medium (P = 0.002) BOD reactors, but not between the low and medium BOD reactors (P = 0.122). In general, then, higher influent BOD resulted in lower average nitrification and higher average N removal in the three reactors, but this effect is largest and only statistically significant comparing the low or medium BOD reactor to the high BOD reactor. Dissolved oxygen remained well above levels which would inhibit nitrification (0.5 mg/L)112 with the lowest reading at any point in any reactor at 2.3 mg/L dissolved oxygen.

It is not apparent that these differences in N metabolism with different influent BOD were mediated by algal population size. The nitrification and N removal trends with higher BOD were the same between Day 21 and Day 30, despite an inversion of the relationship between influent BOD and AI of biomass over this same time-period (Fig. 4.9). To better understand how BOD levels were influencing nitrifying bacterial populations, a mass balance on N was calculated.

Mass Balance on N

At steady state reactor operation (after Day 15), nitrification in the reactors (high BOD, med

BOD, low BOD respectively) was on average (%): 64.4 ± 3.7, 60.7 ± 4.4, 49.2 ± 6.3, and N uptake was on average (%): 32.0 ± 3.5, 36.4 ± 4.5, 50.4 ± 5.3. Additionally, before the reactors were terminated, TSS was recorded by mixing the reactor MLSS well, homogenizing filaments and flocs with a 18G syringe, and then measuring TSS on an aliquot of the homogenized biomass. The TSS values for the high, medium, and low BOD reactors were 1.25 g/L, 1.34 g/L, and 1.55 g/L, respectively. The synthetic wastewater was re-digested at every sampling point, with an average value of 45 mg/L-N. Solids removal was 60-65 mL daily (average 62.5 mL, SRT = 24 d). These parameters, then, were used to create a mass balance on N in the reactors and are summarized in Table 4.3. Denitrification is not expected to contribute to N removal since the reactors were aerated and reached an absolute minimum of 2.3 mg/L dissolved oxygen on one occasion (immediately after the dark cycle in the high BOD feed reactor).

Table 4.3 Parameters used for N mass balance for the low, medium, and high BOD bioreactors.

Finally, to obtain an approximation of the magnitude of N removal by volatilization of ammonia, a simple mass balance was calculated using the conservative assumption that the aeration rate through the reactor dictates overall air exchange in the reactor, and leaves the

system at equilibrium with respect to ammonia. The reactors were aerated at an average rate

of 0.15 L min-1 (measured by air flow meter). If a Henry’s law constant of 27 mol/kg bar is

assumed (NIST 117), the partial pressure of ammonia in the bubbled air stream can be calculated

assuming all air leaving the reactor is saturated with ammonia, and this partial pressure

converted to a concentration of ammonia in the air using the ideal gas law. An upper bound

for the concentration of ammonia in the reactor mixed liquor was calculated using the pKa =

9.24 of ammonium assuming all residual, non-nitrate/nitrite, N in the effluent (average value from Days 15-30) is present as ammonia/ammonium, and the pH is the highest measured at any point in the operation of the reactor (pH 9.07). Using this set of conservative assumptions, the highest possible air stripping rate of ammonia is 0.2 mg/d for Reactor 1. This value represents an upper bound for the amount of ammonia that could be removed via volatilization, and is about 0.4% of the total influent N. Therefore, N removal by volatilization in the reactors was assumed to be negligible. The calculation assumes that all residual non- nitrate + nitrite N measured in effluent exists as either aqueous ammonia or ammonium

[ ] + [ ] = [ ] , (1 % ) (4.2)

𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝑁𝑁𝑁𝑁4 𝑎𝑎𝑎𝑎 𝑁𝑁 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑁𝑁 𝑖𝑖𝑖𝑖 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 The relative distribution of ammonia versus ammonium was calculated using the pKa of

ammonium, with

[ ] [ ] = [ ] (4.3) [ ] + 𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝐻𝐻 𝐾𝐾𝑎𝑎 → 𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝑁𝑁𝑁𝑁4 𝑎𝑎𝑎𝑎 Next, Henry’s law and the ideal gas law were used to solve for the concentration of ammonia

that would exist in equilibrium with air bubbled through the reactors:

[ ] [ ] [ ] = = [ ] = (4.4) 𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝑁𝑁𝑁𝑁3 𝑎𝑎𝑎𝑎 𝐾𝐾𝐻𝐻 → 𝐾𝐾𝐻𝐻 → 𝑁𝑁𝑁𝑁3 𝑔𝑔𝑔𝑔𝑔𝑔 𝑛𝑛𝑛𝑛𝑛𝑛 𝐻𝐻 𝑝𝑝 � � 𝐾𝐾 𝑅𝑅𝑅𝑅 𝑉𝑉 89

Using the concentration of ammonia calculated in the gas phase and the flow rate of air

bubbled through the reactors, the total N removal due to volatilization were then calculated

using

N removal by volatilization = [ ] 0.15 L (4.5) −1 𝑁𝑁𝑁𝑁3 𝑔𝑔𝑔𝑔𝑔𝑔 𝑥𝑥 𝑚𝑚𝑚𝑚𝑚𝑚 Nambiar et al.81 found luxury N uptake of up to 16% in MAB floc systems. Assuming this

value, N recoveries in the mass balances are 96%, 93%, and 84% for the low, medium and

high BOD reactors, respectively. The total N was calculated using the mass balance

mg mg 1L = d L d 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑁𝑁 � � �𝑁𝑁 mg� � 𝑥𝑥 � L % L d

− �𝑇𝑇𝑇𝑇𝑇𝑇 � � 𝑥𝑥mg𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀1L𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 � � 𝑥𝑥 𝑁𝑁 𝑖𝑖𝑖𝑖 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵� (4.6) L d − �𝑁𝑁𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 � � 𝑥𝑥 � The percent recovery was then calculated by

mg d Recovery (%) = (4.7) 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 mg 1L 𝑁𝑁 �L � d 𝑁𝑁𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 � � 𝑥𝑥 The low BOD and medium BOD reactors show a near complete N balance (96% and 93%

recovery), but luxury N uptake would have to be higher still in Reactor 3 to account for all N

removal from the system (only 84% recovery). Luxury N uptake at 16%+ not only creates a

mass balance on N in the reactors, but also could explain how nitrate + nitrite and total N in the effluent could be stable despite increasing algal populations in the reactor. Luxury nutrient uptake is the uptake of nutrients such as N and P at a rate greater than the rate expected to satisfy stoichiometric growth requirements. Luxury P uptake in algae is well documented80,116,

90 but luxury N uptake was identified in a study of MAB floc reactors as well at 16%81.

Additionally, effects of influent BOD on algal luxury N uptake rate could explain decreased nitrification and increased N uptake in the high BOD reactor described in previously in this

Section, a hypothesis further investigated using separate batch N uptake rate experiments.

Nitrification and Uptake Rate Experiments

To further investigate differences in N uptake and nitrification between the three reactors, a series of separate ammonia uptake rate experiments were carried out. First, to determine nitrification rates, biomass from all three reactors was washed with PBS and re-suspended in synthetic wastewater containing 50 mg/L-N of ammonium chloride along potassium phosphate, but without meat extract, peptone, or glucose. To determine N uptake rates, biomass from the reactors was washed and re-suspended with the same ammonia/phosphate synthetic wastewater.

Assuming a luxury N uptake rate of 16% balanced N to 93-96% in the low BOD and medium

BOD reactors, but accounted for only 84% of N in the high BOD reactor for the mass balance. Corroborating this result, the measured ammonia uptake rate was higher in the high

BOD reactor biomass over two hours compared to the low or medium BOD reactors (Table

4.4). These two results suggest additional luxury N uptake (and thus a higher percent N in biomass) in the high BOD reactor compared to the low or medium BOD reactor.

Table 4.4 Nitrate + nitrite production (mg-N/L-d) and N uptake (mg-N/L-d) normalized to biomass for N uptake and nitrification rate experiments conducted over a 2 h period.

light dark Nitrate+Nitrite Production Nitrogen Uptake Nitrate+Nitrite Production Nitrogen Uptake (mg-N/L gTSS d) (mg-N/L gTSS d) (mg-N/L gTSS d) (mg-N/gTSS L d) low bod 2071 4150 2750 n.d. med bod 2430 4130 3080 n.d. high bod 1830 6850 2750 n.d.

To help validate this finding, the %N in growing biomass that would account for N uptake

over the 2 h period was calculated. Algal biomass in the reactors was comprised largely of

Scenedesmus sp. A maximum specific growth rate μm = 0.54/d was determined for S. dimorphus

based on optical density growth curves in batch reactors (Appendix A9). Using an algal growth

rate is appropriate since there was no BOD source in these uptake rate batch reactors, and N

uptake by nitrification alone was not detectable (in the dark reactors nitrification was detected

- via NO3 production, but N removal was less than in the volatilization control). Accounting for volatilization and assuming μm = 0.54/d, a 16% N luxury uptake biomass composition in

newly formed biomass over 2 h would account for 92%, 93%, and 53% of N uptake by the

low, medium, and high BOD biomass in the uptake experiments, respectively. This was

determined by first using the specific growth rate μ to calculate the change in biomass concentration over two hours, with

= ( ) = ( ) (4.8) 𝑋𝑋𝑓𝑓 𝑋𝑋𝑖𝑖 𝜇𝜇𝜇𝜇 → ∆𝑋𝑋 𝑋𝑋𝑓𝑓 − 𝑋𝑋𝑖𝑖 𝑒𝑒

The measured change in total N concentration, and the volume of the reactor, were then used

to calculate the total mass of N taken up by the biomass

mg = [ ] (4.9) L ∆𝑁𝑁 ∆ 𝑁𝑁 � � 𝑥𝑥𝑉𝑉𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 The percent of N recovered through this mass balance was then calculated with

(%) = % (4.10) ∆𝑁𝑁 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 ∆𝑋𝑋 𝑥𝑥 𝑁𝑁 𝑖𝑖𝑖𝑖 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵

The mass balance from the ammonia uptake rate experiment is consistent with luxury N

uptake during biomass growth in the reactors, at the 16% rate described by Nambiar et al.81 or higher (producing 92%, 93%, and 54% N recoveries for low, medium, and high BOD reactor biomass). Ammonia uptake rates by algae in the reactors were higher than ammonia uptake for nitrification under illumination for biomass from all three reactors as shown in Table 4.4,

consistent with published results, indicating that rapid algal nutrient uptake can lead to competitive inhibition of nitrifying bacteria under certain conditions40,41,113. This N uptake is

assumed to be mediated by algae in the reactors, because these batch rate determination

experiments showed that ammonia uptake occurred without glucose, and was not detectible

in the dark.

The high-BOD reactor biomass showed an increased ammonia uptake rate (Table 4.4) in

conjunction with lower average nitrate + nitrite concentrations and higher N removal

efficiencies in 1.5-L CSTR effluent relative to the low and medium BOD reactors, as discussed

previously in this Section. This result suggests that the N uptake rate of algae in MAB reactors

has an effect on both the extent of competitive exclusion of nitrifying bacterial populations

by algae, and the overall nutrient removal efficiencies of the system. The reason for the effect

of BOD on algal N uptake per unit of biomass is unclear, but could be related to physiological

differences in algal activity as a result of a higher growth rate consistent with a higher TSS

concentration in the high BOD reactor (1.25 g/L, 1.34 g/L, and 1.55 g/L TSS, in the low,

medium, and high BOD reactor, respectively).

Nitrogen uptake became undetectable when cells were incubated without light in a separate

incubator shaker, and in fact N removal was higher in the volatilization control sample than

experimental reactor samples in the dark (possibly due to physical interferences of algae with

gas exchange, or the effect of nitrification converting volatile ammonia into non-volatile

nitrate or nitrite). This means that N removal in the illuminated uptake rate experiment was a

conservative estimate of N uptake, since the presence of biomass and nitrifying bacteria

decreased the effective volatilization rate relative to controls, but volatilization was assumed

to account for the same ammonia loss in the experimental samples as in the control sample.

Nitrification rates showed a 25-50% increase in the dark in all three reactors, consistent with published results showing light inhibition of nitrification118. Given the apparent light-

dependency of nitrogen uptake in algae versus AOBs/NOBs, on Day 34 nutrient samples

were taken at the beginning, middle, and end of the 14 h light cycle in the reactors to see if

diel cycles influenced coexistence of algae with nitrifying bacteria over time.

Light Cycle Effects on Full Scale Reactor N Metabolism

From the illuminated/non-illuminated N uptake rate experiments previously in this Section, nitrification was expected to be higher at the termination of the dark cycle since luxury N uptake would be lower, and there would be no competition with nitrifying bacteria for ammonia. As the daylight cycle progressed, a decrease in nitrification was anticipated because luxury N uptake rates eclipsed nitrification rates during the day. However, the rates of nitrification and N uptake did not increase during the day, but instead remained stable over the course of the daylight cycle from readings in the beginning (AM), middle (PM), and end

(night) of the light period (Fig. 4.13). This unexpected stability in N dynamics may be for a number of reasons. It is possible that the very low concentrations of reduced N in the CSTRs change luxury uptake/nitrification dynamics, or that more complex N sources found in peptone and meat extract combine with microbial interactions to change these dynamics.

Overall, though, this result indicates that diel cycles did not play a role in facilitating stable

coexistence between increasing algal populations and nitrifying bacterial populations.

Figure 4.13 Nitrification and N removal as a percentage of total influent N in settled, filtered (0.45 µm) samples taken 30 min after the start of the light period (AM), midway through the light period (PM), and 1 h prior to termination of light period (night).

Light Cycle Effects on Full Scale Reactor P Metabolism

Phosphorus removal was subject to higher daily fluctuations in the reactors than nitrification or N removal (Fig. 4.14 and Table 4.5). Nitrification and N removal had standard deviations of 3.5-6.5% in the reactors across the steady-state (Days 15-30) period of reactor operation,

whereas P removal had day-to-day standard deviations of 11.2%, 8.8%, and 13.6% in the low,

medium, and high BOD reactors, respectively. Unlike N uptake and nitrification, the P uptake

rate did appear to vary over the course of the daylight cycle (Fig. 4.15). In the low, medium,

and high reactor, the P removal rate was progressively higher moving from the start to end of

the 14 h illuminated cycle. Between the beginning and end of the cycle, P removal had

increased by 18% (low BOD), 13.3% (medium BOD), and 59.8% (high BOD) of total

available influent P. Sampling for nutrients was typically done 4-5 h into the light cycle, but on

Day 23 sampling was taken earlier (2 h into the light cycle), and on Day 25 was taken later (8

h into the light cycle). The deviation in sampling time helps explain the variability seen in the

P removal data on Days 23 and 25 (Fig. 4.14). Diel fluctuations in nutrient removal were less

dramatic in other published results using algal wastewater treatment reactors, with reduced carbon storage mitigating the effects of light on nutrient removal by algae82.

Figure 4.14 Total soluble phosphorus removal efficiencies as percentage of influent total P in low, medium, and high BOD reactors after settling and filtration (0.45 µm).

Table 4.5 Standard deviations of daily nitrification, N uptake, and P uptake as a percentage of total influent N/P from Day 15 to 30.

Nitrification N Uptake P Uptake low BOD med BOD high BOD low BOD med BOD high BOD low BOD med BOD high BOD Days 15-30 3.7% 4.4% 6.3% 3.5% 4.5% 5.3% 11.2% 8.8% 13.6%

The changes in P uptake over the daily cycle also agree with results of overall N uptake, in that

there appears to be luxury P uptake which is highest in the reactor biomass fed with highest

influent BOD concentrations. To validate this result, another batch uptake rate experiment was performed to assess P uptake.

Figure 4.15 P removal as a percentage of total influent P in settled, filtered (0.45 µm) samples taken 30 min after the start of the light period (AM), midway through the light period (PM), and 1 h prior to termination of light period (night).

Luxury P Uptake in Bioreactors

To determine if P removal was influenced by BOD directly, or rather was a characteristic of

biomass grown with higher BOD influent, and if the behavior was related to luxury P uptake, high BOD reactor biomass was re-suspended in a synthetic wastewater containing no N source, and only a defined P source (monobasic potassium phosphate) with 0 g/L, 0.05 g/L, or 0.1 g/L dextrose, and incubated over a 24 h period (14 h light, 10 h dark). The initial and

final total P concentrations were then measured. Phosphate removal was greater than 90% in

each of the treatments, substantially higher than average daily P removal for the high BOD

reactor (Fig. 4.16). This suggests a higher affinity for orthophosphate uptake over complex P

in peptone and meat extract.

Figure 4.16 Luxury P uptake over a 24 h period by the high BOD reactor biomass re- suspended in synthetic wastewater with no N source and 0, 0.05, and 0.1g/L glucose.

This luxury P uptake occurred with or without additional BOD, so while the rate may have

differed between glucose treatments, the process was not dependent on the presence of

glucose. The demonstrated luxury P uptake confirm another implication of AOB/NOB versus

algae competition for MAB wastewater treatment: that higher algal populations could result in higher P removal due to algal luxury P uptake in excess of growth requirements.

4.3.4 Nitrate Removal and Biomass Production Rate in MAB Floc Systems

Results described in Section 4.3.3 show that for 1.5-L aerated MAB CSTRs, nitrification was

unaffected by continued algal growth in the reactors over time, and once influent N was

converted to nitrate/nitrite in the reactors, it became recalcitrant to uptake. Nitrate uptake experiments in batch reactors never revealed detectable nitrate uptake rates by biomass in the reactors, and nitrate + nitrite levels in the reactors remained consistent despite continued algal growth. A mass balance on N in the reactors in Section 4.3.3 suggests luxury uptake of reduced

N into biomass, a result substantiated by separate ammonia uptake experiments described also in Section 4.3.3 using biomass from each of the reactors. One hypothesis is that biomass in the reactors were generally scavenging luxury reduced N, but not exhausting these stores for growth, so there may have been little advantage or incentive to also scavenging oxidized N.

To offer support for this hypothesis, and to see if reactor biomass would eventually take up residual nitrate/nitrite, biomass from the medium BOD reactor was re-suspended in duplicate batch reactors to examine cell growth in conjunction with N uptake and nitrification.

Consistent with the hypothesis of rapid reduced N uptake and utilization, >90% of reduced

N was taken up or converted to nitrate or nitrite in both reactors within 24 h (Fig. 4.17). After

24 h, nitrate/nitrite was then utilized and was exhausted within 72 h after inoculation. The biomass specific growth rate over the 4-day period was 0.17-0.17 d-1 (calculated using an

average TSS value over the time period), around four times higher than the biomass

production rates of 0.042 d-1 set by the SRT in the low, medium, and high BOD CSTR reactors. Since there was suspended growth in these batch reactors (in the absence a selection

pressure for settling) biomass production of suspended and settling algae were both measured.

Even if all suspended growth is disregarded, growth rates of the settled biomass alone were

0.04-0.07 d-1, indicating that growth rate was likely not optimized in the 1.5-L CSTRs, with

specific growth rates set by solids removal below the maximum specific growth rate of the

biomass.

The ratio (w/w) of N uptake to new biomass produced was 0.04 for both batch reactors.

Assuming the initial biomass was 16% N81, the net percentage of N in the batch reactor

100 biomass at the end of operation would be 9.7% for both batch reactors. This is lower than original 16% N composition in MAB flocs reported by Nambiar et al. undergoing luxury N uptake81, and the values required to create a mass balance on N in the 1.5-L CSTRs in Section

4.3.3.

The batch reactor data support the hypothesis that residual nitrate + nitrite in the reactor effluent is likely a result of stores of luxury N in biomass (which remains unutilized possibly due to a high SRT and lower-than-optimal growth rate in the biomass). An excess of intracellular reduced N reserves would create little competitive advantage to scavenging additional nitrate or nitrite. Once these stores are exhausted (as seen by the overall decrease in

N composition of biomass in these batch reactors over time), the same biomass will take up nitrate/nitrite for growth.

It is noteworthy that 75% and 58% of new biomass growth in batch reactors 1 and 2, respectively, was found as suspended algae, not associated with flocculated biomass. This suspended algal population grew throughout the course of the 4-day batch reactor operation.

An alternative explanation for enhanced nitrate/nitrite uptake by the MAB biomass in these batch reactors is that nitrate/nitrite uptake is contingent on suspended growth in the reactor.

Because nitrate and nitrite are byproducts of nitrification, nitrifying bacteria have transport systems for clearing produced nitrate and nitrite. Nitrate and nitrite transport may interfere with algal nitrate and nitrite uptake within the floc, thus requiring algae to grow outside of the floc in order to utilize nitrate or nitrite as an N source.

The lower growth rates in the 1.5-L CSTR reactors compared to exponential growth in batch reactors, and corresponding luxury N uptake, may be the mitigating factor allowing for stable coexistence of algal and AOB/NOB populations. SRTs used in this study are in the range of

101

other publications with MAB reactors. One of these studies used minimal aeration resulting

in low nitrate levels in the effluent along with >70% N removal (SRT=20-25 d)26, and the

other maintained DO above 7 mg/L, resulting in high nitrate levels in the effluent along with

<40% N removal (SRT=20 d)42. With aerated MAB floc systems, lowering SRTs may facilitate increased biomass growth rates and more competition between AOB/NOB and algae as nutrient levels come closer to limiting growth.

It is possible that the aerated versus non-aerated approach might be best utilized for different purposes. Non-aerated systems might be best suited for wastes with lower BOD levels, where oxygen from algal photosynthesis alone can satisfy BOD, but diel variation in oxygenation in the reactor could facilitate natural cycles of nitrification and denitrification, enhancing N removal39. Alternatively, introducing some additional mechanical aeration to MAB wastewater

treatment systems might be better suited for treatment of wastes with higher BOD:N/P ratios,

allowing for higher BOD wastewaters to be treated without overwhelming the oxygen supply

by algal photosynthesis. Aerated systems may also support higher growth rates, a notion

supported by the apparently sub-optimal growth rates in the 1.5-L MAB CSTRs operated with

SRT of around 24 d, within the range considered optimum for a set of non-aerated CSTRs

operated by Gutzeit et al.26. To better understand the possible applications and benefits of aerated MAB wastewater treatment, the optimal SRT and HRT operating parameters for these systems must first be determined. Additionally, the relative limits of aerated versus non-aerated

MAB systems for BOD removal should be determined in the future to find the range of

optimal applicability for both MAB cultivation strategies.

102

4.4 Conclusions

A number of publications have described the ability of algae to outcompete nitrifying bacteria through nutrient competition, but several studies using MAB consortia to treat wastewater show release of nitrite or nitrate into effluent, in an apparently DO-dependent manner. To better understand competition between nitrifying bacteria and algae during MAB wastewater treatment, 1.5-L CSTRs were inoculated with activated sludge and fed with synthetic wastewater containing different levels of BOD. Algae began to grow in the reactors starting

10 days after inoculation, and then continued to increase in population over the remainder of reactor operation. However, after an initial 15-day adjustment period, nitrification in the reactors remained stable despite increasing algal populations. Under the reactor operating conditions used, algal and AOB/NOB populations coexisted without competitive exclusion over a period of several weeks.

Despite coexistence within 1.5-L CSTR reactors, ammonia uptake rates in algae were higher than ammonia uptake rates for nitrification under illumination in separate batch experiments.

Moreover, while nitrification remained stable in the reactors over time and with increasing algal populations, higher influent BOD decreased the stable nitrification levels and increased

N uptake. Algae in all MAB reactors demonstrated apparent luxury N uptake, but this uptake was highest in the high BOD reactor biomass, which seems to explain the decreased nitrification levels in this reactor. Luxury N uptake would facilitate higher N removals from wastewater effluent per unit biomass grown in reactors, increasing overall N removal rates relative to biomass without luxury N uptake. The higher ammonia uptake rates compared to nitrification rates in batch experiments, along with the link between lower nitrification and higher N uptake in the high BOD reactor suggest that algae do indeed have the potential to

103 competitively inhibit AOB/NOB under certain conditions, a finding consistent with published results. This competitive effect appears to have been mitigated in the 1.5-L CSTRs, though, allowing for coexistence of both populations over time.

Biomass from all reactors did not readily take up nitrate as an N source, a finding consistent with algal biomass in other published work using MAB consortia for wastewater treatment27,115, though they did appear to take up luxury reduced N. This suggests that the persistence of nitrification in MAB reactors could be detrimental to overall performance by converting reduced N which is readily taken up by algal biomass, to oxidized N which is not. Furthermore, higher measured ammonia uptake rates in the high BOD reactor biomass corresponded to less nitrification and more overall N uptake in the high BOD 1.5-L CSTR compared to low and medium BOD reactors. The high BOD reactor biomass also displayed the ability to uptake luxury P in the absence of BOD or N sources. These results suggest that the competition for nutrients between AOB/NOB and algae has implications for overall reactor function, affecting overall N removal, and the balance between algae which take up luxury P and

AOB/NOB which do not.

The SRT used for the 1.5-L CSTRs produced growth rates in biomass below exponential growth rates seen in separate batch reactors inoculated with biomass from one of the reactors.

Additionally, luxury N was taken up into biomass beyond that which was required for growth.

This lower growth rate and consistent luxury N availability may have been the mitigating factor allowing for the coexistence of nitrifying and algal populations over time. Decreasing SRT would likely lead to more competition by increasing growth rates and by creating reactor conditions closer to nutrient limitation. This could facilitate more competition between nitrifying bacteria and algae in MAB treatment system. Higher N uptake and increased

104

competition in the high BOD reactor could have been a result of higher growth rates in this

reactor relative to the low and medium BOD reactors, as indicated by higher TSS

concentrations.

While non-aerated systems can harness the natural diel cycling of available oxygen from algal

photosynthesis by allowing for cyclical nitrification/denitrification N removal39, or perhaps inhibit nitrification to reduce competition with algae for N uptake, some added mechanical aeration could broaden the applications of MAB consortia for wastewater treatment. MAB reactors with added mechanical aeration could potentially allow for higher biomass productivities (and thus lower HRTs), consistent with the sub-optimal growth rates seen in the 1.5-L CSTRs. Additionally, treatment of wastes with higher BOD:N/P ratios could be possible since algal photosynthetic oxygenation would not limit BOD exertion in the reactors.

Additional research into added aeration for MAB wastewater treatment could help optimize

operating parameters to potentially broaden the applications of MAB treatment systems while

still reaping the benefits of carbon dioxide/oxygen exchange symbiosis and enhanced N and

P removals through algal growth and luxury nutrient uptake.

105

Chapter 5 INVASION IN MICROALGAL COMMUNITIES

5.1 Introduction

Invasion of microbial systems has implications in the fields of ecology, energy production, and

healthcare. In healthcare, microbial invasion has implications for human disease and dysbiosis

as a result of disturbances to the human microbiome119. In MAB systems, invasion is a

potentially useful tool to add capabilities (e.g., increased sludge biomass content) to the existing

microbiome. In natural environmental systems, invasion has relevance to the recent issue of

HABs in the Great Lakes Region, where toxic cyanobacteria dominate the lake during bloom

season, replacing resident microalgal communities present earlier in the season47. There is

some direct evidence that immigration from streams is not a factor in harmful algal bloom

development120. Additionally, the contribution of immigration relative to deterministic factors in dictating community composition is not clear, though there is evidence in some systems,

such as activated sludge waste treatment, that deterministic factors are the most important

drivers of community composition121. Invasion studies can help to uncover the roles of

microbial immigration and deterministic factors in community composition. This is

accomplished when invasion studies address key components of microbial invasion defined

by Kinnunen et al.: dispersal, selection, drift, and diversification119. An example invasion study

in microalgal communities focusing on dinoflagellate bloom development was conducted by

Acosta et al.122. However, in this study, there was no influent or effluent from experimental

reactors, so it is difficult to address dispersal in their reactors. The authors did investigate some

selection effects using a low and high nutrient treatment, but there are additional

configurations for invasion studies which can further test selection effects by creating a range

of conditions which could be favorable to microbial invasion.

106

Information from well-conducted invasion studies could inform control strategies for the

prevention of invasion of communities when invasion causes deleterious outcomes, or for

successful implementation of invasion when the invader brings a desired engineering

functionality to a pre-existing microbial community. The current study represents a simulation of an invasion process to understand the fundamental mechanisms of microbial invasion.

Microalgal pond microcosms were operated over the course of 7 days with feeding along with

immigration of an invading strain at two propagule pressures on Day 3 to investigate dispersal

effects during microalgal invasion of a preexisting phytoplankton community. The invading

microalgal strain was A. flos-aquae (UTEX 1444), a common bloom forming cyanobacteria in

the Great Lakes Region123. These tests provided preliminary data which can help in designing

future microalgal invasion experiments.

5.2 Material and Methods

Microcosms were started using 40 mL of surface water collected from Wade Lagoon in

Cleveland, OH, and amended with 136 μg/L of K2HPO4 and 9.6 mg/L of NH4Cl. The

microcosms were allowed to grow in batch until visible green algal growth developed (3 days),

after which point they were decanted and fed at 9 mL/d with filtered (934-AH glass fiber

filters) Wade Lagoon water collected on the same day as inoculation, amended with the same

concentrations of phosphate and ammonium. The reactors were fed for 4 days to allow for

microbial acclimation and then subjected to immigration from A. flos-aqua at a low and high

propagule pressure, facilitated by a single pulse of A. flos-aquae cells added at one time. Feeding

continued and the concentration of A. flos-aqua in the reactors was recorded by total filament

counts (counting each continuous filament greater than 4 individual cells once). Total algal

population density was tracked by extracting 100 µl of culture into 400 µl 1:1 DMSO:acetone

107 for 20 min in the dark and then reading fluorescence at 340 nm/671 nm ex/em98. The microcosms operated were: three non-invasion controls (C1, C2, C3), three high propagule pressure A. flos-aquae invaded (AH1, AH2, AH3), and three low propagule pressure A. flos- aquae invaded (AL1, AL2, AL3).

5.3 Results and Discussion

Invading A. flos-aquae was eliminated from the reactors at a rate equal to or greater than the washout rate due to dilution without any growth (Figure 5.1). By Day 4 after invasion, A. flos- aquae populations were minimal compared to population initially at invasion (falling from

1.2x104 filaments/mL to ≤125 filaments/mL), after nearly one microcosm residence time. By

Day 4 after invasion, two of the control microcosms had filaments which resembled A. flos- aquae morphologically corresponding to a concentration of 21 filaments/mL.

Figure 5.1 Filament counts of A. flos-aquae after invasion of microcosms. Dotted line indicates the expected filament washout rate without cell growth due to dilution, starting at the average filament concentration at invasion for the three reactors within each propagule pressure treatment.

108

Chlorophyll A content of the microcosms did not monotonically increase or decrease over the course of microcosm operation in experimental or control reactors, but there was variability in day-to-day readings, as shown in Figure 5.2. These day-to-day fluctuations in readings are likely the result of the development of flocs over the course of microcosm operation, which created a non-homogenous culture and thus variability in biomass concentration sampled.

This is a challenge which will have to be addressed in similar future studies, possibly through the use of another biomass measure using a larger sample volume to mitigate this sampling bias. Overall, however, chlorophyll A content shows that the consistent decreases in the concentrations of A. flos-aquae in the reactors were not the result of overall decreases in total phytoplankton abundance.

Figure 5.2 Chlorophyll A fluorescence in microcosms on Days 0, 2, and 4 after invasion by A. flos-aquae.

109

5.4 Conclusions

The experimental invasion microcosms showed that the invading A. flos-aquae strain was not able to establish a population within the existing algal community at either low or high propagule pressures. Future invasion experiments could focus on two key components dealing with deterministic effects and immigration effects on invasion success. One possibility is that timing of nutrient amendment could have an interaction effect with propagule pressure in the microcosms. Presumably, nutrient amendments in microcosms which have resulted in steady state operation means niches are already filled with pre-existing community members.

However, if changes in reactor nutrient levels coincide with high propagule pressure of the invader, this might result in higher invasion success. A continuous input of invading A. flos- aquae instead of a single invading pulse of cells may also change invasion success and increase the likelihood of establishment of the invader into the existing community. Finally, the questions of deterministic factors which influence HAB formation are still not completely clear, and the paradigm of P management plans as a best management practice for HAB prevention is not universally agreed upon124. While research efforts involving field work are crucial, there is also a role for laboratory research to play to isolate variables affecting HAB formation. Examining different combinations of conditions in reactors with even very small propagule pressures (just enough to ensure that the invading organism of interest is present) can be crucial in understanding the deterministic effects of HAB community assembly.

110

Chapter 6 CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS

Demonstrated luxury N and P uptake rates in MAB reactors, and the growth of a settleable

MAB consortia in these reactors, suggest that producing settleable algal biomass from wastewater resources is viable and can be optimized with additional research into reactor operating parameters. Luxury N and P removal could prove an important benefit to incorporating algae into wastewater treatment. The decoupling of growth from N and P removal would lead to faster rates of N and P removal and improve the efficiency of nutrient removal using smaller treatment systems. Altering these parameters could influence the competition between algal and AOB/NOB populations to affect overall nutrient removal. For viability in biofuel production, a high lipid content is important in these MAB consortia in addition to wastewater treatment efficiency. Inoculated MAB floc reactors show that several oleaginous species of algae bioflocculate with activated sludge to produce settleable algal biomass. These MAB aggregates display rapid settling behavior, offering a potential advantage over slower-settling activated sludge. Such strains could be used to assemble a community of algae in MAB flocs, but operating conditions will need to be altered to trigger lipid accumulation in these algae. Alternatively, the experiment described in Chapter 1 with feast- famine reactor cycling offers several possibilities for how this natural selection strategy could be modified in an attempt to increase the lipid content of uncharacterized, natural MAB floc cultures.

111

6.1 Chapter 2: Feast-Famine Reactor Cycling for Natural Selection of Lipid- accumulating Algae Communities

Mixed microalgal-microbial reactors were inoculated with pond water and operated using a

feast-famine feeding regiment at pH 7.5 and pH 9. Feast-famine cycling was expected to create

a selection pressure for increased lipid production in the microbial community in the reactors.

Experimental results, however, demonstrated a decrease in average lipid content in all reactors.

The decrease in lipid content was largest and statistically significant in the pH 9 reactors. The

cause of this decrease in lipid content appears to be the result of interaction effects between a

shift in community composition at pH 9 toward a community dominated by a single OTU as well as physiological downregulation of lipid synthesis in the resident algal population. pH 9 reactors showed several other differences relative to pH 7.5 reactors, including decreased bacterial abundance, decreased diversity, but higher biomass productivity.

There is a theoretical basis for feast-famine selection producing a higher lipid content algal

polyculture. Several bioprocess changes could produce the intended increase in lipid content with nutrient cycling. These changes include altering cycle time, changing influent media or using real wastewater for cultivation, and increasing inoculum diversity. The differences in microbial community dynamics with different pH suggest that this is an important parameter to consider in future iterations of the feast-famine selective reactor operation.

6.2 Chapter 3: MAB Flocs as a Platform for Wastewater Treatment and Biofuel Production

Experiments with bioaugmented MAB flocs and MAB floc of mixed algal populations

revealed insights into interactions between bacteria and algae in general as well as operational

considerations for MAB floc reactors for wastewater treatment and/or biofuel feedstock

112

production. Bioaugmented flocs showed that different species exhibited different degrees of

uptake into activated sludge flocs. This differential bioflocculation efficiency could be

mediated by quorum sensing chemicals. It has important implications for natural ecological roles of different phytoplankton species and for the selection of algal strains for engineered

assembly of MAB floc communities. Preliminary experiments with bioaugmented MAB flocs

inoculated with several strains revealed differences in settleability and P uptake and suggest

that triggering nutrient-limited lipid accumulation could theoretically increase lipid content by up to 25% in the flocs. Polyculture MAB flocs offered a few key insights into operation of

MAB floc reactors. First, luxury P uptake was shown to be rapid and complete (>90% P removal within an hour of incubation with phosphate), a characteristic of MAB flocs that can

allow for enhanced nutrient removal efficiencies from wastewater. Additionally, these

polyculture MAB floc reactors exhibited increasing effluent TSS values over time, which

became larger than typical TSS discharge limits. This deterioration of settleability is

hypothesized to be a result of pH increases from algal photosynthesis resulting in floc

instability.

Several results from MAB floc reactor operation can be examined in future experiments. First, the role of pH in MAB floc settleability is an important research question with implications for the use of MAB flocs for wastewater treatment. N removal in MAB floc reactors was generally higher in aerated versus non-aerated SBRs. Whether or not this effect was mediated by aeration itself, or simply higher levels of turbulent shear in these reactors, was unclear.

Investigating the role of aeration compared to increasing overall mixing rate can help optimize reactor performance relative to energy input. Finally, differences in bioflocculation between algae species is an interesting avenue of research for natural systems. Future research could

113

examine whether or not these differences are indeed mediated by quorum sensing chemicals

and how these differences influence natural ecological niches for species of phytoplankton.

6.3 Chapter 4: Competition Between Nitrification and Algal Growth in MAB Reactors Treating Synthetic Wastewater

Experiments conducted in 1.5-L CSTRs treating synthetic wastewater using a MAB consortia

examined competition between microalgae and nitrifying bacteria and the possible

implications of this competition for wastewater treatment efficiency. These reactors indicated

the ability for stable coexistence between AOB/NOB and algal populations over several

weeks. However, under illumination, algae took up ammonia at a faster rate than nitrifying bacteria, and higher BOD in reactor influent resulted in higher N uptake efficiencies and less nitrification. Mass balance calculations and experimental results suggest that MAB biomass in the reactors scavenged luxury N and P from reactor influent and that biomass production rates were sub-optimal. These factors could have contributed to stable coexistence of nitrifying bacteria and algae. Once nitrate/nitrite is created in MAB reactors, nitrate/nitrite uptake by

algae appears to be minimal. The low nitrate/nitrite uptake by algae and the capacity of algae

for luxury P uptake suggest that the relative abundances of algae versus AOB/NOB in MAB

reactors have implications for nutrient removal efficiencies from wastewater.

These results suggest several research directions for aerated MAB reactors. First, it would be

useful to determine optimal SRT and HRT in aerated MAB reactors and to examine the effects

of lower SRT on the competition between algae and nitrifying bacteria. Additionally, the

relative limits on BOD removal in non-aerated MAB systems should be examined to

determine if aeration can allow for treatment of higher-strength waste streams.

114

6.4 Chapter 5: Invasion in Microalgal Communities

Preliminary experiments were performed regarding invasion of phytoplankton communities

by a bloom forming cyanobacteria through the operation of pond microcosms. These

microcosms were subjected to immigration by the invading cyanobacteria at two propagule

pressures, but the native communities resisted integration of the invader at both low and high

propagule pressures. These preliminary results suggest that different conditions should be

examined in conjunction with invasion pressure, including, for instance, subjecting microcosms to immigration alongside a shift in nutrient composition of influent feed media.

In addition to examining the effects of immigration propagule pressures, there are still open questions regarding deterministic factors influencing HAB formation which can be addressed in microcosm studies. Finally, one challenge for future microcosm invasion studies is natural floc formation which creates sampling bias when analyzing small sample sizes during cell counts.

115

APPENDIX

A.1 Effect of Centrifuging and Resuspension on Algal Growth

Growth curves of algae before and after centrifuging and re-suspending at a lower biomass concentration. Biomass growth indicated by optical density (750nm) measurements.

116

A.2 Quality Control on Modifications to Starch Kit Assay

Repeatability in replicates of flour standards and algae samples as indicated by measured % starch by dry weight (%dw) using scaled-down reaction volumes.

Dry wt (mg) Absorbance 510nm %dw Flour 1 5 0.503 86.2 Flour 2 5.2 0.531 87.5 Flour 3 5.1 0.497 83.1 Flour 4 4.8 0.495 88.3 Algae 1 3.9 0.478 12.3 Algae 2 5.2 0.646 12.4 Algae 3 6 0.704 11.7

Test of linearity of lower-volume Megazyme starch kit analysis using a glucose standard curve.

117

A.3 R Code for Microbial Community Analysis

setwd("C:/Users/Mark/Google Drive/Thesis Work/Data/Experiments/Directed Evolution Project/Microbial Community Data/Analysis")

# Put in correct labels in excel, organized by sample/date, copy and pasted into text document # Import said text document OTUs<-read.table("OTUs_in_samples.txt",stringsAsFactors=FALSE)

# Cut off OTU size from OTU names and then make OTU names rownames of dataset rownames(OTUs)<-do.call('rbind',strsplit(rownames(OTUs),";",fixed=TRUE))[,1]

################## DESEq for Differential Abundance #################

# Create a matrix with sample variable information ph_cycle<-read.table("Sample_info.txt",stringsAsFactors=FALSE) rownames(ph_cycle)<-colnames(OTUs)

library("phyloseq") library("DESeq2")

# Create phyloseq experimental object OTU<-otu_table(OTUs,taxa_are_rows=TRUE) sdata<-sample_data(ph_cycle) pseq<-merge_phyloseq(OTU,sdata)

# Calculate differential abundance using variables cycle #, pH, and interaction:

118 pseqi<-phyloseq_to_deseq2(pseq,~Buffer+Week+Buffer:Week) pseqA<-DESeq(pseqi,test="LRT",reduced=~Buffer+Week)

# Results res<-results(pseqA, cooksCutoff=FALSE)

# Pull out significant OTUs at 95% alpha alpha<-0.05 signif<-res[which(res$padj

# Cross significant OTUs with abundance OTUs_DE<-prop.table(as.matrix(OTUs),2) OTUs_DE<-OTUs_DE[,13:24] OTUs_DE<-OTUs_DE[rownames(OTUs_DE)%in%rownames(signif),] OTUs_sig_tot<-OTUs_DE

# Pick out OTUs with reads >=5% of total read count for pH 9 reactors OTUs_DE[OTUs_DE<.05]<-0

# And OTUs present in more than just initial sampling OTUs_T<-OTUs_DE[rowSums(OTUs_DE[,4:12])>0,]

# Combine OTU prop table values with their p values and output to text file OTUs_T<-cbind(OTUs_T,signif$padj[rownames(signif)%in%rownames(OTUs_T)]) colnames(OTUs_T)[13]<-"Padj" write.table(OTUs_T,file="Significant_OTUs.txt") write.table(OTUs_sig_tot,file="Significant_OTUs_all.txt")

119

##################### Diversity Metrics ######################### library(vegan)

# Shannon index (numerical) div<-data.frame(diversity(t(OTUs))) write.table(div,file="shannon.txt")

# Plots of diversity indices ph_cycle2<-read.table("Sample_info_2.txt",stringsAsFactors=FALSE) rownames(ph_cycle2)<-colnames(OTUs) sdata2<-sample_data(ph_cycle2) pseq2<-merge_phyloseq(OTU,sdata2) plot_richness(pseq2,x="Group",measures=c("Shannon","Simpson"))

################## FOR VISUALIZATION OF DATA ####################

# For each sample, set counts for OTUs not comprising 95% quantile of total counts in sample to 0 for (i in 1:ncol(OTUs)){ OTUs[,i][OTUs[,i]<=quantile(OTUs[,i],0.95)]<-0 }

# Make table % of OTUs, rather than counts OTUs_HT<-prop.table(as.matrix(OTUs[rowSums(OTUs)>0,]),2)

# Make bar plots library("randomcoloR")

120

# Plot parameters: # mar is margins (bottom, top, left, right) # yaxt = 'n' removes y axis label # cex.names changes magnification on x axis labels # space is between bars # args.legend places legend, inset changes position rel to margins (top, right) # mpg middle value moves x-axis labels closer to plot # cex.main is magnification of top label par(mar=c(3,1,1.5,5),mgp=c(1,-.4,.5)) barplot(OTUs_HT, legend.text=TRUE, args.legend=list(x="topright", bty="n",inset=c(-.15,-.05),cex=.5), yaxt='n', space=c(0.5,0,0,0.5,0,0,0.5,0,0,0.5,0,0,2,0,0), names.arg=colnames(OTUs_HT), las=2, cex.names=.6, cex.main = 0.8, col=distinctColorPalette(nrow(OTUs_HT)))

#################### Total OTU reads plot ######################## OTUs_sum<-colSums(OTUs) OTUs_Hsum<-OTUs_sum[1:15] OTUs_Tsum<-OTUs_sum[16:30]

par(mar=c(5,5,2,5),mgp=c(3,.5,0))

121

barplot(OTUs_Hsum, main="Total Reads per Sample pH 7.5", ylab="Total reads", xlab="Sample ID", space=c(0,0,0,0.5,0,0,0.5,0,0,0.5), names.arg=colnames(OTUs_Hsum), las=2, cex.names=.6, cex.main = 0.8, cex.axis=0.7) par(mar=c(5,5,2,5),mgp=c(3,.5,0)) barplot(OTUs_Tsum, main="Total Reads per Sample pH 9", ylab="Total reads", xlab="Sample ID", space=c(0,0,0,0.5,0,0,0.5,0,0,0.5,0,0,0.5), names.arg=colnames(OTUs_Tsum), las=2, cex.names=.6, cex.main = 0.8, cex.axis=0.7)

122

A.4 OTU Information

OTU_291;size=665302; OTU_350;size=235; OTU_581;size=2; OTU_573;size=2; OTU_354;size=85; OTU_582;size=365; OTU_616;size=2; OTU_312;size=1068; OTU_506;size=22; OTU_272;size=316; OTU_513;size=5504; OTU_29;size=10; OTU_339;size=1651; OTU_508;size=13; OTU_345;size=275; OTU_268;size=1625; OTU_560;size=4; OTU_383;size=419; OTU_91;size=4614; OTU_262;size=14; OTU_171;size=633; OTU_10;size=6; OTU_278;size=69; OTU_588;size=16; OTU_509;size=7; OTU_242;size=46370; OTU_4;size=10; OTU_335;size=2; OTU_593;size=2; OTU_313;size=2; OTU_200;size=327636; OTU_347;size=160; OTU_36;size=751; OTU_12;size=9; OTU_484;size=26; OTU_98;size=42921; OTU_352;size=179; OTU_303;size=369; OTU_599;size=4; OTU_526;size=2; OTU_307;size=30126; OTU_439;size=1075; OTU_37;size=956; OTU_583;size=123; OTU_576;size=2; OTU_317;size=2984; OTU_411;size=6; OTU_41;size=33; OTU_477;size=93; OTU_64;size=88; OTU_510;size=26823; OTU_608;size=2; OTU_572;size=2; OTU_458;size=3; OTU_585;size=91; OTU_308;size=19468; OTU_309;size=11958; OTU_105;size=793; OTU_402;size=10; OTU_586;size=18; OTU_190;size=26143; OTU_344;size=357; OTU_176;size=41; OTU_487;size=31; OTU_358;size=52; OTU_34;size=122; OTU_1;size=6; OTU_440;size=530; OTU_579;size=5; OTU_612;size=2; OTU_471;size=1284; OTU_592;size=2; OTU_362;size=45; OTU_578;size=2; OTU_47;size=11; OTU_342;size=744; OTU_575;size=2; OTU_35;size=127; OTU_567;size=21; OTU_160;size=4; OTU_331;size=3584; OTU_231;size=31828; OTU_108;size=89; OTU_624;size=9; OTU_149;size=9; OTU_50;size=765; OTU_343;size=782; OTU_298;size=3; OTU_622;size=4; OTU_590;size=5; OTU_270;size=5226; OTU_276;size=119; OTU_444;size=35; OTU_221;size=16; OTU_46;size=4; OTU_623;size=25; OTU_591;size=2; OTU_478;size=72; OTU_589;size=14; OTU_611;size=2; OTU_26;size=8; OTU_472;size=787; OTU_57;size=199; OTU_121;size=687; OTU_601;size=2; OTU_106;size=291; OTU_45;size=12; OTU_386;size=47; OTU_53;size=2; OTU_602;size=2; OTU_259;size=2732; OTU_65;size=13; OTU_107;size=145; OTU_52;size=18; OTU_604;size=4; OTU_570;size=6; OTU_11;size=8; OTU_448;size=14; OTU_63;size=737; OTU_603;size=2; OTU_31;size=20; OTU_482;size=60; OTU_92;size=382; OTU_559;size=2; OTU_23;size=8; OTU_340;size=1224; OTU_587;size=80; OTU_82;size=8; OTU_40;size=37; OTU_119;size=2; OTU_618;size=2; OTU_79;size=33; OTU_67;size=10; OTU_609;size=2; OTU_613;size=2; OTU_17;size=6; OTU_574;size=2; OTU_49;size=97; OTU_78;size=73; OTU_614;size=2; OTU_269;size=1087; OTU_512;size=115; OTU_274;size=119; OTU_596;size=2; OTU_141;size=2; OTU_365;size=2610; OTU_238;size=1445; OTU_355;size=91; OTU_615;size=5; OTU_568;size=32; OTU_447;size=35968; OTU_39;size=4; OTU_86;size=2; OTU_51;size=17; OTU_571;size=2; OTU_170;size=2214; OTU_273;size=180; OTU_5;size=4; OTU_595;size=8; OTU_620;size=3; OTU_38;size=6; OTU_72;size=11829; OTU_446;size=19; OTU_80;size=27; OTU_619;size=11; OTU_284;size=15; OTU_594;size=10; OTU_353;size=112; OTU_597;size=12; OTU_606;size=2; OTU_474;size=663; OTU_517;size=13; OTU_368;size=19; OTU_598;size=2; OTU_605;size=2; OTU_333;size=5875; OTU_584;size=139; OTU_181;size=11; OTU_457;size=3; OTU_33;size=2; OTU_507;size=35; OTU_110;size=12; OTU_441;size=126; OTU_465;size=2; OTU_625;size=3; OTU_56;size=6; OTU_519;size=5; OTU_553;size=9; OTU_610;size=2; OTU_617;size=3; OTU_338;size=1778; OTU_42;size=90; OTU_118;size=38; OTU_258;size=4250; OTU_32;size=2; OTU_76;size=10; OTU_452;size=21; OTU_122;size=1480; OTU_552;size=35;

123

A.5 Grouping Information for ANOVA

Minitab Output: Grouping Information Using the Tukey Method and 95% Confidence

Cycle N Mean Grouping 1 3 16.179 A 0 3 12.157 B 2 3 9.85 B 3 3 5.905 C 4 3 5.157 C

Means that do not share a letter are significantly different.

124

A.6 Linearity Check for ChlA Fluorometric Analysis

chlA fluorometric detection response with different dilutions of C. vulgaris culture diluted into phosphate buffered saline.

chlA fluorometric detection response with different dilutions of S. dimorphus culture diluted into phosphate buffered saline.

125

A.7 Linearity Check for Colorimetric SPV Lipid Analysis

Colorimetric lipid detection analysis on samples of C. vuglaris and activated sludge at different quantities of extracted biomass.

Total dw (mg) OD530 lipid (mg) % lipid (dry wt) 0.12 0.075 0.03 21.16 C. vulgaris C. vulgaris 0.31 0.167 0.05 16.32 C. vulgaris 0.43 0.245 0.07 16.41 C. vulgaris 0.61 0.356 0.10 16.22

Activated Sludge 0.10 0.046 0.02 19.14 Activated Sludge 0.24 0.094 0.03 12.89 Activated Sludge 0.34 0.151 0.05 13.65 Activated Sludge 0.48 0.213 0.06 12.93

Plots of dry weight extracted versus percent lipids measured shows that between 0.3mg and 0.6mg of dry biomass per extraction assays are repeatable for lipid content determination.

126

A.8 Images of Settleability in MAB Floc Reactors

From left to right: 5:1, 2:1, 1:1, 1:5 From left to right: 5:1, 2:1, 1:1 N. oleoabundans:AS w/w C. vulgaris::AS w/w

From left to right: 5:1, 2:1, 1:1, 1:5 S. dimorphus::AS w/w

127

A.9 S. dimorphus growth curve

Growth curve of S. dimorphus using optical density (750 nm) as a measure of biomass growth.

128

A.10 Microbial Community Analyses Methods Written by Jim Griffin

Amplicon Sequencing and qPCR Methods

By Jim Griffin, PhD Candidate, Northwestern University

Genomic DNA was extracted from replicate 1.5 ml aliquots of mixed biomass using the

FastDNA Spin Kit for Soil (MP Bio, Santa Ana, CA). Following DNA extraction, genomic

DNA concentration was measured via fluorescence using the Quant-It dsDNA quantification kit (Thermo-Fisher Scientific, Rockford, IL) on a 96 well plate reader (BioTek Instruments,

Winoovski, VT). 16S gene copy number was quantified via qPCR and normalized to copy number per nanogram of DNA. For each sample, two 20 uL qPCR reactions were performed using the Eub519F and Eub907R primer set1 in a Biorad CFX Connect thermal cycler (Bio-

Rad, Hercules, CA). Reaction conditions were 95 °C for 5 minutes followed by 40 cycles of:

95 °C (30 s), 55 °C (45 s), and 68 °C (30 s) and a final elongation step at 68°C for 7 minutes.

18S rRNA V8-V9 amplicon libraries were prepared using a two step PCR amplification

protocol. In the first PCR reaction, sequences were amplified using the V8F-1510R primer

set2 tagged with “common sequences” on each primer. The PCR conditions were 95°C for 3

minutes, followed by 25 cycles of 98°C for 20 seconds, 65°C for 15 seconds, and 72°C for 15

seconds, with a final extension at 72°C for 10 min. For the second round of PCR, round 1

PCR product from each sample was amplified using a unique barcoded primer (Fluidigm, San

Francisco, CA). Conditions were 95°C for 5 minutes, followed by 8 cycles of 95°C for 30

seconds, 60°C for 30 seconds, and 68°C for 30 seconds, with a final extension at 68°C for 7

min. A negative PCR control with no template DNA was included for each round of PCR.

DNA sequencing was performed using a Miseq sequencer (Illumina, San Diego CA) at the

University of Illinois Chicago DNA Services Facility.

129

250 base pair paired end 18S amplicon reads were merged using PEAR3, then quality filtered to remove reads with >1 error per 100 nucleotides, checked for chimeras and clustered into

OTUs using Vsearch4. Sequences were clustered into 199 non-chimeric OTUs that were used for downstream analysis. Taxonomy assignment was performed in QIIME5 using the Silva

Eukarya database (v108). To assess differences between samples in terms of composition, pairwise Weighted Unifrac distances were calculated from rarefied OTU abundance counts using QIIME’s “beta_diversity.py” script.

1. Bürgmann, H.; Jenni, S.; Vazquez, F.; Udert, K. M., Regime shift and microbial dynamics in a sequencing batch reactor for nitritation/anammox treatment of urine. Applied and environmental microbiology 2011, AEM. 02986-10.

2. Bradley, I. M.; Pinto, A. J.; Guest, J. S., Design and Evaluation of Illumina MiSeq-

Compatible, 18S rRNA Gene-Specific Primers for Improved Characterization of Mixed

Phototrophic Communities. Applied and environmental microbiology 2016, 82, (19), 5878-5891.

3. Zhang, J.; Kobert, K.; Flouri, T.; Stamatakis, A., PEAR: a fast and accurate Illumina

Paired-End reAd mergeR. Bioinformatics 2013, 30, (5), 614-620.

4. Rognes, T. Vsearch. https://github.com/torognes/vsearch. https://github.com/torognes/vsearch

5. Caporaso, J. G.; Kuczynski, J.; Stombaugh, J.; Bittinger, K.; Bushman, F. D.;

Costello, E. K.; Fierer, N.; Pena, A. G.; Goodrich, J. K.; Gordon, J. I., QIIME allows analysis of high-throughput community sequencing data. Nat Meth 2010, 7, (5), 335-336.

130

REFERENCES

1. Schlagermann, P., Göttlicher, G., Dillschneider, R., Rosello-Sastre, R. & Posten, C. Composition of algal oil and its potential as biofuel. J. Combust. 2012, (2012).

2. Menetrez, M. Y. An overview of algae biofuel production and potential environmental impact. Environ. Sci. Technol. 46, 7073–7085 (2012).

3. Gerardo, M. L., Van Den Hende, S., Vervaeren, H., Coward, T. & Skill, S. C. Harvesting of microalgae within a biorefinery approach: A review of the developments and case studies from pilot-plants. Algal Research (2015). doi:10.1016/j.algal.2015.06.019

4. Chisti, Y. Biodiesel from microalgae beats bioethanol. Trends Biotechnol. (2008). doi:10.1016/j.tibtech.2007.12.002

5. Bernard, O. Policy Analysis Life-Cycle Assessment of Biodiesel Production from Microalgae. 6475–6481 (2009).

6. Gao, H., Scherson, Y. D. & Wells, G. F. Towards energy neutral wastewater treatment: methodology and state of the art. Environ. Sci. Process. Impacts 16, 1223–46 (2014).

7. Hwang, J.-H., Church, J., Lee, S.-J., Park, J. & Lee, W. H. Use of Microalgae for Advanced Wastewater Treatment and Sustainable Bioenergy Generation. 33, (2016).

8. Zhou, W. et al. Environment-enhancing algal biofuel production using wastewaters. Renew. Sustain. Energy Rev. 36, 256–269 (2014).

9. Smith, V. H. & Crews, T. Applying ecological principles of crop cultivation in large- scale algal biomass production. Algal Res. 4, 23–34 (2014).

10. Collotta, M. et al. Comparative LCA of Flocculation for the Harvesting of Microalgae for Biofuels Production. Procedia CIRP 61, 756–760 (2017).

11. Smith, V. H. & Mcbride, R. C. Key ecological challenges in sustainable algal biofuels production. J. Plankton Res. 37, 671–682 (2015).

131

12. Shurin, J. B. et al. Industrial-strength ecology: Trade-offs and opportunities in algal biofuel production. Ecol. Lett. 16, 1393–1404 (2013).

13. Pandhal, J. & Noirel, J. Synthetic microbial ecosystems for biotechnology. Biotechnol. Lett. 1141–1151 (2014). doi:10.1007/s10529-014-1480-y

14. Flynn, K. J., Kenny, P. & Mitra, A. Minimising losses to predation during microalgae cultivation. J. Appl. Phycol. (2017). doi:10.1007/s10811-017-1112-8

15. Carney, L. T. & Lane, T. W. Parasites in algae mass culture. Front. Microbiol. 5, 1–8 (2014).

16. Stockenreiter, M. et al. Functional group richness: Implications of biodiversity for light use and lipid yield in microalgae. J. Phycol. 49, 838–847 (2013).

17. Stockenreiter, M., Graber, A. K., Haupt, F. & Stibor, H. The effect of species diversity on lipid production by micro-algal communities. J. Appl. Phycol. 24, 45–54 (2012).

18. Stockenreiter, M., Haupt, F., Seppälä, J., Tamminen, T. & Spilling, K. Nutrient uptake and lipid yield in diverse microalgal communities grown in wastewater. Algal Res. 15, 77–82 (2016).

19. Shurin, J. B., Mandal, S. & Abbott, R. L. Trait diversity enhances yield in algal biofuel assemblages. J. Appl. Ecol. 51, 603–611 (2014).

20. Jakob O. Nalley, Maria Stockenreiter, and E. L. Community Ecology of Algal Biofuels: Complementarity and Trait-Based Approaches. Ind. Biotechnol. 10, 191–201 (2014).

21. Mooij, P. R., Stouten, G. R., van Loosdrecht, M. C. M. & Kleerebezem, R. Ecology- based selective environments as solution to contamination in microalgal cultivation. Current Opinion in Biotechnology (2015). doi:10.1016/j.copbio.2014.11.001

22. Abdelaziz, A. E. M., Leite, G. B., Belhaj, M. a. & Hallenbeck, P. C. Screening microalgae native to Quebec for wastewater treatment and biodiesel production. Bioresour. Technol. 157, 140–148 (2014).

132

23. Vandamme, D., Foubert, I. & Muylaert, K. Flocculation as a low-cost method for harvesting microalgae for bulk biomass production. Trends Biotechnol. 31, 233–239 (2013).

24. Ummalyma, S. B. et al. Bioflocculation: An alternative strategy for harvesting of microalgae - An overview. Bioresour. Technol. 242, 227–235 (2017).

25. Su, Y., Mennerich, A. & Urban, B. Municipal wastewater treatment and biomass accumulation with a wastewater-born and settleable algal-bacterial culture. Water Res. 45, 3351–3358 (2011).

26. Gutzeit, G., Lorch, D., Weber, A., Engels, M. & Neis, U. Bioflocculent algal-bacterial biomass improves low-cost wastewater treatment. Water Sci. Technol. (2005).

27. Van Den Hende, S., Vervaeren, H., Saveyn, H., Maes, G. & Boon, N. Microalgal bacterial floc properties are improved by a balanced inorganic/organic carbon ratio. Biotechnol. Bioeng. 108, 549–558 (2011).

28. Kim, B.-H. et al. Nutrient Removal and Biofuel Production in High Rate Algal Pond Using Real Municipal Wastewater. 24, 1123–1132 (2014).

29. Xu, Y., Wang, Y., Yang, Y. & Zhou, D. The Role of Starvation in Biomass Harvesting and Lipid Accumulation: Co-Culture of Microalgae-Bacteria in Synthetic Wastewater. Environ. Prog. 35, 103–109 (2016).

30. Liu, L., Fan, H., Liu, Y., Liu, C. & Huang, X. Development of algae-bacteria granular consortia in photo-sequencing batch reactor. Bioresour. Technol. 232, 64–71 (2017).

31. Wang, L., Liu, J., Zhao, Q., Wei, W. & Sun, Y. Comparative study of wastewater treatment and nutrient recycle via activated sludge, microalgae and combination systems. Bioresour. Technol. 211, 1–5 (2016).

32. Chen, Q., Ni, J., Ma, T., Liu, T. & Zheng, M. Bioaugmentation treatment of municipal wastewater with heterotrophic-aerobic nitrogen removal bacteria in a pilot- scale SBR. Bioresour. Technol. 183, 25–32 (2015).

33. Shao, Y. et al. Bioaugmentation in lab scale constructed wetland microcosms for treating polluted river water and domestic wastewater in northern China. Int.

133

Biodeterior. Biodegrad. 95, 151–159 (2014).

34. Wakako, I. O. et al. Bioaugmentation of a wastewater bioreactor system with the nitrous oxide-reducing denitrifier Pseudomonas stutzeri strain TR2. Seibutsu-kogaku Kaishi 115, 37–42 (2013).

35. Djelal, H. & Amrane, A. Biodegradation by bioaugmentation of dairy wastewater by fungal consortium on a bioreactor lab-scale and on a pilot-scale. J. Environ. Sci. (China) 25, 1906–1912 (2013).

36. Su, Y., Mennerich, A. & Urban, B. Synergistic cooperation between wastewater-born algae and activated sludge for wastewater treatment: Influence of algae and sludge inoculation ratios. Bioresour. Technol. 105, 67–73 (2012).

37. Van Den Hende, S. et al. Treatment of industrial wastewaters by microalgal bacterial flocs in sequencing batch reactors. Bioresour. Technol. 161, 245–254 (2014).

38. Van Den Hende, S., Beelen, V., Bore, G., Boon, N. & Vervaeren, H. Up-scaling aquaculture wastewater treatment by microalgal bacterial flocs: from lab reactors to an outdoor raceway pond. Bioresour. Technol. 159, 342–54 (2014).

39. Wang, M., Yang, H., Ergas, S. J. & van der Steen, P. A novel shortcut nitrogen removal process using an algal-bacterial consortium in a photo-sequencing batch reactor (PSBR). Water Res. 87, 38–48 (2015).

40. Choi, O., Das, A., Yu, C. P. & Hu, Z. Nitrifying bacterial growth inhibition in the presence of algae and cyanobacteria. Biotechnol. Bioeng. 107, 1004–1011 (2010).

41. Risgaard-Petersen, N., Nicolaisen, M. H., Revsbech, N. P. & Lomstein, B. A. Competition between Ammonia-Oxidizing Bacteria and Benthic Microalgae Competition between Ammonia-Oxidizing Bacteria and Benthic Microalgae. Appl. Environ. Microbiol. 70, 5528–5537 (2004).

42. Huang, W. et al. Effect of algae growth on aerobic granulation and nutrients removal from synthetic wastewater by using sequencing batch reactors. Bioresour. Technol. 179, 187–192 (2015).

43. Lardon, L., Helias, A., Sialve, B., Steyer, J.-P. & Bernard, O. Life-Cycle Assessment of

134

Biodiesel Production from Microalgae. Environ. Sci. Technol. 43, (2009).

44. Lee, A. K., Lewis, D. M. & Ashman, P. J. Chemical Engineering Research and Design Energy requirements and economic analysis of a full-scale microbial flocculation system for microalgal harvesting. Chem. Eng. Res. Des. 88, 988–996 (2010).

45. Davis, R. et al. Process Design and Economics for the Production of Algal Biomass: Algal Biomass Production in Open Pond Systems and Processing Through Dewatering for Downstream Conversion. Natl. Renew. Energy Lab. 128 (2016).

46. Park, J. B. K., Craggs, R. J. & Shilton, A. N. Wastewater treatment high rate algal ponds for biofuel production. Bioresour. Technol. (2011). doi:10.1016/j.biortech.2010.06.158

47. Davis, T. W., Bullerjahn, G. S., Tuttle, T., McKay, R. M. & Watson, S. B. Effects of Increasing Nitrogen and Phosphorus Concentrations on Phytoplankton Community Growth and Toxicity during Planktothrix Blooms in Sandusky Bay, Lake Erie. Environ. Sci. Technol. 49, 7197–7207 (2015).

48. Mahapatra, D. M., Chanakya, H. N. & Ramachandra, T. V. Bioremediation and lipid synthesis through mixotrophic algal consortia in municipal wastewater. Bioresour. Technol. 168, 142–150 (2014).

49. Fields, M. W. et al. Sources and resources: Importance of nutrients, resource allocation, and ecology in microalgal cultivation for lipid accumulation. Appl. Microbiol. Biotechnol. 98, 4805–4816 (2014).

50. Ibelings, B. W. et al. HOST PARASITE INTERACTIONS BETWEEN FRESHWATER PHYTOPLANKTON AND CHYTRID FUNGI (CHYTRIDIOMYCOTA). doi:10.1111/j.1529-8817.2004.03117.x

51. Grimsley, N. H. et al. Genomics of Algal Host-Virus Interactions. Advances in Botanical Research 64, (Elsevier, 2012).

52. Agha, R. & Quesada, A. Oligopeptides as Biomarkers of Cyanobacterial Subpopulations. Toward an Understanding of Their Biological Role. 1929–1950 (2014). doi:10.3390/toxins6061929

135

53. Pieja, A. J., Sundstrom, E. R. & Criddle, C. S. Cyclic, alternating methane and nitrogen limitation increases PHB production in a methanotrophic community. Bioresour. Technol. 107, 385–392 (2012).

54. Carrino-Kyker, S. R., Smemo, K. a. & Burke, D. J. The effects of pH change and NO−3 pulse on microbial community structure and function: a vernal pool microcosm study. FEMS Microbiol. Ecol. 81, 660–672 (2012).

55. Etheridge, J. R., Birgand, F. & Burchell, M. R. Quantifying nutrient and suspended solids fluxes in a constructed tidal marsh following rainfall: The value of capturing the rapid changes in flow and concentrations. Ecol. Eng. 78, 41–52 (2015).

56. Isles, P. D. F. et al. Dynamic internal drivers of a historically severe cyanobacteria bloom in Lake Champlain revealed through comprehensive monitoring. J. Great Lakes Res. 41, 818–829 (2015).

57. Outram, F. N. et al. High-frequency monitoring of nitrogen and phosphorus response in three rural catchments to the end of the 2011–2012 drought in England. Hydrol. Earth Syst. Sci. 18, 3429–3448 (2014).

58. Montemezzani, V., Duggan, I. C., Hogg, I. D. & Craggs, R. J. A review of potential methods for zooplankton control in wastewater treatment High Rate Algal Ponds and algal production raceways. Algal Res. 11, 211–226 (2015).

59. Evans, L. et al. Effect of organic carbon enrichment on the treatment efficiency of primary settled wastewater by Chlorella vulgaris. Algal Res. 24, 368–377 (2017).

60. Sutherland, D. L., Howard-Williams, C., Turnbull, M. H., Broady, P. a. & Craggs, R. J. Seasonal variation in light utilisation, biomass production and nutrient removal by wastewater microalgae in a full-scale high-rate algal pond. J. Appl. Phycol. 1–13 (2013). doi:10.1007/s10811-013-0142-0

61. Unnithan, V. V., Unc, A. & Smith, G. B. Mini-review: A priori considerations for bacteria-algae interactions in algal biofuel systems receiving municipal wastewaters. Algal Res. 4, 35–40 (2013).

62. Guillard, R. R. L. & Lorenzen, C. J. Yellow-Green Algae With Ghlorophyllide C ’. J.

136

Phycol. 14, 10–14 (1972).

63. Boelee, N. C., Temmink, H., Janssen, M., Buisman, C. J. N. & Wijffels, R. H. Nitrogen and phosphorus removal from municipal wastewater effluent using microalgal biofilms. Water Res. 45, 5925–5933 (2011).

64. McFadden, G.I. and Melkonian, M. Use of Hepes buffer for microalgal culture media and fixation for electron microscopy. Phycologia 25, 551–557 (1986).

65. Fernandes, B. et al. Starch determination in Chlorella vulgaris-a comparison between acid and enzymatic methods. J. Appl. Phycol. (2012). doi:10.1007/s10811-011-9761-5

66. Van Wychen, S. & Laurens, L. M. L. Determination of Total Lipids as Fatty Acid Methyl Esters ( FAME ) by in situ Transesterification - Laboratory Analytical Procedure ( LAP ) - http://www.nrel.gov/docs/fy14osti/60958.pdf. NREL/TP- 5100-60958 (2013).

67. McMurdie, P. J. & Holmes, S. Phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS One 8, (2013).

68. McMurdie, P. J. & Holmes, S. Waste Not, Want Not: Why Rarefying Microbiome Data Is Inadmissible. PLoS Comput. Biol. 10, (2014).

69. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).

70. Oksanen, J. et al. vegan: Community Ecology Package. (2017).

71. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local alignment search tool. J. Mol. Biol. 215, 403–10 (1990).

72. Guest, J. S., van Loosdrecht, M. C. M., Skerlos, S. J. & Love, N. G. Lumped pathway metabolic model of organic carbon accumulation and mobilization by the alga Chlamydomonas reinhardtii. Environ. Sci. Technol. 47, 3258–67 (2013).

73. Lynch, F. et al. Screening native isolates of cyanobacteria and a green alga for integrated wastewater treatment, biomass accumulation and neutral lipid production. Algal Res. 11, 411–420 (2015).

137

74. Hu, Q. et al. Microalgal triacylglycerols as feedstocks for biofuel production: Perspectives and advances. Plant J. 54, 621–639 (2008).

75. Carney, L. T. et al. Microbiome analysis of a microalgal mass culture growing in municipal wastewater in a prototype OMEGA photobioreactor. Algal Res. 4, 52–61 (2014).

76. González-Fernández, C. & Ballesteros, M. Microalgae autoflocculation: An alternative to high-energy consuming harvesting methods. J. Appl. Phycol. 25, 991–999 (2013).

77. Wan, C. et al. Current progress and future prospect of microalgal biomass harvest using various flocculation technologies. Bioresour. Technol. 184, 251–257 (2015).

78. Novoveská, L., Zapata, A. K. M., Zabolotney, J. B., Atwood, M. C. & Sundstrom, E. R. Optimizing microalgae cultivation and wastewater treatment in large-scale offshore photobioreactors. Algal Res. 18, 86–94 (2016).

79. Mujtaba, G. & Lee, K. Treatment of real wastewater using co-culture of immobilized Chlorella vulgaris and suspended activated sludge. Water Res. 120, 174–184 (2017).

80. Powell, N., Shilton, A. N., Pratt, S. & Chisti, Y. Factors influencing luxury uptake of phosphorus by microalgae in waste stabilization ponds. Environ. Sci. Technol. 42, 5958– 5962 (2008).

81. Nambiar, K. R. & Bokil, S. D. Luxury uptake of nitrogen in flocculating algal-bacterial system. Water Res. 15, 667–669 (1981).

82. Gardner-Dale, D. A., Bradley, I. M. & Guest, J. S. Influence of solids residence time and carbon storage on nitrogen and phosphorus recovery by microalgae across diel cycles. Water Res. 121, 231–239 (2017).

83. Gonçalves, A. L., Pires, J. C. & Sim, M. Wastewater polishing by consortia of Chlorella vulgaris and activated sludge native bacteria. J. Clean. Prod. 133, 348–357 (2016).

84. Montemezzani, V., Duggan, I. C., Hogg, I. D. & Craggs, R. J. A review of potential methods for zooplankton control in wastewater treatment High Rate Algal Ponds and algal production raceways. Algal Res. 11, 211–226 (2015).

138

85. Guarnieri, M. T. et al. Examination of triacylglycerol biosynthetic pathways via de novo transcriptomic and proteomic analyses in an unsequenced microalga. PLoS One 6, (2011).

86. Cazzaniga, S. et al. Domestication of the green alga Chlorella sorokiniana: reduction of antenna size improves light-use efficiency in a photobioreactor. Biotechnol. Biofuels 7, 157 (2014).

87. Letcher, P. M. et al. Characterization of Amoeboaphelidium protococcarum, an Algal Parasite New to the Cryptomycota Isolated from an Outdoor Algal Pond Used for the Production of Biofuel. PLoS One 8, (2013).

88. Gong, Y. et al. Vernalophrys algivore gen. nov., sp. nov. (Rhizaria: Cercozoa: Vampyrellida), a new algal predator isolated from outdoor mass culture of Scenedesmus dimorphus. Appl. Environ. Microbiol. (2015). doi:10.1128/AEM.00160-15

89. Krustok, I. et al. Characterization of algal and microbial community growth in a wastewater treating batch photo-bioreactor inoculated with lake water. Algal Res. 11, 421–427 (2015).

90. Mata, T. M., Martins, A. A. & Caetano, N. S. Microalgae for biodiesel production and other applications: A review. Renew. Sustain. Energy Rev. 14, 217–232 (2010).

91. Wang, B. & Lan, C. Q. Biomass production and nitrogen and phosphorus removal by the green alga Neochloris oleoabundans in simulated wastewater and secondary municipal wastewater effluent. Bioresour. Technol. 102, 5639–5644 (2011).

92. Gorman, D. S. & Levine, R. P. Cytochrome F and plastocyanin: their sequence in the photosynthetic electron transport chain of chlamydomonas reinhardi. Proc. Natl. Acad. Sci. 54, 1665–1669 (1965).

93. Nichols, H. W. Handbook of Phycological Methods, Culture Methods and Growth Measurements. (1973).

94. Standard Methods for the Examination of Water and Wastewater.

95. OECD/OCDE. Oecd guideline for the testing of chemicals. (2001).

139

96. Eaton, A. D., Clesceri, L. S., Greenberg, A. E. & Franson, M. A. H. Standard Methods for the Examination of Water and Wastewater. (American public Health Association, American Water Works Association, Water Environment Federation, 1998).

97. Doane, T. a & Horwath, W. R. Spectrophotometric determination of nitrate with a single reagent. Anal. Lett. 36, 2713–2722 (2003).

98. Mayew, P., Cuhel, R. & Nyholm, N. A SIMPLE IN VITRO FLUORESCENCE METHOD FOR BIOMASS MEASUREMENTS IN ALGAL GROWTH INHIBITION TESTS. War. Res 31, 2525–2531 (1997).

99. Mishra, S. K. et al. Rapid quantification of microalgal lipids in aqueous medium by a simple colorimetric method. Bioresour. Technol. 155, 330–333 (2014).

100. Zhou, D. et al. Responses of the microalga Chlorophyta sp. to bacterial quorum sensing molecules (N-acylhomoserine lactones): aromatic protein-induced self- aggregation. Environ. Sci. Technol. acs.est.7b00355 (2017). doi:10.1021/acs.est.7b00355

101. EPA. Report on the performance of Secondary Treatment Technology. (2013).

102. Bossier, P. & Verstraete, W. Triggers for microbial aggregation in activated sludge? Appl. Microbiol. Biotechnol. 45, 1–6 (1996).

103. Adav, S. S., Lee, D. J., Show, K. Y. & Tay, J. H. Aerobic granular sludge: Recent advances. Biotechnol. Adv. 26, 411–423 (2008).

104. Tiron, O., Bumbac, C., Manea, E., Stefanescu, M. & Nita Lazar, M. Overcoming Microalgae Harvesting Barrier by Activated Algae Granules. Sci. Rep. 7, 4646 (2017).

105. Griffiths, M. J., Hille, R. P. Van & Harrison, S. T. L. Lipid productivity , settling potential and fatty acid profile of 11 microalgal species grown under nitrogen replete and limited conditions. 0, 989–1001 (2012).

106. Welter, C., Schwenk, J., Kanani, B., Van Blargan, J. & Belovich, J. Minimal Medium for Optimal Growth and Lipid Production of the Microalgae Scenedesmus dimorphus. Environ. Prog. Sustain. Energy 0, (2013).

107. Liss, S., Droppo, I., Leppard, G. & Milligan, T. Flocculation in Natural and Engineered

140

Environmental Systems. (CRC Press, 2004).

108. Sheng, G.-P., Yu, H.-Q. & Li, X.-Y. Stability of Sludge Flocs Under Shear Conditions: Roles of Extracellular Polymeric Substances (EPS). Wiley Intersci. 93, DO 10.1002/bit (2005).

109. Gardner, J. & Tatterson, G. Characterization of mixing in shaker table containers. Biotechnol. Bioeng. 39, 794–797 (1992).

110. Pereira, J. P. et al. in Single and Two-Phase Flows on Chemical and BIomedical Engineering (eds. Dias, R., Martins, A. A., Lima, R. & Mata, M. M.) 159 (Bentham Science, 2012).

111. de Bruin, L. M. M., de Kreuk, M. K., van der Roest, H. F. R., Uijterlinde, C. & van Loosdrecht, M. C. M. Aerobic granular sludge technology: An alternative to activated sludge? Water Sci. Technol. 49, 1–7 (2004).

112. McCarty, P. & Rittmann, B. Environmental Biotechnology: Principles and Applications. (McGraw-Hill, 2001).

113. Marcilhac, C. et al. Digestate color and light intensity affect nutrient removal and competition phenomena in a microalgal-bacterial ecosystem. Water Res. 64, 278–287 (2014).

114. Aslan, S. & Kapdan, I. K. Batch kinetics of nitrogen and phosphorus removal from synthetic wastewater by algae. Ecol. Eng. 28, 64–70 (2006).

115. Delgadillo-Mirquez, L., Lopes, F., Taidi, B. & Pareau, D. Nitrogen and phosphate removal from wastewater with a mixed microalgae and bacteria culture. Biotechnol. Reports 11, 18–26 (2016).

116. Brown, N. & Shilton, A. Luxury uptake of phosphorus by microalgae in waste stabilisation ponds: Current understanding and future direction. Rev. Environ. Sci. Biotechnol. 13, 321–328 (2014).

117. Sander, R. in NIST Chemistry WebBook (eds. Linstrom, P. J. & Mallard, W. G.) (National Institute of Standards and Technology).

118. Kaplan, D., Wilhelm, R. & Abeliovich, A. Interdependent environmental factors

141

controlling nitrification in waters. Water Sci. Technol. 42, 167–172 (2000).

119. Kinnunen, M. et al. A conceptual framework for invasion in microbial communities. Isme J 10, 1–7 (2016).

120. Kutovaya, O. A. et al. Evidence against fluvial seeding of recurrent toxic blooms of Microcystis spp. in Lake Erie’s western basin. Harmful Algae 15, 71–77 (2012).

121. Griffin, J. S. & Wells, G. F. Regional synchrony in full-scale activated sludge bioreactors due to deterministic microbial community assembly. ISME J. 11, 500–511 (2016).

122. Acosta, F., Zamor, R. M., Najar, F. Z., Roe, B. A. & Hambright, K. D. Dynamics of an experimental microbial invasion. Proc Natl Acad Sci U S A 112, 11594–11599 (2015).

123. Harmful Algal Blooms in the Great Lakes. (2009).

124. Harke, M. J., Davis, T. W., Watson, S. B. & Gobler, C. J. Nutrient-Controlled Niche Differentiation of Western Lake Erie Cyanobacterial Populations Revealed via Metatranscriptomic Surveys. Environ. Sci. Technol. 50, 604–615 (2016).

142