Influence of Native Freshwater Mussel Functional Traits And

INFLUENCE OF NATIVE FRESHWATER MUSSEL FUNCTIONAL TRAITS AND

COMMUNITY STRUCTURE ON NITROGEN REMOVAL IN STREAM SEDIMENTS

ZACHARY LYNN NICKERSON

CARLA L. ATKINSON, COMMITTEE CHAIR BEHZAD MORTAZAVI LISA DAVIS ROBERT H. FINDLAY

A THESIS

Submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Biological Sciences in the Graduate School of The University of Alabama

TUSCALOOSA, ALABAMA

2018

ABSTRACT

Animals physically and chemically modify their environment as a result of their functional traits. These effects are particularly influential in freshwater benthic environments where animal aggregations can impact the recycling and repackaging of major macronutrients. I examined the influence of native freshwater mussels (Bivalvia: Unionidae) on the removal of dissolved inorganic nitrogen (N) via the biogeochemical pathways of denitrification and annamox in freshwater sediments. In one experiment, I used continuous flow-through incubation

+ methods to assess the influence of individual mussel physiological traits (ammonium [NH4 ] excretion, organic matter [OM] biodeposition) on N-removal in stream sediments. In the second experiment of my thesis, I manipulated the biodiversity of mussel aggregations in their natural environment using in-situ stream benthic enclosures to assess the influence of mussel aggregations, associated functional traits, and the effect of mussel biodiversity on N-removal.

+ Incubation results showed NH4 excretion increased the ambient flux of dinitrogen gas (N2) across the sediment-water interface, while OM biodeposition increased the maximum N-removal potential in the sediment. Results of the in-stream enclosure experiment showed there were non- additive effects of mussel biodiversity on N-removal in stream sediments. Results suggested this effect was driven by an increase in biological activity (movement, burrowing), potentially driven by inter-specific competition among species with different niche requirements. My thesis research advances the field by linking specific mussel functional traits to an important ecosystem function, N-removal, and showing the importance of incorporating biodiversity into aggregate- scale studies of organisms’ influences on biogeochemical ecosystem processes.

DEDICATION

To my family, especially my parents Jeff and Sheryl, my siblings Tyler, Emily and

Sydney, and my wonderful wife Anastasia, for their unending love and support. Also, to Drs. Joy

O’keefe and Jennifer Latimer for fueling my passion for research.

iii

LIST OF ABBREVIATIONS AND SYMBOLS

AFDM Ash-free dry mass

AL Alabama

ANCOVA Analysis of covariance

ANOVA Analysis of variance

Annamox Anaerobic ammonium oxidation

ARW Artificial river water

BD-EF Biodiversity-ecosystem function framework

C Carbon

C. asperata Cyclonaias asperata

C.B. Charge balance

Ca2+ Calcium ion

CaCO3 Calcium carbonate

Cl- Chloride ion cm Centimeter cm3 Cubic centimeter

CO2 Carbon dioxide

Con. Conductivity

Coupled-DNF Coupled nitrification-denitrification d Day d-1 Per day

DIN Dissolved inorganic nitrogen

DISL Dauphin Island Sea Lab

DM Soft tissue dry mass

Dmax Strongest effect of complementarity possible for an ecosystem function

DNF Denitrification

DO Dissolved oxygen

DT Strength of complementarity for a specific ecosystem function

EF Ecosystem function e.g. exempli gratia (“for example”)

F F-statistic to determine significantly different means (analysis of variance)

F. flava Fusconaia flava g Gram h-1 Per hour

H2O Water

- NCO3 Bicarbonate ion hr, h Hour i Species i.e. id est (“that is”)

IPT Isotope pairing method

K+ Potassium ion

K2SO4 Potassium sulfate kg Kilogram kg-1 Per kilogram

v km Kilometer

L Liter

L-1 Per liter

L. ornata Lampsilis ornata m meter m2 Square meter m3 Cubic meter m-2 Per square meter mg Milligram

Mg2+ Magnesium ion

MgCl2 Magnesium chloride

MgSO4*7H2O Magnesium sulfate heptahydrate microcosm-1 Per microcosm

MIMS Membrane inlet mass spectrometer min-1 Per minute mL Milliliter mm Millimeter mM Millimolar (Millimoles per liter) mmol Millimole n Sample size

N Nitrogen

N Mussel abundance (individuals m-2)

N2 Dinitrogen gas

N2-N Mole of nitrogen derived from dinitrogen gas

29 N2 Dinitrogen gas derived from anaerobic ammonium oxidation

30 N2 Dinitrogen gas derived from denitrification

N2:Ar method Method to determine the concentration dinitrogen gas in water samples

Na+ Sodium ion

NaCl Sodium chloride

NaHCO3 Sodium bicarbonate

+ NH4 Ammonium

14 + NH4 Naturally-abundant ammonium molecule

- NO2 Nitrite

- NO3 Nitrate

— NO3 N Mole of nitrogen derived from nitrate

15 - NO3 Isotopically-labeled nitrate

N-removal Transformation of dissolved inorganic nitrogen to dinitrogen gas

N-species Any molecule containing a nitrogen atom

NTU Nephelometric Turbidity Unit

O2 Oxygen

OM Organic matter

P Phosphorus p p-value for determining significant results in statistical analyses pi,N Proportion of species i at density-level N

PCA Principal component analysis pCO2 Partial pressure of carbon dioxide

vii

PC1 Principal component one

PC2 Principal component two

R2 Coefficient of determination (regression)

Redox Reduction-oxidation s-1 Per second

S1 Stock solution one for the synthesis of artificial river water

S2 Stock solution two for the synthesis of artificial river water

S3 Stock solution three for the synthesis of artificial river water

SE Standard error of the mean

SLR Simple linear regression

2- SO4 Sulfate ion

Sp. Con. Specific Conductivity species-1 Per mussel species t Symbol representing the beginning of an incubation trial

T0 Start of sediment slurry incubation

T6 End of sediment slurry incubation

Tukey HSD Tukey’s Honest Significant Difference multiple comparison test

UA University of Alabama vol-1 Per volume of solution w/v Percent of molecule in total volume of solution

푥̅ Mean

ZnCl2 Zinc chloride

2D Two dimensional

viii

1.1x One and one tenth times

2x Two times

3x Three times

1000x One thousand times

α Alpha value defining confidence interval for statistical analysis

ΔEF Net effect of biodiversity

ε(EFpoly) Expected effect of biodiversity on an ecosystem function

σ Standard deviation

ΣAZ- Sum of concentrations (in microequivalents) for all anions in solution

ΣCZ+ Sum of concentrations (in microequivalents) for all in solution

µeq Microequivalent

µL Microliter

µm Micrometer

µmol Micromole

µS Microsiemen

°C Degree Centigrade

<< Much less than

< Less than

> Greater than

≥ Greater than or equal to

= Equals

≠ Does not equal

~ Approximately

± Plus or minus

+ Plus

- Negative x By

% Percent

[ ] Concentration in millimoles per liter

# Number of

→ Forward reaction

ACKNOWLEDGEMENTS

I would like to sincerely thank my advisory committee Carla L. Atkinson, Behzad

Mortazavi, Lisa Davis and Robert H. Findlay for their valuable support and advice throughout my time at the University of Alabama. Thank you to the Mortazavi lab at Dauphin Island Sea

Lab, including Alice Kleinhuizen, Dr. Corianne Tatariw, Derek Tollette and Taylor Ledford, for helping me learn how to run chamber incubations, analyze sediment and water samples, and analyze the data that formed the bulk of my thesis. Also, thank you to our collaborators in the

Davis lab at the University of Alabama Department of Geography, especially Matt Koerner, for working hard to make the in-stream enclosure experiment a success. Thank you to my lab mates

Brian van Ee and Monica Winebarger, without whom my fieldwork would not have been possible. I would like to send a huge thank you Anne Bell to whom I owe a debt of gratitude for teaching me the use of and care for the analytical instruments used in our lab, and for helping me analyze hundreds of water and sediment samples. I would like to acknowledge the Birmingham

Audubon Society and the Conchologists of America, Inc. for funding the majority of my thesis research. Also, thank you to the University of Alabama’s Department of Biological Sciences,

College of Arts and Sciences, and Graduate School Association for providing some research and travel funding. Thank you to Dauphin Island Sea Lab for providing lodging while I conducted research in the Mortazavi lab. I would like to thank the Weyerhaeuser Company for allowing us to access their land along the Sipsey River to conduct our in-stream experiment, and to collect mussels, sediment and site water for the incubation trials. Last but not least, thank you to my family and friends for always supporting me even when they found out I studied mussel poop.

CONTENTS

ABSTRACT...... ii

DEDICATION...... iii

LIST OF ABBREVIATIONS AND SYMBOLS...... iv

ACKNOWLEDGEMENTS...... xi

LIST OF TABLES...... xiii

LIST OF FIGURES...... xiv

CHAPTER 1: CONCEPTUAL FRAMEWORK AND SUMMARY OF OBJECTIVES...... 1

CHAPTER 2: USING FUNCTIONAL TRAITS TO DETERMINE THE INFLUENCE OF BURROWING BIVALVES ON NITROGEN REMOVAL IN STREAM SEDIMENTS...... 12

CHAPTER 3: INTEGRATING FUNCTIONAL TRAITS AND BIODIVERSITY TO

ASSESS THE ROLE OF BENTHIC COMMUNITIES ON NITROGEN REMOVAL...... 55

CHAPTER 4: SUMMARY, CONCLUSIONS, AND DIRECTIONS GOING FORWARD...... 99

APPENDIX 1: COMPONENTS AND PROCEDURE FOR SYNTHESIS OF ARTIFICIAL RIVER WATER...... 102

APPENDIX 2: NATURAL MUSSEL SPECIES COMPOSITION AND ABUNDANCES

PRESENT AT STUDY REACH PRIOR TO START OF ENCLOSURE EXPERIMENT...... 106

APPENDIX 3: REACH-SCALE PHYSIOCHEMICAL PARAMETERS AND DISCHARGE AT ENCLOSURE STUDY REACH, AND DEPTH/FLOW MEASUREMENTS FOR EACH ENCLOSURE...... 107

xii

LIST OF TABLES

2.1 Incubation Timeline...... 47

2.2 Mean (± 1 SE) values for mussel parameters and functional traits...... 48

2.3 Mean (± 1 SE) biogeochemical fluxes and potentials from the chamber incubation and

IPT experiments, respectively...... 49

3.1 Mean (± 1 SE) values for the functional trait quantified during the 9-week in-field experiment...... 91

3.2 Mean (± 1 SE) DNF and annamox potentials for each 1- and 2-species community at

high- and low-densities...... 92

A1.1 Major ion composition (μeq L-1) of the Sipsey River near Benevloa, AL, concentration corrections for charge balance (C.B.), and final balanced concentrations used in ARW

synthesis...... 104

A1.2 Recipe for the synthesis of Sipsey River ARW. Three stock solutions (S1, S2, S3) were - Z+ prepared separately and combined to form NO3 -free ARW with major cation (C ) and Z- anion (A ) concentrations equal to those naturally found in the Sipsey River...... 105

A2.1 Unionid mussel species found within our 50-m study reach. Reported are the name of the species and the number found in the sediment while digging to install enclosures...... 106

A3.1 At each visit to the site, we measured dissolved oxygen (DO) both in concentration (mg L-1) and in percent (%), conductivity (µS), specific conductivity (µS cm-1), pH and turbidity (NTU) both upstream and downstream of the enclosure area...... 107

A3.2 At each visit to the study site beginning on 31 August, 2017, we measured the depth (m) of each enclosure and the flow velocity (m s-1) directly above the sediment-water

interface at each enclosure...... 108

xiii

LIST OF FIGURES

1.1 Conceptual basis for my thesis research...... 11

2.1 Continuous flow-through incubation setup modified from Miller-Way and Twilley

(1996)...... 50

+ 2.2 Boxplots highlighting differences in (a) soft tissue dry mass, (b) NH4 excretion and (c) OM biodeposition between the mussel species used in the incubation trials...... 51

+ - 2.3 Mean (a) ambient nutrient (NH4 , NO3 ) flux, (b) ambient gas (N2-N, O2) flux and (c) N- removal potentials (DNF, annamox) between the control and mussel treatments analyzed

in the incubation trials...... 52

2.4 Scatterplots highlighting relationships between the response variables of (a) ambient N2 flux, (b) DNF potential and (c) annamox potential and the explanatory variables of (i) + soft tissue DM, (ii) mussel respiration, (iii) NH4 excretion and (iv) OM biodeposition..53

2.5 (a) Distribution among species of the residual variation present in the ambient N2 flux versus respiration SLR. These data were used as the response variable for SLRs testing + the influence of (b) NH4 excretion and (c) OM biodeposition on ambient N2 flux once the effect (caveat) of respiration had been removed from the data...... 54

3.1 (left) Location of the Sipsey River in northwest Alabama, USA. (right) Outline of the Sipsey River watershed. Our study site is denoted by the large star in the lower 40-km of the watershed...... 93

3.2 Movement was quantified using a 6x6 grid, which was laid atop an enclosure...... 94

3.3 Experiment timeline showing benthic temperature (thin line) and water depth (thick line) recorded by a datalogger located in the middle of our study reach...... 95

3.4 Barplots representing mean values for 1-species F. flava, C. asperata and 2-species treatments for low- (a,c,e,g) and high-density (b,d,f,h) treatments...... 96

3.5 Barplots representing mean values for DNF potentials within (a) low- and (b) high- density mussel aggregations, and annamox potentials within (c) low- and (d) high-density

mussel aggregations...... 97

3.6 Results of principal component analysis (PCA) explaining 76.28% of the variation in our

data...... 98

xiv

CHAPTER 1

CONCEPTUAL FRAMEWORK AND SUMMARY OF OBJECTIVES

In aquatic ecosystems, animals can play an important role in the storage and recycling of the macronutrients carbon (C), nitrogen (N) and phosphorus (P). While these macronutrients are essential to all life on Earth, modern agricultural practices and rapid urbanization have resulted in a drastic increase in the input of C, N and P into aquatic ecosystems (Carpenter et al. 1998).

Increased nutrient pollution, coupled with physical degradation of the landscape (i.e., deforestation, impoundment, dredging, canals) create positive feedback loops that continually weaken the resiliency of aquatic ecosystems, altering the associated floral and faunal communities (Dudgeon et al. 2006). In areas dominated by anthropologic land use, the flux of nutrients transported to aquatic ecosystems often exceed the retention/removal capacity of the biota within watersheds and their associated riparian buffers (Mander et al. 1997). Excess nutrients that enter these systems are transported to reservoirs, estuaries and coasts often causing eutrophication, which stimulates harmful algal blooms and creates hypoxic “dead zones” that can cause massive die-offs of lentic, estuarine and marine fauna (Landsberg 2002, Diaz and

Rosenberg 2008).

The flora and fauna of watersheds play a major role in reducing the downstream transport of nutrients through the retention/removal of C, N and P via biogeochemical cycling (Ranalli and

Macalady 2010, Aufdenkampe et al. 2011). Aquatic animals can influence biogeochemical cycles in many ways due to the variety of functional feeding groups that exist in benthic and pelagic zones, each of which processes organic matter (OM) in different ways

(Wallace and Webster 1996). Aquatic animals act as both sources and sinks of C, N and P

(Vanni 2002, Atkinson et al. 2017). As sources, aquatic animals facilitate bottom-up provisioning of ecosystem function through the input of bioavailable nutrients to the system via excretion and biodeposition, and through the release of nutrients from decomposing tissue following death. As sinks, aquatic organisms assert top-down control on nutrient cycling through secondary production, especially through the incorporation of material into long-term biomass such as bones or shell. The sink effect is intensified when aquatic animals remove nutrients completely from a system via migration, emergence or predation by terrestrial animals.

While the previously-described processes represent direct influences aquatic animals have on nutrient cycling in rivers, researchers are increasingly recognizing the additional role aquatic animals play by indirectly influencing nutrient cycling (Vanni 2002, Atkinson et al.

2017). For example, by controlling prey population size and age distribution, predators and grazers indirectly influence the ecosystem functions provisioned by prey populations (i.e., population-level production, excretion, egestion; Estes et al. 2011). Additionally, predators can indirectly influence source/sink dynamics by modifying feeding behavior and the tissue and excretion stoichiometry of prey through the “ecology of fear” (Schmitz et al. 2008, Dalton and

Flecker 2014). Aquatic animals also indirectly influence nutrient cycling by chemically and physically modifying their environment. This is most evident in the redox-dominated benthic zone, where there can be multiple functional feeding groups of macroinvertebrates coexisting in a relatively small area of streambed (Wallace and Webster 1996). Benthic macrofauna act as indirect sinks by concentrating nutrients in the benthos and stimulating the growth of benthic biofilms which, in turn, increases primary productivity and fosters more complex food-webs

(Spooner and Vaughn 2006, Atkinson et al. 2013).

Benthic macrofauna can also act as indirect nutrient sinks by stimulating the removal of dissolved inorganic nitrogen (DIN) via biogeochemical pathways (hereafter N-removal). N- removal refers to any form of metabolism in chemolithoheterotrophic facultative anaerobes that converts DIN to inert dinitrogen gas (N2) through a series of intermediate reactions (Figure 1.1a).

In aquatic systems, this represents the permanent removal of N, as N2 will diffuse through the water column and escape into the atmosphere. To foster N-removal, three requirements must be met within a system: 1) anoxic conditions, 2) excess reactive N, 3) energy substrate (Knowles

1982). Benthic sediments represent potential hotspots for N-removal as they can contain the required chemical and physical characteristics to activate microbial metabolisms necessary to convert DIN to N2. Previous research has shown that groups of burrowing benthic macrofauna

(e.g., bivalves, chironomid larvae, tubificid worms) chemically and physically alter the sediment, which fosters the conditions necessary to stimulate N-removal (Anschutz et al. 2012, Hölker et al. 2015, Turek and Hoellein 2015, Hoellein et al. 2017).

One can view an organism in terms of the traits that define how that organism interacts with the biotic and abiotic environment. This functional trait approach (Dı́az and Cabido 2001,

Violle et al. 2007) is imperative to understanding an organism’s role in influencing dynamic ecosystem processes such as biogeochemical cycling (Loreau et al. 2001). One group of aquatic organisms, freshwater mussels (Bivalvia: Unionidae), despite being in the same general functional guild (filter-feeding bivalves), possess functional traits that are often species-specific, such as reproductive strategy (Williams et al. 2008, Haag 2012), excretion/egestion rates/stoichiometry (Vaughn et al. 2007, Spooner et al. 2012, Atkinson et al. 2018) and movement/burrowing behavior (Schwalb and Pusch 2007, Allen and Vaughn 2009). Freshwater mussels (hereafter mussels) are long-lived (6-100 years), filter-feeding organisms that thrive in

3 densely-populated, speciose aggregations in lotic ecosystems (Haag 2012). Mussels live partially or completely burrowed in benthic sediments, and utilize their muscular foot and shell morphology to resist the sheer-stress of flowing water and remain anchored in the benthos (Allen and Vaughn 2009). Within a reach, adult mussels move based on changes in flow, erosion or disturbance (Vaughn and Hakenkamp 2001, Haag 2012) with some mussel species being more active than others, resulting in varying degrees of sediment reworking via bioturbation (Allen and Vaughn 2009). Filter-feeding and burrowing represent two defining functional traits that influence ecosystem processes. For example, filter-feeding removes particulate matter from the water column and concentrates it in the benthos, stimulating benthic-pelagic coupling (Vaughn et al. 2008). Burrowing traits alter the redox conditions of the sediment, potentially stimulating N- removal via DNF, as has been shown with other freshwater bivalves (i.e. the invasive Asian clam

[Corbicula fluminea], Turek and Hoellein 2015). Additional species-specific traits such as the rate and stoichiometry (i.e. the ratio of C, N and P) of excretion and biodeposition products have the potential to significantly contribute to the standing-stock and flux of C, N and P in benthic sediments, which can increase local food-web complexity and foster benthic biodiversity

(Vaughn et al. 2008, Atkinson et al. 2014).

By measuring the direct and indirect influence of functional traits on ecosystem processes, one can elucidate the functional role an organism plays in its ecosystem (Vaughn and

Hakenkamp 2001), which can aid in the conservation of imperiled flora and fauna. The goal of my research was to model the influence of native freshwater mussel individuals and aggregations on N-removal processes by integrating the functional traits of movement, burrowing, excretion and biodeposition with an analysis of nitrogen transformations in sediments of an unregulated lowland river in the southeastern United States. Specifically, my research had two objectives:

Objective 1: Microcosm incubation of individual mussels in river sediment to determine the influence of mussel excretion and biodeposition on gas and nutrient flux across the sediment- water interface.

+ Mussels excrete highly reactive N in the form of ammonium (NH4 ). Depending how

+ deep a mussel is buried into the vertical sediment profile, this excreted NH4 can have different

+ fates. If the mussel is completely exposed, it will excrete NH4 directly into the water column,

+ where the NH4 will either be taken up by primary producers (Atkinson et al. 2013), or converted

- to nitrate (NO3 ) via an aerobic microbial metabolism termed nitrification (aerobic oxidation of

+ - NH4 to NO3 , Figure 1.1b). If the mussel is buried and therefore closer to the oxic/anoxic redox

+ boundary, the excreted NH4 can undergo a process termed coupled nitrification-denitrification

(hereafter coupled-DNF, Figure 1.1d). This is a highly redox-sensitive process that occurs

+ + - nearest the oxic/anoxic boundary where the excreted NH4 can be nitrified (NH4 → NO3 ) by

- anaerobes and subsequently undergo denitrification (NO3 → ½ N2, hereafter DNF) by facultative anaerobes on fine spatial and temporal scales. Finally, if the correct conditions are

+ met (i.e., if the mussel is in very close proximity to anoxic sediment), the excreted NH4 could be converted to N2 gas without first being nitrified. This microbially-mediated process is known as

+ anaerobic ammonium oxidation (hereafter annamox), which involves one mole of NH4 -derived

- N combining with one mole of NO3 -derived N through a series of intermediates to form N2

+ (Figure 1.1a). I predicted that mussel NH4 excretion could stimulate N-removal in riverine

+ sediments through coupled-DNF and, if a mussel is buried deep enough in the sediment, NH4 excretion could also stimulate N removal via annamox (Figure 1.1d).

Bivalve biodeposition has been shown to significantly foster microbial colonization in benthic sediments, particularly in estuarine and coastal environments (Grenz et al. 1990, Mirto et

5 al. 2000). OM biodeposition (OM in both feces and pseudofeces) could influence N-removal in multiple ways. First, OM biodeposition could provide energy substrate for the DNF pathway in the form of labile C (Figure 1.1a). While annamox does not require a source of labile C (reactive

N is both the electron donor and acceptor in annamox), the decomposition of OM is an oxygen- consuming process that could create anoxic microenvironments able to sustain both annamox and DNF. Additionally, mussel feces and pseudofeces are covered in a mucus (Nichols et al.

2005), and the mucus’s cohesive nature could also aid in the formation of anoxic microenvironments in the sediment, thus fostering annamox and DNF. I predicted that mussel

OM biodeposition would positively influence N-removal in stream sediments. Mussel feces is released through the supre-anal aperture, which is located near the excurrent siphon where excreta and pseudofeces are released (Nichols et al. 2005). Therefore, it can be assumed that excreta, feces and pseudofeces are all released at the same location in space and could have a significant influence on coupled-DNF and/or annamox if the mussel’s realized niche is near the oxic/anoxic redox boundary.

Objective 2: In-situ community manipulation experiment to assess the integrative effects of mussel community structure and associated functional traits on N removal potential in the sediment of an unregulated lowland river in the southeastern U.S.

This experiment took the chamber incubation hypotheses, scaled them up to the community-level, added the functional traits of movement and burrowing behavior and

+ incorporated the factor of varying biodiversity. In terms of the influence of mussel NH4 excretion and OM biodeposition, the same predictions applied to this experiment as in the

+ chamber incubations with the only difference being that community-scale areal NH4 excretion

6 and OM biodeposition rates were estimated rather than empirically measuring rates of individuals. Additionally, I added the functional traits of horizontal movement and vertical burrowing as explanatory variables against which we modelled N-removal potential in the sediments. Species-specific differences in movement and burrowing traits (Schwalb and Pusch

2007, Allen and Vaughn 2009) integrated with competition for niche space could lead to varying degrees of community-scale movement and burrowing based on the abundance and diversity of mussel species present in the community. As mentioned previously, mussels that burrow are closer to the oxic/anoxic redox boundary. Therefore, I predicted that mussel communities with a high degree of burrowing would foster high rates of N-removal potential in the sediment. While buried mussels have the potential to foster high rates of N-removal potential, the act of burrowing, as well as that of horizontal movement across the benthos, has the potential to significantly disturb the redox gradient in the sediment, increase oxygen penetration into the sediment, and therefore reduce N-removal potential (Figure 1.1c). Therefore, I also predicted that mussel communities with more horizontal movement activity and greater variation in per capita burrowing behavior would foster low rates of N-removal potential.

References

Allen, D. C., and C. C. Vaughn. 2009. Burrowing behavior of freshwater mussels in experimentally manipulated communities. Journal of the North American Benthological Society 28:93-100.

Anschutz, P., A. Ciutat, P. Lecroart, M. Gérino, and A. Boudou. 2012. Effects of tubificid worm bioturbation on freshwater sediment biogeochemistry. Aquatic Geochemistry 18:475- 497.

Atkinson, C. L., K. A. Capps, A. T. Rugenski, and M. J. Vanni. 2017. Consumer‐driven nutrient dynamics in freshwater ecosystems: From individuals to ecosystems. Biological Reviews 92:2003-2023.

Atkinson, C. L., J. F. Kelly, and C. C. Vaughn. 2014. Tracing consumer-derived nitrogen in riverine food webs. Ecosystems 17:485-496.

Atkinson, C. L., B. J. Sansom, C. C. Vaughn, and K. J. Forshay. 2018. Consumer aggregations drive nutrient dynamics and ecosystem metabolism in nutrient-limited systems. Ecosystems 21:521-535.

Atkinson, C. L., C. C. Vaughn, K. J. Forshay, and J. T. Cooper. 2013. Aggregated filter‐feeding consumers alter nutrient limitation: Consequences for ecosystem and community dynamics. Ecology 94:1359-1369.

Aufdenkampe, A. K., E. Mayorga, P. A. Raymond, J. M. Melack, S. C. Doney, S. R. Alin, R. E. Aalto, and K. Yoo. 2011. Riverine coupling of biogeochemical cycles between land, oceans, and atmosphere. Frontiers in Ecology and the Environment 9:53-60.

Carpenter, S. R., N. F. Caraco, D. L. Correll, R. W. Howarth, A. N. Sharpley, and V. H. Smith. 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecological Applications 8:559-568.

Dalton, C. M., and A. S. Flecker. 2014. Metabolic stoichiometry and the ecology of fear in Trinidadian guppies: Consequences for life histories and stream ecosystems. Oecologia 176:691-701.

Diaz, R. J., and R. Rosenberg. 2008. Spreading dead zones and consequences for marine ecosystems. Science 321:926-929.

Dı́az, S., and M. Cabido. 2001. Vive la difference: Plant functional diversity matters to ecosystem processes. Trends in Ecology & Evolution 16:646-655.

Dudgeon, D., A. H. Arthington, M. O. Gessner, Z.-I. Kawabata, D. J. Knowler, C. Lévêque, R. J. Naiman, A. H. Prieur-Richard, D. Soto, and M. L. Stiassny. 2006. Freshwater

biodiversity: Importance, threats, status and conservation challenges. Biological Reviews 81:163-182.

Estes, J. A., J. Terborgh, J. S. Brashares, M. E. Power, J. Berger, W. J. Bond, S. R. Carpenter, T. E. Essington, R. D. Holt, and J. B. Jackson. 2011. Trophic downgrading of planet Earth. Science 333:301-306.

Grenz, C., M. N. Hermin, D. Baudinet, and R. Daumas. 1990. In situ biochemical and bacterial variation of sediments enriched with mussel biodeposits. Hydrobiologia 207:153-160.

Haag, W. R. 2012. North American freshwater mussels: Natural history, ecology, and conservation. Cambridge University Press. New York, New York.

Hoellein, T. J., C. B. Zarnoch, D. A. Bruesewitz, and J. DeMartini. 2017. Contributions of freshwater mussels (Unionidae) to nutrient cycling in an urban river: Filtration, recycling, storage, and removal. Biogeochemistry 135:307-324.

Hölker, F., M. J. Vanni, J. J. Kuiper, C. Meile, H.-P. Grossart, P. Stief, R. Adrian, A. Lorke, O. Dellwig, and A. Brand. 2015. Tube‐dwelling invertebrates: Tiny ecosystem engineers have large effects in lake ecosystems. Ecological Monographs 85:333-351.

Knowles, R. 1982. Denitrification. Microbiological Reviews 46:43.

Landsberg, J. H. 2002. The effects of harmful algal blooms on aquatic organisms. Reviews in Fisheries Science 10:113-390.

Loreau, M., S. Naeem, P. Inchausti, J. Bengtsson, J. Grime, A. Hector, D. Hooper, M. Huston, D. Raffaelli, and B. Schmid. 2001. Biodiversity and ecosystem functioning: Current knowledge and future challenges. Science 294:804-808.

Mander, Ü., V. Kuusemets, K. Lõhmus, and T. Mauring. 1997. Efficiency and dimensioning of riparian buffer zones in agricultural catchments. Ecological Engineering 8:299-324.

Mirto, S., R. Danovaro, and A. Mazzola. 2000. Microbial and meiofaunal response to intensive mussel-farm biodeposition in coastal sediments of the western Mediterranean. Marine Pollution Bulletin 40:244-252.

Nichols, S. J., H. Silverman, T. H. Dietz, J. W. Lynn, and D. L. Garling. 2005. Pathways of food uptake in native (Unionidae) and introduced (Corbiculidae and Dreissenidae) freshwater bivalves. Journal of Great Lakes Research 31:87-96.

Ranalli, A. J., and D. L. Macalady. 2010. The importance of the riparian zone and in-stream processes in nitrate attenuation in undisturbed and agricultural watersheds–a review of the scientific literature. Journal of Hydrology 389:406-415.

Schmitz, O. J., J. H. Grabowski, B. L. Peckarsky, E. L. Preisser, G. C. Trussell, and J. R. Vonesh. 2008. From individuals to ecosystem function: Toward an integration of evolutionary and ecosystem ecology. Ecology 89:2436-2445.

Schwalb, A. N., and M. T. Pusch. 2007. Horizontal and vertical movements of unionid mussels in a lowland river. Journal of the North American Benthological Society 26:261-272.

Spooner, D. E., and C. C. Vaughn. 2006. Context‐dependent effects of freshwater mussels on stream benthic communities. Freshwater Biology 51:1016-1024.

Spooner, D. E., C. C. Vaughn, and H. S. Galbraith. 2012. Species traits and environmental conditions govern the relationship between biodiversity effects across trophic levels. Oecologia 168:533-548.

Turek, K. A., and T. J. Hoellein. 2015. The invasive Asian clam (Corbicula fluminea) increases sediment denitrification and ammonium flux in 2 streams in the midwestern USA. Freshwater Science 34:472-484.

Vanni, M. J. 2002. Nutrient cycling by animals in freshwater ecosystems. Annual Review of Ecology and Systematics 33:341-370.

Vaughn, C. C., and C. C. Hakenkamp. 2001. The functional role of burrowing bivalves in freshwater ecosystems. Freshwater Biology 46:1431-1446.

Vaughn, C. C., S. J. Nichols, and D. E. Spooner. 2008. Community and foodweb ecology of freshwater mussels. Journal of the North American Benthological Society 27:409-423.

Vaughn, C. C., D. E. Spooner, and H. S. Galbraith. 2007. Context‐dependent species identity effects within a functional group of filter‐feeding bivalves. Ecology 88:1654-1662.

Violle, C., M. L. Navas, D. Vile, E. Kazakou, C. Fortunel, I. Hummel, and E. Garnier. 2007. Let the concept of trait be functional! Oikos 116:882-892.

Wallace, J. B., and J. R. Webster. 1996. The role of macroinvertebrates in stream ecosystem function. Annual Review of Entomology 41:115-139.

Williams, J. D., A. E. Bogan, and J. T. Garner. 2008. Freshwater mussels of Alabama and the Mobile basin in Georgia, Mississippi, and Tennessee. University of Alabama Press. Tuscaloosa, Alabama.

Figure 1.1: Conceptual basis for my thesis research. Top: (a) Details of the biogeochemical pathways I studied (nitrification, denitrification, annamox), and theorized influences of freshwater mussel functional traits (NH4+ excretion, OM biodeposition, burrowing) on said pathways. White arrows represent the hypothesized fate of excreted NH4+. Grey arrows represent the hypothesized fate of OM biodeposition. Dashed arrows represent passive diffusion of N-species between oxic and anoxic zones in the sediment. Bottom: Hypothesized effect of mussel functional traits on O2 penetration (black arrow), and N transformation pathways (nitrification = dark grey arrow; DNF and annamox = light grey arrows) for a (b) partially exposed mussel, (c) mussel actively burrowing, or (d) completely buried mussel. The thickness of the N transformation arrows represent the hypothesized magnitude of influence the mussel has on that process between different stages of burrowing.

CHAPTER 2

USING FUNCTIONAL TRAITS TO ASSESS THE INFLUENCE OF BURROWING BIVALVES ON NITRGOEN REMOVAL IN STREAM SEDIMENTS

Abstract

Functional traits define an organism in terms of how it interacts with its environment and determine the influence it has on dynamic ecosystem processes, such as biogeochemical nutrient cycling. Freshwater mussels (Bivalvia: Unionidae) form hotspots of biogeochemical activity in benthic environments by alleviating nutrient limitation at the sediment-water interface; however, little is known about the influence of mussel functional traits on sediment biogeochemical cycles, particularly N-removal via denitrification and annamox. Our aim was to model the influence of

+ two mussel functional traits: ammonium (NH4 ) excretion and organic matter (OM) biodeposition, on N-removal in stream sediments. We quantified mussel excretion and biodeposition, and incubated mussels in microcosms containing river sediment using flow- through methods. We measured nutrient and gas fluxes to quantify real-time ambient N-removal

(denitrification + annamox), and conducted isotope pairing techniques to determine the

+ maximum N-removal potential for denitrification and annamox pathways. NH4 excretion was shown to be a significant predictor of ambient N-removal, whereas OM biodeposition significantly increased the maximum N-removal potential in the sediment. Our study is the first of its kind to link a mussel’s influence on N-removal to specific functional traits and contributes to the growing knowledge of the role these highly imperiled organisms play in directly and indirectly influencing ecosystem-scale processes in lotic systems.

Introduction

Biogeochemical nutrient cycles are dependent on interactions between organisms and the abiotic environment. The magnitude and direction of an organism’s influence on such dynamic ecosystem functions are governed by functional traits (Lavorel and Garnier 2002, de Bello et al.

2010, Flynn et al. 2011, Schmitz et al. 2015), which are characteristics of an organism, expressed as phenotypes, that represent how the organism interacts with its environment (Dı́az and Cabido

2001, Violle et al. 2007). Describing an organism’s functional traits and the associated effects on ecosystem function reveals the functional role an organism plays in its ecosystem (Wallace and

Webster 1996, Rusek 1998, Thrush et al. 2006, Cardinale et al. 2011). Motivated by this framework, there has been great interest in describing functional diversity of a system (i.e., FT richness; Tilman 2001) rather than focusing on biodiversity (i.e., species richness) alone in studies of ecosystem resiliency (Walker 1992, Tilman et al. 1997, Loreau et al. 2001, Petchey and Gaston 2006).

Lotic ecosystems (i.e. streams and rivers) are hotspots of biogeochemical activity due to nutrient loading in the riparian zone (McClain et al. 2003, Vidon et al. 2010), hyporheic exchange of solutes to geochemically and microbially-active sites (Boulton et al. 1998, Lautz and Fanelli 2008) and biotic and abiotic controls of nutrient retention time (Ensign and Doyle

2005, Grimm et al. 2005) and spiraling lengths (Newbold et al. 1981, Ensign and Doyle 2006).

Within the biotic realm of a stream, the coexistence of various groups of aquatic animals that process organic matter in different ways have differing effects on the recycling/repacking dynamics of carbon (C), nitrogen (N) and phosphorus (P) (Grimm 1988, Wallace and Webster

1996, Covich et al. 1999). When density and diversity of aquatic animals are high, functional traits in the community can significantly influence biogeochemical cycling on the ecosystem-

13 scale (Cardinale et al. 2002, Covich et al. 2004, Vaughn 2010). Understanding the functional traits present in an ecological community, their role in provisioning ecosystem function and the persistence of their role over time, we can assess the importance of aquatic fauna on overall ecosystem resiliency (Atkinson et al. 2018). Lotic systems have been modified and degraded globally (Carpenter et al. 1998, Bunn and Arthington 2002, Poff et al. 2002, Blann et al. 2009), and many species have been lost (Richter et al. 1997, Ricciardi and Rasmussen 1999, Strayer and

Dudgeon 2010). Many functional traits provision ecosystem functions that directly or indirectly benefit humans (i.e., ecosystem services; Wilson and Carpenter 1999), and when aquatic species are lost, the functional traits they possess are lost as well (Vaughn 2010, Estes et al. 2011,

Hooper et al. 2012). If we are to understand the cost of species loss, it is in our best interest to quantify the functional role of aquatic fauna and the spatial/temporal significance of their role at the ecosystem-scale.

Functional traits of aquatic fauna influence biogeochemical nutrient cycling directly and indirectly (Vanni 2002, Atkinson et al. 2017). For example, N and P are recycled through ingestion followed by excretion (Vanni and McIntyre 2016) which is a source of bioavailable nutrients for benthic primary producers (Atkinson et al. 2017). N and P can also be repackaged by an animal through secondary production, representing a net sink of nutrients when either assimilated into long-term biomass such as bone or shell (Strayer and Malcom 2007, Vanni et al.

2013) or when biomass is removed from aquatic systems through predation of aquatic animals by terrestrial predators and the emergence of aquatic insects (Nakano and Murakami 2001, Gende et al. 2002, Paetzold et al. 2005). Additionally, biodeposition (i.e., egestion) by aquatic animals, while considered a source of C to the system, also acts as a relatively long-term sink of N and P

(Halvorson et al. 2017). It is recognized that aquatic fauna also indirectly influence

14 biogeochemical nutrient cycling (Vanni 2002, Atkinson et al. 2017). For example, predators not only directly control prey populations, but can also indirectly influence the feeding behavior and excretion and/or tissue stoichiometry of prey through the “ecology of fear”, thus altering local source/sink dynamics (Dalton and Flecker 2014). Additionally, aquatic fauna indirectly influence biogeochemical cycles by physically (Mermillod-Blondin and Rosenberg 2006, Mermillod-

Blondin 2011) and chemically (Matisoff and Wang 1998) altering their environment. This is particularly conspicuous in the benthic N-cycle, where biogeochemical transformations are dominated by redox gradients in the vertical sediment profile (Berner 1981). Multiple groups of aquatic fauna coexist in the benthic zone, and those that burrow vertically into the sediment profile (e.g., bivalves, chironomid larvae, tubificid worms) have been shown to significantly influence redox reactions (Anschutz et al. 2012, Hölker et al. 2015, Turek and Hoellein 2015,

Benelli et al. 2017, Hoellein et al. 2017).

One benthic faunal group known to have direct and indirect effects on ecosystem function are the freshwater mussels (Bivalvia: Unionidae). Freshwater mussels (hereafter mussels) are burrowing, filter-feeding bivalves that represent a link between the pelagic and benthic compartments of a stream as they filter material from the water column and concentrate it in the sediment through production, excretion and biodeposition (Vaughn et al. 2008). By fostering benthic-pelagic coupling, mussels directly concentrate C, N and P in the benthos.

Highly reactive forms of N and P contained in mussel excreta can alleviate nutrient limitation and stimulate biofilm growth (Atkinson et al. 2013, Atkinson et al. 2014). This leads to further indirect effects on biogeochemical cycling by increasing primary production in the benthos and abundance of biofilm-grazing chironomid larvae (Allen et al. 2012, Spooner et al. 2012), which has implications for food web complexity and overall species- and functional-richness. Mussels

15 are ideal model organisms with which to assess the indirect influence of functional traits on biogeochemical nutrient cycling, as they are long-lived (6-100 years), sessile organisms that can dominate the benthic biomass of rivers (Vaughn and Hakenkamp 2001, Vaughn 2017). Mussel aggregations are patchy throughout a watershed, and patches are often dense and speciose, resulting in mussel beds being classified as hotspots for such ecosystem functions as benthic- pelagic coupling (Spooner et al. 2012, Atkinson and Vaughn 2015, Vaughn et al. 2015), alleviation of nutrient limitation (Vaughn et al. 2007, Atkinson et al. 2013) and stimulation of local primary production (Spooner and Vaughn 2006, Atkinson et al. 2013). These effects are largely influenced by two mussel functional traits: excretion and biodeposition (Vaughn and

Hakenkamp 2001, Vaughn 2017).

A less understood indirect ecosystem function provisioned by mussels is their influence on the benthic N-cycle, specifically the removal of dissolved inorganic nitrogen (DIN) via biogeochemical pathways (hereafter N-removal) in stream sediments (Vaughn 2017). Research has shown that native freshwater mussels stimulate significant levels of denitrification (hereafter

DNF) in stream sediments (Benelli et al. 2017, Hoellein et al. 2017, Trentman et al. 2018), but the specific mechanisms are poorly understood. Our research aimed to examine the link between mussel functional traits (excretion, biodeposition) and N-removal via the biogeochemical pathways of DNF and anaerobic ammonium oxidation (hereafter annamox). DNF and annamox refer to microbial metabolisms in which facultative anaerobes (chemolithoheterotrophs) convert

DIN to di-nitrogen gas (N2), which then diffuses through the water column and is released into

- the atmosphere (Knowles 1982). In DNF, two molecules of nitrate (NO3 ) are converted to N2 through a series of intermediate N-species in the absence of oxygen. In annamox, one molecule

- - + of nitrite (NO2 , derived from NO3 ) combines with one molecule of ammonium (NH4 ) through

16 a series of intermediate N-species to form N2 in the absence of oxygen. As a result, DNF and annamox represent the permanent removal of N from an aquatic system. We hypothesized that mussels indirectly stimulate N-removal in stream sediments by physically and chemically altering the benthic environment, and a mussel’s influence on N-removal is contingent on an individual’s excretion and biodeposition rates. N-removal in sediments requires three conditions to foster the microbial metabolisms that represent the DNF and annamox pathways: 1) anoxia, 2) excess reactive N as an electron acceptor and 3) energy substrate as an electron donor (Knowles

1982). We predicted mussel functional traits serve to foster the redox conditions necessary to

+ stimulate biogeochemical N-removal, specifically by excreting NH4 and depositing organic

+ matter (OM), at or near the oxic/anoxic boundary in the sediment. We quantified rates of NH4 excretion and OM biodeposition of three common mussel species in a lowland river in the southeastern U.S. and modelled the influence of the two functional traits on ambient N2 flux across the sediment-water interface and maximum DNF and annamox potentials present in the

+ sediment. We predicted NH4 excretion by a mussel would increase N-removal by introducing a highly reactive form of N to anoxic sites in the sediment, and OM biodeposition would increase

N-removal by providing an energy substrate for the DNF pathway in the form of labile C, or by creating anoxic microenvironments as a result of decomposition of OM or through the cohesive nature of egesta-associated mucus.

Methods and Materials

Experimental Design

We collected mussels, sediment and site-water from the Sipsey River, a fifth-order tributary of the Tombigbee River in the Mobile Basin of Alabama. The main stem of the Sipsey

River flows relatively undisturbed for 180 km, and is characterized as harboring remarkably

17 intact and diverse biological communities (McGregor and O'Neil 1992), including 37 of 41 historically occurring unionid mussel species (Haag and Warren 2010). The undisturbed nature of the river and high native mussel populations throughout make the Sipsey River a good model system for studying ecosystem functions provisioned by mussels.

Three incubations trials took place during consecutive weeks in August 2017. Each incubation trial took approximately five days (Table 1). We used three common species from the

Sipsey River that varied in life history and shell morphology (Williams et al. 2008). Cyclonaias asperata (common name: Alabama orb) belongs to the tribe Quadrulini, is a short-term brooder

(early-spring to mid-summer) and has a rough shell morphology containing many prominent pustules. Fusconaia flava (common name: Gulf pigtoe) belongs to the tribe Pleurobemini, is also short-term brooder (late-spring through summer) and has a satiny-textured shell. Lampsilis ornata (common name: southern pocketbook) belongs to the tribe Lampsilini, is a long-term brooder (August to June of following year) and has a large, smooth shell.

We could only incubate nine microcosms at a time, so the experiment was set-up as a balanced incomplete block design, where each incubation trial consisted of two species treatments and a sediment-only control, all replicated three times (Table 1). Our design allowed us to examine all species combinations with a control, minimizing the risk of pseudoreplication.

A total of six replicate individuals per species treatment were used in the experiment. To reduce variation between trials, we took care to ensure the time between different stages of an incubation matched as closely as possible between the three trials (Table 2.1).

Sediment collection and microcosm preparation

We collected sediment from the Sipsey River 48-hrs before the beginning of an incubation and sieved the sediment into buckets to remove any mussels, as well as large cobble

18 and gravel (≥ 11 mm). We removed large particles for two reasons: 1) smaller particle size in the confined area of the microcosm would still allow the mussel to move and burrow, 2) small particle size maximizes the available sediment surface area to foster microbial growth. We then returned to the lab where we homogenized the sediment and filled ten cylindrical microcosms (9- cm x 30-cm) to a depth of 20-cm. Once filled, the microcosms were fully submerged in oxygenated site water set to site-water temperature (trial 1: 27.3 °C, trial 2: 24.3 °C, trial 3: 26.1

°C). The microcosms sat submerged and in the dark for 24-hrs prior to transport and another 24- hrs once transported to the incubation chamber (Table 2.1). Allowing the microcosms to sit submerged and in the dark for 48-hrs aided in the reestablishment of a redox gradient in the homogenized sediment.

Mussel collection and excretion/biodeposition experiments

One day prior to beginning an incubation trial, we collected mussels that were relatively the same size (shell lengths: F. flava = 48-57 mm, C. asperata = 46-59 mm, L. ornata = 72-86 mm), and that were small enough to fit comfortably in the microcosm environment. We collected mussels from the same reach as the sediment, and collected them in areas 0.5 to 0.7-m deep, just upstream of an established long-term mussel survey site (site 6 in Haag and Warren 2010). F. flava and C. asperata were collected from the near the center of the reach, while L. ornata was collected closer to the bank, which is each species’ preferred habitat (Z.L. Nickerson, personal observation). We measured the shell lengths (mm) for each mussel and lengths were fit into a length-weight regression to estimate soft tissue dry mass in grams (C.L. Atkinson, unpublished data). After the mussels were collected and measured, we characterized mussel excretion and biodeposition with in-field experiments as described in Atkinson et al. (2013). Briefly, nine 1-L chambers were filled with filtered site water (GF/F filter, 0.7 μm pore size, Merck Millipore,

Burlington, MA). Each mussel (n = 3 species-1) was gently scrubbed to remove algae and sediment attached to the shell and placed in a chamber. Cleaned, empty mussel shells collected from the site were used in three control chambers to account for nutrients and solids in the chamber produced by shell-associated algae and microbes. The chambers were sealed and partially submerged in site water to maintain temperature. After an hour, the mussels and control shells were carefully removed. Following the experiment, the mussels were wrapped in damp towels and placed in a cooler for transport. The chambers were immediately placed on ice and brought to the lab where we filtered the contents through a GF/F filter (0.7 μm pore size). The filters and water were promptly frozen until analysis.

Solid material collected on the filter was used to characterize mussel biodeposition (feces

+ pseudofeces). The filters were dried for 48-hrs at 50 °C, weighed on an analytical balance (±

0.0001 g), combusted at 500 °C for 2-hrs and weighed again to determine ash-free dry mass

(AFDM), which was used to quantify the organic matter (OM) content of the biodeposits (mg

-1 + OM h ). We analyzed the filtered chamber water for ammonium (NH4 ) concentration using a

+ Lachat QuickChem flow injection analyzer (Hach Company, Loveland, Colorado, USA). NH4

+ concentrations in the control chambers were averaged and subtracted from the NH4

+ concentrations in each mussel chambers to estimate each mussel’s NH4 excretion rate, reported

+ -1 in moles (µmol NH4 h ).

Chamber incubations

Immediately following the mussel collection and in-field excretion/biodeposition experiment, we capped the microcosms air-tight with a rubber stopper, sealed them with duct tape, placed them upright in large coolers, filled the space in between the microcosms with ice and transported the coolers from the University of Alabama (UA) to Dauphin Island Sea Lab

(DISL) where the incubation experiments took place. The mussels were wrapped in a damp towel and placed in a small cooler to reduce stress (Cope et al. 2003). Upon arrival to DISL, we unpacked the microcosms and placed them in holding tanks of aerated site water set to site temperature. The acclimation and incubation took place within a temperature-controlled environmental chamber set to site-water temperature. Once cooled to site temperature, each mussel was placed in the center of a microcosm (1 mussel microcosm-1) to allow sufficient time for the mussels acclimate to the experimental conditions and the microcosms sat in the dark for an additional 12-hrs prior to the start of an incubation.

Continuous flow-through incubations (Miller-Way and Twilley 1996) were utilized to assess the influence of native mussels on ambient gas and nutrient flux in benthic sediments

+ (Figure 2.1). For the purpose of this study, we focused on the fluxes of ammonium (NH4 ),

- nitrate (NO3 ), oxygen (O2) and dinitrogen gas (N2). Each microcosm was capped completely underwater to ensure no air was introduced to the headspace as to not introduce error into the gas flux measurements. Each microcosm contained 10-cm of overlying water and 20-cm of sediment. We fitted caps to each microcosm that contained a magnetic stir-bar and two valves to which inflow and outflow tubing were connected. While still submerged, a microcosm was placed in a rack which was used to keep the cap securely tightened to the microcosm throughout the incubation trial (Figure 2.1), and each capped microcosm was fitted with inflow and outflow tubing to the valves on the cap.

Once the cap, stand and tubes were fitted, the microcosms were removed from the water bath and placed on a rack in the environmental chamber. The inflow for all nine microcosms was supplied from one 50-L reservoir of aerated, unfiltered site water. Each inflow tubing was connected to a peristatic pump, and the flow for all trials was set at 1.5-mL min-1. As per the

21 design of the experiment, unfiltered site water was constantly flowing into the microcosm, and outflow water flowed into a reservoir. An additional length of tubing was installed to flow directly from the inflow reservoir to the waste reservoir and was used to assess the concentrations of the analytes in the inflow reservoir. The final step was to fit a stir-bar motor into the rack atop each microcosm. The motor fit into the top of the four-pronged rack, and sat atop two plexiglass bars, which were tightened onto the cap (Figure 2.1). The motor turned a magnetic stir-bar which, in turn, caused the magnetic stir-bar inside the microcosm to rotate and simulate flow of the overlying water in the microcosm. The incubation period began once all the motors were installed and the stir-bars were turning. The lights within the environmental chamber were turned off, and the experiment ran undisturbed for 24-hrs.

After the 24-hr dark incubation, we collected water samples from the inflow tubing and from the outflow for each microcosm. First, the specific flow rate for each microcosm was determined by measuring the volume filled in a 10-mL graduated cylinder over a period of one minute. Next, triplicate water samples were collected in 12-mL exetainers to measure dissolved oxygen (DO) concentration, which were used to estimate total respiration (O2 influx) within a microcosm. The exetainers were filled to 3x their volume before measuring for DO. Samples for

DO were analyzed immediately following collection using a Unisense OX-500 O2 microsensor

(Unisense A/S, Aarhus, Denmark). Following DO sample collection, samples were collected for

N2 flux. Triplicate gas flux samples were collected in 12-mL exetainers. Again, the exetainers were filled 3x their volume. For both DO and N2 sample collection, the exetainers were filled inside a beaker, and the overflow was collected and filtered through a GF/F filter (0.7 μm pore size) into two 50-mL centrifuge tubes. These duplicate samples (~30-mL each) were

+ - - immediately frozen until analysis for NH4 , NO2 and NO3 fluxes. Once filled 3x their volume,

22 the exetainers were treated with 250-µL ZnCl2 (50% w/v) to stop all microbial activity, capped air-tight and stored underwater at incubation temperature until analysis.

Ambient dinitrogen gas (N2) fluxes were analyzed at DISL using the N2:Ar method on a membrane inlet mass spectrometer (MIMS, Kana et al. 1994). The MIMS is a modified form of mass spectrometry that specializes in high-precision analysis of dinitrogen gas concentrations in low-volume environmental water samples (Kana et al. 1994). Nutrient samples were transported frozen back to UA and stored at -20 °C until analysis. Samples were analyzed in November 2017 on a Lachat QuickChem flow injection analyzer (Hach Company, Loveland, Colorado, USA).

+ - - Fluxes of O2, N2, NH4 , NO2 , and NO3 for each microcosm were calculated by subtracting the concentration of each analyte in the inflow from the corresponding concentrations from each microcosm’s outflow. Concentrations (µmol L-1) were normalized by flow rate (~0.1 L h-1) and sediment surface area (0.007 m2) to report areal flux rates (µmol m-2 h-1).

N-removal potential

After water samples were collected for nutrient and gas fluxes, we uncapped the microcosms and gently removed the mussels. The mussels remained in a holding tank until transport back to the Sipsey River. We then collected microcosm sediment for analysis of N- removal potentials. The bottom 10-cm of sediment in each microcosm was discarded, and the top

10-cm (the sediment in which the mussel was active) was collected in a sample bag to be analyzed for N-removal potential using isotope pairing techniques (IPT, Thamdrup and

15 - Dalsgaard 2002). In IPT, isotopically-labeled nitrate (98% NO3 ) is introduced to an anaerobic slurry of sediment and incubated for a set amount of time. Subsequent analysis of the

29 30 fractionation of N2 and N2 in the water overlying the slurry determines the maximum potential rates of annamox and DNF, respectively, via the following biogeochemical pathways:

15 − 30 2 푁푂3 → 푁2

15 − 14 + 29 푁푂3 + 푁퐻4 → 푁2

The first equation represents the DNF pathway, where two molecules of the isotopically-labelled

15 - NO3 combine, through a series of intermediates, to form dinitrogen gas with a labelled molar mass of 30N. The second equation represents the annamox pathway, where one molecule of the

15 - isotopically-labeled NO3 combines, through a series of intermediates, with a naturally-present

14 + 29 ammonium molecule ( NH4 ) to form dinitrogen gas with a labelled molar mass of N.

To prepare the slurries, the top 10-cm of sediment from a microcosm was wet-sieved through a 2-mm sieve. All particles ≥ 2-mm were discarded, and only particles < 2-mm were used for IPT slurries. The < 2-mm sediment was homogenized and subsampled into six 12-mL exetainers. Each exetainer received approximately 2-g of wet sediment. The exetainers were then

- filled with anoxic NO3 -free artificial river water (ARW; Appendix I). The ARW was formulated

- to contain the major ions of Sipsey River water, minus NO3 (McGregor and O'Neil 1992), and was synthesized using methods described by Smith et al. (2002). Prior to adding the ARW to the exetainers, we made the solution anoxic by bubbling it with N2 gas for 1-hr. We filled an exetainer with ARW until a meniscus formed, then capped the exetainer air-tight, inverted to inspect for air bubbles, and placed on a shaker table overnight. The following day, the exetainers were removed from the shaker and placed upright to allow the sediment within the slurry to settle to the bottom of the exetainer.

- As the slurries settled, we prepared the labelled NO3 solution by mixing 0.8-mL of a

15 - 232-mM solution of 98% labelled NO3 in 50-mL of ultrapure water, resulting in a

15 - 15 - concentration of [ NO3 ] = 3.712-mM. We added 162-μL of the NO3 solution into each

15 - exetainer, resulting in a concentration of [ NO3 ] = 50-µM in each exetainer. We then

24 immediately added 250-μL solution of ZnCl2 (50% w/v) to three of the six replicate exetainers, ceasing all microbial activity and representing the concentration of N2 at time zero (T0). The exetainers were recapped air-tight and were inverted and inspected closely to ensure there were no air bubbles in the slurry. The exetainers were placed back on the shaker table, and incubated for 6-hrs. After 6-hrs, we removed the exetainers from the shaker table and set them upright to settle. Once settled, the remaining three replicate slurries were spiked with 250-μL of ZnCl2, ceasing all microbial activity and representing the concentration of dinitrogen gas at time final

(T6). The exetainers were recapped and placed back on the shaker table to shake overnight until

29 30 analysis of N2 and N2 using the MIMS. Following analysis on the MIMS, we returned to UA with the sediment. The sediment was dried at 50 °C for 48-hrs and weighed on an analytical

29 30 -1 balance to determine dry mass. Concentrations of N2-N and N2-N (µmol L ) were normalized by slurry incubation time (6 h) and sediment dry mass (kg) to report maximum N-removal

-1 -1 potential as an hourly rate per kg of sediment (µmol N2-N kg h ). Annamox values were low, and some nearly equivalent to the baseline fluctuation (noise) of the MIMS signal. Annamox potential values that had a MIMS signal < 2x the value of noise were considered to be “0”.

Statistical analyses

+ The functional traits of interest in our experiment were NH4 excretion and OM biodeposition, which correspond to our hypotheses that mussels stimulate N-removal by adding excess reactive N and provide labile substrate to foster microbial growth, respectively (see

Figure 1.1a). We used 1-way analysis of variance (ANOVA) to determine if soft tissue DM,

+ NH4 excretion and OM biodeposition differed amongst our species treatments. We also used 1-

+ - way ANOVA to determine if the fluxes of NH4 , NO3 , N2-N and O2, as well as DNF and annamox potentials, differed amongst our species treatments and if species treatments differed

25 from the sediment-only controls. Because we used a balanced incomplete block design experimental design, our ANOVAs were conducted taking into account the variation due to the random effect of incubation trial (block effect) using the R package multcomp version 1.4-8.

Significant ANOVAs were followed by Tukey’s Honestly Significant Difference (HSD) multiple comparisons (α = 0.05) and also took into account the random effect of incubation trial by running the posthoc analyses in the multcomp package using the ghlt() function.

We examined the linear relationship between out N-removal measurements (ambient N2-

+ N flux and DNF and annamox potentials) and mussel soft tissue DM, respiration rates, NH4 excretion and OM biodeposition using simple linear regressions (SLRs). Prior to running SLRs, we conducted an analysis of covariance (ANCOVA) to see if there were species-specific differences within the N-removal variables. Species was the explanatory variable in the

ANCOVA model, and soft tissue DM was the covariate, as it is biologically relevant that larger

+ individuals will be excreting more NH4 and depositing more OM. To test the influence of

+ mussel functional traits (NH4 excretion, OM biodeposition) on N-removal processes, we first needed to address a caveat that goes along with using the microcosm approach with large heterotrophic infauna such as mussels. Due to the relatively small headspace of water (~640 cm3) and slow flow-rate (~1.0 L h-1), N-removal could be heavily influenced by respiration within our experimental system as N-removal occurs under anoxic conditions. A net influx of O2 resulted in an increasingly anoxic environment, which is not biologically relevant as individual mussels likely do not significantly alter O2 concentrations in field conditions. As a system becomes increasingly anoxic, there will inherently be more DNF, as anoxic microenvironments will favor the growth of denitrifying microbes (Nielsen 1992, Rysgaard et al. 1994). Therefore, to assess the influence of mussel functional traits on N-removal, we needed to remove the effect of

26 respiration from the analyses. This was only done for the ambient N2-N flux, as we did not see a strong effect of respiration on DNF and annamox potentials. First, we estimated mussel respiration by subtracting the mean O2 influx of the three sediment-only control microcosms

(respiration due to heterotrophic microbes and meiofauna) in an incubation trial from each mussel microcosms in that same trial. The same was done for N2-N efflux of mussel microcosms by subtracting the mean N2-N efflux of the controls to correct for DNF occurring outside of the presence of a mussel. Second, we fit the control-corrected respiration and DNF estimates into a

SLR with N2-N efflux as the response variable and mussel respiration as the explanatory variable. Then, to remove the effect of respiration from the analysis, the residuals from the above

SLR analysis were used as a response variable to examine the role of specific mussel functional traits on N-removal. To test for outliers among the residuals, we conducted a Bonferroni- adjusted outlier test using the R package car version 2.1-3. If outliers were detected (Bonferroni p < 0.05), they were removed from subsequent analyses. Finally, after correcting for outliers, the

+ residual values were treated as the response variable in SLRs to test for the effects of NH4 excretion and OM biodeposition on ambient N2-N efflux. All analyses were conducted using R version 3.3.1 (R Core Team 2016).

Results

Mussel parameters and functional traits

L. ornata individuals were significantly larger and their resultant soft tissue biomass was greater (F2,13 = 19.28, p < 0.001, Table 2.2, Figure 2.2a) than F. flava and C. asperata. There

+ was a significant difference in NH4 excretion rates among species (F2,19 = 4.47, p = 0.033, Table

+ 2.2, Figure 2.2b) with L. ornata having the highest NH4 excretion rate. Although F. flava contained less soft tissue dry mass than C. asperata (Figure 2.2a) on average, the mean excretion

27 rate of F. flava was higher than that of C. asperata (Figure 2.2b). Additionally, L. ornata deposited significantly more OM as feces and pseudofeces than F. flava and C. asperata (F2,19 =

+ 4.12, p = 0.043, Table 2.2, Figure 2.2c). Unlike NH4 excretion, OM biodeposition followed the trend of heaviest to lightest species with L. ornata depositing the most, followed by C. asperata and F. flava, respectively (Figure 2.2c).

Ambient gas and nutrient flux

The 24-hr chamber incubation allowed us to measure small-scale ambient gas (N2, O2)

+ - and nutrient (NH4 , NO3 ) flux across the sediment-water interface, which were calculated as

µmol m-2 h-1 (Table 2.3). Before taking water samples at the 24-hr mark, we closely examined each microcosm for the presence of bubbles, which would introduce bias into such fine-scale gas flux measurements. Two microcosms, both controls (one from trial 1, one from trial 3), contained microbubbles and were not sampled for ambient fluxes. Our results showed a

+ significant treatment effect of NH4 flux (F3,19 = 3.42, p = 0.038, Table 2.3), but a Tukey’s HSD posthoc analysis showed no individual differences between treatments, and no treatment was significantly different from the control (Figure 2.3a, light grey bars). The control treatments

+ showed a slight average influx of NH4 into the sediment. L. ornata treatments showed a larger

+ average influx of NH4 into the sediment, while F. flava and C. asperata showed an average

- efflux (Figure 2.3). In terms of NO3 flux, there was a significant treatment effect (F3,19 = 4.03, p

= 0.022, Table 2.3), but our posthoc test indicated none of the mussel treatments were

+ significantly different from the control (Figure 2.3a, dark grey bars). Dissimilar to NH4 flux, the

- - control treatments showed an average efflux of NO3 . NO3 flux among mussel treatments

+ showed a similar trend to that of NH4 flux in that L. ornata treatments showed an average influx

- of NO3 , while F. flava and C. asperata showed an average efflux. There was a significant

28 treatment effect on the efflux of N2 gas (F3,19 = 74.73, p << 0.001, Table 2.3) with all mussel treatments showing significantly higher rates of N2 efflux than the controls, and L. ornata showing significantly higher rates than F. flava and C. asperata (Figure 2.3b, dark grey bars). A similar pattern was observed for O2 influx into the sediment (negative flux), with there being a significant treatment effect (F3,19 = 252.3, p << 0.001, Table 2.3) and all mussel treatments showing significantly higher rates of O2 influx than the controls, as well as L. ornata showing significantly higher rates than F. flava and C. asperata (Figure 2.3b, light grey bars).

N-removal potentials

Using the IPT method, we quantified N-removal potentials via two biogeochemical pathways, DNF and annamox (Table 2.3). The two control microcosms that were not sampled for ambient nutrient and gas flux due to the presence of bubbles were included in IPT experiments, as we assumed microbubbles would not significantly influence N-removal potentials in the sediment. Results showed a significant treatment effect on N-removal potential via the DNF pathway (F3,23 = 5.63, p = 0.005, Table 2.3). L. ornata treatments were shown to foster significantly higher average DNF potentials than the controls and F. flava and C. asperata treatments, but F. flava and C. asperata treatments were not significantly different from the controls (Figure 2.3c, dark grey bars). L. ornata treatments showed the highest average annamox potentials and was greater than the control, while mean annamox potentials for F. flava and C. asperata were lower than that of the control treatments (Figure 2.3c, light grey bars). There was not a significant treatment effect for annamox potentials (F3,23 = 2.33, p > 0.1, Table 2.3).

Modelling influence of functional traits on N-removal

The ANCOVA showed soft tissue DM to be a significant indicator of an individual mussel’s influence on ambient N2 flux (F1,14 = 10.35, p = 0.006), DNF potential (F1,14 = 11.46, p

= 0.004) and annamox potential (F1,14 = 4.96, p = 0.043). However, when corrected for soft tissue DM, there were no species-specific differences in ambient N2 flux (F2,14 = 1.66, p > 0.2),

DNF potential (F1,14 = 1.62, p > 0.2) or annamox potential (F1,14 = 0.569, p > 0.5). Thus, all

SLRs were conducted using all mussels in a single data set without taking into effect species- specific relationships. There was a significant positive relationship between soft tissue DM and

2 2 ambient N2 flux (R = 0.335, p = 0.007, Figure 2.4a,i), DNF potential (R = 0.362, p = 0.005,

Figure 2.4b,i) and annamox potential (R2 = 0.120, p = 0.036, Figure 2.4c,i). Additionally, estimated mussel respiration (O2 influx of mussel microcosm – mean O2 influx of control

2 microcosms of the same trial) significantly influenced ambient N2 flux (R = 0.756, p << 0.001,

Figure 2.4a,ii). Respiration was also shown to have a significant relationship with DNF potential

(R2 = 0.207, p = 0.033, Figure 2.4b,ii), but no such relationship existed between respiration and annamox potential (R2 = 0.078, p > 0.1, Figure 2.4c,ii). In terms of the functional traits of

+ interest, results showed a significant positive relationship between NH4 excretion and ambient

2 + N2 flux (R = 0.320, p = 0.008, Figure 2.4a,iii), but no such relationship existed between NH4 excretion and DNF potential (R2 = 0.116, p = 0.09, Figure 2.4b,iii) or annamox potential (R2 =

0.018, p > 0.2, Figure 2.4c,iii). However, results suggest OM biodeposition to be a significant indicator of a mussel’s influence on all N-removal variables, showing significant positive

2 2 relationships with ambient N2 flux (R = 0.592, p < 0.001, Figure 2.4b,iv), DNF potential (R =

0.386, p = 0.005, Figure 2.4b,iv) and annamox potential (R2 = 0.229, p = 0.030, Figure 2.4c,iv).

One data point from the ambient DNF vs. respiration SLR (L. ornata datum; residual variation = -391.88) was determined to be a significant outlier (Bonferroni p < 0.05) and was removed from the following analyses assessing the influence of functional traits on ambient DNF

(Figure 2.5a; note n = 5 for L. ornata). After accounting for respiration in the incubations

+ (~24.4% residual variation remaining in data), NH4 excretion was a significant predictor of

2 + ambient DNF (R = 0.263, p = 0.020, Figure 2.5b), suggesting that NH4 excretion explained a significant amount of the remaining variation in the data. Additionally, results showed a nearly significant positive relationship between OM biodeposition and ambient DNF after accounting for respiration (R2 = 0.187, p = 0.053, Figure 2.5c), suggesting a moderate influence of OM biodeposition on the remaining variation in the data.

Discussion

Our results provide evidence that, within our experimental system, the presence of mussels in riverine sediment stimulated permanent N-removal, and the influence of mussels on

N-removal is biomass-associated rather than species-specific. Additionally, our results describe an additional functional role of mussel physiological traits in the benthos, as the products of excretion and biodeposition were shown to chemically and physically alter sediment and foster

N-removal via the biogeochemical pathways of DNF and annamox. These functional traits have been shown to have significant effects on ecosystem functions such as nutrient limitation

(Vaughn et al. 2007, Atkinson et al. 2013), primary production (Spooner and Vaughn 2006,

Atkinson et al. 2013) and benthic biodiversity (Vaughn et al. 2008, Spooner et al. 2012). Here, our results show that these functional traits also have a significant effect on the ecosystem

+ function of biogeochemical N-removal. We showed that excreta-derived NH4 serves as a bioavailable N source needed for microbially-mediated efflux of N2 across the sediment-water interface, and OM biodeposition plays a significant role in the maximum N-removal potential present in the sediment. Previous studies have shown native freshwater mussels to stimulate N- removal (Benelli et al. 2017, Hoellein et al. 2017, Trentman et al. 2018), but our study is the first to link this stimulation to specific mussel functional traits.

+ The role of NH4 excretion in N-removal

One effect trait we hypothesized would stimulate N-removal was providing a highly

+ reactive form of N through NH4 excretion. We found that, even when accounting for 76% of the

+ variation in the data by removing the effect of respiration, NH4 excretion remained a significant influence on ambient N-removal (DNF + annamox), explaining roughly 26% of the remaining variation in the data. This represents strong evidence that excreta-derived N can be a prominent reactant in N-removal pathways in freshwater sediments containing mussels. Because the incubations ran under dark conditions, we can rule out the possibility of N uptake by autotrophic

+ microbes and assume the fate of excreta-derived NH4 to be uptake by heterotrophic microbes including those that facilitate N-removal. If excreta-derived N underwent the DNF pathway, it would first need to be nitrified before being denitrified. If only nitrification was occurring in our

- system, we would have expected to see higher rates of NO3 efflux in the mussel treatments

+ relative to the controls. Though containing large, NH4 excreting infauna, the three mussel

+ - species treatments did not show significantly different NH4 nor NO3 fluxes relative to the controls. This suggests that excreta-derived N was taken-up into the sediment by nitrifying

- microbes, and the NO3 produced was quickly denitrified, indicating the presence of coupled nitrification-denitrification (hereafter coupled-DNF).

Our results concerning the fate of excreta-derived N contrast with those found in a similar microcosm study by Hoellein et al. (2017) in which the authors observed significant increases in

+ NH4 efflux in mussel treatments relative to sediment-only controls under ambient conditions.

The authors’ study was conducted in an urban stream in the midwestern U.S. with much higher background dissolved inorganic nitrogen (DIN) concentrations than that of the Sipsey River.

With the low background DIN in our study system (McGregor and O'Neil 1992), highly reactive

N from mussel excreta is readily taken-up by microbially-active sediment, even in the absence of autotrophy. The stark differences among our results support the claim that mussel aggregations act as biogeochemical hotspots for alleviating N limitation when DIN is low (Atkinson and

Vaughn 2015), and add to that claim by showing excreta-derived N not only stimulates the growth of primary producers (Spooner and Vaughn 2006, Atkinson et al. 2013), but denitrifying microbes as well.

+ NH4 excretion did not influence DNF potentials nor annamox potentials. This result is not surprising, as in the IPT experiments the sediment slurries were incubated with anoxic ARW

- + on a shaker table for 18-hrs. We assume this resulted in all NO3 and NH4 present in the slurry at the beginning of the IPT experiments to be lost from the system via N-removal processes once uncapped. This would effectively remove any trace of excreta-derived N from the sediment

15 - before we spiked the slurries with isotopically-labelled NO3 .

The role of OM biodeposition in N-removal

We observed significant positive relationships between OM biodeposition and both DNF and annamox potentials. The IPT methodology maximized two of the chemical requirements to

15 - stimulate N-removal by introducing excess reactive N (spiking with NO3 ) to a completely anoxic environment (removing all O2 by bubbling ARW with N2 gas prior to start). When mussel biodeposits were added to this extreme environment, we saw an increase in the maximum N- removal potential via both DNF and annamox pathways. We predicted mussel biodeposits could stimulate N-removal by both chemically and physically altering the benthic environment. OM biodeposition introduces an energy substrate (electron donor) for the DNF pathway in the form of labile C (Grenz et al. 1990, Mirto et al. 2000). Due to the method we used to collect mussel biodeposit, we were not able to differentiate between feces and pseudofeces. Feces is the egesta

33 product of mussels, while pseudofeces is a conglomerate of rejected material captured while filter-feeding (Nichols et al. 2005). Mussel feces is much richer in C, N and P than pseudofeces

(Hoellein et al. 2017), and therefore we expected most of the labile OM we measured in the biodeposits were a product of feces. For OM to influence the DNF pathway, feces would need to be deposited in or near, or transported to, the oxic/anoxic redox boundary where denitrifying microbes can utilize the labile C. Due to the IPT methodology, specifically shaking the sediment and ARW to create a slurry, biodeposits were well-mixed with sediment in an anoxic

15 - environment and, when alleviated with NO3 , there was a significant positive relationship between mussel OM biodeposition rate and the maximum DNF potential in the sediment. This result suggests OM derived from mussel feces chemically alters the benthic environment by providing the energy substrate necessary to stimulate N-removal via the DNF pathway.

We observed a significant positive relationship between mussel OM biodeposition rate and maximum annamox potential in the sediment. As annamox does not require C as an energy substrate as does DNF, this result suggests there could be additional mechanisms by which OM biodeposition could stimulate N-removal, potentially by creating anoxic microenvironments in the sediment via decomposition. Studies of high density marine mussel farms have shown the decomposition of biodeposits to significantly decrease DO in the sediment (Nizzoli et al. 2005,

Carlsson et al. 2010). The population densities of the mussel farms studies were similar to those currently found in natural mussel aggregations in the Sipsey River. Though OM decomposition will occur at different rates in marine and freshwater systems, the evidence of bivalve biodeposits decreasing sediment DO is likely as decomposition is an oxygen-consuming process.

Massive amounts of biodeposit-derived OM decomposition in natural mussel aggregations could satisfy two of the three requirements for N-removal in sediments by both making the

34 environment more anoxic and providing labile C. Although we were not able to elucidate the mechanism by which OM biodeposition stimulates N-removal, our results show the added material to be an important component of overall N-removal potential in freshwater sediments.

OM biodeposition seemed to have only a moderate influence on ambient N2 efflux

(DNF) during the 24-hr chamber incubation. In Carlsson et al. (2010), there was not a measurable amount of biodeposit decomposition (indication of microbial colonization) until >

88-hrs post-deposition. The weak influence of mussel OM biodeposition on ambient DNF in our chamber incubations could be explained by the fact that the 24-hr incubation time did not allow sufficient time for microbial colonization and subsequent utilization and decomposition of mussel feces and pseudofeces. Also, unlike the extreme IPT microenvironment, there may have not been enough available N nor enough anoxic sites in the sediment to foster the necessary colonization of DNF and annamox microbes on mussel biodeposits. Our results suggest that OM biodeposits act as a more long-term influence on sediment biogeochemical processes, whereas highly reactive excreta-derived N stimulates N-removal on much shorter time scales.

Strengths and weaknesses of the microcosm design

It is important to recognize that our experimental design deviates greatly from the natural environment in which mussels thrive. In addition to the respiration caveat previously addressed, we recognized other issues with the microcosm design when studying large benthic infauna such as mussels. A review of the feeding behavior of mussels reported that an adult mussel roughly

61-mm in length can filter water at a rate of ~0.5 to 1-L h-1 (Vaughn et al. 2008, and references within). This estimated filtration-rate is roughly equivalent the flow-rate we used in our continuous-flow design. Additionally, it has been shown that riverine mussels will gape

(indication of filtering activity) roughly 50% of the time under dark laboratory conditions (Chen

1998). If we assume our mussels were actively filtering for 50% of the time during the incubation, and the mussel filtration-rate was roughly 50-100% of the flow-rate within the

+ - chamber, the excreta-derived N, be it in the form of NH4 or NO3 , will have a much longer residence time in a microcosm than it would naturally in the Sipsey River, where reach-scale discharge at the field site is roughly 3x104 L s-1 during baseflow conditions (C.L. Atkinson, unpublished data). Therefore, excreta-derived N from mussels with fully exposed siphons was excreted directly to the overlying water, but not lost “downstream” as quickly as would occur in the natural stream environment, which increased localized DNF beyond what would be expected in the mussel’s natural habitat.

Another component of natural mussel habitat missing from the microcosm design was interstitial flow, which is important as mussels in the Sipsey River will often completely bury themselves. We were not able to quantify how far each mussel buried itself into the sediment, but we did note that no mussel was completely buried, but most F. flava and C. asperata individuals buried to where only their siphons were exposed. Most L. ornata individuals had > 50% of their shell exposed, likely due to their size relative to the sediment surface area keeping them from

+ being able to burrow within the microcosm. We therefore assume NH4 for all mussels was excreted directly into the overlying water and diffused into the sediment to undergo coupled-

DNF. Not much is known about the feeding behavior of buried mussels, but we can assume that unless mussels completely cease metabolism when buried, they must continue to filter-feed, excrete, and produce biodeposits. In a stream, the flow of interstitial water is slower than that of the overlying water column and, therefore, the spiraling length of excreta-derived N from a completely buried mussel would be shorter, potentially increasing local nitrification and coupled-

DNF. Our results support this notion by showing that when excreta-derived N was concentrated

36 locally in the sediment in the absence of light, it significantly increased ambient DNF. As a result, the continuous flow-through design utilized in our chamber incubation experiments could represent a better proxy of what occurs when highly reactive N from mussel excretion is introduced to slow-moving, microbially-active [heterotrophic] interstitial water, rather than directly into the water column. In natural mussel assemblages, especially those found in lowland systems like the Sipsey River where the benthos is characterized as being predominantly composed of gravel and sand, a large proportion of mussels are completely buried beneath the sediment-water interface (Schwalb and Pusch 2007, Allen and Vaughn 2009). In our study system, during quantitative surveys in the summer and fall of 2016, 81% of L. ornata individuals

(n = 222), 52% of F. flava individuals (n = 726) and 51% of C. asperata individuals (n = 1067) were found completely buried (C.L. Atkinson, unpublished data). Though the issue of interstitial flow cannot be addressed with the continuous-flow through microcosm design, future studies should attempt to optimize the flow-rate through the microcosms to closer mimic mussels’ natural habitat. This would reduce the retention time of N within the system, as well as help minimize the influence of respiration on ambient DNF.

Conclusion

Freshwater mussels are one of the most imperiled faunal groups in the world (Strayer et al. 2004). Mussel conservation efforts are hindered by the overall lack of understanding of mussel ecology, behavior, and the provisioning of ecosystem functions. Our results suggest the

+ functional traits of NH4 excretion and OM biodeposition, while provisioning direct ecosystem functions, also contribute to provisioning indirect ecosystem function by stimulating biogeochemical N-removal in benthic sediments. Our microcosm design also allowed us to observe biogeochemical processes on a small scale, model the influence of individual mussels on

37 biogeochemical processes in freshwater sediment, and add to the growing knowledge of the functional roles of mussels in aquatic ecosystems. With growing recognition of the importance of mussels for aquatic ecosystem health and resiliency (see review in Vaughn 2017 and references within), our research indicates that conservation efforts could benefit from viewing functional traits holistically by not only assessing the direct effects, but also the latent effects that can ripple through the biotic and abiotic realms of an aquatic ecosystem.

References

Allen, D. C., and C. C. Vaughn. 2009. Burrowing behavior of freshwater mussels in experimentally manipulated communities. Journal of the North American Benthological Society 28:93-100.

Allen, D. C., C. C. Vaughn, J. F. Kelly, J. T. Cooper, and M. H. Engel. 2012. Bottom‐up biodiversity effects increase resource subsidy flux between ecosystems. Ecology 93:2165-2174.

Anschutz, P., A. Ciutat, P. Lecroart, M. Gérino, and A. Boudou. 2012. Effects of tubificid worm bioturbation on freshwater sediment biogeochemistry. Aquatic Geochemistry 18:475- 497.

Atkinson, C. L., D. C. Allen, L. Davis, and Z. L. Nickerson. 2018. Incorporating ecogeomorphic feedbacks to better understand resiliency in streams: A review and directions forward. Geomorphology 305:123-140.

Atkinson, C. L., K. A. Capps, A. T. Rugenski, and M. J. Vanni. 2017. Consumer‐driven nutrient dynamics in freshwater ecosystems: From individuals to ecosystems. Biological Reviews 92:2003-2023.

Atkinson, C. L., J. F. Kelly, and C. C. Vaughn. 2014. Tracing consumer-derived nitrogen in riverine food webs. Ecosystems 17:485-496.

Atkinson, C. L., and C. C. Vaughn. 2015. Biogeochemical hotspots: Temporal and spatial scaling of the impact of freshwater mussels on ecosystem function. Freshwater Biology 60:563-574.

Benelli, S., M. Bartoli, E. Racchetti, P. C. Moraes, M. Zilius, I. Lubiene, and E. A. Fano. 2017. Rare but large bivalves alter benthic respiration and nutrient recycling in riverine sediments. Aquatic Ecology 51:1-16.

Berner, R. A. 1981. A new geochemical classification of sedimentary environments. Journal of Sedimentary Research 51.

Blann, K. L., J. L. Anderson, G. R. Sands, and B. Vondracek. 2009. Effects of agricultural drainage on aquatic ecosystems: A review. Critical Reviews in Environmental Science and Technology 39:909-1001.

Boulton, A. J., S. Findlay, P. Marmonier, E. H. Stanley, and H. M. Valett. 1998. The functional significance of the hyporheic zone in streams and rivers. Annual Review of Ecology and Systematics 29:59-81.

Bunn, S. E., and A. H. Arthington. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30:492-507.

Cardinale, B. J., K. L. Matulich, D. U. Hooper, J. E. Byrnes, E. Duffy, L. Gamfeldt, P. Balvanera, M. I. O'connor, and A. Gonzalez. 2011. The functional role of producer diversity in ecosystems. American Journal of Botany 98:572-592.

Cardinale, B. J., M. A. Palmer, and S. L. Collins. 2002. Species diversity enhances ecosystem functioning through interspecific facilitation. Nature 415:426.

Carlsson, M. S., R. N. Glud, and J. K. Petersen. 2010. Degradation of mussel (Mytilus edulis) fecal pellets released from hanging long-lines upon sinking and after settling at the sediment. Canadian Journal of Fisheries and Aquatic Sciences 67:1376-1387.

Chen, Y. 1998. The respiratory physiology and energy metabolism of freshwater mussels and their responses to lack of oxygen. Faculty of the Virginia Polytechnic Institute and State University. PhD dissertation, p 88.

Cope, W. G., T. J. Newton, and C. M. Gatenby. 2003. Review of techniques to prevent introduction of zebra mussels (Dreissena polymorpha) during native mussel (Unionoidea) conservation activities. Journal of Shellfish Research 22:177-184.

Covich, A. P., M. C. Austen, F. Bärlocher, E. Chauvet, B. J. Cardinale, C. L. Biles, P. Inchausti, O. Dangles, M. Solan, and M. O. Gessner. 2004. The role of biodiversity in the functioning of freshwater and marine benthic ecosystems. BioScience 54:767-775.

Covich, A. P., M. A. Palmer, and T. A. Crowl. 1999. The role of benthic invertebrate species in freshwater ecosystems: Zoobenthic species influence energy flows and nutrient cycling. BioScience 49:119-127.

Dalton, C. M., and A. S. Flecker. 2014. Metabolic stoichiometry and the ecology of fear in Trinidadian guppies: Consequences for life histories and stream ecosystems. Oecologia 176:691-701. de Bello, F., S. Lavorel, S. Díaz, R. Harrington, J. H. Cornelissen, R. D. Bardgett, M. P. Berg, P. Cipriotti, C. K. Feld, and D. Hering. 2010. Towards an assessment of multiple ecosystem processes and services via functional traits. Biodiversity and Conservation 19:2873-2893.

Dı́az, S., and M. Cabido. 2001. Vive la difference: Plant functional diversity matters to ecosystem processes. Trends in Ecology & Evolution 16:646-655.

Ensign, S. H., and M. W. Doyle. 2005. In‐channel transient storage and associated nutrient retention: Evidence from experimental manipulations. Limnology and Oceanography 50:1740-1751.

Ensign, S. H., and M. W. Doyle. 2006. Nutrient spiraling in streams and river networks. Journal of Geophysical Research: Biogeosciences 111.

Flynn, D. F., N. Mirotchnick, M. Jain, M. I. Palmer, and S. Naeem. 2011. Functional and phylogenetic diversity as predictors of biodiversity–ecosystem‐function relationships. Ecology 92:1573-1581.

Gende, S. M., R. T. Edwards, M. F. Willson, and M. S. Wipfli. 2002. Pacific Salmon in aquatic and terrestrial Ecosystems: Pacific salmon subsidize freshwater and terrestrial ecosystems through several pathways, which generates unique management and conservation issues but also provides valuable research opportunities. BioScience 52:917-928.

Grenz, C., M. N. Hermin, D. Baudinet, and R. Daumas. 1990. In situ biochemical and bacterial variation of sediments enriched with mussel biodeposits. Hydrobiologia 207:153-160.

Grimm, N. B. 1988. Role of macroinvertebrates in nitrogen dynamics of a desert stream. Ecology 69:1884-1893.

Grimm, N. B., R. W. Sheibley, C. L. Crenshaw, C. N. Dahm, W. J. Roach, and L. H. Zeglin. 2005. N retention and transformation in urban streams. Journal of the North American Benthological Society 24:626-642.

Haag, W. R., and M. L. Warren. 2010. Diversity, abundance, and size structure of bivalve assemblages in the Sipsey River, Alabama. Aquatic Conservation: Marine and Freshwater Ecosystems 20:655-667.

Halvorson, H. M., D. J. Hall, and M. A. Evans‐White. 2017. Long‐term stoichiometry and fates highlight animal egestion as nutrient repackaging, not recycling, in aquatic ecosystems. Functional Ecology 31:1802-1812.

Hooper, D. U., E. C. Adair, B. J. Cardinale, J. E. Byrnes, B. A. Hungate, K. L. Matulich, A. Gonzalez, J. E. Duffy, L. Gamfeldt, and M. I. O’Connor. 2012. A global synthesis reveals biodiversity loss as a major driver of ecosystem change. Nature 486:105.

Kana, T. M., C. Darkangelo, M. D. Hunt, J. B. Oldham, G. E. Bennett, and J. C. Cornwell. 1994. Membrane inlet mass spectrometer for rapid high-precision determination of N2, O2, and Ar in environmental water samples. Analytical Chemistry 66:4166-4170.

Knowles, R. 1982. Denitrification. Microbiological Reviews 46:43.

Lautz, L., and R. Fanelli. 2008. Seasonal biogeochemical hotspots in the streambed around restoration structures. Biogeochemistry 91:85-104.

Lavorel, S., and É. Garnier. 2002. Predicting changes in community composition and ecosystem functioning from plant traits: Revisiting the Holy Grail. Functional Ecology 16:545-556.

Matisoff, G., and X. Wang. 1998. Solute transport in sediments by freshwater infaunal bioirrigators. Limnology and Oceanography 43:1487-1499.

McClain, M. E., E. W. Boyer, C. L. Dent, S. E. Gergel, N. B. Grimm, P. M. Groffman, S. C. Hart, J. W. Harvey, C. A. Johnston, and E. Mayorga. 2003. Biogeochemical hot spots and hot moments at the interface of terrestrial and aquatic ecosystems. Ecosystems 6:301- 312.

McGregor, S. W., and P. E. O'Neil. 1992. The Biology and Water-quality Monitoring of the Sipsey River and Lubbub and Bear Creeks, Alabama, 1990-91. Geological Survey of Alabama, Biological Resources Division.

Mermillod-Blondin, F. 2011. The functional significance of bioturbation and biodeposition on biogeochemical processes at the water–sediment interface in freshwater and marine ecosystems. Journal of the North American Benthological Society 30:770-778.

Mermillod-Blondin, F., and R. Rosenberg. 2006. Ecosystem engineering: The impact of bioturbation on biogeochemical processes in marine and freshwater benthic habitats. Aquatic Sciences 68:434-442.

Miller-Way, T., and R. R. Twilley. 1996. Theory and operation of continuous flow systems for the study of benthic-pelagic coupling. Marine Ecology Progress Series 140:257-269.

Nakano, S., and M. Murakami. 2001. Reciprocal subsidies: Dynamic interdependence between terrestrial and aquatic food webs. Proceedings of the National Academy of Sciences 98:166-170.

Newbold, J. D., J. W. Elwood, R. V. O'Neill, and W. V. Winkle. 1981. Measuring nutrient spiralling in streams. Canadian Journal of Fisheries and Aquatic Sciences 38:860-863.

Nielsen, L. P. 1992. Denitrification in sediment determined from nitrogen isotope pairing. FEMS Microbiology Letters 86:357-362.

Nizzoli, D., D. T. Welsh, M. Bartoli, and P. Viaroli. 2005. Impacts of mussel (Mytilus galloprovincialis) farming on oxygen consumption and nutrient recycling in a eutrophic coastal lagoon. Hydrobiologia 550:183-198.

Paetzold, A., C. J. Schubert, and K. Tockner. 2005. Aquatic terrestrial linkages along a braided- river: Riparian arthropods feeding on aquatic insects. Ecosystems 8:748-759.

Petchey, O. L., and K. J. Gaston. 2006. Functional diversity: Back to basics and looking forward. Ecology Letters 9:741-758.

Poff, N., M. M. Brinson, and J. Day. 2002. Aquatic ecosystems and global climate change. Pew Center on Global Climate Change, Arlington, VA 44:1-36.

Ricciardi, A., and J. B. Rasmussen. 1999. Extinction rates of North American freshwater fauna. Conservation biology 13:1220-1222.

Richter, B. D., D. P. Braun, M. A. Mendelson, and L. L. Master. 1997. Threats to imperiled freshwater fauna. Conservation Biology 11:1081-1093.

Rusek, J. 1998. Biodiversity of Collembola and their functional role in the ecosystem. Biodiversity & Conservation 7:1207-1219.

Rysgaard, S., N. Risgaard‐Petersen, S. Niels Peter, J. Kim, and N. Lars Peter. 1994. Oxygen regulation of nitrification and denitrification in sediments. Limnology and Oceanography 39:1643-1652.

Schmitz, O. J., R. W. Buchkowski, K. T. Burghardt, and C. M. Donihue. 2015. Functional traits and trait-mediated interactions: Connecting community-level interactions with ecosystem functioning. Advances in Ecological Research 52:319-343.

Schwalb, A. N., and M. T. Pusch. 2007. Horizontal and vertical movements of unionid mussels in a lowland river. Journal of the North American Benthological Society 26:261-272.

Smith, E., W. Davison, and J. Hamilton-Taylor. 2002. Methods for preparing synthetic freshwaters. Water Research 36:1286-1296.

Spooner, D. E., and C. C. Vaughn. 2006. Context‐dependent effects of freshwater mussels on stream benthic communities. Freshwater Biology 51:1016-1024.

Spooner, D. E., C. C. Vaughn, and H. S. Galbraith. 2012. Species traits and environmental conditions govern the relationship between biodiversity effects across trophic levels. Oecologia 168:533-548.

Strayer, D. L., J. A. Downing, W. R. Haag, T. L. King, J. B. Layzer, T. J. Newton, and S. J. Nichols. 2004. Changing perspectives on pearly mussels, North America's most imperiled animals. BioScience 54:429-439.

Strayer, D. L., and D. Dudgeon. 2010. Freshwater biodiversity conservation: Recent progress and future challenges. Journal of the North American Benthological Society 29:344-358.

Strayer, D. L., and H. M. Malcom. 2007. Shell decay rates of native and alien freshwater bivalves and implications for habitat engineering. Freshwater Biology 52:1611-1617.

Thamdrup, B., and T. Dalsgaard. 2002. Production of N2 through anaerobic ammonium oxidation coupled to nitrate reduction in marine sediments. Applied and Environmental Microbiology 68:1312-1318.

Thrush, S. F., J. E. Hewitt, M. Gibbs, C. Lundquist, and A. Norkko. 2006. Functional role of large organisms in intertidal communities: Community effects and ecosystem function. Ecosystems 9:1029-1040.

Tilman, D. 2001. Functional diversity. Encyclopedia of Biodiversity 3:109-120.

Tilman, D., J. Knops, D. Wedin, P. Reich, M. Ritchie, and E. Siemann. 1997. The influence of functional diversity and composition on ecosystem processes. Science 277:1300-1302.

Trentman, M. T., C. L. Atkinson, and J. D. Brant. 2018. Native freshwater mussel effects on nitrogen cycling: Impacts of nutrient limitation and biomass dependency. Freshwater Science 37:276-286.

Vanni, M. J. 2002. Nutrient cycling by animals in freshwater ecosystems. Annual Review of Ecology and Systematics 33:341-370.

Vanni, M. J., G. Boros, and P. B. McIntyre. 2013. When are fish sources vs. sinks of nutrients in lake ecosystems? Ecology 94:2195-2206.

Vanni, M. J., and P. B. McIntyre. 2016. Predicting nutrient excretion of aquatic animals with metabolic ecology and ecological stoichiometry: A global synthesis. Ecology 97:3460- 3471.

Vaughn, C. C. 2010. Biodiversity losses and ecosystem function in freshwaters: Emerging conclusions and research directions. BioScience 60:25-35.

Vaughn, C. C. 2017. Ecosystem services provided by freshwater mussels. Hydrobiologia 810:15- 27.

Vaughn, C. C., C. L. Atkinson, and J. P. Julian. 2015. Drought‐induced changes in flow regimes lead to long‐term losses in mussel‐provided ecosystem services. Ecology and Evolution 5:1291-1305.

Vaughn, C. C., and C. C. Hakenkamp. 2001. The functional role of burrowing bivalves in freshwater ecosystems. Freshwater Biology 46:1431-1446.

Vaughn, C. C., S. J. Nichols, and D. E. Spooner. 2008. Community and foodweb ecology of freshwater mussels. Journal of the North American Benthological Society 27:409-423.

Vaughn, C. C., D. E. Spooner, and H. S. Galbraith. 2007. Context‐dependent species identity effects within a functional group of filter‐feeding bivalves. Ecology 88:1654-1662.

Vidon, P., C. Allan, D. Burns, T. P. Duval, N. Gurwick, S. Inamdar, R. Lowrance, J. Okay, D. Scott, and S. Sebestyen. 2010. Hot spots and hot moments in riparian zones: Potential for improved water quality management. Journal of the American Water Resources Association 46:278-298.

Violle, C., M. L. Navas, D. Vile, E. Kazakou, C. Fortunel, I. Hummel, and E. Garnier. 2007. Let the concept of trait be functional! Oikos 116:882-892.

Walker, B. H. 1992. Biodiversity and ecological redundancy. Conservation Biology 6:18-23.

Wallace, J. B., and J. R. Webster. 1996. The role of macroinvertebrates in stream ecosystem function. Annual Review of Entomology 41:115-139.

Williams, J. D., A. E. Bogan, and J. T. Garner. 2008. Freshwater mussels of Alabama and the Mobile basin in Georgia, Mississippi, and Tennessee. University of Alabama Press. Tuscaloosa, Alabama

Wilson, M. A., and S. R. Carpenter. 1999. Economic valuation of freshwater ecosystem services in the United States: 1971–1997. Ecological Applications 9:772-783.

Table 2.1: Incubation Timeline - each incubation trial took place during three consecutive weeks, each taking roughly 5 days from start to finish. The “t0” represents the beginning of the 24-hr chamber incubation, while the rows above and below “t0” represent aspects of an incubation trial taking place x number of hours before and after the start of the chamber incubation, respectively.

Incubation Trial 1 (Aug 6-10, 2017): F. flava, L. ornata, Control Incubation Trial 2 (Aug 13-17, 2017): F. flava, C. asperata, Control Incubation Trial 3 (Aug 20-24, 2017): C. asperata, L. ornata, Control Microcosms filled with sieved, homogenized sediment and fully submerged in site t-48 water at site temp t-24 Mussel collection; field excretion/biodeposition experiment

Microcosms capped and stored in cooler with ice; materials transported from UA to t-20 DISL Microcosms uncapped and fully submerged in site water; mussels placed in microcosms t-12 to acclimate t0 Microcosms capped air-tight; start of incubation trial

End of incubation trial; water samples collected for nutrient and gas flux analysis t+24 Microcosms dismantled; IPT slurries prepped and placed on shaker Slurries spiked with 15NO ; T time-point samples spiked with ZnCl ; slurries placed t+42 3 0 2 back on shaker t+48 T6 time-point samples spiked with ZnCl2; slurries placed back on shaker

t+60 Water samples from the incubation and IPT slurries analyzed for [N2] on the MIMS

Table 2.2: Mean (± 1 SE) values for mussel parameters and functional traits. F-statistics and p- values are presented for mussel physiological traits, and are based on results of 1-way ANOVAs (α = 0.05) taking into account the random effect of incubation trial (block effect). Bolded p- values indicate a statistically significant difference among treatments.

Treatment F. flava C. asperata L. ornata F-stat p-value Mussel parameters

Shell length (mm) 52.4 (1.40) 51.2 (1.79) 80.7 (2.74) Soft tissue dry mass (g) 0.94 (0.071) 1.14 (0.121) 2.97 (0.316)

Functional Traits (excretion units: µmol h-1; biodeposition units: mg h-1)

+ NH4 excretion rate 1.84 (0.441) 1.11 (0.153) 3.47 (0.591) 4.47 0.033 OM biodeposition rate 0.95 (0.138) 1.53 (0.171) 2.62 (0.726) 4.12 0.043

Table 2.3: Mean (± 1 SE) biogeochemical fluxes and potentials from the chamber incubation and IPT experiments, respectively. F-statistics and p-values are based on results of 1-way ANOVAs (α = 0.05) taking into account the random effect of incubation trial (block effect). Bolded p- values indicate a statistically significant difference among treatments.

Treatment Sediment F. flava C. asperata L. ornata F-stat p-value Nutrient flux (µmol m-2 h-1)

+ NH4 -N -1.58 (6.67) 50.4 (19.8) 61.3 (20.5) -13.1 (19.4) 3.42 0.038

- NO2 -N 0.107 (0.043) 0.559 (0.244) 0.974 (0.143) 1.59 (0.945) 2.28 0.112

- NO3 -N 14.7 (4.37) 24.3 (13.5) 14.5 (4.96) -24.1 (7.67) 4.03 0.022

Gas flux (µmol m-2 h-1)

N2-N 109 (24.1) 916 (46.2) 990 (49.9) 1434 (135) 74.73 << 0.001

O2 -549 (57.1) -1754 (65.3) -2018 (83.5) -2655 (81.7) 252.3 << 0.001

-1 -1 N-removal Potential (µmol N2-N kg h )

DNF 0.377 (0.142) 0.247 (0.141) 0.401 (0.112) 1.16 (0.243) 5.63 0.005 Annamox 0.118 (0.055) 0.085 (0.048) 0.094 (0.028) 0.225 (0.050) 2.33 0.104

Figure 2.1: Continuous flow-through incubation setup modified from Miller-Way and Twilley (1996). Black lines represent the direction of the flow through the apparatus (~2.5 mL/min). The microcosms were 9 cm in diameter and 30 cm tall (20 cm of sediment, 10 cm of overlying water).

+ Figure 2.2: Boxplots highlighting differences in (a) soft tissue dry mass, (b) NH4 excretion and (c) OM biodeposition between the mussel species used in the incubation trials. The p-values represent the results of 1-way ANOVAs (α = 0.05) taking into account the random effect of incubation trial (block effect). Bolded p-values indicate a statistically significant difference among treatments and lower-case letters over a boxplot represents a significantly different mean determined via Tukey’s HSD posthoc analysis.

+ - Figure 2.3: Mean (a) ambient nutrient (NH4 , NO3 ) flux, (b) ambient gas (N2-N, O2) flux and (c) N-removal potentials (DNF, annamox) between the control and mussel treatments analyzed in the incubation trials. Error bars represent ± 1 SE. The p-values represent the results of 1-way ANOVAs (α = 0.05) taking into account the random effect of incubation trial (block effect). Bolded p-values indicate a statistically significant difference among treatments and letters over a bar represents a significantly different mean determined via Tukey’s HSD posthoc analysis.

Figure 2.4: Scatterplots highlighting relationships between the response variables of (a) ambient N2 flux, (b) DNF potential and (c) annamox potential and the explanatory variables of (i) soft + tissue DM, (ii) mussel respiration, (iii) NH4 excretion and (iv) OM biodeposition. Species are represented by the following symbols: F. flava (○), C. asperata (□), L. ornata (△). Results of SLRs are reported as the coefficient of determination (R2) and p-values. Bolded p-values represent a significant relationship between the explanatory and response variable and are accompanied by the fitted regression line in the plot.

Figure 2.5: (a) Distribution among species of the residual variation present in the ambient N2 flux versus respiration SLR. These data were used as the response variable for SLRs testing the + influence of (b) NH4 excretion and (c) OM biodeposition on ambient N2 flux once the effect (caveat) of respiration had been removed from the data. Species are represented by the following symbols: F. flava (○), C. asperata (□), L. ornata (△). Results of SLRs are reported as the coefficient of determination (R2) and p-value. Bolded p-values represent a significant relationship between the explanatory and response variable and are accompanied by the fitted regression line in the plot. The dashed line in the OM biodeposition plot represents a moderate (p = 0.053) linear relationship.

CHAPTER 3

INTEGRATING FUNCTIONAL TRAITS AND BIODIVERSITY TO ASSESS THE ROLE OF BENTHIC COMMUNITIES ON NITROGEN REMOVAL IN FRESHWHATER SYSTEMS

Abstract

Animal aggregations can have a substantial influence on ecosystem processes such as nutrient recycling and repackaging. Increased biodiversity in a group of organisms facilitating similar ecosystem processes can enhance ecosystem resiliency, as differing functional traits and niche requirements leads to inter-specific competition, altering how organisms physically and chemically impact their environment. Using communities of native freshwater mussels (Bivalvia:

Unionidae), we utilized in-situ community manipulations to assess the influence of mussel functional traits and the effect of mussel biodiversity on nitrogen (N) removal in riverine sediments via the biogeochemical pathways of denitrification and annamox. We conducted a 9- week experiment manipulating mussel community structure in benthic enclosures, tracked mussel functional traits and analyzed sediment for N-removal potential. We found that mussel aggregations can stimulate N-removal, with the largest effect in single-species communities at high-densities. We observed a non-additive effect of species diversity on N-removal, resulting in multi-species aggregations underyielding in regard to N-removal when compared to single- species aggregations; an effect that was driven by an overall increase in biological activity, particularly movement and burrowing. Our study adds to a growing collection of research demonstrating the non-additive effects of biodiversity on important ecosystem processes within aggregations of these functionally-active, yet globally imperiled organisms.

Introduction

Lotic ecosystems (rivers, streams) not only represent a major conduit, transporting essential macronutrients such as carbon (C), nitrogen (N) and phosphorus (P) from land to the oceans and atmosphere (Meybeck 1982, Smith and Hollibaugh 1993, Aufdenkampe et al. 2011), but also represent major sites of biogeochemical activity, fostering high levels of nutrient recycling and repackaging (Meyer et al. 1988, Wagener et al. 1998, Essington and Carpenter

2000). While nutrient dynamics in lotic systems are often controlled by abiotic forces (flow regime, local geology and climate) on large spatial and temporal scales (Frissell et al. 1986,

Thoms and Parsons 2002), biotic contributors have substantial combined effects on the recycling and repackaging of C, N and P, particularly when disturbance is low (Vanni 2002, Atkinson et al.

2017). Riverine benthic environments can harbor multiple functional-feeding groups of aquatic fauna that process organic matter (OM) in different ways, leading to varying effects on local nutrient cycling (Wallace and Webster 1996). Within a functional-feeing group, an organism’s influence on nutrient cycling can reach far beyond the ecosystem functions associated with OM acquisition and processing (i.e., predation/foraging, production, excretion, biodeposition), including activities such as sediment bioturbation (Mermillod-Blondin and Rosenberg 2006,

Mermillod-Blondin 2011) and reproduction (Cederholm et al. 1999, Naiman et al. 2002), among others. The manner and mechanisms by which benthic fauna provision ecosystem function in lotic systems (e.g. biogeochemical nutrient cycling) have long been a subject of interest among aquatic ecologists (Jones et al. 1994, Wallace and Webster 1996, Covich et al. 1999, Vaughn and

Hakenkamp 2001), particularly in response to the massive loss of global aquatic biodiversity as a result of anthropogenic activity (Ricciardi and Rasmussen 1999, Dudgeon et al. 2006).

Linking the role of species functional traits (traits defnining how an organism interacts with its environment; Violle et al. 2007) to ecosystem function is important to understand ecosystem structure, but becomes more complicated when scaled from the individual to the ecosystem due to variation in abiotic factors as well as the integrated effects of intra- and interspecific competition and niche partitioning dynamics that exist in biologically-diverse systems (Hairston and Hairston 1993, Chapin et al. 1997, Cardinale et al. 2002, Schmitz et al.

2008). Interest in exploring the relationship between biodiversity and ecosystem function

(hereafter BD-EF) has increased dramatically in the last three decades (Loreau et al. 2001,

Hooper et al. 2005), starting from early research on terrestrial plant communities (Frank and

McNaughton 1991, McNaughton 1994, Tilman et al. 1997) to more recent studies in terrestrial

(Cadotte et al. 2009, Díaz et al. 2013, Schmitz et al. 2015), marine (Emmerson et al. 2001, Solan et al. 2004, Gamfeldt et al. 2015) and aquatic (Giller et al. 2004, Sandin and Solimini 2009,

Woodward 2009) ecosystems, including studies on BD-EF dynamics in communities of freshwater benthic macrofauna (Covich et al. 2004, Vaughn 2010, Spooner et al. 2012). As biodiversity increases within a functional guild (different organisms provisioning the same EF;

Simberloff and Dayan 1991), the resulting effect on ecosystem function can either be additive or complimentary (Loreau 1998, Loreau and Hector 2001). An additive effect of biodiversity (also called ‘selection effect’) indicates ecological redundancy (Walker 1992), or the fact that, at a given population density, a multi-species community provisions ecosystem function at the same rate as single-species communities (Fox 2005). A complimentary effect of biodiversity suggests a multi-species community provisions ecosystem function either better (overyielding) or worse

(underyielding) than single-species communities of the same density, revealing the importance of

57 functional diversity (Tilman 2001) in determining the ecological role of a functional guild on the ecosystem-scale (Tilman et al. 1997, Petchey and Gaston 2006).

Freshwater mussels (Bivalvia: Unionidae, hereafter mussels) are a group of long-lived (6-

100 years) benthic macrofauna that are functionally-defined as burrowing, filter-feeding organisms (Vaughn and Hakenkamp 2001). Mussels are ideal model organisms for exploring

BD-EF dynamics in freshwater ecosystems as they thrive in dense, speciose, patchy aggregations throughout a watershed (Haag 2012). Additionally, mussels are functionally active by filtering mass amounts of particulate matter from the water column (Welker and Walz 1998, Vaughn et al. 2008), concentrating waste material in the benthos (Vaughn et al. 2008, Atkinson et al. 2013) and disturbing sediment through movement and burrowing activity (Schwalb and Pusch 2007,

Allen and Vaughn 2009). Through the integrative effects of their functional traits on the ecosystem-scale, mussel aggregations create biogeochemical hotspots (Atkinson and Vaughn

2015), provisioning both top-down (Welker and Walz 1998) and bottom-up (Vaughn et al. 2007,

Atkinson et al. 2013) ecosystem functions by removing nutrients from the water column and concentrating them in the benthos. This important functional role of mussels increases the connectivity of a system by coupling the benthic and pelagic zones of a stream (Spooner et al.

2012, Vaughn et al. 2015) which fosters more complex food webs (Vaughn et al. 2008, Allen et al. 2012), making it possible for more functional-feeding groups to coexist (Spooner and Vaughn

2006, Vaughn and Spooner 2006).

More recently, ecologists have begun to investigate the influence of individual mussels

(Benelli et al. 2017, Hoellein et al. 2017) and mussel aggregations (Trentman et al. 2018) on another important ecosystem function in aquatic ecosystems: biogeochemical nitrogen removal

(hereafter N-removal). N-removal refers to microbial metabolisms that convert dissolved

58 inorganic nitrogen (DIN), through a series of intermediate molecules, to dinitrogen gas (N2) that diffuses to the atmosphere, representing the permanent removal of N from the aquatic system

(Knowles 1982). N-removal is an important ecosystem function in lotic systems, as excess anthropogenic N input is a global pollutant degrading freshwaters (Carpenter et al. 1998) and contributing to harmful algal blooms (Landsberg 2002) and coastal dead zones (Diaz and

Rosenberg 2008). Little is known about how communities of interacting species influence N- removal. Attributing such an ecosystem function to mussel aggregations helps to define the functional role these organisms play lotic ecosystems and aids in conservation efforts as unionid mussels are considered to be one of the most imperiled faunal groups in the world (Lydeard et al.

2004, Strayer et al. 2004).

Our objective was to assess the influence of mussel aggregations, associated functional traits and the effect of mussel biodiversity on N-removal potential in stream sediments using an in-situ community manipulation. We focused on two major biogeochemical pathways

- representing microbially-mediated N-removal: dissimilatory nitrate (NO3 ) reduction (known as

+ denitrification, hereafter DNF), and anaerobic ammonium (NH4 ) oxidation (hereafter annamox).

We predicted mussel aggregations satisfy the three environmental requirements necessary to activate microbial metabolisms that result in N-removal. Those requirements are: 1) anoxic environment as N-removing microbes are facultative anaerobes, 2) excess reactive DIN to serve as an electron acceptor (or both acceptor and donor in annamox), and 3) OM to serve as an energy substrate (in the form of labile C) for the DNF pathway (annamox does not require C as energy) or to create anoxic microenvironments via decomposition or increased sediment cohesion (Knowles 1982). As such, we also aimed to elucidate the mechanisms by which mussel aggregations of varying biodiversity either foster or hinder N-removal in sediment by tracking

+ biological activity within a mussel community (here defined as the functional traits of NH4 excretion [excess DIN], OM biodeposition [energy source, increased anoxia], vertical burrowing

[increased anoxia] and horizontal movement), and relating community-scale biological activity to N-removal potential. We hypothesized that the presence of mussels would increase N-removal potential relative to no mussels. Additionally, we hypothesized increasing mussel density would

+ increase N-removal potential as more NH4 and OM was being introduced to microbially-active sites in the sediment, but increasing species diversity would decrease N-removal potential due to differing ecological niches leading to an increase in inter-specific competition measured as an increase in movement and burrowing behavior.

Materials and Methods

Study site

Our study took place in the Sipsey River, a fifth-order tributary of the Tombigbee River in the Mobile Basin of northwestern Alabama (Figure 3.1). The main stem of the Sipsey runs

184-km through the East Gulf Coastal Plain physiographic region of Alabama to its confluence with the Tombigbee River. The Sipsey River flows through mainly forested landscape and is relatively undisturbed by anthropogenic influence. Due to its undisturbed nature, the Sipsey

River watershed contains remarkably high aquatic biodiversity throughout (McGregor and

O'Neil 1992), including dense and diverse mussel populations containing 37 of 41 historically- occurring unionid mussel species (Haag and Warren 2010). Our study site was a 50-m reach of the Sipsey River located ~ 40-km upstream of the river’s confluence with the Tombigbee River

(Figure 3.1). We selected this reach because it was a shallow, straight channel in an otherwise highly meandering river. Our study reach is in close proximity to high-density mussel aggregations surveyed in previous studies (McCullagh et al. 2002, Haag and Warren 2010).

Experimental design

We manipulated mussel abundance and diversity using in-stream enclosures to assess the influence of mussel community structure and associated functional traits on N-removal processes in river sediments. We used the 2nd and 3rd most common mussels in the Sipsey River, Fusconaia flava and Cyclonaias asperata, respectively (the most abundant, Pleurobema decisum, being federally endangered; Haag and Warren 2010). While similar in size, the two mussels vary in life history and shell morphology (Williams et al. 2008). The evolutionary/morphological differences and high natural abundances make experimental communities of F. flava and C. asperata ideal proxies for examining the influence of natural mussel aggregations on ecosystem function. We employed a fully factorial design, crossing two abundance treatments (low density = 24 mussels m-2, high density = 48 mussels m-2; relevant densities for the Sipsey; Haag and Warren 2010) with three diversity treatments (F. flava, C. asperata, F. flava + C. asperata), for a total of six mussel treatments. A sediment-only control was employed. Each mussel treatment and the sediment-only control was replicated four times for a total of 28 experimental communities.

Mussel aggregations were contained within benthic enclosures (50-cm x 50-cm x 20-cm) constructed of untreated lumber (5-cm x 5-cm) and wrapped in galvanized steel mesh (0.6-cm mesh size). The steel mesh covered the four sides and bottom of each enclosure, leaving the top open. The enclosures were installed in our study reach to a depth of 20-cm. We dug sediment from the river leaving a hole into which an enclosure was placed. We took sediment to shore where we sieved it to remove any mussels naturally present within the reach and re-homogenized the sieved sediment in a 100-L plastic bin. Mussels found during sieving were identified to species to provide an estimate of the natural mussel community makeup within our reach

(Appendix 2). Once identified, mussels were placed back into the reach outside of the study area.

Any F. flava and C. asperata individuals found while digging were kept to include in the study.

An enclosure was placed into the dug hole and was filled with the sieved, homogenized sediment to where the open top of the enclosure was flush with the surrounding streambed. The enclosure design kept mussels confined within while also allowing each enclosure to experience natural hyporheic and pelagic flow, as well as natural colonization of benthic microbes, biofilm and macrofauna. We installed enclosures along transects stretching between the banks and spaced 3- m apart from each other. To reduce enclosure-enclosure bias, enclosures were spaced 2-m apart within a transect and offset in a checkerboard fashion between transects. Once all the enclosures were installed, they remained in the reach undisturbed for a week to reestablish a natural benthic environment. We also installed a temperature/pressure logger (Hobo U20L, Onset Corp, Bourne,

MA) in our study reach to track water depth and temperature at the site (Figure 3.3).

We used systematic random sampling (Fortin et al. 1990) to assign treatments in our reach, in which a random order of treatments was assigned, and that same order was repeated throughout the reach. Systematic random sampling allowed us to control for potential spatial autocorrelation bias without the need the implement a blocking design, as DNF and annamox potentials are highly spatially heterogeneous in aquatic sediments (Pina-Ochoa and Alvarez-

Cobelas 2006, Groffman et al. 2009). We collected 216 live mussels (108 each of F. flava and C. asperata) from within our study reach and from sites < 1-km upstream and downstream of our study reach 24-hrs prior to the start of the experiment. Each mussel was tagged with a unique 3- digit number by attaching paper covered in a polymer coating to a length of fly-fishing line and attaching the fishing line to the mussel’s shell using high-strength, waterproof adhesive. We measured total shell length (mm) of each mussel with which we estimated soft tissue dry mass

(g) using length-mass regressions previously established for both species (C.L. Atkinson,

62 unpublished data). At the start of the experiment, enclosures were assigned a treatment and mussels were randomly selected and placed into the enclosures. In all mussel treatments, the communities were stocked in the same position within the enclosure. We used a 6x6 grid to assign the initial position of each mussel, as well as track movement over the course of the experiment (Figure 3.2). The experiment ran for 9-weeks from July 31 – September 29, 2017

(bold, closed arrows in Figure 3.3) with visits occurring weekly, weather permitting. Two weeks into the experiment, a large storm system came through northwestern Alabama, which caused our study reach to swell and peak at >1.25-m in height (Figure 3.3). As a result, we were unable to work in the reach between the dates of August 7-23 (17 days).

Abiotic parameters

During each visit to the study site, we collected reach-scale physiochemical data on the benthos both upstream and downstream of our reach (solid, open arrows in Figure 3.3; Appendix

3). Using a YSI multiparameter probe (YSI Inc., Yellow Springs, Ohio, USA), we measured dissolved oxygen (DO), conductance and pH. Additionally, we measured turbidity upstream and downstream using a AquaFluorTM Handheld Fluorometer/Turbidimeter (Turner Designs,

Sunnyvale, California, USA). Also, beginning on August 31 and through the remainder of the experiment (n = 6 visits), we also measured depth (m) and flow velocity (m s-1) at each enclosure along the downstream margin, so as to not disturb benthic communities inside the enclosure, using a FH950 Handheld Flow Meter (Hach Company, Loveland, Colorado, USA).

Movement and burrowing

We tracked the movement and burrowing behavior of each mussel throughout the study

(solid, open arrows in Figure 3.3). Prior to the study’s start, each mussel was tagged with a length of fishing line that was glued to the shell near the incurrent and excurrent siphons. The

63 length of fishing line that extended past the edge of the shell was measured and recorded, denoting the initial line length. At each visit to the study site, we recorded the length of fishing line exposed for any mussel that was completely buried in the sediment (no exposed siphons visible). We observed the number of mussels fully buried in each enclosure and calculated the depth a mussel was buried (cm) by subtracting the length of fishing line exposed at the time of sampling from the initial length. We used the total number of burrowing occurrences observed for an enclosure throughout the 9-weeks as one of the functional traits of interest, as it represented how often mussels within a community were closer to the oxic/anoxic redox boundary in the sediment.

We tracked mussel movement across the streambed using a grid of 36, 7.16-cm2 cells with a rebar frame. When placed over an enclosure, the grid was raised 2-cm off the sediment surface to ensure we were not disturbing the benthic community. The grid was organized using numbers 1-6 along the upstream axis, and letters A-F along the left bank axis (Figure 3.2). At each visit to the study site, we placed the grid atop an enclosure and recorded the position of each mussel. We calculated movement as the total distance a mussel moved in a straight line from the center of one cell to the center of another. For example, a mussel observed as moving from one cell to an adjacent cell was considered to have moved the length of one cell (movement

= 7.16-cm). Further, a mussel observed to have moved diagonally over two cells and upstream three cells was considered to have moved in a straight line that represents the hypotenuse of a theoretical right triangle with the length of two cells (14.32-cm) and a height of three cells

(21.48-cm), and movement was calculated using the Pythagorean Theorem (movement = 25.82- cm). We used total daily movement (cm d-1) of an enclosure as another functional trait of

64 interest, which we calculated by dividing the total movement (cm) within an enclosure by the number of days (56 d).

Excretion and biodeposition

We performed in-field experiments to characterize community-scale functional traits of

+ NH4 excretion and OM biodeposition (longer solid, open arrows in Figure 3.3). We collected individuals of F. flava and C. asperata directly downstream of the enclosure area and measured excretion and biodeposition rates using in-field chamber incubations (Atkinson et al. 2013) on ten individuals on September 2nd and 25th for a total of 20 replicate samples for C. asperata and

19 for F. flava (contents of one F. flava chamber was spilled during transport to the lab). Both sampling events were during baseflow conditions at our study site (Figure 3.3) and at similar water temperatures (25.61°C at 12:00 on 9/2; 26.20° at 12:00 on 9/25). We filled 25 1-L chambers with filtered site-water (GF/F filter, 0.7 µm pore size, Merck Millipore, Burlington,

MA). Each mussel’s total length was measured to be fit to length-dry mass regressions for both species (C.L. Atkinson, unpublished data), and was then gently scrubbed to remove algae and sediment attached to the shell and placed into a chamber. Additionally, five empty mussel shells were collected, gently scrubbed and placed into individual chambers to act as a control to account for nutrients and solids produced by shell-associated algae and bacteria. Each chamber was capped and were incubated for an hour in the dark. The chambers were partially submerged in site water to maintain temperature. After an hour, the mussels and shell controls were carefully removed using tongs, and chambers were placed on ice and immediately returned to the lab. In the lab, the contents of each chamber were filtered through a GF/F filter (0.7 µm pore size).

Solid material collected on the filter was used to characterize mussel biodeposition (feces

+ pseudofeces). The filters were dried for 48-hrs at 50 °C and then weighed on an analytical balance (± 0.0001 g). The total mass of material collected on the filter was reported as total biodeposition rate (mg h-1). The dried filters were combusted at 500 °C for 2-hrs, and weighed again to determine ash-free dry mass (AFDM), which was used to quantify the organic matter

(OM) content of the biodeposits (also reported as mg h-1) We analyzed the filtered chamber

+ water for ammonium (NH4 ) concentration using a Lachat QuickChem flow injection analyzer

+ (Hach Company, Loveland, Colorado, USA). NH4 concentrations in the control chambers were

+ averaged and subtracted from the NH4 concentrations in each mussel chambers to estimate each

+ + -1 mussel’s NH4 excretion rate, reported in moles (µmol NH4 h ).

We combined data from both sampling events and calculated mean mass-specific hourly

+ + -1 -1 -1 -1 rates of NH4 excretion (µmol NH4 h gDM ) and OM biodeposition (mg OM h gDM ) for each species to estimate the corresponding community-scale rates produced within each

+ enclosure containing mussels. Community-scale areal NH4 excretion and OM biodeposition rates were estimated by multiplying the mass-specific rates from the incubation trials by the soft tissue dry mass for each mussel used in the experiment. To fit better with the timescale of our experiment, values were then multiplied by 24-hrs to report the rates on a per day basis. Rates for each mussel in an enclosure were summed and divided by the surface area of the sediment in the

2 + + -2 -1 enclosure (0.25 m ) to provide total areal daily rates of NH4 excretion (mmol NH4 m d ) and

OM biodeposition (g OM m-2 d-1) for each enclosure containing mussels.

Destructive sampling for N-removal potentials

On September 29, 2017, we destructively sampled each enclosure to assess N removal potential in the sediment. We sampled an enclosure by taking four sediment cores, one in each

66 quadrant of the enclosure, to control for bias related to spatial heterogeneity in the distribution of

DNF and annamox microbes in sediment (Pina-Ochoa and Alvarez-Cobelas 2006, Groffman et al. 2009). We extracted each core to a depth of 20-cm, the bottom 10-cm of which was discarded on shore. The top 10-cm extracted from the four quadrants of an enclosure were combined into one conglomerate sample and stored in a 4-L sample bag. Once sampled, the sediment was immediately placed on ice. This process was repeated for each of the 28 enclosures working downstream up as to not disturb enclosures prior to sampling. After all the enclosures had been sampled, the sediment was transported to Dauphin Island Sea Lab (DISL) for processing and analysis.

We measured N-removal potentials using isotope pairing techniques (IPT, Thamdrup and

15 - Dalsgaard 2002). In IPT, isotopically-labeled nitrate (98% NO3 ) is introduced to an anaerobic slurry of sediment and incubated for a set amount of time. Subsequent analysis of the

29 30 fractionation of N2 and N2 in the water overlying the slurry determines the maximum potential rates of annamox and DNF, respectively, via the following biogeochemical pathways:

15 − 30 2 푁푂3 → 푁2

15 − 14 + 29 푁푂3 + 푁퐻4 → 푁2

The first equation represents the DNF pathway, where two molecules of the isotopically-labelled

15 - NO3 combine, through a series of intermediates, to form dinitrogen gas with a labelled molar mass of 30N. The second equation represents the annamox pathway, where one molecule of the

15 - isotopically-labeled NO3 combines, through a series of intermediates, with a naturally-present

14 + 29 ammonium molecule ( NH4 ) to form dinitrogen gas with a labelled molar mass of N.

To prepare the slurries, a conglomerate sediment sample was wet-sieved through a 2-mm sieve. All particles ≥ 2-mm were discarded, and only particles < 2-mm were used for IPT

67 slurries. The < 2-mm sediment was homogenized and subsampled into six 12-mL exetainers.

Each exetainer received approximately 3-g of wet sediment. The exetainers were then filled with

- anoxic NO3 -free artificial river water (ARW). The ARW was formulated to contain the major

- ions of Sipsey River water, minus NO3 (McGregor and O'Neil 1992), and was synthesized (see

Appendix A) using methods described by Smith et al. (2002). Prior to adding the ARW to the exetainers, we made the solution anoxic by bubbling it with N2 gas for 1-hr. We filled an exetainer with ARW until a meniscus formed, then capped the exetainer air-tight, inverted to inspect for air bubbles, and placed on a shaker table overnight. The following day, the exetainers were removed from the shaker and placed upright to allow the sediment within the slurry to settle to the bottom of the exetainer.

- As the slurries settled, we prepared the labelled NO3 solution by mixing 0.8-mL of a

15 - 232-mM solution of 98% labelled NO3 in 50-mL of ultrapure water, resulting in a

15 - 15 - concentration of [ NO3 ] = 3.712-mM. We added 162-μL of the NO3 solution into each

15 - exetainer, resulting in a concentration of [ NO3 ] = 50-µM in each exetainer. We then immediately added 250-μL solution of ZnCl2 (50% w/v) was introduced to 3 of the 6 replicate exetainers, ceasing all microbial activity and representing the concentration of N2 at time zero

(T0). The exetainers were recapped air-tight, and were inverted and inspected closely to ensure there were no air bubbles in the slurry. The exetainers were placed back on the shaker table, and incubated for 6-hrs. After 6-hrs, we removed the exetainers from the shaker table and set them upright to settle. Once settled, the remaining 3 replicate slurries were spiked with 250-μL of

ZnCl2, ceasing all microbial activity and representing the concentration of dinitrogen gas at time final (T6). The exetainers were recapped and placed back on the shaker table to shake overnight

29 30 until analysis of N2 and N2 using a membrane-inlet mass spectrometer (MIMS, Kana et al.

1994). Following analysis on the MIMS, we returned to UA with the sediment. The sediment was dried at 50 °C for 48-hrs and weighed on an analytical balance to determine dry mass.

29 30 -1 Concentrations of N2-N and N2-N (µmol L ) were normalized by slurry incubation time (~6 h) and sediment dry mass (kg) to report maximum N-removal potential as an hourly rate per kg

-1 -1 of sediment (µmol N2-N kg h ). Annamox values were low, and some nearly equivalent to the baseline fluctuation (noise) of the MIMS signal. Annamox potential values that had a MIMS signal < 2x the value of noise were considered to be “0”.

Statistical Analysis

We used analysis of variance (ANOVA) to explore differences in functional traits (daily

+ movement, the number of burrowing occurrences, daily areal NH4 excretion, daily areal OM biodeposition) and N-removal potentials (DNF, annamox) among our treatments. For movement and burrowing, we conducted 2-way ANOVA to test the influence of density, diversity and the interaction effect of the two factors. If we observed a significant effect of density among the two functional traits, we then explored treatment effects further by separating low- and high-density treatments and conducting 1-way ANOVAs to test for differences among mussel treatments at

+ each density level. Because NH4 excretion and OM biodeposition were estimated based on biomass, we did not conduct 2-way ANOVAs for the two functional traits. Rather, we conducted

1-way ANOVAs to test for a treatment effect at each density level separately. In a similar manner to movement and burrowing functional traits, we conducted 2-way ANOVAs testing the influence of density, diversity and the interaction effect on both DNF and annamox potentials.

Additionally, we conducted 1-way ANOVAs at each density level to test for treatment effects plus differences from the sediment-only control for both DNF and annamox potentials.

Statistically significant (p < 0.05) 1-way ANOVAs at each density level were followed by

Tukey’s Honestly Significant Difference (HSD) multiple comparisons to test for individual differences among means. ANOVAs and multiple comparisons were conducting using R version

3.3.1 (R Core Team 2016).

We used principal component analysis (PCA) to visualize the integrated influence of functional traits and community structure on N-removal potential. Conducting a PCA allowed us to reduce the measured predictor and response variables to a 2D plane with which we interpreted collinearity among variables as well as groupings and patterns that existed in our mussel treatments. The measured variables we included in the PCA were the functional traits of interest

+ (daily movement, total burrowing occurrences, daily areal NH4 excretion and OM biodeposition), and N-removal potentials (DNF, annamox). Prior to conducting the PCA, we standardized each variable to 푥̅ = 0 and σ = 1 to give each variable equal weight in the analysis.

We then calculated loading coefficients for each standardized variable to determine their relative contribution to PC1 and PC2. The magnitude and direction of the variables’ influences on each axis were plotted on an x-y plane with PC1 representing the horizontal axis and PC2 representing the vertical axis. We then calculated where each mussel treatment fit into our PCA. We determined a score for PC1 and PC2 for each enclosure based on the integration of variable loading coefficients specific to that enclosure. We plotted the enclosures on the same plot as the loadings forming a “biplot”, which simultaneously visualizes variable loading arrows and enclosure scores to assess patterns and relationships. To further assess differences in out treatments, we conducted 1-way ANOVAs to test for differences in standardized PC1 and PC2 scores between all mussel treatments. Significant ANOVAs (p < 0.05) were followed by a Tukey

HSD multiple comparison posthoc test. The PCA, ANOVAs and multiple comparisons were conducted in R version 3.3.1 (R Core Team 2016).

In addition to interpreting the influence of community-scale functional traits on N- removal, we tested the net effect of biodiversity (ΔEF) within low and high diversity treatments on DNF and annamox potentials within a BD-EF framework presented by Petchey (2003).

Within this framework, an ecosystem function (EF) is measured in “monocultures” (our two 1- species treatments) as well as in a “polyculture” containing each species represented by monocultures (our 2-species treatment). For our experiment, the expected effect of biodiversity on the EF of a polyculture (ε(EFpoly)) of a certain density (N) was the sum of the mean EFs provisioned by each monoculture multiplied by their proportional abundance in the polyculture, and was calculated using the following equation:

̅̅̅̅ ̅̅̅̅ 휀(퐸퐹푝표푙푦) = 훴(퐸퐹푖,푚표푛표 × 푝푖,푚표푛표,푁푝표푙푦)

Where i = species and pi,N = the proportion of species i at the density-level of the polyculture

(Npoly). The ΔEF equation determines if a polyculture provisions an EF in an additive (ΔEF = 0), or complimentary (ΔEF ≠ 0) fashion. Complimentary effects can further be split into overyielding (polyculture performs better than would be expected based on monocultures; ΔEF >

0) or underyielding (polyculture performs worse than would be expected based on monocultures;

ΔEF < 0) effects of biodiversity in terms of the EF provisioned (Petchey 2003). We calculated

ΔEF by subtracting the EF of the polyculture from the mean expected EF calculated above, or:

∆퐸퐹 = 퐸퐹푝표푙푦 − 휀(퐸퐹̅̅̅̅푝표푙푦)

If we observed a complimentary effect of biodiversity, we calculated additional parameters

(Dmax, DT) to assess how strong the effect of complementarity was on the EF of interest (Loreau

1998, Petchey 2003). Calculating Dmax revealed the weighted proportional difference between the EF of a polyculture and the maximum EF provisioned by any monoculture at that same

71 density, thus revealing the strongest effect of complementarity possible for a given EF at each density level. We calculated Dmax via the following equation:

(퐸퐹푝표푙푦 − max (퐸퐹푖,푚표푛표,푁푝표푙푦)) 퐷푚푎푥 = max (퐸퐹푖,푚표푛표,푁푝표푙푦)

Calculating DT revealed the weighted proportional deviation of the observed EF in a polyculture from the EF that would be expected given an additive effect of biodiversity, thus revealing the strength of complementarity for a given EF in a specific polyculture. We calculated Dmax via the following equation:

(퐸퐹푝표푙푦 − ε(퐸퐹̅̅̅̅푝표푙푦)) 퐷푇 = ε(퐸퐹̅̅̅̅푝표푙푦)

While ΔEF provided insight into the net effect of biodiversity (additive vs. complimentary effects), assessing Dmax and DT separately provided insight into the potential and actual strength of complementarity within a specific polyculture, respectively. Additionally, comparing how close DT is to Dmax revealed how sizable the observed complementarity effect was within our polyculture communities. We calculated ΔEF, Dmax and DT of DNF and annamox potentials for each 2-species enclosure (polyculture) at low- and high-densities. We then averaged each value to provide one mean (±SE) value of ΔEF, Dmax and DT for DNF and annamox potentials at low- and high-densities.

Results

Community-scale functional traits

There was an effect of mussel density (2-way ANOVA: F1,18 = 8.02, p = 0.011) on mean daily movement, with high-density mussel aggregations moving more, on average than low- density aggregations (Table 3.1). We did not observe a diversity (2-way ANOVA: F2,18 = 2.16, p

> 0.1) or interaction (2-way ANOVA: F2,18 = 0.32, p > 0.7) effect on mean daily movement.

Testing treatment effects at each density, we observed no significant differences in mean daily movement between mussel treatments at low- (1-way ANOVA: F2,9 = 0.80, p > 0.4; Figure 3.4a) or high-densities (1-way ANOVA: F2,9 = 1.40, p > 0.2; Figure 3.4b).

There was an effect of both density (2-way ANOVA: F1,18 = 9.80, p = 0.006) and diversity (2-way ANOVA: F2,18 = 10.60, p < 0.001) on the number of burrowing occurrences observed throughout the experiment (Table 3.1), but no interaction effect was observed (2-way

ANOVA: F2,18 = 0.47, p < 0.6) in burrowing behavior. In further exploring treatment effects at each density level, results showed differences in the mean number of burrowing occurrences observed between 1-species treatments of C. asperata and F. flava at both low- (1-way ANOVA:

F2,9 = 5.69, p = 0.025; Figure 3.4c) and high-densities (1-way ANOVA: F2,9 = 5.47, p = 0.028;

Figure 3.4d) with C. asperata burrowing more often than F. flava, but neither 1-species treatment was significantly different from the 2-species treatments at each density level.

+ Mean estimated areal NH4 excretion (Table 3.1) was not significantly different among treatments in low-density aggregations (1-way ANOVA: F2,9 = 2.40, p > 0.1; Figure 3.4e), but was different among treatments in high-density aggregations (F2,9 = 10.69, p = 0.004; Figure

+ 3.4f), with 1-species C. asperata communities having the potential to excrete more NH4 than both 1-species F. flava and 2-speices communities. Contrarily, there was no treatment effect for mean estimated areal OM biodeposition (Table 3.1), between low- (1-way ANOVA: F2,9 = 0.64, p > 0.5; Figure 3.4g) or high-density (1-way ANOVA: F2,9 = 2.50, p > 0.1; Figure 3.4h) aggregations.

N-removal potentials

Results of 2-way ANOVAs for N-removal potentials showed that, for DNF potentials

(Table 3.2), there was no effect of density (F1,18 = 3.06, p > 0.1), diversity (F2,18 = 1.87, p > 0.1)

73 or an interaction between the two factors (F2,18 = 2.04, p > 0.1). We observed the same trend for annamox potentials as well, with results of a 2-way ANOVA showing no effect of density (F1,18

= 1.33, p > 0.2), diversity (F2,18 = 3.01, p > 0.2), or an interaction between the two factors (F2,18

= 2.61, p > 0.1). In exploring density-specific treatment effects and differences from the sediment-only control for DNF potentials, we observed no significant differences among the low-density treatments or between the low-density treatments and the sediment-only control (1- way ANOVA: F3,12 = 0.27, p > 0.8; Figure 3.5a). There was, however, a significant treatment effect for DNF potential in high-density treatments (1-way ANOVA: F3,12 = 3.98, p = 0.035;

Figure 3.5b). Results of the Tukey’s HSD posthoc comparison showed no statistically significant differences among means, but the high-density C. asperata treatment, on average, showed moderately higher DNF potential than both the control and the high-density 2-species treatments

(Tukey’s adjusted p-values = 0.098 and 0.065, respectively; Figure 3.5b). Annamox potentials

(Table 3.2) showed the same trend as DNF potentials for low- and high-density treatments

(Figure 3.5). There was not a significant treatment effect among low-density treatments for annamox potentials (1-way ANOVA: F3,12 = 0.82, p > 0.5, Figure 3.5c). There was, however, a significant treatment effect among high-density treatments (1-way ANOVA: F3,12 = 4.15, p =

0.031, Figure 3.5d), with high-density C. asperata treatments showing moderately higher annamox potentials, on average, than both the control and the high-density 2-species treatments

(Tukey’s adjusted p-values = 0.099 and 0.090, respectively; Figure 3.5d).

Principal component analysis

The two principal component axes PC1 and PC2 explained a combined total of 76.28% of the variation in our data (Figure 3.6). Similar loading coefficient magnitudes along the horizontal axis in the PCA biplot (Figure 3.6a) shows that not a single measured variable

74 dominates the makeup of first principal component (PC1), which explained 49.76% of the variation on our data. We see a clear separation between low- (closed symbols) and high-density

(open symbols) treatments along PC1 (Figure 3.6a) suggesting a large portion of the variation in our measured variables was contributable to the soft tissue biomass present within our experimental communities. This interpretation is supported by the results of a 1-way ANOVA comparing PC1 scores by treatment. There was a significant treatment effect among PC1 scores

(F5,18 = 12.22, p < 0.001) with both high-density 1-species treatments showing significantly higher means than all low-density mussel treatments (Figure 3.6b).

The second principal component (PC2) explained an additional 26.52% of the variation in our data (Figure 3.6). PC2 revealed high collinearity between functional traits and between the two N-removal pathways, and showed a clear separation between the functional traits and N- removal potentials (Figure 3.6a). Grouping the highly-correlated functional traits and terming them “biological activity”, PC2 indicated an increase in biological activity resulted in a decrease in overall N-removal potential. Conversely, PC2 suggested an experimental mussel community with low biological activity fostered higher rates of N-removal potential. There was a significant treatment effect among PC2 scores (F5,18 = 2.97, p = 0.040), with high-density 2-species treatments showing, on average, significantly higher PC2 scores than low-density F. flava treatments, and moderately (0.05 < Tukey’s adjusted p < 0.1) higher PC2 scores than high- density F. flava treatments (Figure 3.6b).

Effect of biodiversity on N-removal potentials

In terms of DNF potential, there was a weak ΔEF in low-density treatments and a strong

ΔEF in high-density treatments (Table 3.2). This suggests there was an additive effect of biodiversity on DNF when densities were low (but note high SE), but the effect of biodiversity

75 became complimentary as mussel density increased. The negative (-) sign of ΔEF (Table 3.2) indicates the complimentary effect of biodiversity resulted in an underyielding of DNF potential.

Increasing biodiversity strongly increased the magnitude of underyielding of DNF potential in high-density treatments, as DT was within 10% of Dmax (Table 3.2), suggesting the mean effect of complementarity for DNF potential in high-density communities almost equaled the maximum amount of complementarity possible given our data.

The ΔEF for annamox potential showed a similar trend, with the degree of complementarity (underyielding) increasing in magnitude with increasing density (Table 3.2). In low-density communities, there was a more negative ΔEF for annamox potential than DNF potential (Table 3.2), suggesting that even at low densities, increased biodiversity can lead to underyielding of annamox potentials relative to 1-species communities, with the effect of biodiversity being moderately strong (DT within 30% of Dmax, Table 3.2). Conversely, at high- densities there was a high degree of underyielding in terms of annamox potential (DT within 9% of Dmax, Table 3.2) suggesting overall that, in conjunction with DNF potential, an increase in biodiversity in high-density communities results in strong underyielding of overall N-removal potential in the sediment.

Discussion

Our results show that mussel aggregations have the potential to increase N-removal in stream sediments, with the largest effect found in 1-species communities at high-densities. Our

BD-EF analysis indicated a strong complimentary effect of biodiversity on N-removal potential, especially among high-density communities, with 2-species communities underyielding compared to 1-species communities in terms of both DNF and annamox potential. Additionally, the PCA revealed potential underlying mechanisms that determine the influence of mussel

76 aggregations on N-removal by suggesting an increase in biological activity, particularly movement and burrowing behavior, decreased overall N-removal potential. Previous research has shown individual mussels stimulate N-removal in sediment via the DNF pathway (Benelli et al. 2017, Hoellein et al. 2017). These studies used microcosm incubation designs and, while these designs are good for measuring fine-scale nutrient and gas fluxes, they represent a highly- modified benthic environment far from the mussels’ natural habitat. Our research, along with a study conducted by Trentman et al. (2018) in a southeastern Oklahoma river, are the first to assess the influence of native freshwater mussel aggregations on N-removal potential in sediment using in-situ sampling methods. Whereas the Oklahoma study analyzed N-removal potential in sediment samples collected from natural mussel aggregations (Trentman et al. 2018), our study was particularly unique in that we experimentally manipulated mussel biodiversity in natural benthic conditions. This method allowed us to assess aspects of mussel ecosystem ecology, specifically mussel-sediment interactions, only speculated on in previous research, namely BD-

EF dynamics (see Vaughn 2010, Trentman et al. 2018) and the effect of bioturbation (see Turek and Hoellein 2015, Benelli et al. 2017, Hoellein et al. 2017, Vaughn 2017).

The total soft-tissue biomass of an aggregation explained the majority of the variation in our data, as was evident by the separation of low- and high-density treatments along PC1.

Despite the underyielding impact of species diversity, biomass positively influenced N-removal potential in single-species communities, as both DNF and annamox potentials increased, on average, between low- and high-density aggregations of C. asperata and F. flava treatments.

This confirmed our hypothesis that increased mussel density would increase N-removal potential. We observed the opposite trend between biomass and N-removal potential in 2-species treatments, confirming another of our initial hypotheses. We predicted an increase in mussel

77 diversity would decrease N-removal potential, and that is indeed what we observed in our high- density treatments, suggesting biomass alone cannot fully explain an aggregation’s influence on

N-removal. Past studies have attempted to extrapolate the influence of mussels on N-removal to the aggregation-scale, essentially assuming an additive BD-EF effect based on biomass, and independent of species (Hoellein et al. 2017, Trentman et al. 2018). Positive relationships between total biomass (or bivalve density) and N-removal have been observed in aggregate communities of such bivalves as the eastern oyster (Crassostrea virginica; Newell et al. 2005,

Hoellein and Zarnoch 2014) and the invasive zebra mussel (Dreissena polymorpha; Bruesewitz et al. 2006, Bruesewitz et al. 2008) in estuarine/coastal and freshwater ecosystems, respectively.

Informed by these observations, ecologists have suggested this relationship can be applied to understand the influence of freshwater mussel aggregations on N-removal (Hoellein et al. 2017,

Vaughn 2017). Our results suggest the contrary by showing biologically-diverse aggregations of bivalves, rather than those composed of a single-species as in oyster and zebra mussel aggregations, have a non-additive effect on sediment biogeochemical processes, potentially driven by inter-specific competition among species with differing functional niches.

While oysters and zebra mussels are functionally-similar to unionid mussels in that they are filter-feeding organisms that concentrate nutrients and energy in the benthos, the major functional difference between the bivalves is that oysters and zebra mussels are reef-building epifauna that are completely sessile throughout their adult life cycle while unionid mussels are infauna that remain mobile as adults, as is evident by our study and others (Schwalb and Pusch

2007, Allen and Vaughn 2009). Thus, unionid mussel aggregations provision the ecosystem function of sediment bioturbation that is absent in communities of reef-building bivalves.

Sediment bioturbation can significantly impact O2 penetration, DIN flux and OM transport in

78 freshwater benthic environments, influencing N-removal (Mermillod-Blondin and Rosenberg

2006, Mermillod-Blondin 2011). Therefore, while our experimental design and analysis methodology have been inspired by research on marine and invasive freshwater bivalves, we must apply these methods, in combination with BD-EF calculations and emerging revelations into unionid ecology (i.e., movement and burrowing behavior), to determine the influence of this highly imperiled group of organisms on important ecosystem functions like N-removal.

Incubating individual mussels serves as a good indicator of the influence of physiological traits (excretion, biodeposition) on N-removal (Benelli et al. 2017, Hoellein et al. 2017), and in- situ sediment sampling from mussel aggregations serves as a good snapshot of community-scale

N-removal potential in a natural setting (Trentman et al. 2018), but our study further advances the field by showing, in a controlled experiment, that community-scale N-removal potential cannot be explained by biomass alone. Other studies have also shown complimentary effects of mussel aggregation biodiversity on various ecosystem functions including movement and burrowing behavior (Allen and Vaughn 2009), benthic primary production (Vaughn et al. 2007) and algae and macroinvertebrate colonization (Spooner and Vaughn 2006, Spooner et al. 2012).

As our experiment represents another example of complementarity in BD-EF dynamics of mussel aggregations, we suggest that determining the influence of individual mussels on ecosystem function and extrapolating based on biomass or density is not sufficient for determining the functional role of natural mussel aggregations in lotic ecosystems. Rather, future studies should continue to explore the effect of varying mussel density and diversity on ecosystem function to elucidate BD-EF dynamics driven by inter-specific competition, utilizing such methods as in-field benthic enclosures (as in our study; also see design of Vaughn et al.

2007, Spooner et al. 2012), mesocosms (see design of Allen and Vaughn 2009) or community-

79 scale incubations (as has been done with oyster reef communities; see design of Kellogg et al.

2013).

The results of the BD-EF analysis showing a high degree of underyielding for high- density 2-species treatments suggests there were inter-specific interactions between species causing overall N-removal potential in the sediment to decrease relative to 1-species treatments.

Additionally, the fact that we observed a weak effect of biodiversity on N-removal potential in low-density mussel treatments and a strong effect in high-density treatments suggests a density threshold in our study system at which interspecific competition for niche space in diverse aggregations begins to have a significant effect on sediment microbial communities and associated ecosystem functions. The PCA showed an increase in overall biological activity, influenced most by movement and burrowing activity, decreased total N-removal potential in the sediment. Among low-density treatments, we saw a pattern of increasing biological activity that reflects the results of our movement and burrowing observations throughout the study. The F. flava treatment had the most negative PC2 score among low-density treatments, which corresponds to the fact that F. flava moved and burrowed the least, and fostered the highest mean

N-removal potentials. The C. asperata treatment, the most positive PC2 score among low- density treatments, showed the opposite relationship by moving and burrowing the most, but fostering the lowest mean N-removal potentials. The low-density 2-species treatment lies equidistant between each 1-species treatment along PC2, reflecting the weak (nearly additive) effect of biodiversity we observed, a pattern reflected by the fact that mean burrowing and movement in the low-density 2-species treatment are also nearly equidistant between means for each 1-species treatment.

Among high-density treatments, our PCA agrees with both the results of the BD-EF and

N-removal potential analyses, but does not agree with our observations of movement and burrowing functional traits, hinting at inter-specific interactions among species not revealed to us based on our experimental design. Both high-density 1-species treatments have nearly the same

PC2 scores as their low-density counterparts, which is a good indicator of how biomass dominated the influence of functional traits and N-removal potentials in our 1-species treatments.

The high-density 2-species treatment had a much higher PC2 score than both high-density 1- species treatments, indicating a much higher rate of biological activity and lower N-removal potentials. This clear visualization of the strong complimentary effect of biodiversity we observed is reflected well in the N-removal potential results, with the high-density 2-species treatment fostering the lowest N-removal of all mussel treatments. Interestingly, the high PC2 score for the high-density 2-species treatment is not reflected well in our observations of movement and burrowing functional traits. Viewing the PCA, we would expect the high-density

2-species treatment to have shown much higher rates of movement and burrowing relative the each 1-species treatment, but this is not what we observed. Rather, just as with the low-density treatments, mean movement and burrowing for the high-density 2-species treatment was equidistant between each 1-species treatment. This result suggests the resolution of our movement and burrowing data was not high enough to capture the potential inter-specific interactions resulting from competition for niche space and leading to more movement and burrowing activity in the 2-species treatment. Our interpretation contradicts (while the observational data agrees with) a study conducted by Allen and Vaughn (2009) where one of the major conclusions was that community structure (i.e., species diversity) did not affect species- specific movement and burrowing behavior. While Allen and Vaughn (2009) had higher

81 resolution data than ours (aggregations were observed every three days), the mesocosm design and short trial period (11-d) make comparing results and interpretations difficult. Our study shows a clear underyielding of N-removal potential as aggregation biodiversity increases, but we could not elucidate the mechanism causing this complimentary effect of biodiversity given our experimental design. Future work on the influence of mussel aggregations on N-removal in sediment would benefit from collecting higher-resolution data on the mussel functional traits that chemically and physically alter the benthic environment, specifically movement and burrowing behavior in diverse communities that perturbs the sediment, altering redox conditions and, inevitably, N-removal potential.

Conclusion

Freshwater mussel aggregations provision many ecosystem functions due to the fact that these functionally-active organisms thrive in dense and speciose aggregations (Vaughn and

Hakenkamp 2001, Vaughn 2017). Mussels are highly imperiled organisms (Lydeard et al. 2004,

Strayer et al. 2004), but conservation efforts have been impeded by an overall lack of understanding of the functional role mussel aggregations play in their natural environment. We showed, through direct manipulation biodiversity and in-situ observations of functional traits and analysis of sediment properties, that mussel aggregations have a complimentary effect of biodiversity on the permanent removal of DIN in their natural environment via the biogeochemical pathways of DNF and annamox. Additionally, because our BD-EF analysis and

PCA results did not correspond to our observed movement and burrowing behavior, our results open new questions into the underlying mechanisms driving BD-EF interactions in freshwater mussel aggregations. Our field-based study, along with others of similar design (Spooner and

Vaughn 2006, Vaughn et al. 2007, Spooner et al. 2012), have consistently revealed the non-

82 additive effect of mussel biodiversity on important ecosystem functions in lotic systems. Our study, in particular, demonstrates the importance of using the BD-EF framework when examining the influence of mussel aggregations on sediment biogeochemical properties rather than assuming such provisioning of ecosystem function can be extrapolated based on biomass alone. In the face of global unionid species loss and anthropogenic alteration of lotic ecosystems, our study aids in directing future research to realize the wholistic functional role of freshwater mussels in their natural environment, encouraging stakeholders to invest resources in conservation and remediation efforts.

References

Allen, D. C., and C. C. Vaughn. 2009. Burrowing behavior of freshwater mussels in experimentally manipulated communities. Journal of the North American Benthological Society 28:93-100.

Allen, D. C., C. C. Vaughn, J. F. Kelly, J. T. Cooper, and M. H. Engel. 2012. Bottom‐up biodiversity effects increase resource subsidy flux between ecosystems. Ecology 93:2165-2174.

Atkinson, C. L., K. A. Capps, A. T. Rugenski, and M. J. Vanni. 2017. Consumer‐driven nutrient dynamics in freshwater ecosystems: From individuals to ecosystems. Biological Reviews 92:2003-2023.

Atkinson, C. L., and C. C. Vaughn. 2015. Biogeochemical hotspots: Temporal and spatial scaling of the impact of freshwater mussels on ecosystem function. Freshwater Biology 60:563-574.

Bruesewitz, D. A., J. L. Tank, and M. J. Bernot. 2008. Delineating the effects of zebra mussels (Dreissena polymorpha) on N transformation rates using laboratory mesocosms. Journal of the North American Benthological Society 27:236-251.

Bruesewitz, D. A., J. L. Tank, M. J. Bernot, W. B. Richardson, and E. A. Strauss. 2006. Seasonal effects of the zebra mussel (Dreissena polymorpha) on sediment denitrification rates in Pool 8 of the Upper Mississippi River. Canadian Journal of Fisheries and Aquatic Sciences 63:957-969.

Cadotte, M. W., J. Cavender-Bares, D. Tilman, and T. H. Oakley. 2009. Using phylogenetic, functional and trait diversity to understand patterns of plant community productivity. PloS One 4:e5695.

Cardinale, B. J., M. A. Palmer, and S. L. Collins. 2002. Species diversity enhances ecosystem functioning through interspecific facilitation. Nature 415:426.

Cederholm, C. J., M. D. Kunze, T. Murota, and A. Sibatani. 1999. Pacific salmon carcasses: Essential contributions of nutrients and energy for aquatic and terrestrial ecosystems. Fisheries 24:6-15.

Chapin, F. S., B. H. Walker, R. J. Hobbs, D. U. Hooper, J. H. Lawton, O. E. Sala, and D. Tilman. 1997. Biotic control over the functioning of ecosystems. Science 277:500-504.

Diaz, R. J., and R. Rosenberg. 2008. Spreading dead zones and consequences for marine ecosystems. Science 321:926-929.

Díaz, S., A. Purvis, J. H. Cornelissen, G. M. Mace, M. J. Donoghue, R. M. Ewers, P. Jordano, and W. D. Pearse. 2013. Functional traits, the phylogeny of function, and ecosystem service vulnerability. Ecology and Evolution 3:2958-2975.

Dudgeon, D., A. H. Arthington, M. O. Gessner, Z.-I. Kawabata, D. J. Knowler, C. Lévêque, R. J. Naiman, A. H. Prieur-Richard, D. Soto, and M. L. Stiassny. 2006. Freshwater biodiversity: Importance, threats, status and conservation challenges. Biological Reviews 81:163-182.

Emmerson, M. C., M. Solan, C. Emes, D. M. Paterson, and D. Raffaelli. 2001. Consistent patterns and the idiosyncratic effects of biodiversity in marine ecosystems. Nature 411:73.

Essington, T. E., and S. R. Carpenter. 2000. Mini-review: Nutrient cycling in lakes and streams: Insights from a comparative analysis. Ecosystems 3:131-143.

Fortin, M. J., P. Drapeau, and P. Legendre. 1990. Spatial autocorrelation and sampling design in plant ecology. Progress in Theoretical Vegetation Science 11:209-222.

Fox, J. W. 2005. Interpreting the ‘selection effect’of biodiversity on ecosystem function. Ecology Letters 8:846-856.

Frank, D., and S. McNaughton. 1991. Stability increases with diversity in plant communities: Empirical evidence from the 1988 Yellowstone drought. Oikos 62:360-362.

Frissell, C. A., W. J. Liss, C. E. Warren, and M. D. Hurley. 1986. A hierarchical framework for stream habitat classification: Viewing streams in a watershed context. Environmental Management 10:199-214.

Gamfeldt, L., J. S. Lefcheck, J. E. Byrnes, B. J. Cardinale, J. E. Duffy, and J. N. Griffin. 2015. Marine biodiversity and ecosystem functioning: What's known and what's next? Oikos 124:252-265.

Giller, P. S., H. Hillebrand, U. G. Berninger, M. O. Gessner, S. Hawkins, P. Inchausti, C. Inglis, H. Leslie, B. Malmqvist, and M. T. Monaghan. 2004. Biodiversity effects on ecosystem functioning: Emerging issues and their experimental test in aquatic environments. Oikos 104:423-436.

Groffman, P. M., K. Butterbach-Bahl, R. W. Fulweiler, A. J. Gold, J. L. Morse, E. K. Stander, C. Tague, C. Tonitto, and P. Vidon. 2009. Challenges to incorporating spatially and temporally explicit phenomena (hotspots and hot moments) in denitrification models. Biogeochemistry 93:49-77.

Haag, W. R. 2012. North American freshwater mussels: Natural history, ecology, and conservation. Cambridge University Press. New York, New York.

Haag, W. R., and M. L. Warren. 2010. Diversity, abundance, and size structure of bivalve assemblages in the Sipsey River, Alabama. Aquatic Conservation: Marine and Freshwater Ecosystems 20:655-667.

Hairston, N. G., and N. G. Hairston. 1993. Cause-effect relationships in energy flow, trophic structure, and interspecific interactions. The American Naturalist 142:379-411.

Hoellein, T. J., and C. B. Zarnoch. 2014. Effect of eastern oysters (Crassostrea virginica) on sediment carbon and nitrogen dynamics in an urban estuary. Ecological Applications 24:271-286.

Hooper, D. U., F. Chapin, J. Ewel, A. Hector, P. Inchausti, S. Lavorel, J. Lawton, D. Lodge, M. Loreau, and S. Naeem. 2005. Effects of biodiversity on ecosystem functioning: A consensus of current knowledge. Ecological Monographs 75:3-35.

Jones, C. G., J. H. Lawton, and M. Shachak. 1994. Organisms as ecosystem engineers. Pages 130-147 Ecosystem Management. Springer. New York, New York.

Kellogg, M. L., J. C. Cornwell, M. S. Owens, and K. T. Paynter. 2013. Denitrification and nutrient assimilation on a restored oyster reef. Marine Ecology Progress Series 480:1-19.

Knowles, R. 1982. Denitrification. Microbiological Reviews 46:43.

Landsberg, J. H. 2002. The effects of harmful algal blooms on aquatic organisms. Reviews in Fisheries Science 10:113-390.

Loreau, M. 1998. Separating sampling and other effects in biodiversity experiments. Oikos 82:600-602.

Loreau, M., and A. Hector. 2001. Partitioning selection and complementarity in biodiversity experiments. Nature 412:72-76.

Lydeard, C., R. H. Cowie, W. F. Ponder, A. E. Bogan, P. Bouchet, S. A. Clark, K. S. Cummings, T. J. Frest, O. Gargominy, and D. G. Herbert. 2004. The global decline of nonmarine mollusks. BioScience 54:321-330.

McCullagh, W. H., J. D. Williams, S. W. McGregor, J. M. Pierson, and C. Lydeard. 2002. The Unionid (Bivalvia) fauna of the Sipsey River in northwestern Alabama, an aquatic hotspot. American Malacological Bulletin 17:1-15.

McNaughton, S. J. 1994. Biodiversity and function of grazing ecosystems. Pages 361-383 Biodiversity and ecosystem function. Springer. Berlin, Heidelberg.

Mermillod-Blondin, F., and R. Rosenberg. 2006. Ecosystem engineering: The impact of bioturbation on biogeochemical processes in marine and freshwater benthic habitats. Aquatic Sciences 68:434-442.

Meybeck, M. 1982. Carbon, nitrogen, and phosphorus transport by world rivers. American Journal of Science 282:401-450.

Meyer, J. L., W. H. McDowell, T. L. Bott, J. W. Elwood, C. Ishizaki, J. M. Melack, B. L. Peckarsky, B. J. Peterson, and P. A. Rublee. 1988. Elemental dynamics in streams. Journal of the North American Benthological Society 7:410-432.

Naiman, R. J., R. E. Bilby, D. E. Schindler, and J. M. Helfield. 2002. Pacific salmon, nutrients, and the dynamics of freshwater and riparian ecosystems. Ecosystems 5:399-417.

Newell, R. I., T. Fisher, R. Holyoke, and J. Cornwell. 2005. Influence of eastern oysters on nitrogen and phosphorus regeneration in Chesapeake Bay, USA. Pages 93-120 The Comparative Roles of Suspension-Feeders in Ecosystems. Springer. Dordrecht.

Petchey, O. L. 2003. Integrating methods that investigate how complementarity influences ecosystem functioning. Oikos 101:323-330.

Petchey, O. L., and K. J. Gaston. 2006. Functional diversity: Back to basics and looking forward. Ecology Letters 9:741-758.

Pina-Ochoa, E., and M. Alvarez-Cobelas. 2006. Denitrification in aquatic environments: A cross-system analysis. Biogeochemistry 81:111-130.

Ricciardi, A., and J. B. Rasmussen. 1999. Extinction rates of North American freshwater fauna. Conservation Biology 13:1220-1222.

Sandin, L., and A. G. Solimini. 2009. Freshwater ecosystem structure–function relationships: From theory to application. Freshwater Biology 54:2017-2024.

Schwalb, A. N., and M. T. Pusch. 2007. Horizontal and vertical movements of unionid mussels in a lowland river. Journal of the North American Benthological Society 26:261-272.

Simberloff, D., and T. Dayan. 1991. The guild concept and the structure of ecological communities. Annual Review of Ecology and Systematics 22:115-143.

Smith, E., W. Davison, and J. Hamilton-Taylor. 2002. Methods for preparing synthetic freshwaters. Water Research 36:1286-1296.

Smith, S., and J. Hollibaugh. 1993. Coastal metabolism and the oceanic organic carbon balance. Reviews of Geophysics 31:75-89.

Solan, M., B. J. Cardinale, A. L. Downing, K. A. Engelhardt, J. L. Ruesink, and D. S. Srivastava. 2004. Extinction and ecosystem function in the marine benthos. Science 306:1177-1180.

Spooner, D. E., and C. C. Vaughn. 2006. Context‐dependent effects of freshwater mussels on stream benthic communities. Freshwater Biology 51:1016-1024.

Spooner, D. E., C. C. Vaughn, and H. S. Galbraith. 2012. Species traits and environmental conditions govern the relationship between biodiversity effects across trophic levels. Oecologia 168:533-548.

Thamdrup, B., and T. Dalsgaard. 2002. Production of N2 through anaerobic ammonium oxidation coupled to nitrate reduction in marine sediments. Applied and Environmental Microbiology 68:1312-1318.

Thoms, M. C., and M. Parsons. 2002. Eco-geomorphology: An interdisciplinary approach to river science. International Association of Hydrological Sciences 276:113-119.